**Advances in Knee Surgery**

Editors

**Yong In In Jun Koh**

Basel • Beijing • Wuhan • Barcelona • Belgrade • Novi Sad • Cluj • Manchester

*Editors* Yong In The Catholic University of Korea Seoul, Republic of Korea

In Jun Koh The Catholic University of Korea Seoul, Republic of Korea

*Editorial Office* MDPI St. Alban-Anlage 66 4052 Basel, Switzerland

This is a reprint of articles from the Special Issue published online in the open access journal *Medicina* (ISSN 1648-9144) (available at: https://www.mdpi.com/journal/medicina/special issues/ Knee Surgery ).

For citation purposes, cite each article independently as indicated on the article page online and as indicated below:

Lastname, A.A.; Lastname, B.B. Article Title. *Journal Name* **Year**, *Volume Number*, Page Range.

**ISBN 978-3-0365-9608-2 (Hbk) ISBN 978-3-0365-9609-9 (PDF) doi.org/10.3390/books978-3-0365-9609-9**

© 2023 by the authors. Articles in this book are Open Access and distributed under the Creative Commons Attribution (CC BY) license. The book as a whole is distributed by MDPI under the terms and conditions of the Creative Commons Attribution-NonCommercial-NoDerivs (CC BY-NC-ND) license.

## **Contents**



Reprinted from: *Medicina* **2022**, *58*, 1226, doi:10.3390/medicina58091226 .............. **221**

## **About the Editors**

#### **Yong In**

Yong In, MD, Ph.D., is a professor at the Orthopedic Surgery Department of the College of Medicine at the Catholic University of Korea, Seoul, Republic of Korea. He is currently a chief knee surgeon and the former president of the Orthopedic Surgery Department of Seoul St. Mary's Hospital, Seoul, Republic of Korea. He is a member of multiple national and international orthopedic societies including ISAKOS, AAHKS, the Korean Knee Society, and the Korean Orthopaedic Association and an editorial board member of the *Journal of Arthroplasty* and *Clinics in Orthopedic Surgery*. In addition, he is a member and trustee of multiple Korean societies related to orthopedics and stem cells. He has authored or co-authored numerous orthopedic journal articles and consulted multiple international knee arthroplasty companies.

#### **In Jun Koh**

In Jun Koh, MD, Ph.D., is currently the Director of the Joint Reconstruction Center and a chief surgeon in the Knee Surgery and Sports Medicine Division at Eunpyeong St. Mary's Hospital. In addition, he is currently a professor and teaches with the affiliated Orthopedic Surgery Department at the College of Medicine, Catholic University of Korea, Seoul, Republic of Korea. He is currently a member of multiple committees of the Korean Orthopaedic Association, the Korean Knee Society, and the Korean Arthroscopy Society. Moreover, he is a member of multiple advisory boards of national healthcare agencies and an editorial board member of *Clinics in Orthopedic Surgery*, *Knee Surgery & Related Research*, and *Medicina*. He is the author or co-author of more than 100 peer-reviewed, highly-ranked orthopedic journal articles and has received numerous awards.

*Systematic Review*

### **Bone Marrow Aspirate Concentrate versus Human Umbilical Cord Blood-Derived Mesenchymal Stem Cells for Combined Cartilage Regeneration Procedure in Patients Undergoing High Tibial Osteotomy: A Systematic Review and Meta-Analysis**

**Dojoon Park 1, Youn Ho Choi 1, Se Hyun Kang 1, Hae Seok Koh <sup>1</sup> and Yong In 2,\***


**Abstract:** *Background and objectives:* Cartilage regeneration using mesenchymal stem cells (MSCs) has been attempted to improve articular cartilage regeneration in varus knee osteoarthritis (OA) patients undergoing high tibial osteotomy (HTO). Bone marrow aspirate concentrate (BMAC) and human umbilical cord blood-derived MSCs (hUCB-MSCs) have been reported to be effective. However, whether BMAC is superior to hUCB-MSCs remains unclear. This systematic review and meta-analysis aimed to determine the clinical efficacy of cartilage repair procedures with BMAC or hUCB-MSCs in patients undergoing HTO. *Materials and Methods:* A systematic search was conducted using three global databases, PubMed, EMBASE, and the Cochrane Library, for studies in which the clinical outcomes after BMAC or hUCB-MSCs were used in patients undergoing HTO for varus knee OA. Data extraction, quality control, and meta-analysis were performed. To compare the clinical efficacy of BMAC and hUCB-MSCs, reported clinical outcome assessments and second-look arthroscopic findings were analyzed using standardized mean differences (SMDs) with 95% confidence intervals (CIs). *Results:* The present review included seven studies of 499 patients who received either BMAC (BMAC group, *n* = 169) or hUCB-MSCs (hUCB-MSC group, *n* = 330). Improved clinical outcomes were found in both BMAC and hUCB-MSC groups; however, a significant difference was not observed between procedures (International Knee Documentation Committee score; *p* = 0.91, Western Ontario and McMaster Universities OA Index; *p* = 0.05, Knee Society Score (KSS) Pain; *p* = 0.85, KSS Function; *p* = 0.37). On second-look arthroscopy, the hUCB-MSC group showed better International Cartilage Repair Society Cartilage Repair Assessment grade compared with the BMAC group (*p* < 0.001). *Conclusions:* Both BMAC and hUCB-MSCs with HTO improved clinical outcomes in varus knee OA patients, and there was no difference in clinical outcomes between them. However, hUCB-MSCs were more effective in articular cartilage regeneration than BMAC augmentation.

**Keywords:** high tibial osteotomy; bone marrow aspirate concentrate; human umbilical cord blood-derived mesenchymal stem cells; knee osteoarthritis

#### **1. Introduction**

High tibial osteotomy (HTO) is a reliable surgical method for physically active or young patients with medial compartment knee osteoarthritis (OA) with varus deformity [1–3]. The HTO procedure reduces the weight loading of the medial compartment through alteration of the weight-bearing axis, and healing of the damaged cartilage can be expected due to reduced stress [4,5]. Satisfactory clinical and radiological improvement after HTO at shortterm and mid-term follow-up have been reported in several studies [6,7]; however, longterm follow-up data revealed a troubling trend of deteriorating outcomes in some patients,

**Citation:** Park, D.; Choi, Y.H.; Kang, S.H.; Koh, H.S.; In, Y. Bone Marrow Aspirate Concentrate versus Human Umbilical Cord Blood-Derived Mesenchymal Stem Cells for Combined Cartilage Regeneration Procedure in Patients Undergoing High Tibial Osteotomy: A Systematic Review and Meta-Analysis. *Medicina* **2023**, *59*, 634. https://doi.org/ 10.3390/medicina59030634

Academic Editor: Cory Xian

Received: 6 February 2023 Revised: 20 February 2023 Accepted: 17 March 2023 Published: 22 March 2023

**Copyright:** © 2023 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

necessitating revision total knee arthroplasty (TKA) due to unsatisfactory results [8,9]. Therefore, a combination of additional cartilage regeneration procedures with HTO has been attempted to improve articular cartilage regeneration and surgical outcomes [10]. Among additional procedures, mesenchymal stem cell (MSC) enhancement is an emerging option for cartilage regeneration procedures for OA [11,12], and its effectiveness has been demonstrated in several clinical trials [13–15].

Bone marrow aspirate concentrate (BMAC) and human umbilical cord blood-derived mesenchymal stem cells (hUCB-MSCs) are representative stem cell-based orthobiologics [16]. BMAC augmentation relies on the inclusion of many growth factors and pluripotent stromal cells that induce MSC differentiation into chondrocytes, potentially leading to the production of native, hyaline-like cartilage [17,18]. hUCB-MSCs have low immunogenicity and can be collected using a noninvasive method. In addition, hUCB-MSCs have a good expansion capacity to provide sufficient cells for treatment [19,20].

The superior results of the combination of HTO and cartilage regeneration procedures compared with HTO alone have been reported in several studies [21,22]. Despite promising results regarding cartilage regeneration procedures using MSC-based orthobiologics, the optimal cartilage regeneration procedure that can be performed with HTO remains unclear. Direct comparison of the efficacy of BMAC or hUCB-MSCs combined with HTO has been conducted in few studies to date [23,24].

The purpose of this meta-analysis was to identify available studies on the clinical efficacy of HTO with cartilage regeneration procedure using BMAC or hUCB-MSCs in varus knee OA patients, and to compare the efficacy of the two procedures for clinical improvement and cartilage regeneration. To the best of our knowledge, this is the first meta-analysis in which this topic was investigated. We hypothesized equivalent clinical effects of BMAC and hUCB-MSCs with HTO in varus knee OA patients.

#### **2. Materials and Methods**

#### *2.1. Meta-Analyses Principles*

This analysis was conducted under the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) principle [25]. Studies in which the clinical effects of BMAC and hUCB-MSCs in patients undergoing HTO for varus knee OA were systematically reviewed. Ethical approval was unnecessary because all analyses were conducted using existing literature.

#### *2.2. Search Strategy*

A systematic search was conducted by two reviewers using three global online databases (PubMed, EMBASE, and Cochrane Library) for studies published by 22 September 2022, in which clinical effects after the use of BMAC or hUCB-MSCs in patients undergoing HTO for varus knee OA were investigated. The publication language was restricted to English.

A search for relevant articles was conducted using various combinations of the following keywords: "osteoarthritis", "knee", "high tibial osteotomy", "bone marrow aspirate concentrate", and "umbilical cord blood-derived mesenchymal stem cells." Details of the search terms and strategy are presented in Supplemental Table S1.

A secondary manual search was performed on the related reviews and meta-analyses and their reference lists to retrieve relevant articles that were not identified using the databases.

#### *2.3. Study Criteria and Screening Process*

Two independent reviewers evaluated the eligibility of potentially relevant articles retrieved after removing duplicates based on the predefined criteria. The review was conducted on the title and abstract of the study and, in cases of uncertainty, was performed on the entire text. Discrepancies were discussed with a third reviewer.

The inclusion criteria were as follows: (1) studies using BMAC or hUCB-MSCs with HTO for treatment of varus knee OA, (2) studies investigating the clinical effects of BMAC or hUCB-MSCs with HTO on postoperative outcome, and (3) studies with a mean follow-up period >18 months. The exclusion criteria were as follows: (1) reviews, meta-analyses, case reports, and letters; (2) duplicates of previously published articles.

#### *2.4. Data Extraction and Quality Control*

Data extraction was performed using standardized protocols. The following variables of the included studies were collected: first author, publication year, study design, level of evidence, type of osteotomy, type of intervention, sex, age, body mass index (BMI), sample size, preoperative Kellgren–Lawrence grade [26], preoperative International Cartilage Repair Society (ICRS) grade [27], cartilage defect size, mean follow-up duration, preoperative hip-knee-ankle (HKA) angle, clinical assessment, postoperative ICRS-Cartilage Repair Assessment (CRA) grade [28], postoperative Koshino stage [29], a brief description of the preparation method of intervention, postoperative rehabilitation protocol, and reported adverse events.

Because the final eligible studies did not include randomized controlled trials, two independent reviewers evaluated the bias and risk of all eligible observational studies using the Methodological Item for Non-Randomized Studies (MINORS) [30], with 12 categories for comparative studies and 8 categories for noncomparative studies. Each category was rated 0 (not reported), 1 (reported but inadequate), or 2 (reported and adequate).

#### *2.5. Statistical Analyses*

Descriptive statistics including the mean and standard deviation for any numerical variable were recorded. Any missing standard deviation was estimated using preestablished methodologies [31]. The 95% confidence interval (CI) of the appropriate variable was reported. Heterogeneity was determined using the I2 statistic; if I<sup>2</sup> < 50% (low heterogeneity), a fixed-effects model was used; otherwise, a random-effects model was used, and a funnel plot was used to assess the existence of publication bias (Supplemental Figure S1). A *p*-value < 0.05 was considered statistically significant.

Clinical outcome was evaluated using the mean International Knee Documentation Committee (IKDC) score [32], Western Ontario and McMaster Universities OA Index (WOMAC) score [33], and other clinical outcome measures performed before and after surgery. The quality of articular cartilage regeneration was determined using the ICRS-CRA on second-look arthroscopy. Clinical efficacy based on the type of treatment was estimated using the standardized mean difference (SMD) and was analyzed to compare the clinical effects between patients who received BMAC (BMAC group, *n* = 169) or hUCB-MSCs (hUCB-MSC group, *n* = 330). Statistical significance was set at *p* < 0.05. All statistical analyses were conducted using Review Manager software (version 5.3; The Nordic Cochrane Centre, Copenhagen, Denmark).

#### **3. Results**

#### *3.1. Study Characteristics*

Details of the search process, including literature identification and verification, are presented in Figure 1. A total of 33 potentially relevant studies were retrieved from 3 databases, of which 23 remained after 10 duplicate articles were removed. During the review process, the titles and abstracts of the studies were reviewed, and for three uncertain cases, the review was extended to include full texts for identification. One article was added after a related review article search. After comprehensive screening, 7 studies involving 499 patients were included in this systematic review.

The outcomes of HTO with BMAC were reported in two studies, the outcomes of HTO with hUCB-MSCs were reported in three studies, and comparative outcomes between HTO with BMAC and HTO with hUCB-MSCs were reported in two studies. The baseline characteristics of the studies are summarized in Tables 1–3.

#### *3.2. Methodological Quality*

The level of evidence was 3 in three studies [23,24,34] (42.9%) and 4 in four studies [35–38] (57.1%). The mean MINORS score for comparative studies (three studies, maximum of 24 points) was 14.7 ± 2.4 (range 13–18), and noncomparative studies (four studies, maximum of 16 points) had a mean MINORS score of 9 ± 1.5 (range, 6–11; Supplemental Table S3).


**Table 1.**

Characteristics

 of the included studies.


**Table 2.**

Preoperative

 status and outcomes of

interventions

 of the included studies.

International

 Cartilage Repair Society Cartilage Repair Assessment;

 SF-36, Short Form 36 Health Survey.


 **3.** Preoperative preservation, rehabilitation and complications of the included

 studies.

**Table**

#### *3.3. Meta-Analysis Results*

The study population consisted of 499 patients treated with HTO and either BMAC (169 patients) or hUCB-MSCs (330 patients). The mean age of subjects was not significantly different between the BMAC and hUCB-MSC groups (57.7 ± 7.3 and 57.8 ± 5.4 years, respectively; *p* = 0.86). In all studies except one (6/7; 85.7%), preoperative BMI was reported. Based on the available data, the BMAC group had a higher BMI than the hUCB-MSC group (26.3 ± 3.4 vs. 25.9 ± 2.7 kg/m2, *<sup>p</sup>* = 0.02).

#### 3.3.1. Clinical Outcomes

• **IKDC**: In six [23,34–38] of the seven studies, the IKDC score used to evaluate the clinical effects of BMAC or hUCB-MSCs with HTO in patients with varus knee OA was reported. The IKDC subgroup consisted of 127 patients treated using BMAC, and 298 patients treated using hUCB-MSCs. Significant heterogeneity was found (*p* < 0.001, I <sup>2</sup> = 94%), and the random effects model was used. Patients treated with BMAC had significantly improved IKDC scores (SMD, 4.13; 95% CI, 1.23–7.00), as did those in the hUCB-MSC group (SMD, 3.92; 95% CI, 3.65–4.20). However, based on IKDC score, the clinical effects of BMAC vs. hUCB-MSCs for combined cartilage regeneration in patients undergoing HTO were equivalent (*p* = 0.91; Figure 2).


**Figure 2.** Estimated clinical effects of bone marrow aspirate concentrate (BMAC) and human umbilical cord blood-derived mesenchymal stem cells (hUCB-MSCs) in knee osteoarthritis (OA) patients treated with high tibial osteotomy (HTO) based on the International Knee Documentation Committee (IKDC) score [23,34–38].

• **WOMAC**: In 5 studies [24,34,36–38], the WOMAC score was reported, and 90 and 275 participants were included in the BMAC and hUCB-MSC groups, respectively. Since only one [38] of five studies reported WOMAC subscales, while the other four reported only a total WOMAC score, and only the total WOMAC score was compared. Significant heterogeneity was found (*p* < 0.001; I<sup>2</sup> = 93%), and a random effects model was used. Based on the WOMAC index, both BMAC and hUCB-MSCs had a significant clinical effect compared with the preoperative status (SMD 2.09, 95% CI, 1.25–2.93 for BMAC and 3.39, 95% CI, 2.39–4.397 for hUCB-MSCs). However, the difference between groups did not reach statistical significance (*p* = 0.05; Figure 3).


**Figure 3.** Estimated clinical effects of bone marrow aspirate concentrate (BMAC) and human umbilical cord blood-derived mesenchymal stem cells (hUCB-MSCs) in knee osteoarthritis (OA) patients treated with high tibial osteotomy (HTO) based on the Western Ontario and McMaster Universities OA Index (WOMAC) [24,34,36–38].

• **Other reported clinical outcomes**: Other reported outcomes were KSS score (three studies), VAS (three studies), KOOS (two studies), HSS (two studies), SF-36 (one study), and Tegner activity score (two studies). The mean preoperative and postoperative values for the outcome scales are shown in Table 4. KSS was the only measure used in more than one study to evaluate BMAC and hUCB-MSCs; it was used in three studies [24,34,38]. Mean KSS subscale (pain and function) values were reported in all studies; statistically significant heterogeneity was detected (KSS pain, *p* = 0.01, I <sup>2</sup> = 73%; KSS function, *p* < 0.001, I2 = 84%). In a random effects model, patients treated with BMAC or hUCB-MSCs showed improved clinical outcome after surgery (KSS pain SMD, 1.51, 95% CI, 0.43–2.59 for BMAC and 1.40; 95% CI, 1.12–1.68 for hUCB-MSCs; KSS function SMD, 1.99, 95% CI, 0.63–3.34 for BMAC and 1.35; 95% CI, 1.08–1.63 for hUCB-MSCs). However, differences in KSS pain and function scores were not found between the two treatment groups (*p* = 0.85 and *p* = 0.37, respectively; Figure 4).

#### 3.3.2. Second-Look Arthroscopic Findings

In six studies [23,24,34,36–38], second-look findings were reported, and four studies [23,24,34,37] compared preoperative ICRS grade with postoperative ICRS-CRA grade. Significant heterogeneity was found, and a random effects model was used. Greater improvement in articular cartilage regeneration was observed in patients treated with hUCB-MSCs (SMD, 4.18; 95% CI, 3.61–4.75) than in patients treated with BMAC (SMD, 1.81; 95% CI, 1.10–2.53; *p* < 0.001, Figure 5).


**Table 4.** Reported clinical outcome measures of the included studies.



**Figure 4.** Estimated clinical effects of bone marrow aspirate concentrate (BMAC) and human umbilical cord blood-derived mesenchymal stem cells (hUCB-MSCs) in knee osteoarthritis (OA) patients treated with high tibial osteotomy (HTO) based on Knee Society Scores of (**a**) pain (KSS-pain) and (**b**) function (KSS-function) [24,34,38].


**Figure 5.** Estimated clinical effects of bone marrow aspirate concentrate (BMAC) and human umbilical cord blood-derived mesenchymal stem cells (hUCB-MSCs) in knee osteoarthritis (OA) patients treated with high tibial osteotomy (HTO) based on second-look arthroscopic articular cartilage status [23,24,34,37].

#### **4. Discussion**

The main finding of this study was that patients who underwent HTO for varus knee OA had improved clinical outcomes regardless of whether BMAC or hUCB-MSCs were used for cartilage regeneration. Postoperative clinical outcomes were significantly improved in both groups compared with preoperative baseline measurements, and both treatments provided reliable results in terms of functional score improvement and pain relief, without group differences. However, in comparison of second-look arthroscopic findings, hUCB-MSCs were more effective in articular cartilage regeneration than BMAC.

HTO provides a mechanical environment that can prevent the progression of degenerative changes by correcting the axis in which the weight is focused on the medial side of the knee of OA patients. Several studies have reported successful short- and medium-term outcomes after HTO [6,7]. However, with long-term follow-up, a trend for worsening outcomes was observed, with some patients requiring revision TKA due to poor outcomes [8,9]. It is unclear if deterioration of HTO results over time is associated with inappropriate cartilage regeneration of medial osteoarthritic cartilage [4,39].

Therefore, many attempts have been made to combine HTO with several types of cartilage regeneration procedures to improve long-term outcomes. Among such procedures, augmentation using MSCs that can differentiate into chondrocytes and produce extracellular matrix molecules important for cartilage regeneration is a promising option for managing cartilage defects [40]. Because the etiology of knee OA involves both biomechanical and biochemical changes in knee articular cartilage, combining HTO and MSC augmentation procedures appears promising.

MSCs concentrated after extraction from autologous bone marrow were introduced as a next-generation therapy for cartilage disease [34]. This procedure has the advantage of obtaining MSCs quickly and easily, and all processes from harvesting to transplantation can be performed in a single operation [41]. Considering the cell concentration, the most commonly used method for collecting BMAC is to collect aspirate from the iliac crest [42], but BMAC can also be collected from other sites, such as the proximal tibia [43]. Because only approximately 0.001% of MSCs of nucleated cells are obtained from aspirate, they are concentrated using density-gradient centrifugation to increase the number [44]. The concentrated BMAC is then injected directly into the affected joint or used in combination with other surgical procedures, such as proximal tibial osteotomy, to promote cartilage regeneration. Various growth factors included in BMAC induce cartilage regeneration and provide a favorable environment for MSC adhesion [45]. The immune control and anti-inflammatory effects of BMAC also help restore cartilage [40]. In patients receiving HTO, microfractures treated with BMAC were reported to have better arthroscopic findings regarding cartilage recovery than microfractures not treated with BMAC [34]; however, limitations exist. Because of the systems used in the field, including centrifuges, obtaining uniform cell numbers and concentrations is difficult. In addition, the amount of ideal BMAC required per unit defect size has not been established [46].

hUCB-MSCs are mesenchymal stem cells that are derived from human umbilical cord blood. Recently, hUCB-MSCs have been selected as a new treatment option for cartilage regeneration, and improved clinical results of knee OA with application of hUCB-MSCs have been reported in several studies [47,48]. These cells have been shown to have a high potential for cartilage regeneration, as well as anti-inflammatory and immunomodulatory properties [36,49]. hUCB-MSCs can be obtained from a donor, and they are typically expanded in culture before being injected directly into the affected joint or used in combination with other surgical procedures. hUCB-MSCs have several advantages over BMAC. First, hUCB-MSC showed higher proliferation rates and more than 1000-fold greater expansion capacity [49]. Second, because hUCB-MSCs are easy to obtain through cord blood banks, donor site morbidity is not a concern. However, the average additional cost of approximately \$5300 or more compared with BMAC is a major obstacle to hUCB-MSC treatment.

The specific surgical method for using BMAC or hUCB-MSCs in combination with proximal tibial osteotomy will depend on the individual patient's condition and the surgeon's preferences. However, in general, these procedures involve the transplantation of stem cells with cartilage preparations such as microfracture before or after osteotomy [10].

Various treatment options that can be combined with HTO for cartridge regeneration have been reviewed; however, direct comparisons between BMAC and hUCB-MSCs are limited. To the best of our knowledge, this study is the first meta-analysis of the clinical outcome of HTO combined with BMAC or hUCB-MSCs. In the present study, both groups showed significantly improved postoperative clinical outcomes compared with preoperative baseline measurements. Basically, HTO results in clinical improvement because the joint reaction force is transferred to the lateral side of the knee via mechanical axis transfer, which is located lateral to the intercondylar eminence or at least to the center of the knee [50]. In a meta-analysis of seven studies to confirm the effectiveness of concurrent procedures during HTO [10], the combined cartilage regeneration procedure reportedly had a slightly beneficial effect on clinical outcome. However, in subgroup analysis of that meta-analysis, similar to the results obtained in the present study, the clinical results were significantly improved in the MSC subgroup. These positive synergistic effects are likely a result of the more effective cartilage regeneration potential and pain-reducing anti-inflammatory properties of MSCs due to paracrine effects. Magnetic resonance imaging (MRI) and histological examinations have been used to evaluate the improvement in cartilage status. MRI or histological examination was not used in any of the seven studies reviewed in the present systematic analysis. However, improvements in cartilage condition were observed in four studies [23,24,34,37,38] in which comparable preoperative and postoperative ICRS grades were reported, and hUCB-MSCs were significantly superior to BMAC in the meta-analysis performed.

The potential side effects associated with MSC-based treatments include local reactions, such as pain and swelling, as well as the potential for cells to differentiate into inappropriate cell types and the risk of tumor growth [51]. However, tumor growth is not a common side effect of MSC-based treatments [52]. Among the studies included in this analysis, swelling was the most common adverse event, and no adverse events met the criteria for being classified as serious according to the individual study protocols. However, the follow-up period in these studies was not sufficiently long to assess the risk of tumorigenicity.

The current review is the first systematic review and meta-analysis comparing the results of HTO combined with BMAC or hUCB-MSCs. However, the present review had several limitations. First, the studies included in the review were of a retrospective nature with levels of evidence of 3 or 4. Due to the absence of randomized controlled trials and prospective comparative studies, further studies with higher power are needed to control for possible confounding factors and to strengthen or refute current conclusions. However, due to the inherent strength of meta-analyses, an analysis including the single-arm study was conducted to provide useful information to clinicians. Second, sensitivity analysis confirmed high heterogeneity between studies, for which a random effects model was adopted when integrating the results. In addition, an independent analysis on various scoring scales used to evaluate clinical results was performed to accommodate the diversity of outcome variables. Third, the follow-up period of the reviewed studies was not sufficiently long to evaluate long-term clinical outcomes and survival rates. The follow-up period is an important factor in evaluating the stability of orthobiologics, so the results of this study should be interpreted carefully. In addition, the healing and maturation processes of articular cartilage may vary depending on the timing of second-look arthroscopy, resulting in potential bias. Finally, the absence of a control group that underwent HTO alone may bias the interpretation of the results.

In future studies, it is recommended to consider standardized outcome measures for assessing clinical improvement, as well as imaging techniques such as MRI or histological examination to evaluate the quality and quantity of cartilage regeneration.

#### **5. Conclusions**

Clinical outcomes were improved for both BMAC and hUCB-MSCs combined with HTO in varus knee OA patients, and postoperative outcomes did not differ between the two groups. However, hUCB-MSCs showed better articular cartilage regeneration in ICRS-CRA grade compared with BMAC. Further verification of the results requires larger, well-designed randomized controlled trials that include assessments of long-term follow-up.

**Supplementary Materials:** The following supporting information can be downloaded at https: //www.mdpi.com/article/10.3390/medicina59030634/s1, Supplemental Table S1: Keywords and search details; Supplemental Figure S1: Funnel plot of comparison: BMAC and hUCB-MSC, outcome: IKDC score; Supplemental Table S2: Risk-of-bias assessment performed using the MINORS score.

**Author Contributions:** Conceptualization, Y.I. and D.P.; methodology, Y.I., D.P. and Y.H.C.; software, D.P.; validation, Y.I., D.P., Y.H.C. and H.S.K.; formal analysis, Y.I., D.P., Y.H.C., S.H.K. and H.S.K.; investigation, Y.I., D.P. and H.S.K.; resources, Y.I., D.P. and H.S.K.; data curation, D.P., Y.H.C., S.H.K. and H.S.K.; writing—original draft, D.P.; writing—review and editing, Y.I.; visualization, Y.I. and D.P.; supervision, Y.I.; project administration, Y.I. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Institutional Review Board Statement:** Ethical review and approval were waived for this study due to the nature of the systematic review article.

**Informed Consent Statement:** Patient consent was waived due to the nature of the systematic review article.

**Data Availability Statement:** The data published in this research are available on request from the first author (D.P.).

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


**Disclaimer/Publisher's Note:** The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

#### *Systematic Review*

### **Cartilage Regeneration Using Human Umbilical Cord Blood Derived Mesenchymal Stem Cells: A Systematic Review and Meta-Analysis**

**Dong Hwan Lee 1, Seon Ae Kim 2, Jun-Seob Song 3, Asode Ananthram Shetty 4, Bo-Hyoung Kim <sup>1</sup> and Seok Jung Kim 2,\***


**Abstract:** *Background and Objectives:* Human umbilical-cord-blood-derived mesenchymal stem cells (hUCB-MSCs) have recently been used in clinical cartilage regeneration procedures with the expectation of improved regeneration capacity. However, the number of studies using hUCB-MSCs is still insufficient, and long-term follow-up results after use are insufficient, indicating the need for additional data and research. We have attempted to prove the efficacy and safety of hUCB-MSC treatment in a comprehensive analysis by including all subjects with knee articular cartilage defect or osteoarthritis who have undergone cartilage repair surgery using hUCB-MSCs. We conducted a meta-analysis and demonstrated efficacy and safety based on a systematic review. *Materials and Methods:* This systematic review was conducted following the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) guidelines. For this study, we searched the PubMed, Embase, Web of Science, Scopus, and Cochrane Library literature databases up to June 2022. A total of seven studies were included, and quality assessment was performed for each included study using the Newcastle–Ottawa Quality Assessment Scale. Statistical analysis was performed on the extracted pooled clinical outcome data, and subgroup analyses were completed. *Results:* A total of 570 patients were included in the analysis. In pooled analysis, the final follow-up International Knee Documentation Committee (IKDC) score showed a significant increase (mean difference (MD), −32.82; 95% confidence interval (CI), −38.32 to −27.32; *p* < 0.00001) with significant heterogeneity (I2 = 93%, *p* < 0.00001) compared to the preoperative score. The Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) scores at final follow-up were significantly decreased (MD, 30.73; 95% CI, 24.10–37.36; *p* < 0.00001) compared to the preoperative scores, with significant heterogeneity (I<sup>2</sup> = 95%, *p* < 0.00001). The visual analog scale (VAS) score at final follow-up was significantly decreased (MD, 4.81; 95% CI, 3.17–6.46; *p* < 0.00001) compared to the preoperative score, with significant heterogeneity (I<sup>2</sup> = 98%, *p* < 0.00001). Two studies evaluated the modified Magnetic Resonance Observation of Cartilage Repair Tissue (M-MOCART) score and confirmed sufficient improvement. In a study analyzing a group treated with bone marrow aspiration concentrate (BMAC), there was no significant difference in clinical outcome or M-MOCART score, and the post-treatment International Cartilage Repair Society (ICRS) grade increased. *Conclusion:* This analysis demonstrated the safety, efficacy, and quality of repaired cartilage following hUCB-MSC therapy. However, there was no clear difference in the comparison with BMAC. In the future, comparative studies with other stem cell therapies or cartilage repair procedures should be published to support the superior effect of hUCB-MSC therapy to improve treatment of cartilage defect or osteoarthritis.

**Citation:** Lee, D.H.; Kim, S.A.; Song, J.-S.; Shetty, A.A.; Kim, B.-H.; Kim, S.J. Cartilage Regeneration Using Human Umbilical Cord Blood Derived Mesenchymal Stem Cells: A Systematic Review and Meta-Analysis. *Medicina* **2022**, *58*, 1801. https://doi.org/10.3390/ medicina58121801

Academic Editor: Yong In

Received: 29 October 2022 Accepted: 5 December 2022 Published: 6 December 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

**Keywords:** human umbilical cord blood derived mesenchymal stem cells; cartilage regeneration; cartilage repair; osteoarthritis treatment; stem cell therapy

#### **1. Introduction**

Osteoarthritis is one of the main diseases of the modern aging population, and available methods for its treatment are gradually expanding with the development of arthroplasty, oral medications, and physiotherapy [1]. Recently, with improvements in quality of life, the target and favored treatment methods of osteoarthritis have changed. Patient needs for early diagnosis and treatment of osteoarthritis are gradually increasing, and many studies are being conducted on joint-preserving surgery. With the introduction of magnetic resonance imaging (MRI), it became possible to evaluate the condition of articular cartilage more precisely before surgery [2], and treatments for articular cartilage defects have advanced. However, due to the limited regenerative capacity of cartilage, no treatment method has achieved a dramatic effect. As cartilage regeneration through microfracture surgery is regenerated by fibrocartilage, its limitations are already well known [3,4]. Accordingly, various methods such as autologous matrix-induced chondrogenesis, autologous chondrocyte implantation, and osteochondral autograft transfer system surgery have been tested for hyaline-like cartilage regeneration [5]. However, problems such as donor site morbidity and the need for multiple operations remain, and problems also persist postoperatively in the long term due to limited cartilage regeneration [6]. In addition to surgical treatment, studies using small-molecule drugs to enhance cell regeneration are also active, which are used not only in cartilage but also in various tissues [7,8].

Basic and clinical research efforts on various types of stem cell therapy are now active. Mesenchymal stem cells (MSCs) have a high regeneration capacity and are most suitable for cartilage repair, and many associated studies have been conducted. MSCs can be obtained from various sources, such as bone marrow, adipose tissue, placenta, and umbilical cords. Methods for extracting MSCs from bone marrow or adipose tissue, such as bone marrow aspiration concentrate (BMAC), adipose-tissue-derived MSCs (Ad-MSCs), and adiposetissue-derived stromal vascular fraction (ADSVF), have been studied, and their efficacy has been demonstrated [9,10]. However, this treatment approach also involves the need for invasive collection of stem cells, and there are issues with the quantity or quality of stem cells that can be obtained [10]. In contrast, there are fewer problems with the human umbilical cord blood derived MSC (hUCB-MSC) collection process, there are fewer ethical concerns that may arise when using embryonic stem cells, and these cells have a superior differentiation capacity compared to adult stem cells. Accordingly, researchers in various medical fields are increasingly focusing in this direction [11–13]. The usage for cartilage repair has also been analyzed in many animal studies [14], and it has been approved as a medical product and is used in clinical practice. Cartistem® (Medipost Inc., Sungnam, Gyeonggi-do, South Korea) was approved by the Korea Food and Drug Administration and is being used in practice, and clinical studies on this product are currently underway in the United States [9]. However, Cartistem® has not yet been used in countries other than Korea, and long-term follow-up results after use are insufficient, indicating the need for additional data and research. Therefore, we aimed to support the efficacy of the cartilage regeneration procedure using hUCB-MSCs as a treatment for cartilage defects through this systematic review. The objective of this systematic review was to prove the efficacy and safety of hUCB-MSCs treatment in a comprehensive analysis by including all subjects with knee articular cartilage defect or osteoarthritis who have undergone cartilage repair surgery using hUCB-MSCs. We included all cartilage repair surgeries without dividing them into arthroscopic assist and open procedure and excluded subjects with injection and other treatments. A meta-analysis synthesizing the clinical outcomes of studies published thus far and an analysis of the collective results were conducted. In some of the included studies, BMAC and microfracture procedures were used as comparison targets for the treatment

of cartilage defects, and the results were analyzed. It is expected that the reliability of hUCB-MSC therapy will increase, and more studies will be conducted to achieve more effective treatment methods.

#### **2. Materials and Methods**

This systematic review was conducted following the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) guidelines [15].

#### *2.1. Search Strategy*

Two reviewers independently conducted an electronic literature search on the treatment of knee cartilage defects and osteoarthritis using umbilical-cord-blood-derived stem cells. The PubMed, Embase, Web of Science, Scopus, and Cochrane Library databases were searched by two reviewers up to June 2022. The main search keywords were (MeSH term "Cartilage, Articular" or "knee joint" or "Osteoarthritis, knee") AND ("Umbilical cord blood-derived mesenchymal stem cell" or the MeSH term "Cord Blood Stem Cell Transplantation"), and the search was conducted including additional keywords related to these terms. After the initial search, duplicates were deleted, and each reviewer verified that there were no missing articles from the electronic search (Supplementary Materials File S1).

#### *2.2. Study Selection with Eligibility Criteria*

The inclusion and exclusion criteria are shown in Table 1. We searched all studies using hUCB-MSCs as a treatment for knee cartilage defects and osteoarthritis. Due to the small number of studies, randomized controlled trials (RCTs), prospective cohort studies, retrospective cohort studies, and case–control studies were all included. Only cases reporting on direct surgical treatment of cartilage lesions using hUCB-MSCs were included, and those involving intra-articular injection were excluded. Animal studies, phase I/II clinical trials, case reports, technical notes, review articles, and articles without accessible full-text versions were excluded, and when two or more studies were published by a single center, any with overlapping patient groups were excluded.

**Table 1.** Inclusion and exclusion criteria.

#### Inclusion criteria


Exclusion criteria


#### *2.3. Data Extraction and Quality Assessment*

Two reviewers extracted the following data from the included studies: first author, publication year, inclusion criteria, number of participants, age, body mass index, defect size, follow-up duration, outcome, concomitant intervention, and study design. Each extracted data point was verified by the rest of the reviewers. If additional information was needed from the included studies, the author of the study was contacted, and information was obtained.

Quality assessment of the included studies was conducted independently by two reviewers using the Newcastle–Ottawa scale [16]. The study by Lim et al. [17] was evaluated using the Newcastle–Ottawa scale because of the analysis of the clinical outcome during extended follow-up after their RCT. Three quality parameters of the Newcastle–Ottawa scale were evaluated, selection, comparability, and outcome. When the opinions of the reviewers differed, the final decision was achieved through discussion. The final evaluation was divided by the number of stars, and a score ≥ 7 points was indicative of high quality, that of 5–6 points was indicative of moderate quality, and that of four or fewer points was indicative of low quality.

#### *2.4. Data and Statistical Analyses*

In the included studies, statistical analysis was performed by extracting information on the preoperative clinical outcome and clinical outcome at final follow-up in all cases where hUCB-MSCs were used for treatment. Meta-analysis was performed on the measurement method used simultaneously in three or more studies among the clinical outcome scoring methods. In the study by Lim et al. [17], extended follow-up was performed up to 5 years after the RCT, but the analysis was conducted using the outcome measured at the 3-year follow-up, in line with the follow-up period of other included studies. In the study by Song et al. [18], analysis was performed by dividing the participants into two groups according to the presence or absence of trochlea lesions. Six studies evaluating participants using the International Knee Documentation Committee (IKDC) score, six studies evaluating participants using the Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC), and five studies evaluating participants using the visual analog scale (VAS) were assessed with each measurement method. Each outcome was a continuous variable and was measured as the mean difference (MD) with 95% confidence interval (CI). Heterogeneity was assessed using the I<sup>2</sup> test. When I<sup>2</sup> < 50%, a fixed-effects model was used; in other cases, a random-effects model was used. All statistical analyses were performed using RevMan version 5.4 (The Cochrane Collaboration, London, UK).

#### **3. Results**

#### *3.1. Identification of Studies*

Figure 1 presents the search information and shows a PRISMA flow diagram of the study selection process. An electronic literature search found 202 articles, and 53 duplicate articles were removed. After checking the titles/abstracts of the remaining 149 articles, those that met the exclusion criteria were removed, and the full-text versions of the remaining articles (n = 14) were analyzed. Among them, seven articles were additionally removed based on the inclusion/exclusion criteria, and seven articles were finally reviewed and meta-analyzed.

#### *3.2. Characteristics of Included Studies*

Table 2 summarizes the characteristics of the seven studies included in this review, including inclusion criteria, number of participants, age, body mass index, defect size, followup duration, outcome, concomitant intervention (performance of high tibial osteotomy (HTO)), and study design. In four studies, HTO was performed concomitantly [18–21], HTO was not performed in two studies [17,19], and one study did not consider HTO [20]. In two studies, only kissing lesions in the medial compartment were used as inclusion criteria [21,22]. In two studies, the results were evaluated using a modified Magnetic Resonance Observation of Cartilage Repair Tissue (M-MOCART) score generated with MRI [19,20]. All seven studies included in the analysis scored ≥ 6 points when evaluated using the Newcastle–Ottawa scale. Four studies scored 6 points (moderate quality), two studies scored 7 points, and one study scored 8 points (high quality) (Table 3).

**Figure 1.** PRISMA flow diagram for the systematic review.


**Table 2.**

Characteristics

 of included studies.


 meansthesubject gets pointsinthatarea

**Table**

**2.**

*Cont.*

chung 2021 [23]

song 2020 (1) [19]

song 2020 (2) [18]

song 2020 (3) [21]

Lee 2021 [22]

Ryu 2022 [20] Lim 2021 [17]

 \*

 \*

 \*

 \*

 \*

 \*

 \*

 0

 0

 0

 0

 \*

 0

 \*

> \*

 \*

 \*

 \*

 \*

 \*

 \* \*

 \*

 \*

 \*

 \*

 \*

 \*

 \*

 \*

 \*

 \*

 \*

 \*

 \*

 \*

 0

 0

 0

 0

 0

 0

 0

 \*

 \*

 \*

 \*

 \*

 \*

 \*

 \*

 \*

 \*

 \*

 \*

 \*

 \*

 0

 0

 0

 0

 0

 \*

 \*

 6

 6

 6

 6

 7

 7

 8

 high

 high

 high

 moderate

 moderate

 moderate

 moderate

#### *3.3. Subgroup Analysis of Included Studies*

Subgroup analysis was performed in five studies, which are described in Table 2. Analysis of these studies showed that age had no effect on outcome [17–20], except in one study [21]. In addition, subgroup analysis of lesion size revealed that it did not affect the outcome in three studies [17–19] but did so in one study [21]. Unlike in other studies, subgroup analysis revealed that a younger age and a larger lesion size led to significantly greater improvements in the 2-year outcome in the study by Song et al. [21]. However, as the inclusion criteria of this study included age >60 years and kissing lesions in the medial compartment, the age cutoff for dividing the younger/older age groups was 65 years. As the regenerative ability of those > 65 years of age is significantly reduced, this result should be considered separately from the results of the other four studies. In addition, the number of patients included in this study was rather small (n = 25 participants). Therefore, it would be reasonable to conclude that age and lesion size have little effect on outcome based on the collected research results. In two studies [18,19], subgroup analyses were performed on other factors, such as lesion location and number, and neither the presence of trochlear lesions nor the number of lesions had a significant effect on the outcome. However, medial femoral condyle (MFC) lesions led to a significantly worse outcome than trochlear lesions [19]. In addition, subgroup analysis was performed considering obesity in two studies [18,21], showing that obesity did not have a significant effect on the outcome.

#### *3.4. Study Outcome*

The outcomes of each included study are summarized in Table 4. Publication bias was evaluated using a funnel plot, which showed a symmetrical appearance, suggesting low possibility of bias (Supplementary Materials File S2).

#### 3.4.1. IKDC Score

The preoperative and final follow-up scores were extracted from six studies evaluating IKDC scores and analyzed. Totals of 441 patients with preoperative scores and 431 patients with final follow-up scores were included. The shortest final follow-up period was a mean of 1.7 years, and most studies presented 2–3 years of follow-up data (Table 2). However, in the study by Lim et al. [17], 3-year follow-up data were extracted and used for analysis. In the study by Song et al. [18], the divided groups were evaluated according to the presence or absence of trochlear lesions. In pooled analysis, the final follow-up IKDC score was significantly increased (MD, −32.82; 95% CI, −38.32 to −27.32; *p* < 0.00001) compared to the preoperative score, with significant heterogeneity (I2 = 93%, *p* < 0.00001) (Figure 2).

#### 3.4.2. WOMAC Score

The preoperative and final follow-up WOMAC scores were extracted from six studies and analyzed. For this, 446 patients with preoperative scores and 436 patients with final follow-up scores were included. The shortest final follow-up period was a mean of 15.6 months, and most studies presented 2–3 years of data (Table 2). As before, 3-year follow-up data were used from the study by Lim et al. [17], and the data of Song et al. [18] were divided into two groups and analyzed. In pooled analysis, WOMAC final follow-up scores were significantly decreased (MD, 30.73; 95% CI, 24.10–37.36; *p* < 0.00001) compared to preoperative scores, with significant heterogeneity (I2 = 95%, *p* < 0.00001) (Figure 3).



25

Y = year, M = month, hUCB-MSC

International

pre = preoperative,

 Knee

 final = final follow-up.

57.8)

64.1)

59.3)

 = human umbilical cord blood derived

Documentation

 Committee score, WOMAC = Western Ontario and McMaster Universities Osteoarthritis

60.7)

28.9)

31.0)

34.9)

mesenchymal

 stem cell, BMAC = bone marrow aspiration concentrate,

33.5)

31.0)

38.2)

42.3)

35.8)

 IKDC =

 Index, VAS = visual analogue scale,

**Figure 2.** Forest plot of mean differences in IKDC subjective scores between pre-operation and at final follow-up.

**Figure 3.** Forest plot of mean differences in WOMAC scores pre-operation and at final follow-up.

#### 3.4.3. VAS Score

The preoperative and final follow-up VAS scores were extracted from five studies and analyzed. Here, 348 patients with preoperative scores and 338 patients with final follow-up scores were included. The final follow-up period ranged from a minimum of 2 years to a maximum of 36.1 months (Table 2). The data of the studies [17,21] in which VAS scores were evaluated based on a total of 100 points were divided by 10, and the standard was unified into a total of 10 points for analysis. As before, 3-year follow-up data were used from the study by Lim et al. [17], and the data of Song et al. [18] were divided into two groups and analyzed. In pooled analysis, the VAS final follow-up scores were significantly decreased (MD, 4.81; 95% CI, 3.17–6.46; *p* < 0.00001) compared to preoperative scores, with significant heterogeneity (I2 = 98%, *p* < 0.00001) (Figure 4).


**Figure 4.** Forest plot of mean difference in VAS scores between pre-operation and final follow-up.

#### 3.4.4. M-MOCART Score

The M-MOCART score was measured in two of the seven studies included in the analysis (Supplementary Materials File S3). In the study by Song et al. [19], 34 patients received M-MOCART scores, and the mean score at 3–6 months after surgery was 30.58 points. According to MRI scans performed ≥ 1 year after surgery (mean, 21.2 months), the mean score increased to 55.44 points. Ryu et al. [20] measured the M-MOCART score using MRI scans at 1 year and 2 years after surgery in 27 patients, recording mean respective scores

of 69.63 and 73.7 points. In both studies, the M-MOCART score after ≥1 year was sufficient compared to the general results of other cartilage repair surgeries [24,25]. However, evidence confirming whether the M-MOCART score correlates with clinical outcome is insufficient, and the M-MOCART score should be considered an auxiliary indicator of repaired cartilage quality [25]. In the study by Ryu et al. [20], the M-MOCART scores of 25 patients in the BMAC group were measured and compared with those of the hUCB-MSC group. Ultimately, the results of the BMAC group at 1 year and 2 years were 65.4 and 70.2 points, with the hUCB-MSC group showing better results, although there was no significant difference.

#### 3.4.5. Comparison with BMAC and Microfracture Procedures

Comparisons with a control group were performed in three studies. In one study, the group only treated with microfracture surgery was set as a control group [17]; in two studies, the group treated with BMAC procedures was set as a control group [20,22]. The clinical outcome data of the control group are included in Table 4. In the study by Lim et al. [17], compared to the microfracture group, IKDC scores at 48 weeks did not show a significant difference in the hUCB-MSC group, with a mean of 53.4 points versus a mean of 53.5 points. However, at 3 years of follow-up, the hUCB-MSC group had a mean score of 57.4 points and the microfracture group had a mean score of 49.0 points, confirming that better results were obtained in the hUCB-MSC group. Moreover, this trend continues in the 4- and 5-year follow-up comparisons. WOMAC scores also showed no significant difference between the groups, with a mean of 24.7 points for the hUCB-MSC group and a mean of 26.2 points for the microfracture group at 48 weeks; in contrast, at 3 years of follow-up, the hUCB-MSC group had a mean score of 25.4 points versus the microfracture group mean score of 34.5 points, indicating better results in the hUCB-MSC group. Again, this trend persists in the 4- and 5-year follow-up comparisons. In addition, considering the VAS scores, the results of the hUCB-MSC group appear better in 3-, 4-, and 5-year follow-up comparisons with the microfracture group. Regardless of the scoring system, a divergence between groups seems to appear from the third year onward; although the microfracture group had good results until 48 weeks, this good result was not maintained thereafter. In the study by Ryu et al. [20], the outcome of hUCB-MSC therapy at 2 years of follow-up was compared with that of BMAC, and the hUCB-MSC group had a mean IKDC score of 81.35 points compared to the 80.287 points of the BMAC group, showing no significant difference. There was also no significant difference in VAS scores, with means of 0.85 and 0.92 points, respectively. In the study by Lee et al. [22], the outcome of the hUCB-MSC group was evaluated at a mean of 15.6 months, while that of the BMAC group was evaluated at a mean of 20.7 months. WOMAC scores were slightly better in the hUCB-MSC group, with a mean of 19.5 points, compared to the BMAC group with a mean of 23.4 points, but there was no significant difference. In addition to the WOMAC score, the Hospital for Special Surgery score and the pain and function Knee Society scores were also calculated but did not show significant differences between groups, although the results of the hUCB-MSC group were slightly better.

#### **4. Discussion**

In the past, arthroplasty was the most common treatment for osteoarthritis; recently, joint-preserving surgery has been developed. In particular, in Korea, the cartilage regeneration procedure performed together with correction of varus or valgus alignment through corrective osteotomy is gradually increasing in popularity [26]. In line with this, stem cell therapy for cartilage defects and osteoarthritis has developed through many studies. Previously, stem cell therapy mainly used BMAC, AD-MSCs, and ADSVF, but research on hUCB-MSCs has been active since Cartistem® was approved by Korea's Food and Drug Administration. As there is a large number of MSCs in umbilical cord blood, its use has been expanded in various fields [27]. To date, a disadvantage to using stem cells has been that the procedure is invasive, and the cells are difficult to obtain in sufficient number from

bone marrow and adipose tissue. On the other hand, sourcing hUCB-MSCs can result in collection of a constant and sufficient number of stem cells in a non-invasive manner. In addition, these cells are hypoimmunogenic and do not cause immune-related problems. In addition, unlike with embryonic stem cells, there are no ethical problems [5,28]. Finally, hUCB-MSCs boast higher proliferation, karyotype stability after prolonged expansion, and easier chondrogenic differentiation compared to MSCs extracted from bone marrow and adipose tissue [29]. Therefore, cartilage regeneration therapy using hUCB-MSCs is considered more effective than conventional treatment and is being used in clinical trials. Many retrospective studies have been announced and usually report satisfactory results. Several case reports have also been published, one of which reported satisfactory results using hUCB-MSCs as a treatment for juvenile osteochondritis dissecans [30].

Therefore, we summarized the published therapeutic effects of hUCB-MSCs and present our opinions. First, hUCB-MSC therapy is a new clinical treatment, and its safety must be confirmed. There appear to be no severe adverse effects (AEs) among the studies published to date; however, there were reports of general AEs in three studies. In the study by Park et al. [28], there were no serious AEs, and one patient experienced an elevation of antithyroglobulin antibody level, which was reported as a treatment-emergent AE, but it spontaneously normalized without any special treatment. Ryu et al. [20] reported three cases of adhesions, but this is a minor complication of arthroscopic surgery, and the frequency in this study was not particularly high. Chung et al. [23] reported that temporary knee swelling was observed up to 1 month after surgery in some patients, but it was self-limiting, and there were no other serious AEs. In the study by Lim et al. [17], there were no other surgical-related serious AEs other than surgical site pain, the final reported serious AEs were also not related to the use of hUCB-MSCs, and there were no immunological reactions. Therefore, the safety of hUCB-MSCs is supported by the available research results.

In several studies included in this review, subgroup analyses were conducted. In four studies with subgroup analyses according to age [17–20], excluding one other study [21], age did not affect the results. In three studies [17–19], lesion size did not affect the results. Considering the process of obtaining stem cells from bone marrow or adipose tissue, there may only be a small number of stem cells extracted from a patient of old age, and there may be an overall decrease in their quality with a low differentiation potential. However, as mentioned earlier, hUCB-MSC therapy involves using a product of a certain quality and number of stem cells, and the results can be seen above. However, implanted MSCs are known to promote the regeneration of host cells with a paracrine effect rather than turning them into cartilage, so it may not be described only by stem cell quality [31,32]. In addition, some studies have reported good pain level and functional outcome when using high-dose MSCs [33], indicating the need for additional research.

In pooled data analysis of seven studies included in this systematic review, IKDC, WOMAC, and VAS scores all showed statistically significant improvements at final followup compared to the preoperative evaluation. This means that clinical outcomes can be effectively improved through cartilage regeneration surgery using hUCB-MSCs. The final follow-up duration of the included studies varied, but the mean is ≥15.6 months, and considering that most of the studies included a follow-up period of about 2 years, this means that the clinical outcome of the hUCB-MSC therapy was sufficiently improved and maintained during the short-term follow-up. It is known that the clinical symptoms recur before two years if there is a failure in cartilage regeneration. Therefore, a favorable shortterm outcome is implied when the improvement of clinical outcome is maintained during the 2-year follow-up [34]. In addition, three studies [17–19] showed good results in clinical outcomes measured after a mean period of ≥3 years. Several studies reported good quality of regenerated cartilage for up to 2 years after microfracture surgery and then gradually deteriorated [35,36]. In the study by Lim et al. [17], the microfracture group had a worse outcome from the third year onward, while the hUCB-MSC group maintained an improved outcome until 5 years of follow-up. With hUCB-MSC therapy, it can be inferred that the

quality of regenerated cartilage is good, and regenerated cartilage is more hyaline-like than microfracture alone. In addition, Suh et al. [37] compared radiological changes after surgery between a group that received hUCB-MSC therapy and a group that received microfracture surgery, with both groups receiving HTO as concomitant surgery, and found that the hUCB-MSC group had a significant increase in joint space width increment (*p* < 0.05) of 0.6 mm compared to that of just 0.1 mm in the microfracture group.

Among the studies included in this systematic review, two [20,22] considered BMAC, and no significant differences were seen in the IKDC, WOMAC, VAS, and M-MOCART scores. However, Lee et al. performed second-look arthroscopy in all patients, and the hUCB-MSC group had better ICRS grade results in the medial femoral condyle and medial tibial condyle, so it seems that hUCB-MSC therapy facilitates better cartilage regeneration. Ryu et al. also performed second-look arthroscopy in some patients and measured the ICRS grade, and they reported no significant difference. However, the study by Ryu et al. includes a difference in the mean age of the hUCB-MSC group and BMAC group (53.93 vs. 39.64 years). In other words, the BMAC group may have experienced better results because younger people have better healing potential. Considering these points, more data must be accumulated to compare the clinical outcomes of cartilage regeneration. In the study by Yang et al. [38], which was excluded from this analysis because there was a risk of overlapping patient groups, the results of hUCB-MSC and BMAC treatments were compared by performing propensity score matching for sex, age, body mass index, and lesion size. No significant differences between the two groups in the measurement of IKDC score, Knee Injury and Osteoarthritis Outcome Score, Short-Form 36 score, or Tegner Activity Score were found. However, following ICRS Cartilage Repair Assessment grading, which was carried out through second-look arthroscopy, the score of the hUCB-MSC group (n = 44) was 9.2 ± 2.2 points, and that of the BMAC group was (n = 37) 7.2 ± 3.0 points. In other words, the hUCB-MSC group experienced significantly better cartilage regeneration. Based on these results, hUCB-MSC therapy can be considered to lead to slightly betterquality cartilage regeneration than BMAC. However, there is still little comparative research with BMAC, and more is needed, with long-term results ≥3 years considered crucial.

Our study has some limitations. First, among the studies included in our analysis, six were retrospective studies, and one was an extended follow-up data analysis after RCT. Meta-analysis based on RCT studies could not be performed due to the small number of studies. In order to form a more scientific basis for the effect of hUCB-MSCs, studies analyzing multiple RCTs will be needed. Second, in the studies included in our investigation, several patient characteristics such as age, sex, defect size, and follow-up duration varied, and it is believed that this may have affected the heterogeneity of the results. Lastly, three of the studies included in this meta-analysis were published by a single center, and two of the remaining studies were also published by another single center. We checked the inclusion criteria with the study authors to confirm that there were no overlapping patients and proceeded with our analysis; however, this is a potential cause of bias. All of these limitations are considered to be caused by the insufficient number of studies published so far.

#### **5. Conclusions**

Cartilage regeneration surgery using hUCB-MSCs is expected to lead to better results than conventional stem cell therapy or other cartilage repair procedures. Through this review, the safety, efficacy, and quality of repaired cartilage associated with this procedure were demonstrated. In the future, a number of comparative studies considering other stem cell therapies or cartilage repair procedures will be published, and we expect that the relatively superior effect of hUCB-MSC therapy will be demonstrated. Based on this, we hope that the treatment of cartilage defects and osteoarthritis will be improved.

**Supplementary Materials:** The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/medicina58121801/s1, Supplementary File S1: Keywords and search results. Supplementary File S2: Funnel plot of IKDC, WOMAC, and VAS scores. Supplementary File S3: M-MOCART score and second-look arthroscopic findings of included studies.

**Author Contributions:** Conceptualization, D.H.L. and S.J.K.; methodology, D.H.L., S.A.K. and J.-S.S.; data curation, D.H.L., S.A.K. and J.-S.S.; writing—original draft preparation, D.H.L. and B.-H.K.; writing—review and editing, D.H.L. and S.J.K.; visualization, D.H.L.; supervision, S.J.K. and A.A.S.; project administration, D.H.L. and S.J.K. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** The data presented in this study are available in the main article and Supplementary Materials Files.

**Acknowledgments:** We thank Seung Jae Lee, librarian of the Medical Library of The Catholic University of Korea, Seoul, Republic of Korea, for helping with the data search.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


### *Review* **Autologous Collagen-Induced Chondrogenesis: From Bench to Clinical Development**

**You Seung Chun 1, Seon Ae Kim 1, Yun Hwan Kim 1, Joong Hoon Lee 1, Asode Ananthram Shetty <sup>2</sup> and Seok Jung Kim 1,\***


**Abstract:** Microfracture is a common technique that uses bone marrow components to stimulate cartilage regeneration. However, the clinical results of microfracture range from poor to good. To enhance cartilage healing, several reinforcing techniques have been developed, including porcine-derived collagen scaffold, hyaluronic acid, and chitosan. Autologous collagen-induced chondrogenesis (ACIC) is a single-step surgical technique for cartilage regeneration that combines gel-type atelocollagen scaffolding with microfracture. Even though ACIC is a relatively new technique, literature show excellent clinical results. In addition, all procedures of ACIC are performed arthroscopically, which is increasing in preference among surgeons and patients. The ACIC technique also is called the Shetty–Kim technique because it was developed from the works of A.A. Shetty and S.J. Kim. This is an up-to-date review of the history of ACIC.

**Keywords:** atelocollagen; microfracture; cartilage repair; knee; ACIC

#### **1. Introduction**

The articular cartilage is a thin layer of viscoelastic connective tissue that complicates repair of the cartilage. Studies have emphasized the importance of repairing articular cartilage to delay arthroplasty. Multiple drilling techniques have been used to treat articular cartilage defects. In 1959, Pridie [1] said, "If these sclerotic areas were drilled and the holes were not too far apart, smooth fibro-cartilage would spread over the surface." He introduced the drilling of the subchondral bone to heal cartilage defects.

Steadman et al. [2] modified this method to avoid thermal necrosis from drilling, using bone marrow from the subchondral bone to form a clot on the defect holes (Figure 1). The clot contained mesenchymal stem cells (MSCs) and abundant growth factors, inducing repair by fibrous and hyaline-like cartilage [3] (Figure 2).

Multiple drilling techniques produce predominantly fibrocartilage [4]; therefore, several methods have been introduced to produce hyaline-like cartilage. Autologous chondrocyte implantation is considered an ideal procedure to induce hyaline-like cartilage. However, it requires a two-stage procedure, damages the donor site cartilage, and has a high cost.

The limitation of the marrow stimulation technique is that bone marrow stem cells and growth factors are released into the joint rather than remaining at an articular surface. Collagen gel or scaffold has been used to overcome this limitation by providing mechanical stability to form clots [5]. Cell-free type I collagen gels or scaffolds combined with marrow stimulation, such as multiple drilling, have shown good outcomes regarding the induction of hyaline-like cartilage [6–8]. Autologous matrix-induced chondrogenesis (AMIC) uses a porcine collagen matrix to provide a biological scaffold.

**Citation:** Chun, Y.S.; Kim, S.A.; Kim, Y.H.; Lee, J.H.; Shetty, A.A.; Kim, S.J. Autologous Collagen-Induced Chondrogenesis: From Bench to Clinical Development. *Medicina* **2023**, *59*, 530. https://doi.org/10.3390/ medicina59030530

Academic Editors: Yong In and In Jun Koh

Received: 29 January 2023 Revised: 4 March 2023 Accepted: 7 March 2023 Published: 8 March 2023

**Copyright:** © 2023 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

**Figure 1.** (**a**) Multiple holes drilled into a cartilage defect; (**b**) second-look arthroscopy 2 years later.

**Figure 2.** (**a**) Microfracture of a cartilage defect of the trochlea; (**b**) second-look arthroscopy 2 years later.

Despite its popularity since 2000, the AMIC technique requires an open surgical incision for the preparation of the defect and the application of the collagen membrane. To overcome this weakness, autologous collagen-induced chondrogenesis (ACIC) was developed, for which the procedures are performed arthroscopically. For ACIC, atelocollagen gel is used as a scaffold instead of the membrane used in AMIC. Atelocollagen is a highly purified type I collagen obtained following the treatment of skin dermis with pepsin and telopeptide removal, which reduces immunogenicity.

This article reviews the developmental history with a basic scientific rationale and the various techniques and results of ACIC for the repairing of knee cartilage.

#### **2. Basic Science and Methods for Chondrogenesis**

#### *2.1. Basic Science of Cartilage Injury*

Articular cartilage is a thin, viscoelastic layer of connective tissue 2–3 mm thick [9] (Figure 3). It is of mesodermal origin and is characterized by a cellular component immersed within an extracellular matrix composed of polysaccharides, fibrous protein, and interstitial fluid [10]. The cartilage has no direct supply of blood, nerve signals, or nutrition and relies on diffusion through the surrounding tissues.

**Figure 3.** The structure of articular cartilage, comprised of four zones: superficial, middle, deep, and calcified. Among zones, there are differences in collagen fibers, arrangement of chondrocytes, and distribution of proteoglycans and glycosaminoglycans (GAGs). Hematoxylin and eosin staining. Scale bar, 200 μm.

Cartilage can be damaged easily by acute trauma or repetitive microtrauma and is exposed to mechanical stress during active daily living. Such stress can be increased to 10 or 20 times the body weight during sports activities [9]. Acute cartilage injury initiates a repair process that starts with the formation of a blood clot containing bone marrow cells that form fibrocartilaginous tissue. However, repeated microtraumas damage chondrocytes, decrease production of proteoglycan, and damage collagen meshwork. As this meshwork limits water penetration, damage leads to swelling and stiffness of the tissue. Responses to cartilage injury involve both anabolic and catabolic reactions. Aggrecan-degrading enzyme disintegrin, metalloproteinase with thrombospondin motif 5 (ADAMTS-5), and collagenase matrix metalloproteinase 13 (MMP-13), which degrades type II collagen, all contribute to the breakdown of cartilage. However, the induction of chondroprotective genes leads to anabolic effects on the cartilage [11] (Figure 4).

**Figure 4.** Mechanism leading to the breakdown of cartilage.

#### *2.2. Microfracture and Enhanced Microfracture*

Microfracture is a minimally invasive technique that uses an arthroscope to drill small, equidistant holes in subchondral bone, at least 3–4 mm apart and 4 mm in depth, with 3–4 holes per a 1 cm area [12]. This procedure induces the migration of MSCs from the bone marrow to the cartilage defect to allow the formation of fibrocartilage [13]. Kruez et al. reported that microfracture showed good short-term results in small cartilage defects but poorer results at 18 months after surgery, as reflected by Cincinnati Knee Rating System and International Cartilage Repair Society (ICRS) scores. This effect is clearer in large defects and in defects of the patellofemoral joint [14], where subchondral osteophytes develop in 20–50% of cases [15].

AMIC adds a collagen scaffold to multiple drilled holes in a cartilage defect. The scaffold of the collagen binds MSCs and growth factors to the cartilage defect, enhancing regeneration [16,17]. MSCs from microfracture have the same phenotypic plasticity as chondrogenic cells in the cartilage basal zone. With AMIC, MSCs are distributed on the membrane, which acts as the roof of a "biological chamber" [18]. A systematic review conducted in 2022 by Migliorini et al. reported that AMIC showed better clinical scores and a lower rate of revision, compared to microfracture [19].

AMIC requires an open procedure to attach the collagen matrix to the cartilage defect (Figure 5). Even a small defect in articular cartilage needs a large incision for the AMIC procedure, which can delay patient recovery.

**Figure 5.** (**a**) Cartilage defect, (**b**) after defect preparation, (**c**) membrane-covered defect, and (**d**) second-look arthroscopy 2 years later. Image courtesy of Prof. Sung-Hwan Kim.

To overcome this, ACIC was introduced as an alternative in 2009. Together, Shetty and Kim developed the ACIC technique to be performed with arthroscopy, leaving wounds only for the arthroscopic portal incisions [20]. In this procedure, atelocollagen gel is mixed with fibrin and thrombin and used as a scaffold. The mixture is applied on the cartilage defect after microdrilling the defect under CO2 gas insufflation to maintain the collagen clot in the chondral defect area.

#### *2.3. Autologous Collagen-Induced Chondrogenesis (ACIC)*

#### 2.3.1. Basic Science of ACIC

Atelocollagen is a porcine type I collagen that has been treated to detach the telomeres (Figure 6). This process suppresses immunologic reactions, increasing the effects of the surrounding environment on the collagen.

Atelocollagen induces MSCs to differentiate into chondrocytes (Figure 7). The process is examined by the expression of genes such as Sox9, type II collagen, and aggrecan. Jeong et al. conducted an animal study with rabbits involving a circular, articular cartilage defect 4 mm in diameter in the trochlear region. The 10 rabbits in the control group were not treated, and the 10 rabbits in the experimental group underwent injections of atelocollagen mixed with fibrin. After 12 weeks, the cartilage was examined. The experimental group had regenerated smooth, hyaline-like cartilage, whereas the control group showed incomplete and irregular cartilage [21] (Figure 8).

**Figure 6.** Overview of collagen production. Procollagen chains are synthesized and form a helix in the endoplasmic reticulum (ER). Propeptides are removed by proteinases, and the fibrils are assembled into collagen fibers in the extracellular space (EC).

Many clinical trials have used atelocollagen as an enhancing material for microfractures in a cartilage defect [20–22]. A recent study showed that atelocollagen promotes the chondrogenic differentiation of human adipose-derived MSCs. Chondrogenic genes and proteins were evaluated by RT-qPCR and ELISA, showing differentiation on day 21 [23]. MRI follow-up was conducted at 1 year and showed good cartilage defect filling [20].

**Figure 7.** Transmission electron microscopy of differentiated mesenchymal stem cells (MSCs) into chondrocytes and extracellular collagen fibrils (box). (**Left**) Scale bar, 2000 nm; (**Right**) scale bar, 1000 nm.

**Figure 8.** (**a**) Injection of atelocollagen mixed with fibrin into the trochlear defect of a rabbit knee. (**b**) Atelocollagen injection group after 12 weeks. (**c**) Control group without injection after 12 weeks.

#### 2.3.2. Surgical Technique

ACIC is performed in the same manner as routine knee arthroscopy (Figure 9). The articular cartilage is evaluated and mapped according to ICRS guidelines. The lesion is debrided until the margins show stable vertical cartilage. The surface is lightly abraded with a curette or burr to debride sclerotic subchondral bone. Microdrilling is performed, and the subchondral bone is drilled with a 3.5 mm diameter bit up to a depth of 6–10 mm, with a 3 mm interval. After that, the arthroscopic water of the knee joint is drained, and the joint is insufflated with CO2 (Figure 9). The surface of the subchondral bone is dried with cotton to promote adhesion by the collagen mixture. Atelocollagen gel mixed with fibrinogen and thrombin (ratio—fibrin 1: thrombin 0.2 and atelocollagen 0.8) is prepared and applied under arthroscopy. The applied mixture is assessed for stability in the cartilage defect by observing the range of motion of the knee after 2 min [20,24] (Figure 10).

**Figure 9.** Arthroscopic settings for ACIC.

**Figure 10.** (**a**) Cartilage defect, (**b**) defect preparation and multiple drilling, (**c**) collagen gel injection into the defect under CO2 gas insufflation, and (**d**) second-look arthroscopy at 2 years later.

#### 2.3.3. Post-Operative Protocol

Patients are recommended to undergo continuous passive motion (CPM) rehabilitation for 4 h/day post-operatively for 4–6 weeks. Patients with a femoral condyle lesion are allowed partial weight-bearing at 6 weeks, and progression to full weight-bearing is encouraged at 3 months. Cartilage being repaired in patellar and trochlear lesions is protected with a knee brace locked at 0–20% movement, which is gradually increased to 90% over 6 weeks. Full weight-bearing is encouraged with protected knee motion right after the surgery [24].

#### 2.3.4. Clinical Results

Clinical studies of ACIC were collected through PubMed, the Cochrane library, and ScienceDirect, with keywords "autologous collagen-induced chondrogenesis" and "porcine, collagen, chondrogenesis"; there have been several studies using ACIC on the knee or the talus. Excluding studies about the talus, five studies were analyzed. Two studies were randomized, controlled trials (RCTs), one study was a matched, comparative study, and two were noncomparative studies. (Table 1)



KOOS: Knee Injury and Osteoarthritis Outcome Score, MOCART: Magnetic Resonance Observation of Cartilage Repair Tissue, IKDC: International Knee Documentation Committee.

Kim M.S. et al. compared porcine-derived collagen-augmented chondrogenesis to microfracture in 2020 in a multicenter, randomized control study. One hundred patients were randomly assigned to a microfracture or an investigational group. Clinical and MRI outcomes were assessed at 12 and 24 months post-operatively. Magnetic resonance observation of cartilage repair tissue (MOCART) assessment was used to analyze cartilage tissue repair. MOCART score, VAS score, and KOOS pain score were significantly improved in the test group. In addition, the investigational group showed better filling of the cartilage defect in the knee joint [25].

Silva et al. compared ACIC to microfracture in 2020 based on clinical scores at 6 months and 24 months. Eleven patients who underwent ACIC were compared with 11 age- and sex-matched patients who underwent a microfracture-only procedure. The ACIC group showed a significantly better SF36 mental function, International Knee Documentation Committee (IKDC) score, and VAS score at 24 months [26].

Kim S.J. et al. evaluated 30 patients with ICRS grade III/IVa symptomatic knees who were treated with ACIC. Patients were followed for 6 years, and the Lysholm score, Knee Injury and Osteoarthritis Outcome Score (KOOS), and IKDC score were significantly improved. Radiological evaluation was performed using MRI at 6 months, 1 year, and 3 years using the MOCART scoring system. The mean MOCART score was 78, similar to those of other successful cartilage repair techniques [24].

Shetty et al. evaluated 10 patients with symptomatic chondral defects who were treated with ACIC. Morphological and biochemical MRIs were performed at the 1-year follow-up, and the Lysholm score was assessed at the 2-year follow-up. MRIs showed good cartilage defect filling and suggested hyaline-like repair tissue, and the Lysholm score was significantly improved [20].

#### **3. Discussion**

Mithoefer et al. conducted a systematic review about microfracture in 2009. Microfracture was a good first-line treatment for cartilage defect but did have a few disadvantages. The short-term outcome was good, but the long-term outcome was inconclusive, due to insufficient data. Shortcomings of microfracture included limited regeneration of hyalinelike cartilage, variable volume of cartilage repair, and functional deterioration [15]. As a consequence, many cartilage regeneration techniques were designed to overcome these limitations of microfracture. ACIC not only improved short-term clinical outcomes, it also significantly improved long-term clinical outcomes at 6 years [24].

ACIC is one of the methods for enhancing cartilage repair; it has no donor site comorbidity and is performed using arthroscopy in a single stage. An animal study and an in vitro study showed that atelocollagen promotes hyaline-like cartilage regeneration [21,23]. These encouraging outcomes and the concept of a single-stage cartilage resurfacing technique are attractive for many surgeons.

Though ACIC is performed mostly under arthroscopy, one study involved a mini open procedure. A multicenter study by Kim M.S. et al. involved an open procedure applying atelocollagen at the chondral defect, whereas S Kim S.J. used CO2 gas to inflate the joint space while applying atelocollagen. CO2 gas allows the gravity-independent application of the gel mixture without opening the joint. Thus, all procedures can be conducted under arthroscopy, minimizing soft tissue damage. The method of microfracture may influence the outcomes of cartilage regeneration. Kim S.J. used a 3.5 mm diameter drill up to a depth of 10 mm, with a 3 mm interval. On the other hand, the multicenter study of Kim M.S. did not specify the technique of microfracture. A fibrin and thrombin mixture is used to stabilize the collagen scaffold; however, the mixture ratio varies among studies. Since the collagen scaffold plays a major role in the good outcome of the microfracture technique, the durability of the collagen scaffold may influence the outcomes of cartilage regeneration.

Clinical trials about ACIC are limited. Silva et al. compared ACIC with microfracture in an RCT, another of which was performed by Kim M.S. et al. In addition, Kim S.J. et al. and Shetty et al. found better clinical results and superior MRI findings to microfracture.

This review showed the need to standardize reporting of the AMIC technique to enable future comparisons of efficacy and determine the effects of various technical variations [5]. Some of the technical factors that should be reported are as follows:


In addition, we suggest that the measures for outcomes be standardized. The most common outcome instruments are KOOS, Lysholm, and IKDC scoring. We suggest that these three instruments be used in studies with follow-ups of at least 1–2 years for adequate comparison. MRI protocols for cartilage assessment, such as the modified MOCART score suggested by Marlovits et al., [27], should be routine to allow the comparison of follow-up radiographic studies.

An advantage of ACIC is its cost-effectiveness, compared to ACI. In the UK, the cost of ACI is three times that of ACIC [28]. In addition, ACIC has no donor site morbidity and is performed in one stage.

There are some limitations of ACIC. The few prospective studies of ACIC each had a small sample size. While the results were positive and showed the usefulness of ACIC, comparison studies to other enhanced microfracture techniques are needed.

#### **4. Conclusions**

Our review of the ACIC technique suggests it is a promising cartilage repair technique. The outcome scores and MRI results are promising, but there are few comparative studies with other cell-based cartilage methods. Ideal conditions for chondrogenesis should be studied.

**Author Contributions:** Conceptualization, S.J.K. and A.A.S.; investigation, Y.H.K. and J.H.L.; writing—draft preparation, Y.S.C.; description of surgical methods, S.J.K.; description of basic science, S.A.K.; administration, Y.S.C.; supervision, S.J.K. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Data Availability Statement:** Previous studies included in this study can be found on PubMed.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


**Disclaimer/Publisher's Note:** The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

### *Article* **Allogeneic Umbilical Cord-Blood-Derived Mesenchymal Stem Cells and Hyaluronate Composite Combined with High Tibial Osteotomy for Medial Knee Osteoarthritis with Full-Thickness Cartilage Defects**

**Yong-Beom Park 1,\*, Han-Jun Lee 2, Hyun-Cheul Nam <sup>2</sup> and Jung-Gwan Park <sup>3</sup>**


**Abstract:** *Background and Objectives*: Although the effects of cartilage repair in patients who are undergoing high tibial osteotomy (HTO) remains controversial, cartilage repair may be required for the full-thickness cartilage defect because of a concern of lower clinical outcome. The purpose of this study was to investigate clinical outcome and cartilage repair following implantation of allogeneic umbilical cord-blood-derived MSCs (UCB-MSCs)–hyaluronate composite in patients who received HTO for medial knee osteoarthritis (OA) with full-thickness cartilage defect. *Materials and Methods*: Inclusion criteria were patients with a medial knee OA, a full-thickness cartilage defect (International Cartilage Repair Society (ICRS) grade IV) <sup>≥</sup> 3 cm<sup>2</sup> of the medial femoral condyle, and a varus deformity ≥ 5◦. The full-thickness cartilage defect was treated with implantation of an allogeneic UCB-MSCs–hyaluronate composite following medial open-wedge HTO. Visual analogue scale for pain and Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) score were assessed at each follow-up. Cartilage repair was assessed by the ICRS cartilage repair assessment system at second-look arthroscopy when the plate was removed. *Results*: Twelve patients (mean age 56.1 years; mean defect size: 4.5 cm2) were included, and 10 patients underwent second-look arthroscopy during plate removal after a minimum of 1 year after the HTO. At the final follow-up of mean 2.9 years (range; 1–6 years), all clinical outcomes had improved. At second-look arthroscopy, repaired tissue was observed in all cases. One case (10%) showed grade I, seven (70%) cases showed grade II, and two (20%) cases showed grade III according to ICRS cartilage repair assessment system, which meant that 80% showed an overall repair assessment of "normal" or "nearly normal". *Conclusion*: Allogeneic UCB-MSCs-HA composite implantation combined with HTO resulted in favorable clinical outcome and cartilage repair in all cases. These findings suggest that UCB-MSCs-HA composite implantation combined with HTO would be a good therapeutic option for patients with knee OA and full-thickness cartilage defects.

**Keywords:** osteoarthritis; cartilage; high tibial osteotomy; stem cells; umbilical cord blood

#### **1. Introduction**

High tibial osteotomy (HTO) is a well-established treatment option in knee osteoarthritis (OA) with varus deformity [1–3]. Shifting the mechanical axis of the lower extremity to the lateral side decreases the contact pressure on the medial compartment, which can provide the biological environment to prevent further degenerative changes [4,5]. Although several studies reported favorable short-term and mid-term outcomes after HTO, favorable long-term outcomes may be associate with adequate cartilage repair [6,7].

**Citation:** Park, Y.-B.; Lee, H.-J.; Nam, H.-C.; Park, J.-G. Allogeneic Umbilical Cord-Blood-Derived Mesenchymal Stem Cells and Hyaluronate Composite Combined with High Tibial Osteotomy for Medial Knee Osteoarthritis with Full-Thickness Cartilage Defects. *Medicina* **2023**, *59*, 148. https:// doi.org/10.3390/medicina59010148

Academic Editors: Yong In and In Jun Koh

Received: 23 November 2022 Revised: 31 December 2022 Accepted: 5 January 2023 Published: 11 January 2023

**Copyright:** © 2023 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

Although several therapeutic approaches to restore the cartilage have been investigated, the currently available options are not optimal for cartilage repair in knee OA. For the treatment of osteoarthritic cartilage defects, however, microfracture has been used [8,9], but it showed the deterioration of clinical outcomes over time [10].

Recently, mesenchymal stem cells (MSCs) have been gaining attention as a potential cell source for cartilage repair in older patients because of their unique properties, including immunomodulatory and anti-inflammatory capacities, and paracrine activity [11–16]. Allogeneic umbilical cord-blood-derived MSCs (UCB-MSCs) have advantages of non-invasive cell collection, high expansion capacity, hypo-immunogenicity, and immunomodulatory capacity [17–19]. Allogeneic UCB-MSCs implantation showed the safety and efficacy of cartilage repair in older patients with knee OA [20–24]. However, clinical outcomes and cartilage repair after allogeneic UCB-MSCs implantation combined with HTO have been reported rarely [25–27].

The purpose of this study was to investigate clinical outcome and cartilage repair following implantation of allogeneic UCB-MSCs–hyaluronate (UCB-MSCs-HA) composite combined with HTO for older patients who had medial knee OA with large full-thickness cartilage defect. It was hypothesized that an implantation of UCB-MSCs-HA composite would result in good cartilage remodeling and favorable clinical outcomes.

#### **2. Materials and Methods**

A total of 62 patients who underwent HTO between 2016 and 2021 were retrospectively reviewed for study inclusion. HTO was indicated for patients who had isolated medial compartmental OA and varus malalignment of the lower extremity (varus deformity ≥ 5◦) without ligament instability. Patients who underwent HTO and concomitant implantation of a UCB-MSCs-HA composite (Cartistem®) for medial compartmental OA and a fullthickness cartilage defect (International Cartilage Repair Society (ICRS) grade IV) ≥ 3 cm<sup>2</sup> of the medial femoral condyle were included. Patients with a history of previous knee surgery, other cartilage repair procedure such as microfracture, and follow-up loss were excluded. Finally, 12 patients were included for this study (Figure 1). This study was approved by the institutional review board of our hospital (IRB No. 2109-025-19385).

**Figure 1.** Flow diagram of the included patients.

#### **3. UCB-MSCs-HA Composite**

The medicinal product (UCB-MSCs-HA composite, Cartistem®) of this study was approved by the Korea Food and Drug Administration in 2012. Allogeneic UCB-MSCs were taken from donor UCB stored at a cord blood bank and were produced according to good manufacturing practice guidelines by Medipost (Seoul, South Korea). This product comprises 1.5 mL of UCB-MSCs (7.5 × 106) and 4% HA.

#### **4. Surgical Technique**

A standard arthroscopy was performed to assess cartilage defects, and arthroscopic procedures including debridement of the cartilage flaps, meniscectomy, or meniscal repair were performed if necessary. Biplanar open-wedge HTO was then performed according to preoperative planning to achieve a valgus alignment of 3–5◦.

After HTO, the UCB-MSCs-HA composite was implanted following the previously reported technique [20,21]. A small arthrotomy was made to expose the cartilage defect on the femoral condyle. Cartilage defects was prepared for healthy underlying bone and peripheral margin. Multiple drill holes (4 mm × 4 mm (diameter × depth)) in the subchondral bone were made to place the UCB-MSCs-HA composite. In addition, multiple drill holes with small diameter of 1.4 mm were made between the larger drill holes for better integration. The UCB-MSCs-HA composite was implanted into the drill holes carefully (Figure 2).

**Figure 2.** Surgical procedure of implantation for the allogeneic UCB-MSCs-HA composite and combined HTO. (**A**) Preoperative whole-lower-extremity radiograph showed varus limb alignment. (**B**) Postoperative whole-lower-extremity radiograph showed valgus limb alignment after HTO. (**C**) Arthroscopic inspection and confirmation of ICRS 3 and 4 cartilage defect. (**D**) Cartilage defect preparation for underlying healthy subchondral bone and peripheral margin. (**E**) Multiple drill holes in the cartilage defect site. (**F**) Implantation of the UCB-MSCs-HA composite into drill holes and cartilage defect surface (Cartistem®).

#### **5. Postoperative Rehabilitation**

Venous impulse pumps were prescribed to prevent deep vein thrombosis. Quadricepsstrengthening and straight-leg-raising exercises were performed immediately after surgery. Additionally, a range-of-motion (ROM) exercise was allowed from postoperative day 1 and progressed as tolerated. Partial-weight bearing with crutch ambulation was allowed during 6 weeks, and full-weight bearing was allowed after 12 weeks. The second arthroscopy was performed during plate removal after the union of the osteotomy site.

#### **6. Outcome Measures**

Clinical and radiological evaluation were performed preoperatively at 1, 3, and 6 months; postoperatively at 1 year; and annually thereafter. A 100 mm visual analogue scale (VAS) for pain and Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) [28] were evaluated. Anteroposterior, lateral, and merchant views and

whole-lower-extremity radiographs were obtained for radiological evaluation, including lower-limb alignment and Kellgren–Lawrence grade. The cartilage repair of the medial femoral condyle was assessed visually during second-look arthroscopy using ICRS Macroscopic Assessment of Cartilage Repair [29]. ICRS Macroscopic Assessment of Cartilage Repair consists of three items: degree of defect repair, integration to border zone, and macroscopic appearance, which is graded as the following: normal as grade I, nearly normal as grade II, abnormal as grade III, and severely abnormal as grade IV.

#### **7. Statistical Analysis**

VAS and WOMAC score for pain and Wilcoxon signed-rank test were used. All statistical analyses were executed using IBM SPSS statistics version 23.0 (IBM Corp., Armonk, NY, USA); a *p*-value < 0.05 was considered significant.

#### **8. Results**

Twelve patients (9 women and 3 men) were included in this study. A mean age was 54.3 ± 7.8 years (range, 42–66 years). Seven patients had meniscal problems, which was treated by meniscectomy. The mean follow-up was 2.9 years (range, 1 to 6 years). The mean cartilage defect size of the medial femoral condyle was 4.5 cm2 (range, 4 to 6.9 cm2) (Table 1).


**Table 1.** Demographics and Baseline Characteristics of patients of this study.

Kellgren–Lawrence (K-L) grade II or III, sustaining typical bipolar lesions with varying degrees of severity. \* Pain on the 100 mm visual analogue scale (VAS) ranges from 0 to 100, with higher score indicating worse results. † Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) ranges from 0 to 92, with higher score indicating worse results.

At final follow-up, the VAS pain score was significantly improved from 61.6 to 11.4 and the WOMAC score from 46.6 to 12.3 (*p* < 0.05, Figure 3). The lower-limb alignment was changed from varus 6.7 to valgus 2.2.

At an average of 18 months after a UCB-MSCs-HA composite implantation combined with HTO, 10 patients underwent second-look arthroscopy during plate removal. At second-look arthroscopy, repaired tissue was observed in all cases. The mean total score of ICRS Macroscopic Assessment of Cartilage Repair was 8.4 (range, 5 to 12 points) (Table 2). One case (10%) showed grade I, seven (70%) cases showed grade II, and two (20%) cases showed grade III (Figure 4), which meant that 80% of the repaired cartilage was classified as "normal" or "nearly normal".

**Figure 3.** Clinical outcomes after the UCB-MSCs-HA composite implantation combined with HTO. (**A**) Pain on the 100 mm visual analogue scale (VAS) score; (**B**) Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) score.

**Table 2.** Cartilage repair assessment of second arthroscopic findings.


\* Grade was classified according to the total score.

**Figure 4.** Evaluation of the repaired cartilage according to the ICRS Macroscopic Assessment of Cartilage Repair during second look arthroscopy. (**A**) A case of ICRS cartilage repair assessment score of 12, which was grade I in overall repair assessment, i.e., "normal". (**B**) A case of ICRS cartilage repair assessment score of 9, which was grade II in overall repair assessment, i.e., "nearly normal". (**C**) A case of ICRS cartilage repair assessment score of 5, which was grade II in overall repair assessment, i.e., "abnormal".

#### **9. Discussion**

This study demonstrated that implantation of an allogeneic UCB-MSCs-HA composite combined with HTO provides favorable clinical outcome and cartilage restoration for fullthickness cartilage defects in knee OA with varus deformity. The consistent regenerative response was observed in all cases despite large full-thickness cartilage defects in knee OA, which may suggest that UCB-MSCs-HA implantation is required for better outcomes in patients who planned to undergo HTO for medial compartmental OA with large fullthickness cartilage defects and varus deformity.

The repaired cartilage was observed in all cases during second-look arthroscopy after UCB-MSCs-HA composite implantation with concomitant HTO. A limited number of studies has reported that only HTO without cartilage repair procedure could provide cartilage repair potentially in some cases due to the change of the biomechanical environment [4,5]. However, the quality and quantity of the repaired cartilage were still insufficient. In addition, a randomized controlled trial regarding MSCs implantation combined with HTO reported a higher proportion of cartilage repair compared to HTO only [30]. A recent meta-analysis including four comparative studies reported that intra-articular MSCs administration combined with HTO showed better cartilage repair compared with the HTO alone [31]. In line with previous a meta-analysis, intra-articular UCB-MSCs-HA composite implantation combined with HTO in this study showed cartilage repair in all patients with medial compartmental OA with large full-thickness cartilage defects and varus deformity. Taken together with the results of this study and those of the meta-analysis, intra-articular MSCs administration could enhance cartilage repair in patients who underwent HTO for the treatment of knee OA and varus deformity.

Interestingly, regardless of the status of cartilage repair, pain and function at final follow-up was significantly improved compared to preoperative evaluation. Several studies have demonstrated that HTO could provide satisfactory pain and functional improvements [5,32,33], which may be induced by the decrease of contact pressure on the medial compartment by shifting the load from the medial to lateral compartment [4,5]. A recent meta-analysis study reported that intra-articular MSCs administration combined with HTO may improve clinical outcomes as compared to HTO alone [31]. In line with a previous meta-analysis, intra-articular UCB-MSCs-HA composite implantation combined with HTO in this study showed significant improvement in pain and function in all patients with knee OA with large full-thickness cartilage defects and varus deformity. Taken together, cartilage repair with MSCs administration in patients who underwent HTO could be a viable option for improved clinical outcomes at long-term follow-up.

To date, there has been no reliable cartilage repair procedure for favorable outcomes in osteoarthritic cartilage defects [34]. Microfracture, the most common of a small cartilage defects, generally leads to fibrous repair tissue with unsatisfactory durability [8,9]. Autologous chondrocyte implantation (ACI) is usually recommended for younger patients with large focal chondral defects [35]. Both procedures are generally limited to restoring cartilage in large defects of older patients, with outcomes tending to deteriorate over time [10]. Some recent studies have demonstrated that surgical implantation of UCB-MSCs-HA composite could result in reliable cartilage repair in osteoarthritic cartilage defects. In this regard, surgical implantation of the UCB-MSCs-HA composite was selected for the cartilage repair procedure in this study. In accordance with previous studies with UCB-MSCs-HA composite, cartilage repair was observed in all cases despite full-thickness large cartilage defects more than 4 cm2. Therefore, surgical implantation of UCB-MSCs-HA composite could be a reliable option for cartilage repair in osteoarthritic knees.

Some limitations of this study need to be addressed. First, this study was retrospective and did not include a control group. However, clinical outcome including cartilage repair after allogeneic UCB-MSCs implantation combined with HTO has rarely been reported [25–27]. Second, a small number of patients were included in this study. However, the regenerated cartilage was evaluated via second-look arthroscopy. In addition, only patients with a large full-thickness osteoarthritic cartilage defects were included in this study. Finally, magnetic resonance or biopsy for histological evaluation was not performed in this study, which would be a more reliable means for determining the properties of repaired cartilage. However, a direct visual evaluation of the repaired cartilage is one of the most reliable validated assessment tools for cartilage repair.

In conclusion, this study showed that allogeneic UCB-MSCs-HA composite implantation combined with HTO resulted in favorable clinical outcome and cartilage repair in all cases. These findings suggest that UCB-MSCs-HA composite implantation combined with HTO would be a good therapeutic option for patients with knee OA and full-thickness large cartilage defects.

**Author Contributions:** Y.-B.P., study concepts/design and manuscript drafting/revision; H.-J.L., data acquisition/analysis and manuscript revision; H.-C.N., data acquisition/analysis and manuscript revision; J.-G.P. and H.-C.N., data acquisition/analysis and manuscript revision. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Institutional Review Board Statement:** This study was approved by the institutional ethical review board.

**Informed Consent Statement:** Patient consent was waived because this study is a retrospective record review.

**Data Availability Statement:** Not applicable.

**Acknowledgments:** Authors would like to acknowledge support from Yoo-Sun Won at our institution for obtaining clinical data acquisition.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


**Disclaimer/Publisher's Note:** The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

### *Article* **Relationship between Outerbridge Scale and Chondropathy Femorotibial Joint in Relation to Gender and Age—The Use of 1.5T and 3.0T MRI Scanners**

**Dominik Siero ´n 1,\*, Izabella Jabło ´nska 2, Paweł Niemiec 3, Dawid Lukoszek 4, Karol Szyluk 5,6, Ivan Platzek 7, Hugo Meusburger 1, Georgios Delimpasis <sup>1</sup> and Andreas Christe <sup>1</sup>**


**Abstract:** *Background and Objective*: Magnetic resonance imaging (MRI) enables the effective evaluation of chondromalacia of the knee joint. Cartilage disease is affected by many factors, including gender, age, and body mass index (BMI). The aim of this study was to check the relationship between the severity of chondromalacia of the femoro-tibial joint and age, gender, and BMI assessed with 1.5T and 3.0T MRI scanners. *Materials and Methods*: The cross-observational study included 324 patients—159 (49%) females and 165 (51%) males aged 8–87 (45.1 ± 20.9). The BMI of study group was between 14.3 and 47.3 (27.7 ± 5.02). 1.5T and 3.0T MRI scanners were used in the study. The articular cartilage of the knee joint was assessed using the Outerbridge scale. *Results*: The age of the patients showed a significant correlation with Outerbrige for each compartment of the femorotibial joint (Spearman's rank correlation rho: 0.69–0.74, *p* < 0.0001). A higher correlation between BMI and Outerbridge was noted in the femur medial (rho = 0.45, *p* < 0.001) and the tibia medial (rho = 0.43, *p* < 0.001) than in the femur lateral (rho = 0.29, *p* < 0.001) and the tibia lateral compartment (rho = 0.34, *p* < 0.001). *Conclusions*: The severity of chondromalacia significantly depends on age and BMI level, regardless of gender.

**Keywords:** femorotibial joint; chondromalacia; aging; body mass index; magnetic resonance imaging

### **1. Introduction**

Chondromalacia is a disease affecting the hyaline cartilage that covers the articular surfaces of bones. It causes the cartilage to soften and often leads to tearing and erosion of the cartilage. The environment and physical stress both have an effect on this cartilage. Degeneration of the cartilage also occurs in response to microtraumatic wear. Repeated activities that cause compressive stress on the joint or increased loads on the joint can lead to chondromalacia [1–3]. Aging also affects the hyaline cartilage. As we age the number of chondrocytes in cartilage decreases, which correlates with a decrease in the number of proteoglycans produced. This reduction leads to a decrease in the water content of cartilage. A loss of elastic properties develops in the cartilage due to the gradual loss of collagen fibril cross-linking, which also occurs with age. With age, the superficial zone is the first to be damaged [4].

**Citation:** Siero ´n, D.; Jabło ´nska, I.; Niemiec, P.; Lukoszek, D.; Szyluk, K.; Platzek, I.; Meusburger, H.; Delimpasis, G.; Christe, A. Relationship between Outerbridge Scale and Chondropathy Femorotibial Joint in Relation to Gender and Age—The Use of 1.5T and 3.0T MRI Scanners. *Medicina* **2022**, *58*, 1634. https://doi.org/10.3390/ medicina58111634

Academic Editors: Yong In and In Jun Koh

Received: 29 September 2022 Accepted: 9 November 2022 Published: 12 November 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

The Outerbridge scale is most commonly used to assess chondromalacia. This classification allows us to categorize cartilage degeneration into four degrees of advancement. It is common to find several degenerative processes at varying levels of severity in the same knee joint [5,6]. MRI is an effective and non-invasive method for evaluating the articular cartilage of the knee joint. This method can be used to detect and monitor degenerative changes that may lead to osteoarthritis [7]. Typical MRI scans (PD and fat-suppressed T2-weighted) can assess the characteristics of cartilage pathology [8]. Together with existing grading scales that assess articular cartilage, MRI can be considered a highly accurate and non-invasive cartilage diagnostic tool [9], with an accuracy of up to 91% [10].

Obesity has a direct correlation with the degeneration of the osteo-articular system. Being overweight causes increased pressure on the articular surfaces of small and large joints, which results in faster wear of the articular surfaces, including cartilage, and its accelerated degeneration [11].

Osteoarthritis (OA) is the most common form of arthritis, affecting approximately 90 million adults (36.8% of the adult population) in the United States alone [12] and hundreds of millions of people worldwide. The disease primarily affects the articular hyaline cartilage in stressed joints, such as the knee joint. Other tissues such as the synovial membrane of the joint capsule and subchondral bone are also affected and contribute to disease progression. As the disease advances, severe cartilage degeneration, joint space narrowing, subchondral bone thickening, osteophyte [13] or bone spur formation, and joint inflammation with associated swelling and pain occur [14]. Increased risk factors for OA are numerous and include obesity, being of the female gender, age, congenital structural defects of the joint, and acute joint trauma [15]. Scientific evidence from systematic reviews show that the progression of knee degenerative changes increases with age and excess weight. In addition, studies indicate that BMI is a significant indicator of the degeneration of articular cartilage in individual compartments of the same knee [16–18]. Furthermore, we were interested in the normal range of chondromalacia depending on age, gender and BMI using the Outerbridge classification as quantification of cartilage degeneration. Age and BMI could be used as a direct replacement for the Outerbridge score without the need to perform an MRI.

Due to the large amount of research on patellofemoral chondromalacia, we wanted to improve our understanding of femorotibial cartilage degeneration and check its characteristics. The aim of the study was to investigate the correlation between the severity of chondromalacia of the femorotibial joints and BMI, broken down into categories based on age and gender. Additionally, the differences between using a 1.5T and a 3.0T MRI scanner in the assessment of the knee cartilage were checked.

#### **2. Materials and Methods**

#### *2.1. Study Design*

The IRB (Institutional Review Board) states that the presented retrospective studies containing irreversibly anonymized data in our institution do not require the approval of the bioethics committee. In the current observational cross-sectional study, we analyzed the effect of BMI and demographic variables (including age and gender of patients) on the severity of femorotibial joint chondromalacia in consecutive incoming patients undergoing evaluations of knee joint lesions from 2018 to 2019. Patients were recruited from community and clinical hospitals as well as private facilities in Zamo´s´c Elblag, Jelenia Góra, and Bielsko Biala (Poland). The medial (femur and tibia separately) and lateral (femur and tibia separately) compartments of the knee joint were evaluated.

The following work is the most recent in a series of scientific publications on chondromalacia of the knee joint. Earlier work included radiological measurements to assess chondromalacia of the knee joint [19]. A paper on chondromalacia of the patellofemoral joint in correlation with BMI depending on age and gender is currently being published. All work was performed based on the same study group and their respective MRI scans.

#### *2.2. Description of the Participants*

The study group included 324 patients, 159 (47.1%) women and 165 (52.9%) men. A total of 155 (47.8%) patients, including 70 (45.2%) women and 85 (54.8%) men, were examined on the 1.5T scanner, while 169 (52.2%) patients, including 89 (52.7%) women and 80 (47.3%) men, were examined on the 3.0T scanner. Four age classes were defined for the study: <30 years (94 participants); 30–45 years (61 participants); 46–60 years (78 participants); >60 years (91 participants).

#### *2.3. Inclusion and Exclusion Criteria*

Inclusion criteria: patients with pain or suspected osteoarthritis or post-traumatic lesions who on the order of an orthopedist, surgeon, or physiotherapist submitted for an MRI. Several patients also submitted for a private scan at their own request due to complaints of joint pain.

Exclusion criteria: previous knee surgery or chronic post-traumatic changes of the knee.

#### *2.4. Evaluation of Cartilage Chondromalacia*

To assess cartilage chondromalacia, we used the 5-level Outerbridge classification (Table 1) using fat-saturated proton density sequences—a modified classification for arthroscopic cartilage assessment [16–19]. Three board-certified radiologists with 12, 25 and 32 years of experience in musculoskeletal imaging classified the MR images according to the Outerbridge score. Cases were randomly and equally distributed among the radiologists. The MRI scans reported by the radiologists did not overlap. For this reason, consistency between radiologists was not assessed.

**Table 1.** Outerbridge classification [19–23].


#### *2.5. Image Acquisition*

The study was evaluated on an iMac pro (Apple, Cupertino, CA, USA) using the FDA approved OsiriX MD software (version 11.0, Pixmeo SARL, Bernex, Switzerland). MRIs were performed on 3.0T scanners (Ingenia 3.0T, Philips, Amsterdam, The Netherlands) and on 1.5T GE scanners (SIGNA, GE, Milwaukee, WI, USA) GE at different facilities located in clinical hospitals and private facilities in Zamo´s´c, Elbl ˛ag, Jelenia Góra, and Bielsko-Biala. All MRI studies were irreversibly anonymized [16].

The following diagnostic sequence protocol was used in the study: axial, sagittal and coronal PD FS, sagittal and coronal T1 (all with a slice thickness of 3 mm), 3D highresolution PD FS with a slice thickness from 0.8 to 1 mm. The same protocol we used in the earlier publication [19].

#### *2.6. Statistical Analysis*

A Chi-square (χ2) test of independence was used to analyze the difference of chondromalacia in the various age, sex and MRI-unit groups. The correlation between the age subgroups and Outerbridge scale chondromalacia lesion scores, the C contingency co-efficient, was calculated, and the R-Spearman rank correlation was checked. The comparison of correlation coefficients was performed using the z statistic.

Subsequently, the correlation between BMI level and Outerbridge chondromalacia lesion scores was tested using the Kruskal–Wallis test, ANOVA, post-hoc analysis (Conover test), the trend for mean values (Me) of BMI level in each Outerbridge scale grade, the Jockheere–Terpstra trend test, and the R-Spearman (rho). The analysis was performed for the whole group, 1.5T and 3.0T, for females and males—separately for 1.5T and 3.0T—and a pooled analysis of Outerbridge score 0/1/2 vs. 3/4 was performed. The continuity coefficient was used as a measure of the relationship between the age subgroup and Outerbridge scale grade. The use of the χ<sup>2</sup> test provided the basis for the calculation of the C coefficient. Our table had dimensions of 4 × 4; in this case Cmax = 0.866. To compare the impact of the different variables (age, BMI, sex and MRI scanner type) on the Outerbridge score, a logistic regression model was used with the dependent dichotomous variable being no or slight degeneration (Outerbridge score 0/1/2) and severe degeneration (Outerbridge score 3/4).

#### **3. Results**

*3.1. Age*

Age demonstrated significant correlation coefficients (Spearman rank correlation, *p* < 0.001) with Outerbridge for the femur lateral (rho = 0.69), tibia lateral (rho = 0.71), femur medial (0.72), tibia medial (rho = 0.74) (Table 2, Figure 1).

**Table 2.** Relationship between the Outerbridge scale and age for particular compartments of the femorotibial joint.


**Figure 1.** Relationship between the Outerbridge scale and age for femur lateral (**A**), femur medial (**B**), tibia lateral (**C**), and tibia medial (**D**).

#### *3.2. BMI*

The mean BMI for the entire study group was 27.7; SD: 5.02. The highest values of BMI were noted in the patients diagnosed with Outerbridge Scale 2 for the tibia lateral and Outerbridge Scale 4 for the femur medial and tibia medial. For the individual parameters of the Outerbridge scale, the mean BMI is presented in Table 3.



#### 3.2.1. Femur Lateral

A statistically significant positive correlation was found between BMI level and Outerbridge parameter score (rho = 0.29, *p* < 0.001) (Figure 2A).

**Figure 2.** Relationship between BMI and Outerbridge scores using both scanners (1.5T and 3.0T) in the entire study group for femur lateral (**A**), femur medial (**B**), tibia lateral (**C**), and tibia medial (**D**).

There was a statistically significant difference in BMI level according to the Outerbridge parameter score (KW, *p* < 0.001) in the entire study group of patients (scanner 1.5T and 3.0T).

For the femur lateral, there were no significant statistical differences between the Outerbridge subgroups (1 and 2 vs. 3 and 4) in terms of the BMI level (U-test, *p* = 0.39).

The grade 0 Outerbridge parameter score showed a statistically significant lower mean BMI level compared to grades 1, 2 and 3 (*p* < 0.05).

#### 3.2.2. Femur Medial

There was a statistically significant positive correlation between the BMI level and Outerbridge parameter score (rho = 0.45, *p* < 0.001) for both scanners (Figure 2B).

There was a statistically significant difference in BMI level between the Outerbridge scores (*p* < 0.001) in the entire study group.

For the femur medial, there were significant statistical differences between the Outerbridge subgroups (1 and 2 vs. 3 and 4) in terms of the BMI level (*p* = 0.01).

The grade 0 Outerbridge parameter score showed a statistically significant lower mean BMI level compared to grades 2, 3 and 4 (*p* < 0.05).

#### 3.2.3. Tibia Lateral

A significant correlation was found between the BMI level and Outerbridge parameter score (rho = 0.34, *p* < 0.001) for both scanners (Figure 2C).

The parameters' Outerbridge scores significantly differed statistically in terms of the BMI (*p* = 0.0001) for both scanners.

For the tibia lateral, no significant differences were found between the Outerbridge subgroups (1 and 2 vs. 3 and 4) in terms of the BMI level (*p* = 0.72)

The grade 0 Outerbridge parameter score showed a statistically significant lower mean BMI level compared to the level of the rest of the parameters (*p* < 0.05).

#### 3.2.4. Tibia Medial

There was a statistically significant relationship between the BMI level and Outerbridge parameter score (rho = 0.42, *p* < 0.001) at this location (Figure 2D).

The Outerbridge score parameters were significantly statistically different in terms of the BMI level (*p* < 0.001) at this location.

The Outerbridge subgroups (1 and 2 vs. 3 and 4) differed significantly in terms of the BMI level for the tibia medial (*p* = 0.017).

The grade 0 Outerbridge parameter score showed a statistically significant lower mean BMI level compared to the level of the rest of the parameters (*p* < 0.05).

#### *3.3. Gender*

There were no significant differences between women and men in the Outerbridge assessment for each knee joint compartment in each age subgroup (test χ2, *p* > 0.05).

#### 3.3.1. Women

For women, age demonstrated the following correlation coefficients (Spearman rank correlation, *p* < 0.001) with Outerbridge: for the femur lateral (rho = 0.72), the tibia lateral (rho = 0.71), the femur medial (rho = 0.70), and the tibia medial (rho = 0.70).

#### 3.3.2. Men

For men, age demonstrated the following correlation coefficients (Spearman rank correlation, *p* < 0.001) with Outerbridge: for the femur lateral (rho = 0.66), the tibia lateral (rho = 0.71), the femur medial (rho = 0.74), and the tibia medial (rho = 0.77).

#### *3.4. Type of Scanners (1.5T vs. 3.0T)*

The demographics of the two groups were as follows: A total of 70 (45.2%) females and 85 (54.8%) males were examined on the 1.5T scanner, while 89 (52.7%) females and 80 (47.3%) males were examined on the 3.0T scanner. The average age of the 1.5T and 3.0T group was 43.8 ± 18.7 and 46.9 ± 23.5 years, while the mean BMI of the 1.5T and 3.0T group added up to 28.2 ± 5.0 and 27.1 ± 5.2.

Significant differences were found between the scanner type (1.5T vs. 3.0T) in the Outerbridge assessment for the femur lateral (χ2, *p* = 0.002), tibia lateral (χ2, *p* = 0.006), femur medial (χ2, *p* = 0.034), and tibia medial (χ2, *p* = 0.007) in the entire study group:

For the 1.5T scanner, the Outerbridge scale scores were in general higher than for the 3.0T scanner (Table 4).

**Table 4.** Outerbridge Scale for each compartment of femorotibial joint.


When all joint compartments were evaluated together, the correlation coefficient of the Outerbridge and BMI at 1.5T/3.0T was 0.3997/0.3303 (*p* = 0.1314) and the correlation of the Outerbridge and age at 1.5T/3.0T was 0.687/0.777 (*p* < 0.001).

#### *3.5. Logistic Regression of Age, BMI, Sex and MRI Scanner Type on Outerbridge Score*

Age, BMI and MRI scanner type played a significant role in predicting the severity of knee degeneration (Outerbridge 3 and 4) with the following logistic regression formula:

Logit (Outerbridge 3/4) = 0.0888 × A + 0.0864 × B + 0.0604 × M − 0.717 × T − 6.58 (1)

A (age), B (BMI), M (male patient) and T (1.5 Tesla MRI) demonstrated odd ratios of 1.093 (95% CI: 1.073 to 1.113), 1.090 (95% CI: 1.026 to 1.159), 1.062 (95% CI: 0.578 to 1.953) and 0.488 (95% CI: 0.265 to 0.901), respectively. According to the *p*-value, the influence was strongest for the age (*p* < 0.001), followed by BMI (*p* = 0.005) and MRI type (*p* = 0.022); sex did not have a significant impact on degeneration prediction (*p* = 0.85).

#### **4. Discussion**

The results showed a significant correlation between the Outerbridge chondromalacia score and BMI and age. A similar correlation was noted in a study by Matada, who additionally noted that a significant increase in larger knee cartilage lesions occurs in individuals > 50 years and BMI > 25 [24]. Chondromalacia changes are noted when assessing structural changes in osteoarthritis [21]. Risk factors for symptomatic knee joint changes due to osteoarthritis include obesity and age [25]. The relationship between BMI and knee OA is mainly linear [26].

#### *4.1. Gender*

Previous studies have noted differences in the extent of cartilage loss of the femorotibial joint between men and women, with greater degeneration observed in women than in men [27–29]. Despite these reports lacking clarity, the hypothesized mechanism for the significant progression of knee cartilage disorders involves biomechanical and hormonal factors that distinguish women from men [28]. In women, the progression of degenerative changes of the knee joint significantly increases after the age of 50 in the postmenopausal period [27]. A cross-sectional study found that women taking hormone therapy in the postmenopausal period exhibit greater cartilage volume [30]. Possible biomechanical differences in gait and the knee joint in women may also accelerate the development of OA [31,32]. Our study showed no significant difference in the Outerbridge scores between women and men. Additionally, the correlation of Outerbridge and age/BMI was the same between both genders.

#### *4.2. Aging*

The development of knee osteoarthritis is closely related to aging [33,34]. The catabolic– anabolic imbalance of cartilage causes matrix destruction through excessive oxidation of antioxidant systems in chondrocytes, including glutathione and peroxiredoxins [35]. Joint components that undergo changes due to aging also contribute to the degeneration of hyaline cartilage. Osteoporosis or the weakening of the quadriceps muscle of the thigh leads to the dysfunction of the femorotibial joint, increasing the maximum stresses on the cartilage [36–38]. In the analysis of this study, a significant relationship was found between the presence of greater chondromolytic changes and increasing age for this group.

#### *4.3. BMI*

For the entire group, a positive correlation was found between BMI and Outerbridge scores. A higher correlation between BMI and Outerbridge was found in the medial compartments, femur medial (rho = 0.45) and tibia medial (rho = 0.42), than the lateral compartments, femur lateral (rho = 0.29) and tibia lateral (rho = 0.34). It should be noted that obesity is not only associated with osteoarthritic changes of the knee joint in the mechanical but also in the metabolic background [39–41]. Through research, adipokines, leptin, and resistin, which have endocrine functions in adipose tissue, have been identified [38]. In vivo, findings have indicated that there is a detrimental effect on chondrocyte proliferation as well as the initiation of extracellular matrix metalloproteinase expression, resulting in reduced cartilage volume [42].

#### *4.4. MRI and the Assessment of Cartilage Chondromalacia*

A comparison of knee joint images from the same patients on the 1.5T and 3.0T apparatus in Wang's study, followed by arthroscopy, concluded that the visualization of anatomical structures and the confidence in making diagnoses of cartilage lesions are both improved when using 3.0T scanners. The reliability (88.2% vs. 86.4%) and sensitivity (51.3% vs. 42.9%) of the 3.0T device was improved compared with the 1.5T. The correct assessment of cartilage damage differed in favor of the 3.0T apparatus (51.3 vs. 42.9%) [43]. A study by Mandell, who used a similar methodology to Wang in examining a larger group, inferred no significant differences between the 1.5T and 3.0T [44]. In contrast, a systematic review and meta-analysis concluded that the 3.0T scanners were significantly more reliable for imaging articular cartilage [10] with age-related degeneration. In our study, the 1.5T scanner scored higher Outerbridge levels in all compartments compared to the 3.0T unit, but the correlation coefficients of Outerbridge and BMI did not differ, and the correlation coefficient of Outerbridge and age was even significantly higher than with the 3.0T unit. Therefore, one explanation for the higher Outerbridge scores of the1.5T scanner could be a possible geographical inclusion inhomogeneity. From our results, it seems that both the 1.5T and the 3.0T apparatuses are effective methods of evaluating Outerbridge cartilage according to the BMI level. On the other hand, the 3.0T apparatus may have an advantage in terms of age correlation.

#### *4.5. Limitations*

One major drawback of our study is the lack of a comparison of the same group of subjects using the 1.5T and 3.0T apparatus. The lack of a characterization of the group in terms of additional injuries such as ACL, meniscus, fat pad, and other factors such as physical activity or quadriceps thigh muscle strength made it impossible to test for differences in Outerbridge scores. The most important limitation of the study is the BMI used, which shows low sensitivity in obesity research and non-specificity in chondromalacia studies [45].

#### **5. Conclusions**

Assessing the degree of cartilage degenerative changes with 1.5T and 3.0T MRI is an effective form of lesion classification using the Outerbridge scale.

A positive correlation was observed between the degree of chondromalacia in both compartments of the femorotibial joint and the body mass index and age of the subject.

No differences were noted between men and women in the assessment of articular cartilage.

Our suggestion for further research is a longitudinal follow-up to evaluate the effect of the duration of obesity during life on cartilage changes and to see which apparatus, the 1.5T or 3.0T, is more effective in monitoring patients with OA. In order to provide a complete answer to the question of whether age and BMI can be used as a variable in the Outerbridge scale without the use of MRI, a predictive model study with multivariable analysis should be performed, taking into account many factors causing a predisposition to chondropathy of the femorotibial joint.

**Author Contributions:** Conceptualization, D.S.; I.J.; methodology, D.S.; A.C.; software, D.L.; K.S.; P.N.; I.P.; H.M.; G.D.; validation, D.S.; A.C.; P.N.; I.P.; H.M.; G.D.; formal analysis, D.L.; K.S.; D.S.; A.C.; I.P.; P.N.; H.M.; G.D.; investigation, D.S.; I.J.; I.P.; H.M.; G.D.; resources, D.S.; data curation, D.S.; I.J.; writing—original draft preparation, D.S.; I.J.; writing—review and editing, D.L., K.S.; I.J.; A.C.; P.N.; I.P.; H.M.; G.D.; visualization, D.L.; K.S.; supervision, A.C.; K.S. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Institutional Review Board Statement:** The IRB approval could be waived due to the retrospective and irreversibly anonymized patent data.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** Data are available upon special request.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


### *Article* **Knee Diameter and Cross-Sectional Area as Biomarkers for Cartilage Knee Degeneration on Magnetic Resonance Images**

**Elias Primetis 1,†, Dionysios Drakopoulos 1,†, Dominik Sieron 1, Hugo Meusburger 1, Karol Szyluk 2,3, Paweł Niemiec 4, Verena C. Obmann 5, Alan A. Peters 5, Adrian T. Huber 5, Lukas Ebner 5,†, Georgios Delimpasis 1,† and Andreas Christe 1,5,\*,†**

	- <sup>4</sup> Department of Biochemistry and Medical Genetics, Faculty of Health Sciences in Katowice, Medical University of Silesia in Katowice, 40-752 Katowice, Poland
	- <sup>5</sup> Department of Diagnostic, Interventional and Pediatric Radiology, Inselspital, Bern University Hospital, University of Bern, 3010 Bern, Switzerland
	- **\*** Correspondence: andreas.christe@insel.ch
	- † These authors contributed equally to this work.

**Abstract:** *Background and Objectives*: Osteoarthritis (OA) of the knee is a degenerative disorder characterized by damage to the joint cartilage, pain, swelling, and walking disability. The purpose of this study was to assess whether demographic and radiologic parameters (knee diameters and knee crosssectional area from magnetic resonance (MR) images) could be used as surrogate biomarkers for the prediction of OA. *Materials and Methods*: The knee diameters and cross-sectional areas of 481 patients were measured on knee MR images, and the corresponding demographic parameters were extracted from the patients' clinical records. The images were graded based on the modified Outerbridge arthroscopic classification that was used as ground truth. Receiver-operating characteristic (ROC) analysis was performed on the collected data. *Results*: ROC analysis established that age was the most accurate predictor of severe knee cartilage degeneration (corresponding to Outerbridge grades 3 and 4) with an area under the curve (AUC) of the specificity–sensitivity plot of 0.865 ± 0.02. An age over 41 years was associated with a sensitivity and specificity for severe degeneration of 82.8% (CI: 77.5–87.3%), and 76.4% (CI: 70.4–81.6%), respectively. The second-best degeneration predictor was the normalized knee cross-sectional area, with an AUC of 0.767 ± 0.04), followed by BMI (AUC = 0.739 ± 0.02), and normalized knee maximal diameter (AUC = 0.724 ± 0.05), meaning that knee degeneration increases with increasing knee diameter. *Conclusions*: Age is the best predictor of knee damage progression in OA and can be used as surrogate marker for knee degeneration. Knee diameters and cross-sectional area also correlate with the extent of cartilage lesions. Though less-accurate predictors of damage progression than age, they have predictive value and are therefore easily available surrogate markers of OA that can be used also by general practitioners and orthopedic surgeons.

**Keywords:** Outerbridge; chondromalacia; aging; body mass index; degeneration; magnetic resonance imaging

#### **1. Introduction**

Osteoarthritis (OA) of the knee is a very common pathology affecting the whole joint but in particular the cartilage [1]. It is classified as primary if its origin is unknown, or secondary if it is subsequent to a specific condition or event like trauma, repetitive micro stress, surgery, or malalignment [2]. OA is a degenerative disease that usually becomes manifest in the elderly [1]. The medial compartment is the most affected, though also

**Citation:** Primetis, E.; Drakopoulos, D.; Sieron, D.; Meusburger, H.; Szyluk, K.; Niemiec, P.; Obmann, V.C.; Peters, A.A.; Huber, A.T.; Ebner, L.; et al. Knee Diameter and Cross-Sectional Area as Biomarkers for Cartilage Knee Degeneration on Magnetic Resonance Images. *Medicina* **2023**, *59*, 27. https:// doi.org/10.3390/medicina59010027

Academic Editors: Yong In and In Jun Koh

Received: 24 November 2022 Revised: 16 December 2022 Accepted: 20 December 2022 Published: 23 December 2022

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

the lateral and the patellofemoral compartments are also affected. The cartilage changes characteristic of this condition are accompanied by marginal osteophytes, subchondral bone cyst formation (geodes) and sclerosis, buttressing, and soft tissue changes, namely ganglion cyst formation, and periarticular soft tissue edema. Reactive synovium thickening and intra-articular fluid can also be observed [3,4]. The symptoms, including pain, stiffness and walking disability can increase to the point to deteriorate life quality, leading in some cases to depression [5]. The pathology is prevalent in the obese female population, and correlates with knee diameter [6–9]. Disease frequency and severity increase with increasing age [10,11].

Obesity, previous knee trauma, biomechanical factors, female gender and older age are common known risk factors for the development of osteoarthritis [12–14]. As no drugs are available to treat this condition, the focus is on prevention and management of joint damage progression through interventions on the modifiable risk factors, like the diet and physical exercise [13].

MR imaging (MRI) with 1.5 and 3 Tesla systems has proven to be a suitable technology for both the quantitative and qualitative assessment of joint cartilages [15,16], especially of the knee, and is clearly superior to plain radiography for the evaluation of the hyaline cartilage, as the last methodology provides only indirect signs of chondromalacia, actually not depicting the cartilage itself [17–20]. The Outerbridge scoring system is widely used to evaluate the extent of cartilage degeneration. It is a 5-grade scale: grade 0—normal articular cartilage, grade 1—softening of the cartilage due to biochemical modifications of the cartilage composition, grade 2—extent of cartilage loss <50%, grade 3—extent of cartilage loss >50%, and grade 4—complete cartilage loss accompanied by subchondral bone changes [21].

In spite of the high accuracy of MRI techniques for the assessment of the disease extent, this methodology is expensive and its use is restricted to equipped radiologic facilities and experienced medical staff. Therefore, more easily available tools to detect the disease and monitor its progression are needed.

Starting from the evidence that obese individuals with large knee cross sections often present with extensive cartilage degeneration (chondromalacia) [6], we hypothesized that both knee diameter and cross-sectional area might be employed as easily accessible surrogate biomarkers of chondromalacia. Similarly, given the high prevalence of OA in the obese female population as well as in the elderly [1,7], further parameters like age, gender, and BMI might also be suitable predictors of OA.

#### **2. Materials and Methods**

#### *2.1. Ethics*

The responsible ethics commission waived the requirement to obtain informed consent from the patients included in this study due to the study's retrospective nature and the irreversible anonymization of patients' identifying information.

#### *2.2. Study Design*

Among patients undergoing knee MRI between 2018 and 2019, 481 patients were retrospectively selected from the archives of community and clinical hospitals as well as private clinics in Zamo´s´c Elblag, Jelenia Góra and Bielsko-Biala (Poland) based on the inclusion and exclusion criteria listed below. The study was performed according to the STROBE guidelines. A number of 120–121 patients for each of four age groups were selected: <30 years (120); 30–45 years (120); 46–60 years (120); >60 years (121). Demographic data were retrieved from the MRI safety questionnaires of the corresponding clinics (age, sex, height, weight and BMI). The MR images used for this study had been acquired on either 1.5 T (SIGNA, GE, Milwaukee, WI, USA) or 3 T (Ingenia 3.0 T, Philips, Amsterdam, Netherlands) scanners using the following diagnostic sequence protocols: axial, sagittal and coronal PD FS, sagittal and coronal T1 (all with a slice thickness of 3 mm), 3D high-resolution PD FS with a slice thickness from 0.8 to 1 mm. All MRI data were irreversibly anonymized, and

evaluated using iMac pro (Apple, Cupertino, CA, USA) with FDA-approved OsiriX MD software (version 11.0, Pixmeo SARL, Bernex, Switzerland). The radiological evaluation of cartilage chondromalacia included the medial, lateral and retropatellar compartments of the knee joint.

#### *2.3. Inclusion and Exclusion Criteria*

All patients undergoing an MRI of the knee in the period from 2018 to 2019 at the above-mentioned radiological clinics were eligible for this study. MRI referral was mostly due to knee pain, suspicion of arthrosis or post-traumatic lesions.

Patients with previous surgery, chronic post-traumatic changes, tumorous lesions in the knee joint, and examinations lacking the PD sequences for any reason (abortion of the MRI scan) were not included in the study.

#### *2.4. Image Analysis*

Four radiologists with radiology board examination and 10, 12, 17 and 25 years of experience in musculoskeletal radiology analyzed the images. Each radiologist assessed 30 patients from each age group. No double reading was performed, thus inter- and intrareader agreement could not be assessed. To evaluate cartilage chondromalacia, the modified Outerbridge classification (Table 1) for arthroscopic cartilage evaluation was used, and grading scores were attributed based on fat-saturated proton density sequences [22,23]. Besides cartilage assessment using Outerbridge scoring system, the Insall-Salvati index, cross-sectional area of the knee at the level of the upper pole of the patella as well as diameters at the same level were calculated [24].

**Table 1.** Modified Outerbridge classification.


The following parameters of the knee joint were measured for the subsequent analysis:


In addition, some individuals tend to use the right more than the left side, or vice versa, and/or may be or may have been active in a sport discipline involving overuse of one side and thus are suffering from repetitive microtrauma. This is potentially related with a differential Outerbridge grading for the two body sides, and therefore the analysis did include specification of the investigated side.

Furthermore, the MRI resolution of images acquired with a 3 T unit is higher than that of images from examinations performed at a 1.5 T unit, suggesting the possibility of earlier

detection of (severe) degeneration related with the use of the more powerful 3 T unit [19]. Therefore, the used MRI strength (1.5 T vs. 3 T) Tesla was also included in the evaluation.

**Figure 1.** Normal Insall-Salvati index. For the knee in the figure, an Insall Salvati index of 1.18 was obtained by dividing the length of the patellar ligament (51.8 mm) by the patellar pole distance (43.9 mm).

**Figure 2.** Knee diameters were measured on axial PD-weighted images at the exact level of the patella upper pole, indicated by the green line on the right image. Radiologists measured strictly antero-posterior = vertical and medio-lateral = horizontal (red arrows). The maximal diameter of the knee was not measured (green line on left image).

**Figure 3.** The largest axial femur diameter was measured at the level of the condyles as indicated in the figure.

#### *2.5. Statistical Analysis*

The Outerbridge scores were assessed in the lateral, medial and retropatellar compartment. The maximal score of all three compartments was pooled into zero or slight degeneration (Outerbridge 0–2) and severe degeneration (Outerbridge 3 and 4) and was defined as the classifying variable (outcome). A logistic regression was performed with the binary variables (gender (f, m), examination side (right, left), MRI strength (1.5 or 3 T)) to retrieve regression coefficients and odds ratios. Measurements were normalized for 122 patients: the maximal knee diameter and cross-sectional area were normalized by dividing their value by the max femoral diameter to level out the fact that larger patients have larger knees without necessarily suffering from knee degeneration. The non-binary variables (age, height, weight, BMI, Insall-Salvati index, knee cross-sectional area, vertical, horizontal and maximal diameter as well as normalized maximal diameter and cross-sectional area) were analyzed using receiver operating characteristic curves (ROC). The respective areas under the curve (AUC) were used as comparators to establish the most accurate parameter for the prediction of severe cartilage degeneration. AUC comparison was performed based on pairwise comparison of ROC curves. MedCalc® (Version 19.3, Medcalc Software Ltd, Ostend, Belgium) was used for statistical analysis. The significance level was set to *p* < 0.05.

Sample size calculation. The null hypothesis is that an AUC of 0.5 corresponds to random detection probability. The significance level and power of the test were set to 0.05 and 0.8, respectively. The estimated ratio between slight and severe degeneration was 3:1. To reach an AUC of 0.6 and higher (at least 0.1 higher than chance) a sample size of 352 patients was needed.

#### **3. Results**

#### *3.1. Demographics and Average Measurements*

Gender (f:m) and examination side (right:left) were equally distributed (Table 2). The average age and BMI of the study patients was 45.3 ± 22 years and 27.2 ± 5.4 kg/m2, respectively. The retropatellar compartment presented the highest level of cartilage degeneration with an average Outerbridge score of 1.9 ± 1.5, followed by the medial and lateral compartment with scores of 1.7 ± 1.5 and 1.5 ± 1.4, respectively. The average maximal knee diameter and cross-sectional area of the knee were 13.8 ± 1.4 cm and 131.6 ± 30.2 cm2, respectively. The examined parameters and corresponding results are summarized in Table 3.

**Table 2.** Binary variables distribution.


**Table 3.** Summary of non-binary variable statistics.


N: number of cases; SD: Standard Deviation; P: Percentile.

#### *3.2. Logistic Regression for Binary Variables*

Gender, examined side and MRI scanner strength did not correlate with the evaluated outcome (severe degeneration = Outerbridge grades 3 and 4), proving therefore useless in predicting degeneration, with the following logistic regression formula:

Logit (severe degeneration) = −0.264\*M+ 0.068 \* R − 0.18 \* T + 0.25

M: male patient = 1, female patient = 0; R: right knee = 1, left knee = 0; T: 1.5 T scanner = 1, 3 T scanner = 0. These variables demonstrated non-significant odd ratios for degeneration around 1 (Table 4).

**Table 4.** Logistic regression of dichotomous variables for severe knee degeneration (Outerbridge > 2).


CI: Confidence Interval, m: male; MRI: Magnetic Resonance Imaging; T: Tesla.

*3.3. Receiver Operating Characteristic Curves (ROC) for Non-Binary Variables*

The parameter that allowed for the most accurate prediction of severe degeneration (Outerbridge grades 3 and 4) was the age, with an AUC of 0.865 ± 0.02 on the specificitysensitivity plot. For an age over 41 years, sensitivity and specificity for severe degeneration

were 82.8% (CI: 77.5–87.3%) and 76.4% (CI: 70.4–81.6%), respectively. The next best predictor for degeneration was the normalized cross section area of the knee (AUC = 0.767 ± 0.04), followed by BMI (AUC = 0.739 ± 0.02) and the normalized maximal diameter of the knee (AUC = 0.724 ± 0.05). The absolute diameters and cross-sectional areas showed AUCs in the range of 0.653–0.685. The Insall-Salvati index proved to be useless for making predictions. The AUC, sensitivity, specificity and relative criteria for all non-binary variables are listed in Table 5 and the ROC curves are depicted in Figure 4.

**Table 5.** Accuracy (AUC) of demographic and radiologic parameters for the prediction of severe knee degeneration (Outerbridge > 2).


AUC: Area under the curve; CI: Confidence Interval; BMI: Body Mass Index.

**Figure 4.** Areas under the curve of the non-binary variables. Top predictors for severe Outerbridge score before normalization are age, BMI and maximal knee diameter.

#### *3.4. Comparisons of ROC Curves*

Age has a significantly higher AUC for predicting severe degeneration than BMI (*p* < 0.0001), maximal diameter (*p* < 0.0001) and knee cross-sectional area (*p* < 0.0001). BMI showed a significantly better accuracy for detecting degeneration than maximal diameter (*p* = 0.0049) and knee cross-sectional area (*p* = 0.0008).

For the 122 patients that have been normalized by the maximal femur diameter (Figure 5), age did not demonstrate a significant difference in AUC compared to the normalized maximal diameter (*p* = 0.363) or the normalized knee area (*p* = 0.846). No combination of variables for a multicriteria ROC analysis reached a higher AUC than age alone.

**Figure 5.** Normalizing the knee cross-sectional area and maximal knee diameter by dividing the measurements by the maximal femur diameter (at the level of the condyles) results in an increase in the respective AUCs.

#### **4. Discussion**

The purpose of our study was to assess the areas under the curve of multiple variables (age, height, weight, BMI, Insall-Salvati index, cross section area of the knee, vertical diameter, horizontal diameter, maximal diameter) as well as normalized cross-sectional area and normalized maximal diameter as biomarkers for the prediction of knee cartilage degeneration. The knee images used for this study were acquired using 1.5 T and 3 T MRI scanners. To grade cartilage changes, the modified Outerbridge classification was used. Both 1.5 T and 3.0 T scanners have been proven effective in detecting not only advanced cartilage loss but also mild cartilage changes [25], although previous studies have stressed the superiority of 3.0 T scanners for cartilage assessment [19,26].

Degenerative cartilage changes and subsequent loss in joint cartilage volume are among the first structural changes observed during OA progression, and they are always present in patients with advanced osteoarthritis [21].

Age and obesity are known risk factors for knee pain and osteoarthritis. Previous studies showed that degenerative knee changes increase with increasing obesity measured in terms of BMI and that cartilage loss extent is age-dependent and more pronounced in individuals aged 50+ years [9,19,27–29].

Our study results provide evidence for the fact that knee cartilage changes begin earlier in life with a cut-off for prediction of cartilage degeneration set to 41 years of age, making age the best surrogate biomarker for severe cartilage degeneration.

Many age-dependent factors such as gradual strength loss of quadriceps muscle and increased oxidative stress in the knee joint microenvironment contribute to this process [30–34].

As for obesity, body weight excess promotes knee joint degeneration in a multifactorial way, including both mechanical and metabolic components. On the one side indeed, obesity constitute a chronic overload for the legs' joints, and it also modifies the biomechanics and the axes load of the joints in the lower extremities. On the other side, metabolic factors are affected by obesity, e.g., the proliferation of chondrocytes, which is impaired in overweight individuals, or leptin secretion, a hormone regulating the food intake habits that was shown to also play a role in OA [10,35,36].

Previous attempts have been made to quantify the amount of adipose tissue surrounding the knee, beyond BMI, e.g., by measuring the anteroposterior and sinistro-dexter (medio-lateral) knee diameter as well as by measuring the surface of the knee cross-sectional area, and to evaluate these measurements as possible prognostic indicators for cartilage loss [8]. The rationale behind the evaluation of these parameters with respect to their ability to be prognostic markers was their potential use by clinicians as a simplified way to predict cartilage degeneration, performing the relevant measurements on the x-ray images, where the cartilage is not visible, thus avoiding a more expensive MRI examination, or even by measuring the diameters or the circumference of the knee directly on the patient's body, making imaging unnecessary, at least at the moment a first risk assessment is indicated. Only a weak association was observed between the extent of cartilage loss, as graded by the Outerbridge scale, and the anteroposterior knee diameter, the sinistro-dexter diameter or the knee surface area.

In the present study, we included in our evaluation also the normalized cross-sectional area and the normalized maximal diameter of the knee by dividing the cross-sectional area and the maximal knee diameter by the maximal femoral diameter in a subgroup of 122 patients. The area under the curve was used to evaluate the different parameters as possible predictors. No combination of these parameters proved to be a better predictor of degeneration than age alone. In the 122 patients in which measurements were normalized to the maximal femur diameter, normalized diameter and area measurement showed the same ability to predict severe cartilage degeneration as the age variable did, whereas the absolute measurements of the diameter and area (*n* = 481) were significantly inferior as surrogate biomarkers than the variable age alone. This signifies that in order for the measurements to be as efficient as age in predicting cartilage degeneration, the measurements need to be normalized.

#### *4.1. Strengths*

The large number of patients (481), the inclusion of all knee compartments in the cartilage assessment, the further analysis of knee cross-sectional area and maximal diameter upon normalization and the calculation of the ROC AUC values for all parameters are some of the strengths of our study, making the statistical results of the study more reliable and strengthening the conclusions.

#### *4.2. Limitations*

Axial images were not perpendicular to leg axis, which might have caused an overestimation of the diameters, circumference and knee cross-sectional areas. However, we counteracted the above-mentioned bias by simplifying the readout: the strictly vertical and horizontal diameters of the knee were measured and the knee cross-sectional surface was measured automatically by Osirix, allowing greater reproducibility. Furthermore, the Outerbridge classification is a radiological score and not a reliable direct measurement. Although the correlation between the Outerbridge score and arthroscopic or surgical knee findings is high [21], the clinical impact would still be the ground truth. Another limitation is that inter- and intra-reader variability assessments are missing due to the fact that no double reading was performed because of the large size of the dataset under investigation.

#### **5. Conclusions**

Our hypothesis that larger knee diameters and cross-sectional area but also gender, obesity (BMI), and the older age are correlated with a larger extent of knee cartilage degeneration proved correct. These parameters can therefore be used as surrogate biomarkers for knee cartilage loss in OA (Outerbridge grades 3 and 4), with age representing the most

accurate biomarker, but not the only one. The measurement of the knee diameter directly on the patient's body or on the patient's knee x-ray images can substitute for the use of lengthier and more expensive imaging methodologies and constitute an easily accessible diagnostic tool that not only radiologists but also general practitioners or orthopedic surgeons can use, with a maximal knee diameter over 13 cm being an alarm bell for severe knee cartilage degeneration.

**Author Contributions:** Conceptualization, A.C., D.S. and H.M.; methodology, A.C., D.S. and H.M.; software, A.C.; Validation, E.P., D.D., G.D., K.S., P.N., A.C. and D.S.; formal analysis, E.P., G.D., K.S., P.N., A.C. and D.S.; investigation, A.C. and D.S.; resources, A.C., L.E., D.S. and H.M.; data curation, D.S., K.S. and P.N.; writing—original draft preparation, E.P., D.D., G.D., D.S. and A.C.; writing—review and editing, E.P., D.D:, H.M., D.S., G.D., K.S., P.N., V.C.O., A.A.P., A.T.H., L.E. and A.C.; visualization, A.C., V.C.O., A.A.P. and A.T.H.; supervision, A.C. and D.S.; project administration, A.C. and L.E. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Institutional Review Board Statement:** Ethical review and approval were waived for this study due to the retrospective nature of the study without participation of the patients and the irreversible anonymization of the image data (SIL.KB.1076.2022.MP).

**Informed Consent Statement:** The responsible ethics commission waived the requirement to obtain informed consent from the patients included in this study due to the study's retrospective nature and the irreversible anonymization of patients' identifying information.

**Data Availability Statement:** Data are available upon special request.

**Acknowledgments:** We would like to thank Grazia Maria Cereghetti De Marchi for the distinguished manuscript preparation.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


**Disclaimer/Publisher's Note:** The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

### *Article* **Adjustable-Loop Cortical Suspensory Fixation Results in Greater Tibial Tunnel Widening Compared to Interference Screw Fixation in Primary Anterior Cruciate Ligament Reconstruction**

**Tae-Jin Lee 1, Ki-Mo Jang 2, Tae-Jin Kim 1, Sang-Min Lee <sup>1</sup> and Ji-Hoon Bae 1,\***


**Abstract:** *Background:* Although the use of adjustable-loop suspensory fixation has increased in recent years, the influence of the shortcomings of suspensory fixation, such as the bungee-cord or windshield-wiper effects, on tunnel widening remains to be clarified. *Hypothesis/Purpose:* The purpose of this study was to compare adjustable-loop femoral cortical suspensory fixation and interference screw fixation in terms of tunnel widening and clinical outcomes after anterior cruciate ligament reconstruction (ACLR). We hypothesized that tunnel widening in the adjustable-loop femoral cortical suspensory fixation (AL) group would be comparable to that in the interference screw fixation (IF) group. *Methods:* This study evaluated patients who underwent primary ACLR at our institution between March 2015 and June 2019. The femoral and tibial tunnel diameters were measured using plain radiographs in the immediate postoperative period and 2 years after ACLR. Tunnel widening and clinical outcomes (Lysholm score, 2000 International Knee Documentation Committee subjective score, and Tegner activity level) were compared between the two groups. *Results:* There were 48 patients (mean age, 29.8 ± 12.0 years) in the AL group and 44 patients (mean age, 26.0 ± 9.5 years) in the IF group. Tunnel widening was significantly greater in the AL group than that in the IF group at the tibia anteroposterior (AP) middle (2.03 mm vs. 1.32 mm, *p* = 0.017), tibia AP distal (1.52 mm vs. 0.84 mm, *p* = 0.012), tibia lateral proximal (1.85 mm vs. 1.00 mm, *p* = 0.001), tibia lateral middle (2.36 mm vs. 1.03 mm, *p* < 0.001), and tibia lateral distal (2.34 mm vs. 0.85 mm, *p* < 0.001) levels. There were no significant differences between the two groups with respect to femoral tunnel widening and clinical outcomes. *Conclusions:* Tibial tunnel widening was significantly greater in the AL group than in the IF group at 2 years after primary ACLR. However, the clinical outcomes in the two groups were comparable at 2 years.

**Keywords:** anterior cruciate ligament; reconstruction; bone tunnel widening; adjustable-loop device; interference screw; hamstring tendon; autograft

#### **1. Introduction**

Anatomical anterior cruciate ligament (ACL) reconstruction (ACLR) is being performed more frequently than in the past, and the use of adjustable- or fixed-loop devices has increased with the development of suspensory fixation. With suspensory fixation, the graft has full contact within the bone tunnel without any foreign material, which may allow early graft integration. However, femoral cortical suspensory fixation may increase the risk of tunnel widening due to micro-movements at the bone-tendon interface, such as the bungee-cord and windshield-wiper effects along the longitudinal and transverse axes, respectively [1,2].

Micro-movements at the tendon-bone interface, such as the bungee-cord and windshieldwiper effects, have resulted in tunnel widening in experimental animal studies [3]. A

**Citation:** Lee, T.-J.; Jang, K.-M.; Kim, T.-J.; Lee, S.-M.; Bae, J.-H.

Adjustable-Loop Cortical Suspensory Fixation Results in Greater Tibial Tunnel Widening Compared to Interference Screw Fixation in Primary Anterior Cruciate Ligament Reconstruction. *Medicina* **2022**, *58*, 1193. https://doi.org/10.3390/ medicina58091193

Academic Editor: Vassilios S. Nikolaou

Received: 8 August 2022 Accepted: 30 August 2022 Published: 1 September 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

recent meta-analysis revealed greater femoral tunnel widening with suspensory fixation using a fixed-loop device for ACLR than that with transfemoral cross-pin fixation [4]. In comparison, interference screw fixation has the advantage of reducing graft movement in the tunnel and synovial fluid influx, although there is a risk of graft and tunnel damage during screw insertion. A recent study also reported that tunnel widening with interference screws was less than that with cortical suspensory fixation [5,6].

The use of adjustable-loop devices has increased to compensate for the micro-movement due to cortical suspensory fixation. Several studies have reported that adjustable-loop devices can reduce micro-movements compared to fixed-loop devices after ACLR [7–10]. However, studies have reported that the adjustable-loop device may cause tunnel widening after surgery [11,12].

Although most previous studies did not report an association between tunnel widening and adverse clinical outcomes, large tunnels may compromise graft fixation during revision surgery or may necessitate two-stage surgery [13–17]. Therefore, the purpose of our study was to compare tunnel widening and clinical outcomes after ACLR with adjustable-loop femoral cortical suspensory fixation to those with interference screw fixation. We hypothesized that tunnel widening in the adjustable-loop femoral cortical suspensory fixation (AL) group would be comparable to that in the interference screw fixation (IF) group.

#### **2. Methods**

#### *2.1. Participants*

This study retrospectively analyzed data obtained from our prospective longitudinal observational study of primary ACLR between March 2015 and June 2019 at our institutions, namely Korea University Anam Hospital and Korea University Guro Hospital, Seoul, Republic of Korea. All patients were provided written information about the study and informed consent was obtained. Ethical approval was obtained from the institutional review board (IRB) of our hospital (IRB No. 2021GR0105). Patients who underwent primary ACLR were included, and demographic data were collected from electronic medical records.

Patients who met the following inclusion criteria were included: (1) anatomical singlebundle ACLR using hamstring-tendon autograft (outside-in technique); (2) the same tibial fixation technique (interference screw fixation with additional fixation); and (3) a followup of at least 2 years. The exclusion criteria were as follows: (1) other ligament injuries requiring surgical treatment, (2) previous knee surgeries, and (3) subsequent injuries (ACL graft failure, meniscus or cartilage injury, or contralateral ACL rupture).

Of the 352 patients who underwent primary ACLR in our institution, 109 met the inclusion and exclusion criteria for the study. The study was conducted on 92 patients who completed serial tests. The AL group included 48 patients and 44 patients were allocated to the IF group. The patient's flow diagram is shown in Figure 1.

#### *2.2. Surgical Technique and Rehabilitation*

All patients underwent anatomical single-bundle ACLR. Surgeries were performed by two senior surgeons. One surgeon used an adjustable-loop cortical suspensory device for femoral fixation, and the other surgeon used an interference screw for femoral fixation. Tibial fixation was performed in the same way in both groups. If there was a meniscus tear, concomitant meniscus repair or a meniscectomy was performed, according to meniscal repairability, prior to ACLR.

A two-incision outside-in technique was used for anatomical femoral tunnel placement in all cases. In the AL group, the femoral part of the graft was prepared using adjustableloop suspensory devices (TightRope [Arthrex, Naples, FL, USA]), and the femoral tunnel was made by retrograde drilling. In the interference screw fixation group (IF group), antegrade drilling and bioabsorbable interference screw (GENESYS Matryx [ConMed, Utica, NY, USA]) were used for femoral fixation. In both groups, a bioabsorbable interference screw was used for tibial fixation after cyclic loading. Additionally, a 6.5-mm cancellous

screw and spiked washer (Arthrex, Naples, FL, USA) were used on the tibial side. The femoral and tibial tunnels were of the same size as the graft diameter, and bioabsorbable interference screws of the same size were used. In both group, grafts were inserted up to the distal end of the tibial tunnel. For the femoral tunnel, the suspensory group inserted the grafts into the femoral tunnel 5 mm away from proximal end and the interference group inserted the grafts up to the proximal end of the femoral tunnel (Figure 2).

**Figure 1.** Patient flow diagram. AL, adjustable-loop suspensory fixation; IF, interference screw fixation.

**Figure 2.** Surgical technique.

After ACLR, all patients underwent the same rehabilitation protocol. A return to previous sports activities was allowed if the patient achieved >80–90% muscle strength compared with that of the contralateral uninjured leg.

#### *2.3. Radiographic Evaluation*

Anteroposterior (AP) and lateral radiographs were taken immediately after surgery and at 12 and 24 months after surgery. On the AP and lateral radiographs, the diameters of the femoral and tibial tunnels were measured between the two sclerotic margins perpendicular to the longitudinal axis at the proximal, middle, and distal sites of the tunnel (Figure 3) [11]. Tunnel widening was calculated by subtracting the diameter measured immediately after surgery from that measured 24 months after surgery.

**Figure 3.** Anteroposterior and lateral radiographs of a right knee showing measurement of each tunnel at proximal (P), middle (M), and distal (D) locations.

All radiographic measurements were performed on the picture archiving and communications system (PACS [INFINITT Healthcare, Seoul, Korea]) using a mouse cursor with automated distance calculation. Two orthopedic surgeons, who were trained in identifying each variable, independently reviewed the radiographs. They performed measurements on two separate occasions, 4 weeks apart.

#### *2.4. Clinical Evaluation*

All patients were evaluated during their regular visits to the outpatient clinic at 6, 12, and 24 months after surgery. The independent staff (experienced athletic trainers and physical therapists) from our sports rehabilitation center conducted all the tests and obtained the data. Patients completed three patient-reported outcomes, including the Lysholm Knee Scoring Scale, the 2000 International Knee Documentation Committee (IKDC) subjective scale, and Tegner activity level [18,19].

#### *2.5. Second-Look Arthroscopy and Graft Evaluation Method*

Second-look arthroscopy was performed for skin irritation from the tibial fixation screw at least one year after ACLR. During the second-look arthroscopy, newly identified problems that were not present at the time of ACLR were investigated. In addition, a modified graft maturation scoring system (Korea University Medical Center score; KUMC score) was used for graft evaluation based on previous studies [20,21].

Two blinded observers (orthopedic surgeons) scored each parameter, and interobserver and intraobserver reliabilities were evaluated. Total graft maturation scores (KUMC score) and individual parameter scores were compared between the two groups.

#### *2.6. Statistical Analysis*

In a pilot study of 10 patients from both groups, tunnel widening was measured, and the mean and standard deviation were calculated. To achieve a power of 80% using a two-group *t*-test with a two-sided significance level of *p* < 0.05, a sample size of 44 in each treatment group was required when calculated using the value measured at the middle of the lateral tibial view as a variable. Data for the final follow-up were available for 48 patients in the AL group and 44 patients in the IF group. Therefore, a power of 80% was reached.

Demographic variables (including patient sex, age, and body mass index), clinical outcomes, and tunnel widening calculated using radiologic measurements were compared between the two groups. Independent *t*-tests and Mann–Whitney U tests were used for continuous variables. Chi-square tests and Fisher's exact tests were used for categorical variables. Paired-samples *t*-tests and Wilcoxon signed-rank tests were applied to continuous variables within each group.

For each variable, the intraclass and interclass correlation coefficients were calculated to quantify the agreement between the measurements. Correlation coefficients >0.75 and <0.4 represent good and poor agreement, respectively [22]. In addition, all variables were expressed using mean values and standard deviations.

Statistical significance was set at *p* < 0.05. Statistical analysis was performed using SPSS for Windows (version 20.0; IBM Corp., Armonk, NY, USA).

#### **3. Results**

Demographic data of the 92 patients are presented in Table 1. No significant differences in age, sex, or body mass index were observed between the two groups.


**Table 1.** Demographic factors.

Values are given as mean and standard deviation. BMI, body mass index. \* Comparisons between the AL and IF groups. AL, adjustable-loop suspensory fixation; IF, interference screw fixation.

#### *3.1. Radiologic Outcomes*

Significant tunnel widening was observed at all measured sites between the immediate postoperative period and 2 years after ACLR in both groups (Tables 2 and 3). All correlation coefficients ranged from 0.813 to 0.935, indicating good reliability.



Bold indicates statistical significance (*p* < 0.05). Values are given as mean and standard deviation. \* Comparisons between the immediate and 2-yearpostoperative values. Lat, lateral. AP, anterior posterior; Lat, lateral.

The tunnel widening in the AL group was significantly greater than that in the IF group at the tibia AP middle, tibia AP distal, tibia lateral proximal, tibia lateral middle, and tibia lateral distal (Table 4).


Bold indicates statistical significance (*p* < 0.05). Values are given as mean and standard deviation. \* Comparisons between the immediate and 2-year postoperative values. Lat, lateral.



Bold indicates statistical significance (*p* < 0.05). Values are presented as mean and standard deviation. \* Comparisons between the AL and IF groups. Lat, lateral.

#### *3.2. Clinical Outcomes*

There were no significant differences in patient-reported outcomes (Lysholm score, 2000 IKDC subjective score, and Tegner activity level) between the two groups at 2 years after ACLR (Table 5).


**Table 5.** Clinical outcomes: 2-year postoperative patient-reported outcomes.

Values are presented as mean and standard deviation. \* Comparisons between the AL and IF groups. \*\* Data are presented as a frequency distribution. IKDC, international knee documentation committee.

#### *3.3. Second-Look Arthroscopic Evaluation*

Twenty-four patients in each group underwent second-look arthroscopic evaluation, and the operation was performed at an average of 24.5 and 26.8 months after ACLR in the AL and IF groups, respectively. After the second-look arthroscopic examination, total graft maturation scores (KUMC score) were compared between the two groups. In patients undergoing second-look arthroscopy, there was no significant difference in the KUMC scores between the two groups (Table 6). All correlation coefficients ranged from 0.838 to 0.905, indicating good reliability. Regarding newly identified problems in group 1, two meniscal tears were identified. In group 2, two cyclops lesions and two meniscal tears were identified.


**Table 6.** Second-look arthroscopic evaluation: graft maturation scores (KUMC scores).

\* Comparisons between the AL and IF groups. \*\* Data are presented as a frequency distribution. KUMC, Korea University Medical Center.

#### **4. Discussion**

The main findings of this study were that ACLR using interference screw fixation was associated with less postoperative tibial tunnel widening than that using adjustableloop femoral cortical suspensory fixation at the 2-year follow-up and that there were no significant differences between the two groups with regard to femoral tunnel widening or clinical outcomes. However, it was difficult to clearly explain the difference in tunnel widening between the two groups. The exact reason was not found, but various factors are presumed to be involved.

Several potential reasons for the differences between the two groups can be considered with regard to tibial tunnel widening. Interference screw fixation has the advantage of reducing micro-movements compared to adjustable-loop femoral cortical suspensory fixation. In a randomized controlled trial, Fauno and Kaalund [14] found that tunnel widening is influenced by the mechanical properties of the implants. They found that extracortical fixation of the graft was associated with more laxity than close-to-joint fixation. Similarly, Giorgio et al. [23] reported that the use of a more elastic fixation system (TightRope) is associated with more laxity and results in greater tunnel widening. They explained that graft motion, such as the bungee-cord and windshield-wiper effects, is caused by lowstiffness systems such as those using extracortical fixation. In the study by Sabat et al. [24], tunnel widening was significantly lesser in the transfix group than in the EndoButton group. They thought that this was because the fixation point was farther away from the aperture. Interestingly, Lind et al. [25] suggested that the femoral fixation technique could potentially affect the biomechanical and biological environment in the tibial tunnel, thus affecting tunnel widening in the tibia. Accordingly, we thought that the difference in the femoral fixation method in our study could have influenced the tunnel widening in the tibia.

However, unlike tibial tunnel widening, a significant difference in femoral tunnel widening was not seen in the present study. We attributed this to the difference between the characteristics of the distal femur and the proximal tibia. Chang et al. [26] reported the mechanical and structural properties of distal femur and proximal tibial bones in vivo. Their study showed that the stiffness of the distal femur was greater than that of the proximal tibia. Therefore, the widening of the femoral tunnel could be expected to be relatively small. Kuskucu [27] reported that a graft that becomes free because of fixation in the tunnel at a site distant from the tunnel opening in the joint will result in greater micro-movement with knee motion. Since micro-movement in the femoral tunnel was less than that in the tibial tunnel, the difference in the femoral tunnel width was not significant.

In contrast to the results of our study, some previous studies have reported that greater femoral tunnel widening occurred when using a suspensory device on the femoral side. Mayr et al. [5] reported that ACLR using adjustable-loop cortical button fixation was associated with greater femoral tunnel enlargement than that with biodegradable interference screw fixation. However, their study differed from our study in that they used an all-inside ACLR. In the suspensory group, patients were treated with all-side ACLR using adjustable-loop button fixation on both the femoral and tibial sides. Suspensory tibial fixation may have influenced whether tunnel widening occurred in the femur or tibia. In a similar study, Baumfeld et al. [15] reported a comparison between double cross-pin and suspensory graft fixation. Femoral tunnel widening associated with the use of the EndoButton suspensory fixation system was significantly greater than that with the use of double cross-pins for fixation within the tunnel. However, their results may be different from ours because they did not standardize the tibial fixation method.

In this study, both groups showed excellent overall clinical outcomes. Patient-reported outcomes, including the Lysholm score, 2000 IKDC subjective score, and Tegner activity level, improved over a postoperative period of 2 years. However, there were no significant differences between the two groups. In addition, in patients undergoing second-look arthroscopy, there was no significant difference in the graft maturation scoring system (KUMC) scores between the two groups. Raj et al. [28] evaluated the correlation between the bone tunnel diameter following ACLR using a hamstring-tendon autograft and functional outcomes. They concluded that neither the diameter nor widening of the bone tunnel during the follow-up period was correlated with functional outcomes. Several studies have reported that there were no clinical implications of tunnel widening after ACLR [12,29–32]. However, the absence of clinical differences may be attributed to the small difference in tunnel widening between the two groups. Therefore, the results of long-term follow-up may vary. Tunnel widening should be considered in revision ACLR, but a difference of approximately 1 mm will not have significant implications.

This study has several limitations. First, this study was designed retrospectively, and the operations were performed by two surgeons. Although the same outside-in technique was used, the difference between antegrade and retrograde drilling for the femoral tunnel could affect the results. Second, the use of an interference screw can affect the width of the tunnel. There was a risk of graft and tunnel damage during screw insertion. However, the analysis showed that there was no significant difference in tunnel widening by the interference screw in the present study. Third, the minimum follow-up period was only 2 years. Although there was no difference in clinical outcomes in this study, further studies are needed as outcomes may vary at the mid- to long-term follow-up.

#### **5. Conclusions**

In conclusion, tibial tunnel widening after ACLR using a hamstring-tendon autograft was significantly greater with adjustable-loop femoral cortical suspensory fixation than with interference screw fixation at the 2-year follow-up. However, the clinical outcomes in the two groups were comparable at the 2-year follow-up.

**Author Contributions:** T.-J.L.: study concepts/design, manuscript drafting/revision, K.-M.J.: data acquisition/analysis, manuscript revision, T.-J.K.: data acquisition/analysis, S.-M.L.: data acquisition/analysis, J.-H.B.: study concepts/design, manuscript drafting/revision. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Institutional Review Board Statement:** This study was approved by the institutional ethical review board (2021GR0105).

**Informed Consent Statement:** The informed consent was obtained from each patient.

**Acknowledgments:** Authors would like to acknowledge support from Sports Medical Center at our institution for obtaining clinical measurements and Medical Science Research Center for statistical analysis.

**Conflicts of Interest:** Each author certifies that he has no commercial associations (e.g., consultancies, stock ownership, equity interest, patent/licensing arrangements, etc.) that might pose a conflict of interest in connection with the submitted article.

#### **References**


### *Article* **Beneficial Effect of Curved Dilator System for Femoral Tunnel Creation in Preventing Femoral Tunnel Widening after Anterior Cruciate Ligament Reconstruction**

**O-Sung Lee 1, Joong Il Kim 2, Seok Hyeon Han <sup>3</sup> and Joon Kyu Lee 4,\***


**Abstract:** *Backgrounds and objectives:* A prevalent concern in anterior cruciate ligament (ACL) reconstruction is postoperative tunnel widening. We hypothesized that employing a curved dilator system (CDS) for femoral tunnel creation can reduce this widening after ACL reconstruction compared to the use of a conventional rigid reamer. *Materials and Methods:* A retrospective study was conducted involving 56 patients who underwent primary ACL reconstruction between January 2012 and July 2013. The patients were categorized into two groups: the reamer group (*n* = 28) and CDS group (*n* = 28). All participants were followed up for a minimum of 2 years. Clinical assessment included the Lachman test and pivot-shift test, and the Lysholm score and subjective International Knee Documentation Committee scores. Radiographic evaluation covered the tunnel widening rate, represented as the ratio of the tunnel diameter 2 years after surgery to the tunnel diameter immediately after surgery, and the ratio (A/B) of femoral tunnel (A) to tibial tunnel (B) diameters at respective time points. *Results:* No significant disparities were found between the two groups in terms of clinical outcomes. However, the reamer group exhibited a greater femoral tunnel widening rate compared to the CDS group (reamer group vs. CDS group: 142.7 ± 22.0% vs. 128.0 ± 19.0% on the anteroposterior (AP) radiograph and 140.8 ± 14.2% vs. 122.9 ± 13.4% on the lateral radiograph; all *p <* 0.05). Two years post-operation, the A/B ratio rose in the reamer group (0.96 ± 0.05→1.00 ± 0.05 on the AP radiograph and 0.94 ± 0.03→1.00 ± 0.0.04 on the lateral radiograph; all *p* < 0.05), while it decreased in the CDS group (0.99 ± 0.02→0.96 ± 0.05 on the AP radiograph and 0.97 ± 0.03→0.93 ± 0.06 on the lateral radiograph; all *p* < 0.05). *Conclusion:* The use of CDS for femoral tunnel creation in primary ACL reconstruction provides a potential advantage by limiting tunnel widening compared to the conventional rigid-reamer approach.

**Keywords:** bone tunnel widening; bone tunnel enlargement; anterior cruciate ligament; anterior cruciate ligament reconstruction

#### **1. Introduction**

The issue of tunnel widening following anterior cruciate ligament (ACL) reconstruction is a well-documented complication, which could potentially lead to surgical failure. [1–4]. It not only jeopardizes the graft success but also requires staged management in revision ACL reconstruction [5,6]. While the precise cause of tunnel widening remains elusive, contributing factors have been suggested to be biological, biomechanical, and mechanical [7–10]. Biological factors include healing at the graft–tunnel interface and

**Citation:** Lee, O.-S.; Kim, J.I.; Han, S.H.; Lee, J.K. Beneficial Effect of Curved Dilator System for Femoral Tunnel Creation in Preventing Femoral Tunnel Widening after Anterior Cruciate Ligament Reconstruction. *Medicina* **2023**, *59*, 1437. https://doi.org/10.3390/ medicina59081437

Academic Editor: In Jun Koh

Received: 17 July 2023 Revised: 5 August 2023 Accepted: 7 August 2023 Published: 8 August 2023

**Copyright:** © 2023 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

increased cytokine levels leading to inflammatory responses, infection, and cell apoptosis. Biomechanical factors include foreign body reactions and necrosis due to heat during drilling for tunnel creation. And, mechanical factors include the graft fixation method, the graft motion within the tunnel, the position of the tunnel, and accelerated rehabilitation.

Among these mechanical factors, different outcomes related to tunnel widening have been reported with varying graft fixation methods, graft types, tunnel locations, and femoral tunnel creation methods [11–16]. However, the literature offers scarce comparative data on tunnel widening after using a dilator versus a conventional reamer in ACL reconstruction. We introduce the curved dilator system (CDS) in order to overcome the weakness of the anteromedial (AM) portal technique with a conventional reamer, which may hinder the attainment of an anatomical position for the femoral tunnel [17]. By using CDS for femoral tunnel creation, we theorize that it might condense the inner wall of the tunnel and hence mitigate tunnel widening. Consequently, we posit that utilizing the CDS in ACL reconstruction could potentially reduce both mechanical and biomechanical tunnel widening in comparison to the conventional rigid reamer. Therefore, this study aims to examine tunnel enlargement following ACL reconstruction conducted using the CDS versus the conventional reamer.

#### **2. Materials and Methods**

We examined individuals who underwent initial ACL reconstruction from January 2012 to July 2013. The inclusion criteria were primary ACL reconstruction using the allogenous tibialis tendon. Only those patients who underwent both clinical and radiological assessments at a 2-year postoperative follow-up were incorporated in the study. The exclusion criteria were as follows: (1) previous intra-articular ligament reconstruction, (2) multiple-ligament injury, and (3) previous osseous procedures. A concomitant meniscal injury was not an exclusion criterion. The institutional review board granted approval for this research.

#### *2.1. Surgical Procedure*

Surgical procedures exhibited no variations, with the exception of the technique employed for femoral tunnel creation between the groups. The portals utilized included anterolateral, AM, and accessory anteromedial (AAM). Notably, the AM portal was positioned slightly more proximal than the standard approach. In contrast, the AAM portal was created 10–15 mm medial to the medial border of the patellar tendon, representing a notably lateral shift from the usual AAM portal placement and situated distally at the level immediately proximal to the lateral meniscal superior surface [17].

In the reamer group, the procedure involved inserting a guide pin through a drill guide, targeting the center of the ACL footprint. A conventional rigid reamer with the same diameter as the graft was utilized to create the femoral tunnel. On the other hand, in the CDS group, a 4.5 mm-diameter curved guide trocar with a sharp end was introduced at the anatomical footprint of the ACL. Subsequently, another 4.5 mm-diameter curved guide trocar with a sharp end was carefully inserted at a marked point while the knee was flexed to slightly over 90◦, which was the optimal angle to prevent damage to the medial condyle cartilage during trocar passage. The trocar was then gently hammered until it completely penetrated the far cortex of the lateral condyle. Afterward, the trocar was removed, and the tunnel was gradually widened in incremental steps, increasing by 1 mm in diameter at a time to match the graft's diameter.

For creating the tibial tunnel, the entry point was positioned at the level of the tibial tubercle, 2–3 cm medial to the tubercle, just above the attachment site of the pes anserinus. A guide pin was inserted at a 55◦ angle to the tibial plateau, guided by a tibial drill, aiming at the center of the ACL footprint. The tibial tunnel was then drilled along the guide pin using a standard reamer with a diameter that corresponded to that of the graft.

All patients in both groups conducted the same standardized home-based rehabilitation protocol. Routine follow ups were conducted at 2 weeks, 6 weeks, 3 months, 6 months, 1 year, and annually thereafter to ensure appropriate rehabilitation for each phase.

#### *2.2. Clinical and Radiological Assessment*

Clinical and radiological assessments were conducted prior to the surgery, at 3 months, 12 months after the operation, and annually thereafter. The clinical and radiological outcomes achieved 2 years after the operation were utilized for comparing the groups.

The assessment of knee joint stability included the Lachman test and pivot-shift test. The Lachman test measured anterior translation compared to the uninvolved side and was graded from 0 to 3 (0, <2 mm; 1, 2–5 mm; 2, 5–10 mm; 3, >10 mm). The pivot-shift test was evaluated against the uninvolved side and graded from 0 to 3 (0, same as uninvolved side; 1, gentle gliding; 2, clunk; 3, locking). Clinical status was evaluated 1 day before the surgery and annually thereafter using the Lysholm score and subjective International Knee Documentation Committee (IKDC) scores. A Lysholm score consists of eight items, including limping, locking, pain, stair climbing, the use of supports, instability, swelling, and squatting. Subjective IKDC scores include 18 items covering the domains of symptoms (7 items), sports activities (10 items), and function (1 item). The IKDC-SKF ranges from 0 to 100; higher scores indicate higher levels of function and fewer symptoms.

For radiological assessment, standardized anteroposterior (AP) and lateral knee radiographs were taken to evaluate tunnel widening. The diameter of the femoral and tibial tunnels was measured based on the method by L'Insalata et al. [2]. This involved measuring the length between the inner borders of the sclerotic margins perpendicular to the tunnel axis on the AP and lateral radiographs (Figure 1). Radiographs obtained immediately after the surgery and 2 years postoperatively were used for evaluation.

**Figure 1.** Tunnel diameter was determined as the widest distance between the sclerotic margins of the tunnel perpendicular to the axis of the tunnel based on the method by L'Insalata et al. The diameters were measured from the AP and lateral radiograph of the operated knee at 2 years post-operation.

All radiological parameters were measured twice, with a 2-week interval and using a Picture Archiving and Communication System, by two orthopedic surgeons who were not part of the surgical team. To assess tunnel widening, a tunnel widening rate was calculated, representing the ratio of the tunnel diameter after 2 years of surgery to the tunnel diameter immediately after surgery. Additionally, the ratio (A/B) of the diameter of the femoral tunnel (A) and tibial tunnel (B) was calculated at each point to account for slight differences in the graft diameter among cases.

#### *2.3. Statistical Analysis*

Continuous variables were expressed as means with standard deviations, and their comparison was conducted using Student's *t*-test. Categorical variables, on the other hand, were analyzed using the Pearson chi-square test. The inter- and intraobserver reliabilities of radiological measurements were determined to be acceptable based on the calculated intraclass correlation coefficients. Two orthopedic surgeons working in the knee division of the orthopedic department performed two measurements, at 1-week intervals, and the interand intra-observer reliability for the radiologic measurements were 0.89 (range, 0.87–0.97) and 0.91 (range, 0.85–0.96), respectively. A post hoc power analysis was conducted using G power calculator 3.1.9.2 to ensure that the sample size for this retrospective study was sufficient. All statistical analyses were carried out using SPSS software (ver. 25.0.0), and statistical significance was set at a *p*-value of less than 0.05.

#### **3. Results**

Among a total of 60 consecutive patients who underwent primary ACL reconstruction, 4 patients were excluded due to not meeting the selection criteria or inadequate follow up. These exclusions were based on the following reasons: less than 2 years of follow up for two patients, one patient experienced postoperative complications related to an infection, and one patient had missing clinical data or radiographs.

Ultimately, 56 patients with a minimum 2-year follow up were included in the study and divided into two groups based on the surgical instrument used for femoral tunneling: the reamer group (standard rigid reamer, *n* = 28), which involved a standard rigid reamer, and the CDS group (curved dilator system, *n* = 28), which involved a curved dilator system. Patients' demographic data, including age, sex, and body mass index, were retrieved from medical records and are presented in Table 1. Demographic characteristics showed no significant differences between the two groups.


**Table 1.** Demographic data of patients in the reamer and CDS groups.

BMI, body mass index; CDS, curved dilator system. The values are presented as the mean ± standard deviation with the range in parentheses. † Derived using Student's *t*-test. \* Derived using the Pearson chi-square test.

#### *Clinical and Radiological Results*

Clinical evaluations of knee joint stability, including the Lachman test and pivot-shift test, revealed no significant differences between the two groups. Similarly, no significant differences were found in the Lysholm score and subjective IKDC score between groups (Table 2).


**Table 2.** Comparison of the clinical outcomes between the reamer and CDS groups 2 years after surgery.

CDS, curved dilator system; IKDC, International Knee Documentation Committee. The values are presented as the mean ± standard deviation with the range in parentheses. † Derived using Student's *t*-test. \* Derived using the Pearson chi-square test.

The femoral tunnel widening rate on the AP radiograph was greater in the reamer group than in the CDS group (reamer group, 142.7 ± 22.0 vs. CDS group, 128.0 ± 19.0; *p =* 0.014). The femoral tunnel widening rate on the lateral radiograph was also larger in the reamer group than in the CDS group (reamer group, 140.8 ± 14.2 vs. CDS group, 122.9 ± 13.4; *p =* 0.001). However, the tibial tunnel widening rate on both the AP and lateral radiographs showed no statistically significant differences in both groups (Table 3).

**Table 3.** Comparison of tunnel widening rate (percent) between the reamer and CDS groups using the plain radiograph 2 years after surgery.


CDS, curved dilator system. The tunnel widening rate from immediately to 2 years after surgery was calculated. The values are presented as the mean ± standard deviation. The statistical significance was set at *p* < 0.05 and derived using Student's *t*-test.

Regarding the ratio (A/B) of the diameters of the femoral tunnel (A) and tibial tunnel (B), the ratio increased in the reamer group from immediately after surgery to the last follow up (AP radiograph: 0.96 ± 0.05 to 1.00 ± 0.05, *p* < 0.001; lateral radiograph: 0.94 ± 0.03 to 1.00 ± 0.04, *p* < 0.001). Conversely, in the CDS group, the ratio decreased (AP radiograph: 0.99 ± 0.02 to 0.96 ± 0.05, *p* = 0.001; lateral radiograph: 0.97 ± 0.03 to 0.93 ± 0.06, *p* = 0.020) (Table 4).

**Table 4.** Ratio (A/B) of the femoral tunnel (A) and tibia tunnel (B) immediately and 2 years after surgery.


CDS, curved dilator system. The values are presented as the mean ± standard deviation. The statistical significance was set at *p* < 0.05 and derived using Student's *t*-test.

#### **4. Discussion**

The main findings of our investigation are as follows: (1) Femoral tunnel widening was significantly larger in the reamer group than in the CDS group 2 years following primary ACL reconstruction. (2) The diameter ratio of the femoral to tibial tunnel decreased in the CDS group but increased in the reamer group 2 years after surgery. These findings suggest that using a curved dilator system (CDS) to create the femoral tunnel in ACL reconstruction could be advantageous in preventing femoral tunnel widening, compared to the conventional rigid reamer.

Tunnel enlargement is a widely recognized complication after ACL surgery since the early 1990s [1,2,18]. The prevalence of tunnel widening has been documented, ranging in variability from 20% to 100% from several studies [11,13,14,19–21]. This complication is concerning as it can lead to ACL reconstruction failure due to unsuitable conditions for graft fixation. Moreover, it may necessitate complex, staged management for revision ACL surgery, including initial bone grafting procedures to fill the widened tunnels [10,22,23]. This complication is not only related to surgical techniques but also to the type of fixation devices, graft selection, and demographic factors [12,15,16,19,24,25]. The relation between tunnel widening and clinical outcomes remains uncertain. Some studies have hypothesized that tunnel widening potentially serves as an early finding of graft failure. However, many recent studies could not find any relationship between tunnel enlargement and the clinical results after surgery, and only a few have demonstrated the impact on the clinical outcome of tunnel widening [15,16,21,25–27].

While the precise causes of tunnel enlargement are very partially understood, it is generally agreed upon that multiple factors, including biological, biomechanical, and mechanical, contribute to its occurrence. Several studies reported which factors affect tunnel widening, such as the types of grafts, the type of fixation device, the position and size of the tunnel holes, the micro-motion of the graft in the tunnel, and altered cytokine levels in the synovial joint fluid [28–33]. Biological factors such as graft healing at the tunnel surface, inflammatory reactions, abnormal activity of the osteoclast, and infection are implicated in tunnel widening after ACL reconstruction. A biomechanical factor is heat necrosis from drilling and foreign bodies. A certain amount of heat by drilling may induce the necrosis of the bone and secondary inflammation. And, several studies have reported that various types of allogenic grafts may raise the risk for tunnel enlargement with the foreign-body immune response [1,20,34,35]. Several studies reported the graft type, cyclic loading of the knee joint, and motion of the graft within the tunnel as mechanical factors after ACL reconstruction [26,29,36]. The longitudinal motion of the graft along the tunnel, known as the bungee effect, and the transverse motion of the graft perpendicular to the axis of the tunnel known as the windshield-wiper effect may lead to subsequent bone resorption.

Recently, surgeons suggested some possible treatments in order to mitigate the tunnel widening. A preclinical study showed that the utilization of mesenchymal stem cells seeded in a collagen type-I scaffold enhanced the biological healing of ACL grafts in a rabbit model [37]. Additionally, Robinson J. et al. reported that a poly-L-lactic acidhydroxyapatite-blended bio-absorbable screw reduced the tunnel widening after ACL surgery using the hamstring tendon [19]. Yamazaki S. et al. demonstrated in their study that the utilization of transforming growth factor-beta 1 substantially enhanced the production of collagen fibers linking the tendon graft to bone in dogs [38]. And, Hashimoto Y. et al. reported that recombinant BMP-2 treatment resulted in the successful regeneration of the tendon–bone junction in a rabbit model [5]. Additionally, hyperbaric oxygen significantly promoted the incorporation of bone to tendon, and rose the tensile strength of the graft by enhancing the amount of trabecular bone around the graft [17].

Our hypothesis is that continuous dilators for the creation of the femoral tunnel might reduce the tunnel widening compared to conventional drilling with a rigid reamer by creating a more condense tunnel wall. Previous studies have reported varying results regarding tunnel widening according to the type of reamer used for tunnel drilling [13,17]. Justin R. Knight et al. observed significantly greater tibial tunnel deformation when using the acorn reamer compared to the mono-fluted reamer [17]. However, Rainer Siebold et al. did not find a significant reduction in postoperative tunnel widening after ACL reconstruction using compaction drilling with a stepped router [17]. In our investigation, we sought to determine if the creation of the femoral tunnel using CDS could effectively decrease femoral bone tunnel widening following ACL reconstruction, and our hypothesis was strongly supported.

Despite these findings, our study has some limitations. Firstly, it was retrospective and lacked randomization, potentially leading to selection bias. However, we carefully compared the two groups, ensuring no significant differences in basic demographic data, thereby minimizing confounding factors. Secondly, the evaluation period was limited to 2 years after surgery, and further investigation is required to assess whether changes in tunnel diameter affect long-term clinical outcomes after ACL reconstruction. Thirdly, we did not employ three-dimensional imaging modalities like CT or MRI to evaluate tunnel diameters. Nonetheless, the method used in our study, as established by L'Insalata et al., has been widely accepted as a standard approach for measuring tunnel diameter after ACL reconstruction in numerous published studies.

#### **5. Conclusions**

The femoral tunnel widening rate was larger in the reamer group than in the CDS group on both the AP and lateral radiographs, and the ratio (A/B) of the diameter of the femoral tunnel (A) and tibial tunnel (B) decreased in the CDS group but increased in the reamer group 2 years after ACL reconstruction. These results suggest that using CDS in primary ACL reconstruction may help prevent femoral tunnel widening, compared to the conventional rigid reamer. Nevertheless, there were no notable disparities in clinical results among the two groups during the 2-year follow up. Additional investigations with extended follow-up durations are warranted to comprehensively understand the long-term implications of our findings.

**Author Contributions:** Conceptualization, J.K.L.; Methodology, J.K.L. and O.-S.L.; Formal Analysis, J.I.K. and O.-S.L.; Investigation, S.H.H.; Resources, J.K.L.; Data Curation, S.H.H. and J.I.K.; Writing—Original Draft Preparation, O.-S.L.; Writing—Review and Editing, J.K.L.; Visualization, O.-S.L.; Supervision, J.K.L. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Institutional Review Board Statement:** This study was approved by the Institutional Review Board of Hallym University Sacred Heart Hospital (IRB no.: 2016-I037 issued on the 8 March 2016).

**Informed Consent Statement:** Patient consent was waived due to anonymity of the data used in the study. It was approved by the Institutional Review Board of Seoul National University Hospital.

**Data Availability Statement:** The data presented in this study are available on request from the corresponding author. The data are not publicly available due to privacy.

**Conflicts of Interest:** The authors declare that they have no competing interest.

#### **References**


**Disclaimer/Publisher's Note:** The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

### *Article* **Changes in Bone Marrow Lesions Following Root Repair Surgery Using Modified Mason–Allen Stitches in Medial Meniscus Posterior Root Tears**

**Kyu Sung Chung 1,\*, Jeong Ku Ha 1, Jin Seong Kim <sup>2</sup> and Jin Goo Kim <sup>3</sup>**


**Abstract:** *Background and Objectives*: Root repair can prevent osteoarthritis (OA) by restoring hoop tension in medial meniscus posterior root tears (MMPRTs). This study aims to investigate bone marrow edema (BME) lesions known to be associated with OA following MMPRTs. *Methods*: Thirty patients with transtibial pull-out repair were recruited. Subchondral BME lesions were evaluated using magnetic resonance imaging (MRI) at 1-year follow-ups. Participants were categorized into three groups: no change of BME lesions (group one), improved BME lesions (group two) and worsened BME lesions (group three). Clinical scores and radiological outcomes, specifically Kellgren– Lawrence grade, medial joint space width and cartilage grade and meniscal extrusion were evaluated and compared between groups. *Results*: After surgery, twenty-three patients with no BME, three patients with BME lesions on the medial femoral condyle, one patient with BME lesions on the medial tibia plateau and three patients with BME lesions on both were investigated. A total of 20 patients in group one (66.7%) showed no change in BME lesions. In group two, seven patients (23.3%) presented with improved BME lesions. Only three patients (10%) showed worsened BME lesions (group three). Moreover, Lysholm scores and the rate of progression of cartilage grades were significantly worse in group three patients. Meniscal extrusion was significantly reduced in group two, whereas extrusion was significantly progressed in group three. *Conclusions*: Patients with worsened BME lesions showed less favorable outcomes than other patients. A decrease in meniscal extrusion can have a positive effect on BME lesions after root repair.

**Keywords:** bone marrow lesion; knee; meniscus; root tear; root repair

### **1. Introduction**

Bone marrow edema (BME) is defined as an area of low signal intensity on T1-weighted images that is associated with intermediate or high signal intensity findings on T2-weighted images. BME lesions are a risk factor for structural deterioration in knee osteoarthritis and are strongly associated with the progression of osteoarthritis (OA) [1–3].

The meniscus is composed of an interconnecting network of collagen fibers, proteoglycans and glycoproteins [4,5]. Collagen fibers stretch under axial-load-increasing internal hoop stress which absorbs and redistributes forces transmitted to the joint [6–10]. Medial meniscus posterior root tears (MMPRTs) are defined as an avulsion injury or radial tear occurring in posterior bone attachment [11–16]. MMPRTs can predispose patients to degenerative OA due to the loss of hoop tension and load-sharing ability, which results in unsupportable peak pressures on the knee joint [14]. Root repair can restore the meniscal function; thus, it can prevent or delay osteoarthritic changes [17–19]. Root repair shows superior mid- [20] and long-term [21] clinical outcomes when compared with meniscectomy.

**Citation:** Chung, K.S.; Ha, J.K.; Kim, J.S.; Kim, J.G. Changes in Bone Marrow Lesions Following Root Repair Surgery Using Modified Mason–Allen Stitches in Medial Meniscus Posterior Root Tears. *Medicina* **2022**, *58*, 1601. https:// doi.org/10.3390/medicina58111601

Academic Editor: In Jun Koh

Received: 31 August 2022 Accepted: 1 November 2022 Published: 4 November 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

One of the main goals of root repair is the prevention and delay of the onset of osteoarthritic changes through the restoration of meniscal hoop tension. If root repair is not effective or successful, BME lesions will worsen following surgery. Several studies of both the clinical and radiological outcomes of root repair have reported favorable results [22–27]. However, there is little evidence or investigation into the change of BME lesions following root repair.

It has been shown that greater meniscus extrusion is a significant predictor of the progression of arthritic changes in osteoarthritic knees [18]. Therefore, it seems logical that if meniscus extrusion can be eliminated or reduced, the MMPRT will be successfully repaired, and the chance of subsequent degenerative arthritis will be reduced. However, there are no studies that have investigated the correlation between meniscus extrusion and BME lesions.

As such, this study aims to clarify the treatment effects of root repair through investigating the change of BME lesions following root repair and comparing these with both preand post-operative 1-year follow-up results in MMPRTs. In addition, this study aims to investigate the correlation of meniscus extrusion and BME lesions after root repair. We hypothesize that patients with worsened BME lesions will show less-favorable outcomes than other patients and that patients with more reduced meniscal extrusions can demonstrate positive effects on BME lesions after root repair.

#### **2. Materials and Methods**

#### *2.1. Study Population*

This study protocol was approved by our institutional review board (PAIK-08-003), and all patients signed an approved written consent form. Between 2017 and 2020, patients who consented underwent magnetic resonance imaging (MRI) scans both prior to surgery and 1-year following surgery. MMPRT was defined as a complete radial tear on the medial meniscus posterior bone. This was diagnosed by MRI scans (Intera Achieva; Philips, Eindhoven, Netherlands) indicated via an absence of an identifiable meniscus or a highintensity signal replacing the normal dark meniscal signal (ghost sign) in the sagittal plane, a vertical linear defect at the root on the coronal plane, and a radial linear defect at the posterior insertion point in the axial plane [28,29].

The inclusion criteria for the study were: first, MMPRT shown on an MRI scan in a patient with persistent knee pain; second, patients who underwent arthroscopic pullout fixation by modified Mason–Allen stitches; third, patients with a Kellgren–Lawrence (K-L) score of grade two or less. Study exclusion criteria were: first, patients whose MMPRT was combined with a high tibial osteotomy; second, patients with a concomitant ligament injury; and finally, patients who did not want an MRI scan follow-up 1-year after surgery (Figure 1).

Participants were identified using medical records. Prior to surgery, participants were interviewed in the outpatient department and their clinical histories were reviewed and collated in a database.

#### *2.2. Evaluation of Bone Marrow Lesions*

The MRI scans were evaluated both prior to surgery and 1-year following surgery. The evaluations of the BME lesions at both the medial femoral condyle (MFC) and medial tibial plateau (MTP) were conducted. Changes of the BME lesions were assessed by comparing the statuses of the BME lesions between MRI scans prior to surgery and 1-year post surgery. Participants were categorized into 3 groups: no change of BME lesions (group 1), improved BME lesions (group 2) and worsened BME lesions (group 3), through comparisons between pre- and post-operative BME lesion statuses (Figures 2 and 3). The cartilage grade in the medial compartment was graded according to the modified Outerbridge classification by MRI. The healing status was assessed by checking continuity between the bone bed and meniscus on post-operative 1-year follow-up MRI scans.

**Figure 1.** Flow chart of included participants.

**Figure 2.** Magnetic resonance imaging (MRI) scans of a patient with a medial meniscus posterior root tear. (**A**) Pre-operative MRI view with a bone marrow edema lesion on the medial tibial plateau (yellow arrow). (**B**) One-year follow-up MRI view showing no bone marrow edema lesion on the medial tibial plateau.

**Figure 3.** Magnetic resonance imaging (MRI) scans of a patient with a medial meniscus posterior root tear. (**A**) Pre-operative MRI view with no bone marrow edema lesion on the medial femoral condyle. (**B**) One-year follow-up MRI view showing a bone marrow edema lesion on the medial femoral condyle (yellow arrow).

#### *2.3. Surgical Technique for Pullout Fixation*

All surgical procedures were performed by one surgeon (K.S.C.) and involved transtibial pull-out repair using modified Mason–Allen stitches [30]. When MMPRT was confirmed on arthroscopic examination, the superficial medial collateral ligament (sMCL) was released by periosteal stripping on the distal attachment area of the sMCL. This created a sufficient working space depending on the patient's status [31]. A meniscus resector and shaver (Linvatec; Largo, FL, USA) was used to remove the fibrous tissue and obtain fresh meniscal tissue. A curette (Linvatec; Largo, FL, USA) was then inserted through the anteromedial (AM) portal, forming a bone bed at the native root insertion site. A suture hook (Linvatec; Largo, FL, USA) loaded with No. 1 polydioxanone (PDS; Ethicon; Somerville, NJ, USA) was then passed and PDS inserted at a point 5 mm medial to the torn edge in a vertical direction. One more PDS was placed in a position inside that of the first suture, in an identical manner. The shuttle relay method completed the horizontal loop. One or two simple vertical stitches were made which overlayed and crossed the horizontal suture. After making a suitable bone tunnel by anterior cruciate ligament reconstruction tibial tunnel guide (Linvatec; Largo, FL, USA), the ends of the sutures were pulled through the tibial tunnel. Finally, the meniscus was reduced and stabilized by tying the suture over an Endo-button (Smith & Nephew; Andover, MA, USA).

#### *2.4. Postoperative Rehabilitation*

After three weeks of immobilization, range of motion (ROM) exercises were started and progressed to up to 90◦ flexion until 6-weeks after surgery. Toe touch weight-bearing using crutches commenced immediately after surgery, with the brace locked to allow for full extension of the knee joint for the first 3-weeks after surgery. Progressive partial weight-bearing exercises commenced 3-weeks following surgery. Full weight-bearing and progressive closed kinetic chain strengthening exercises were permitted 6-weeks after surgery. Light running was permitted after 3-months and sports participation after 6-months. Permanent lifestyle modifications aimed at avoiding deep knee flexion were recommended

#### *2.5. Clinical Outcomes*

The Lysholm score, Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) and Knee injury and Osteoarthritis Outcome Score (KOOS) were evaluated as the clinical outcomes both prior to and 1-year after surgery.

#### *2.6. Radiological Outcomes*

The Rosenberg 45◦ posteroanterior standing view was used to assess the K-L arthritis grade and measure the medial joint space both prior to surgery and 1-year after surgery [32]. The K-L grade (0/1/2/3/4) was defined as follows: grade 0 indicated no degenerative change. Grade 1 indicated questionable osteophytes and no joint space narrowing. Grade 2 showed definite osteophytes with possible joint space narrowing. Grade 3 showed definite joint space narrowing with moderate multiple osteophytes and some sclerosis. Finally, grade 4 indicated severe joint space narrowing with cysts, osteophytes and sclerosis [33]. The medial joint space was measured from the center of the medial femoral condyle to the center of the medial tibial plateau using a picture-archiving and communication system (PACS, Marotech; Infinitti, Seoul, Korea).

Each MRI scan was checked and the BME lesion, cartilage grade, healing status and meniscal extrusion were evaluated both prior to surgery and 1-year following surgery. Extrusion of the medial meniscus (in mm) was defined as the amount of meniscal displacement from the superomedial aspect of the tibial plateau to the periphery of the meniscal body at the level of the medial collateral ligament (MCL) in the coronal plane.

Radiographic images were examined independently by two authors who were blinded to the procedures used, in consultation with a single experienced musculoskeletal radiologist. All radiographic measurements were documented three times at 2-week intervals

using PACS. The averages of these measurements were used in our analyses. Intraclass correlation coefficients (ICCs) were calculated to determine the interobserver reliability of the differences in radiological outcome measurements. All measurements that allowed 1 decimal value were documented three times at 2-week intervals to assess their test– retest reliability.

#### *2.7. Statistical Analysis*

Statistical analyses were performed using SPSS software (ver. 20.0 for Windows; SPSS Inc., Chicago, IL, USA). Statistical significance was set at 5% (*p* < 0.05). The final clinical outcomes, clinical scores and radiological outcomes for each group were compared. The Kruskal–Wallis test was used to compare variables which were not normally distributed between groups and other non-parametric values. The chi-squared (χ2) test was used to compare categorical data. If more than 20% of the expected frequencies were >5, Fisher's exact test was used.

The statistical power of the study was calculated retrospectively at 79% to compare the rate of progression of the cartilage grade between group 1 (25%) and group 3 (100%) at the 0.05 significance level using Fisher's exact test.

#### **3. Results**

In total, 30 patients were enrolled. The mean age of the patients was 58.7 ± 5.7 years (Table 1). Demographics and clinical characteristics prior to surgery are shown in Table 1. This was similar between all three groups, with no significant differences (*p* > 0.05).


**Table 1.** Pre-operative Demographics and Characteristics of Patients Included.

<sup>a</sup> The point of diagnosis was defined as the time when painful symptoms originally occurred. <sup>b</sup> The values were acquired through posteroanterior 45◦ flexion weight-bearing radiographs. <sup>c</sup> The values were acquired through magnetic resonance imaging. Data are presented as mean ± standard deviation.

The results of subchondral BME lesions after surgery are described in Table 2. A total of 20 patients had no change of BME lesion, 66.7% (group one). There were seven patients with improved BME lesion, 23.3% (group two). Finally, there were three patients with worsened BME lesions, 10% (group three). All ICCs ranged from 0.90 to 0.96, indicating an excellent reliability (ICCs > 0.9) of the current study [34,35]. The outcomes both before and after surgery in each group are shown in Table 3. All clinical scores following surgery were significantly improved in all groups compared with the pre-operative scores (*p* < 0.001).


#### **Table 2.** Bone Marrow Lesions (BME) of Included Participants.

MFC: medial femoral condyle. MTP: medial tibial plateau.

**Table 3.** Pre-operative and Post-operative Clinical and Radiological Outcomes in Each Group.

**BME Lesion (***n* **= 29)**


<sup>a</sup> The values were acquired through posteroanterior 45o flexion weight-bearing radiographs. <sup>b</sup> The values were acquired through magnetic resonance imaging. Data are presented as mean ± standard deviation.

When comparing the post-operative clinical and radiological outcomes between groups, the overall clinical and radiological outcomes showed a tendency for group two to be higher than other groups (Table 4). Clinically, the Lysholm score of group three was significantly lower than those of the other groups. Regarding cartilage status, the number of patients with progression of cartilage grade after surgery was five (25%) in group one, zero (0%) in group two and three (100%) in group three, showing group three to have the worst cartilage grades following surgery.

In terms of meniscal extrusion, there was no significant difference in the results in group one. Meniscal extrusion was significantly reduced in group two, whereas extrusion significantly progressed in group three (Tables 3 and 4). Thus, a decrease in meniscal extrusion can have a positive effect on BME lesions after root repair in MMPRTs.


**Table 4.** Postoperative Clinical and Radiological Outcomes Between Groups a.

<sup>a</sup> Values are expressed as mean ± standard deviation. <sup>b</sup> Kruskal–Wallis test. <sup>c</sup> Fisher exact test. <sup>d</sup> The value was taken from magnetic resonance imaging checked at postoperative 1-year follow-up.

#### **4. Discussion**

The study showed 90% of the included participants to have either no change in BME lesions or an improvement of BME lesions. Only 10% of participants had worsened BME lesions. Patients with worsened BME lesions showed less favorable outcomes than other patients. Patients with improved BME lesions showed more reduced extrusion after surgery, whereas patients with worsened BME lesions showed more progressed extrusion.

BME lesions are known to be associated with the progression of osteoarthritis and a potential risk factor for structural deterioration in knee osteoarthritis [1–3]. BME lesions can fluctuate in size within a short time, and are associated with the progression of arthritic changes and fluctuations of pain in knee OA [1–3]. Thus, non-functioning meniscus by loss of hoop tension can lead to progression of BME lesions on the tibiofemoral joint.

MMPRTs are defined as an avulsion injury or radial tear occurring in the posterior bony attachment [11–13], and they can lead to arthritic changes due to the loss of meniscal hoop tension and load-sharing ability on the tibiofemoral joint [14]. There are presently two surgical options for MMPRT: meniscectomy [26,36] and root repair [22–27]. The 'traditional' gold standard treatment for MMPRT is meniscectomy, which is widely used. However, meniscectomy cannot restore meniscal hoop tension; thus, most meniscectomized cases ultimately progress to degenerative arthritis [36–38]. However, this study shows that better outcomes are seen after meniscal repair compared with partial meniscectomy for MMRTs. This is demonstrated through greater improvements in Lysholm scores, lower rates of progression to knee OA and a lower re-operation rate [39]. In a recently published long-term study comparing meniscectomy and root repair surgery, root repair surgery was more effective than meniscectomy in terms of both clinical outcomes and survival rate. In terms of clinical failure, defined as conversion to TKA, the 10-year survival rate of meniscectomy patients was 44.4%, whereas that of root repair surgery patients was 79.6% [21]. Thus, management with restoring meniscal hoop tension is critical to delay arthritis in MMPRTs. Moreover, it can prevent further progression of both newly formed BME lesions or existing BME lesions in the tibiofemoral joint. However, to our knowledge, there is little existing evidence to clarify the treatment's effects of root repair through the investigation of BME lesion change following surgery.

In the current study, only three patients (10%) showed worsened BME lesions after surgery; thus, it is assumed that in almost all included participants, root repair surgery restored the meniscal hoop tension and load-bearing ability on tibiofemoral joints. This suggests root repair to be a favorable surgical procedure to prevent or delay the progression of osteoarthritis in MMPRTs, an aspect of BME lesions.

In this study, all clinical scores following surgery were significantly improved when compared with the pre-operative scores in all groups, even in worsened BME lesion patients. This suggests that root repair has better clinical outcomes following surgery in MMPRTs patients. Moreover, the overall clinical and radiological outcomes showed a tendency for group two to score higher than groups one and three. However, the Lysholm score was significantly lower in the worsened BME lesion group, group three. Radiologically, regarding cartilage status, the number of patients with a progression of cartilage grade after surgery was five patients (25%) in group one. In group two, this was 0 (0%). However, all patients with worsened BME lesions showed a worsened cartilage status after surgery, suggestive of a link between cartilage injury and BME lesions. In a previously reported study, Outerbridge grade ≥ three chondral lesions were found to be independent negative prognostic factors following root repair [23]. This suggests worsened BME lesions may be negative prognostic factors.

The results of the current study show worsened BME lesions to have a clinically unfavorable effect following root repair surgery. In MMPRTs, it is important to restore well-functioned meniscus by root repair. This study indicates that one of the goals of root repair surgery is to prevent BME lesions as much as feasibly possible. This is the clinical relevance of the study.

To our knowledge, there are no studies that have investigated the correlation between meniscus extrusion and the BME lesions. It would be worthwhile to note that this is the first study to investigate the correlation between meniscal extrusion and BME lesions. In this study, patients with improved BME lesions showed more reduced extrusion after surgery, whereas patients with worsened BME lesion showed more progressed extrusion. This means that reduced meniscal extrusion correlates with more favorable clinical outcomes compared with more progressed extrusion. Thus, more reduced meniscal extrusion after root repair can have a positive effect on BME lesions. Chung et al. reported that patients with a decreased meniscus extrusion at 1-year post-operative have more favorable clinical scores and radiographic findings at midterm follow-ups than those with increased extrusion at 1-year post-operative [40]. Meniscal extrusion has a critical impact on the results; thus, surgeons should reduce meniscus extrusion as much as possible when performing root repair. However, this result might be caused by the small sample size of this study. In the future, a larger cohort study will be needed to identify the role of meniscal extrusion in BME lesions.

This study has several limitations. First, it is a retrospective and non-randomized study; thus, there may be selection bias in the selection of participant enrollment. However, basic demographic and preoperative data were similar between groups. Therefore, any selection bias would not have any significant influence on BME lesions following surgery. Secondly, the present study performed was an in vivo study; thus, it has the potential weaknesses of an in silico study [41]. Thirdly, the sample size was relatively small, and the number of patients with worsened BME lesions was less than those of the other two groups. However, a good statistical power was achieved to compare the rates of progression of cartilage grade between groups one and three. Fourth, the follow-up duration was short at only 1-year. Finally, although the meniscal healing status was assessed by evaluating continuity between the bone bed and meniscus proper, the actual restoration of hoop tension and the healing status of the fixed meniscus was not assessed. This was due to patients not consenting to further arthroscopies.

#### **5. Conclusions**

Patients with worsened BME lesions have poor clinical outcomes compared with other patients. A decrease in meniscal extrusion can have a positive effect on BME lesions after root repair. Surgeons should reduce meniscus extrusion as much as possible when performing root repair. In the future, a larger prospective cohort study will be needed to identify the relevant role of meniscal extrusion in BME lesions.

**Author Contributions:** Conceptualization, K.S.C. and J.K.H.; methodology, K.S.C.; formal analysis, K.S.C.; investigation, J.S.K.; data curation, J.K.H. and J.S.K.; writing—original draft preparation, K.S.C.; writing—review and editing, J.S.K.; supervision, J.K.H. and J.G.K. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work was supported by the National Research Foundation of Korea (NRF) grant funded by the Korea government (MSIT) (No. 2021R1G1A1014579).

**Institutional Review Board Statement:** The study was conducted in accordance with the Declaration of Helsinki and approved by the Institutional Review Board of Inje University Seoul Paik Hospital (PAIK 2022-08-003).

**Informed Consent Statement:** The Institutional Review Board waived the requirements of informed consent as all data were anonymous. Therefore, this study was performed without prior informed consent.

**Data Availability Statement:** The data presented in this study are available on request from the corresponding author.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


### *Review* **Medial Meniscus Posterior Root Tear: How Far Have We Come and What Remains?**

**Hyun-Soo Moon 1,2, Chong-Hyuk Choi 1,3, Min Jung 1,3, Kwangho Chung 1,4, Se-Han Jung 1,5, Yun-Hyeok Kim <sup>3</sup> and Sung-Hwan Kim 1,5,\***


**Abstract:** Medial meniscus posterior root tears (MMRTs), defined as tears or avulsions that occur within 1 cm of the tibial attachment of the medial meniscus posterior root, lead to biomechanically detrimental knee conditions by creating a functionally meniscal-deficient status. Given their biomechanical significance, MMRTs have recently been gaining increasing interest. Accordingly, numerous studies have been conducted on the anatomy, biomechanics, clinical features, diagnosis, and treatment of MMRTs, and extensive knowledge has been accumulated. Although a consensus has not yet been reached on several issues, such as surgical indications, surgical techniques, and rehabilitation protocols, this article aimed to comprehensively review the current knowledge on MMRTs and to introduce the author's treatment strategies.

**Keywords:** meniscus root; medial meniscus posterior root; root tear; medial meniscus posterior root tear; meniscus root repair; transtibial pull-out repair

#### **1. Introduction**

Over the past few decades, perceptions of the clinical importance of menisci have changed [1]. The meniscus was once considered a functionless remnant vestige [2]; however, as a result of extensive research over a long time, it is now accepted as one of the most important structures of the knee joint. The meniscus is responsible for approximately 40–80% of the load transmission in the knee joint [3], and damage to this structure increases the peak contact pressure and decreases the contact area of the articular surface of the knee [4]. In addition, the meniscus plays several vital biomechanical roles, such as shock absorption, joint stabilization, lubrication, and proprioception [5]. Accordingly, a paradigm of treatment for meniscus tears has changed from resection to preservation [1,6].

There are various types of meniscus tears based on their location and patterns, each with different characteristics [7–9]. Among these, a medial meniscus posterior root tear (MMRT) is considered a detrimental injury because it causes conditions similar to those of total meniscectomy [10]. Although an MMRT was not well known until approximately a decade ago, numerous studies conducted with increasing attention to its biomechanical and clinical importance have been reported [11]. MMRTs account for approximately 10% of all types of meniscus tears and 22–28% of medial meniscus tears [12–16]. Given that MMRTs commonly precede advanced knee osteoarthritis [17], the actual incidence

**Citation:** Moon, H.-S.; Choi, C.-H.; Jung, M.; Chung, K.; Jung, S.-H.; Kim, Y.-H.; Kim, S.-H. Medial Meniscus Posterior Root Tear: How Far Have We Come and What Remains? *Medicina* **2023**, *59*, 1181. https://doi.org/ 10.3390/medicina59071181

Academic Editors: Yong In and In Jun Koh

Received: 4 May 2023 Revised: 16 June 2023 Accepted: 19 June 2023 Published: 21 June 2023

**Copyright:** © 2023 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

is considered higher than the reported one. Although much remains to be studied and understood, the multifaceted findings on MMRTs currently form the basis of their treatment strategies. Several review papers have been published on meniscus root tears; however, only a few articles have provided a comprehensive summary of the latest knowledge specifically focusing on MMRTs. Therefore, we aimed to comprehensively review the current knowledge on MMRTs, including their anatomy, biomechanics, clinical presentation, treatment options, and clinical outcomes, and we provide our treatment strategies.

#### **2. Medial Meniscus Posterior Root Tear**

#### *2.1. Anatomy*

The medial meniscus posterior root (MMPR), the tibial attachment of the medial meniscus posterior horn, is a ligamentous structure that anchors the meniscus to the tibia. The structural transition from a fibrocartilaginous meniscus body to a ligament-like structure facilitates the load transmission from the meniscus to the bone [18]. The MMPR is mainly composed of collagen fibers running parallel to the longitudinal axis of the meniscus, which also contributes to resisting the tensile force applied to the meniscus [19]. Owing to these mechanical characteristics, the MMPR stabilizes the meniscus between the femoral condyle and tibial plateau and allows effective load transmission [10]. The MMPR is located approximately 9.6 mm posterior and 0.7 mm lateral from the medial tibial eminence apex, with an area of the attachment site of approximately 30.4 mm<sup>2</sup> [20] (Figure 1A). In addition, the MMPR has diagonally-oriented posterior fibrous expansions called shiny white fibers, which increase the attachment area of the MMPR to the tibial plateau [20,21]. Shiny white fibers are used as anatomical landmarks during surgical procedures, involving the MMPR or posterior cruciate ligament [21,22].

**Figure 1.** (**A**) An illustration showing the medial meniscus posterior root and its surrounding anatomical structures. (**B**) The occurrence of a medial meniscus posterior root tear leads to the loss of hoop tension in the meniscus, resulting in meniscus extrusion. PCL: posterior cruciate ligament; MMPR: medial meniscus posterior root; MTE: medial tibial eminence; MMRT: medial meniscus posterior root tear; ACL: anterior cruciate ligament.

#### *2.2. Definition and Classification*

An MMRT is defined as a tear or avulsion that occurs within 1 cm of the tibial attachment of the MMPR [23] (Figure 1B). This type of meniscal injury can be classified according to the system by LaPrade et al. as follows: (1) type 1, partial stable tear; (2) type 2, complete radial tear; (3) type 3, bucket-handle tear with meniscus root detachment; (4) type 4, complex oblique tear extending into the root attachment; and (5) type 5, avulsion fracture of the root attachment [23]. This concise classification system is clinically beneficial because it provides criteria for distinguishing MMRTs from other types of meniscus tears and helps determine treatment options. Furthermore, this classification system provides potential criteria for predicting the chronicity of an MMRT. Types 1, 2, and 4 are associated

with acute or chronic tears, whereas types 3 and 5 are usually found in acute traumatic tears [23,24]. Among the five types of MMRTs, type 2 is the most common [23,24].

#### **3. Biomechanics**

The MMPR plays an important role in transmitting and distributing the load applied to the meniscus, thereby preserving the adjacent articular cartilage. The hoop tension generated by the collagen fibers that make up the meniscus and its root attachment counteracts the compressive forces applied to the knee joint while maintaining the meniscus between the femur and tibia [5]. A radial tear of these structures disrupts hoop tension. This disruption pushes the meniscus out between the two bones, eventually causing the meniscus to lose its function (Figure 1B). In a cadaveric study, Allaire et al. reported that an MMRT increased the contact pressure in the medial compartment of the knee by 25%, equivalent to that after total medial meniscectomy [10]. Subsequent biomechanical studies have consistently shown that an MMRT increases contact pressure while decreasing the contact area in the medial compartment of the knee [25–27]. On the other hand, the repair of an MMRT restores joint mechanics to a condition similar to that of an intact knee under time-zero conditions [10,25–27]. However, the non-anatomical repair may not restore the native biomechanics of the knee [28]. In addition to loading profiles, MMRTs have been reported to directly affect knee stability by causing rotation and translation of the joint [10].

#### **4. Clinical Presentation**

#### *4.1. Pathophysiology and Clinical Features*

MMRTs are caused by acute traumatic events or degenerative changes [29]. Although there are no clear criteria for differentiating them, degenerative MMRTs are thought to occur during light daily activities or without accompanying acute injuries to other knee structures.

Most traumatic MMRTs occur in association with ligament injuries of the knee. Anterior cruciate ligament injuries can accompany MMRTs [30]. In particular, MMRTs can be observed in cases of multi-ligament injuries [31–33], and varus injury patterns may further increase the possibility of MMRTs [32]. The incidence of traumatic MMRTs associated with ligament injuries of the knee ranges from 3% to 14% [31–33].

Degenerative MMRTs are relatively more common than traumatic MMRTs and are more susceptible to tears than root tears in other locations [29]. The medial meniscus posterior horn carries a large portion of the load applied to the knee joint, including compressive and shear forces [3]. Furthermore, the MMPR shows the lowest mobility of all meniscus roots, making it vulnerable to degenerative tears [34]. Park et al. reported that the degree of MMPR degeneration was highly associated with the degree of an MMRT and that degeneration, accompanied by fibrocartilage metaplasia or calcification, may precede an MMRT [35]. Possible risk factors for these types of MMRTs include older age, female sex, increased body mass index (BMI), increased varus alignment of the lower extremity, and decreased sports activity levels [14]. MMRTs tend to occur during light daily activities, such as squatting, rising from a chair, or going up or down stairs [12,36]. Interestingly, compared with other meniscus tears, MMRTs showed a relatively clear onset of symptoms. The onset of acute symptoms is commonly expressed as painful popping [13], and approximately 85–91% of patients state that they clearly recall the time of symptom onset [37,38]. According to Bae et al., a single painful popping sensation could be a highly predictive clinical sign for identifying MMRTs, with a positive predictive value of 96.5% [13]. In terms of physical examination, several parameters, such as pain on full flexion, joint line tenderness, and the McMurray test, can be used; however, there are no specific physical findings to diagnose MMRTs [12,39]. The Akmese sign, with severe medial joint line tenderness in near extension and minimal or no tenderness in knee hyperflexion, has recently been proposed as a possible physical examination to distinguish MMRTs from other medial meniscus pathologies, with high sensitivity and specificity [40].

#### *4.2. Imaging*

Magnetic resonance imaging (MRI) is the gold standard for diagnosing MMRTs. Although the quality of imaging and the ability of the observer may affect the accuracy of image interpretation, MRI is known to have a high accuracy in identifying MMRTs. According to Lee et al., the sensitivity, specificity, and accuracy of MRI for the detection of MMRTs are 86–90%, 94–95%, and 94%, respectively [41]. MMRTs show characteristic radiographic findings in each plane on MRI, which are as follows: (1) vertical high signal at the MMPR on the coronal image (truncation or cleft sign), (2) vertical high signal at the MMPR on the axial plane, and (3) absence of identifiable MMPR on the sagittal plane, but the meniscus reappears immediately in adjacent images (ghost sign) (Figure 2A–C) [42–44]. In addition, medial meniscus extrusion in the coronal plane, a medial displacement of the medial meniscus with respect to the outer border of the medial tibial plateau, can be used to predict MMRTs (Figure 2D) [45–47]. Choi et al. reported that a medial meniscus extrusion >3 mm could be associated with the occurrence of an MMRT [47]. The comprehensive use of the abovementioned radiographic findings can facilitate the accurate diagnosis of MMRTs [42,43]. In addition, T2-weighted sequences are considered most useful for identifying MMRTs [41].

**Figure 2.** Characteristic magnetic resonance imaging findings, suggestive of a medial meniscus posterior root tear. (**A**) Vertical high signal on the coronal image (truncation or cleft sign; white arrow). (**B**) Vertical high signal on the axial image (white arrow). (**C**) Loss of identifiable medial meniscus posterior root on the sagittal image (ghost sign; white arrow). (**D**) Meniscus extrusion on the coronal image (the length of the white arrow).

Plain radiography or ultrasonography can be used as screening tools for diagnosing MMRTs. Patients with MMRTs reportedly show a relatively decreased medial joint space of the knee on weight-bearing knee radiographs, possibly attributable to osteoarthritis and medial meniscus subluxation [48–50]. The peripheral medial joint space-width ratio, which compares the peripheral medial joint space width between the affected and unaffected knees, may be smaller in patients with MMRTs than in those without MMRTs on standing knee radiographs [49]. Similarly, Kodama et al. found that patients with MMRTs showed decreased medial joint space width and increased distance between the medial tibial eminence and medial femoral condyle on weight-bearing posterior-anterior knee radiographs [50]. In the ultrasound examination, medial meniscus extrusion is used to predict the occurrence of MMRTs. Although medial meniscus extrusion is not a pathognomonic sign of MMRTs, parameters indicating the degree of extrusion and the amount of change can be used to predict these tears. On ultrasonographic evaluation, patients with MMRTs have greater medial meniscus extrusion than those without MMRTs [51]. Furthermore, dynamic medial meniscus extrusion, in accordance with changes in weight-bearing conditions, is reduced in patients with MMRTs [52].

#### **5. Treatment**

Treatment options for MMRTs include non-operative treatment, meniscectomy, surgical repair, and high tibial osteotomy. Although non-operative treatment and partial meniscectomy were the most commonly used treatment options for MMRTs until a decade ago, the frequency of surgical repair has increased as the biomechanical importance of the MMPR has been elucidated [10,25–27]. Surgical repair is not appropriate for some cases of MMRTs, and non-operative treatment and meniscectomy remain available in some situations. However, based on the findings of numerous clinical studies, surgical repair has become the treatment of choice for MMRTs.

#### *5.1. Non-Operative Treatment*

Non-operative treatments include medication, intra-articular injection, and unloader knee braces primarily aimed at relieving the symptoms and also include lifestyle modification and supervised exercise for long-term results [53–56]. These kinds of treatments can be applied to the following patients who are not suitable for surgical treatments: (1) older patients, (2) those with significant comorbidities, (3) those with advanced knee osteoarthritis (Kellgren-Lawrence grade ≥ 3), and (4) those unwilling to comply with a strict postoperative rehabilitation protocol. In certain groups of patients, symptomatic and functional improvements can be achieved with non-operative treatment [53,54]. However, since meniscal pathology remains unresolved in these cases, clinical improvement is mostly limited to the short term, and degenerative changes in the knee joint are reported to progress [53–55,57,58].

#### *5.2. Meniscectomy*

A torn MMRT tissue can cause impingement between the tibiofemoral joints, leading to pain and mechanical symptoms [12]. Meniscectomy can be a treatment option to relieve these symptoms. Arthroscopic partial meniscectomy for MMRTs can lead to clinical improvement and is reported to show a particularly good prognosis in patients with well aligned non-arthritic knees [12,59,60]. However, the mid-to-long-term clinical results of partial meniscectomy for MMRTs are generally poor, and degenerative changes in the knee joint also progress [61,62]. In addition, arthritic changes in the knee after partial meniscectomy for MMRTs have been reported to be more severe than those after non-operative treatment and surgical repair [58,63,64]. According to a recent study by Yang et al., meniscectomy for MMRTs results in a more significant stress concentration in the tibial cartilage compared with that in untreated MMRTs under dynamic loading conditions [65]. Therefore, owing to its limited clinical benefit, meniscectomy should be performed cautiously in selective cases [29].

#### *5.3. Surgical Repair*

MMRTs lead to biomechanically detrimental conditions in the knee by creating a functionally meniscal-deficient status, but surgical repair is reported to normalize it [10,25–27]. According to a cadaveric study by Padalecki et al., the repair of MMRTs could restore the loading profiles of the knee, being indistinguishable from those conditions with an intact meniscus [26]. Although this finding is based on a time-zero study, it provides a rationale for surgical repair of an MMRT. Along with scientific evidence, unsatisfactory clinical outcomes of meniscectomy have led to increasing interest in performing surgical repair for MMRTs [61–64].

#### 5.3.1. Indication and Prognostic Factors

The indications for surgical repair of MMRTs are continuously expanding on the basis of numerous studies. Although a consensus has not yet been reached, patients who are suitable for surgical repair of MMRTs are those who meet most of the following conditions: (1) young and physically active, (2) without severe arthritic changes in the knee, (3) without severe varus malalignment of the lower extremity, (4) without high BMI, and (5) willingness to comply with a strict postoperative rehabilitation program [66–73]. In general, older age, high-grade cartilage lesions of the tibiofemoral joint, and varus alignment of the lower extremities are typical poor prognostic factors for surgical repair of MMRTs [66–69]. Since chronological age differs from biological age, no clear criteria exist for age-related surgical indications. Although 65 years of age is sometimes used as a criterion [38,56,74], several patient characteristics, such as activity level and comorbidities, should be comprehensively considered when deciding on surgery. Arthritic knee changes can be evaluated by plain radiography, MRI, or arthroscopy. It has been reported that radiographic osteoarthritis grade ≥3, according to the Kellgren-Lawrence grade system, and cartilage lesions in the medial tibiofemoral joint of grade ≥3, according to the Outerbridge classification system, are associated with unfavorable prognoses [66–70,75]. Concerning the lower limb alignment, varus >5◦ is known to be a poor prognostic factor [66–69]. However, findings from a recent study suggest that the surgical outcomes of patients with a hip-knee-ankle angle between 5◦ and 10◦ varus were not inferior to those of patients with a hip-knee-ankle angle <5◦ varus [74]. Therefore, rather than taking the angle of the lower limb alignment as an absolute criterion, evaluating whether the varus alignment is due to constitutional varus or pathological causes may also be required. High BMI is also reported to be a risk factor for poor outcomes after the surgical repair of MMRTs [70,71]. As in severe varus alignment of the lower extremities, a high BMI may adversely affect the repaired MMRT by increasing its load [75]. In addition, given the relatively low failure load of suture fixation and the limited healing rate of MMRTs [70,72,73], patients should comply with strict postoperative rehabilitation programs to obtain an optimal healing environment [76]. Furthermore, the preoperative symptom duration was found to affect surgical outcomes [38]. A careful patient selection process that considers the aforementioned factors may lead to relatively favorable surgical outcomes. Further research is required to establish optimal surgical indications for the repair of MMRTs.

#### 5.3.2. Surgical Techniques

The repair methods for MMRTs include two techniques: one uses a suture anchor (suture anchor repair), and the other uses a transosseous tunnel (transtibial pull-out repair). Suture anchor repair is performed by placing the suture anchor on the region of an MMRT above the posterior tibial plateau [36,77–80]. This method has the advantage of reducing the risk of the bungee effect, micromotion between the meniscus-suture complex, and abrasion of the suture material, which may be found in long meniscus-suture constructs, resulting from transtibial pull-out repair [81,82]. However, this surgical method is technically demanding and requires an additional high posteromedial working portal and specialized instruments. It also involves the risk of damaging the cartilage and neurovascular structures [75]. In transtibial pull-out repair, after the suture strands stitched on the torn edge

of an MMRT are pulled out through a transosseous tunnel, created from the footprint of the MMRT, to the tibia outer cortex, fixation is made above the anterior tibial cortex. As mentioned, this surgical method may have potential disadvantages, such as micromotion and suture abrasion [81,82]. However, this method is technically less challenging and has a relatively low risk of damage to the vital structures of the knee joint [75]. It also avoids potential complications caused by the loosening of the suture anchor [29]. Therefore, most surgeons use transtibial pullout repair for MMRTs [83]. Although there are few related studies, a paper reported that there are no differences in surgical outcomes between the two surgical techniques for MMRTs [36]. In addition to these two representative techniques, various additional procedures have been used to minimize meniscal extrusion that may persist after surgical repair, including peripheral release, centralization suture, and whip-running suture [84–87]. However, no well designed clinical study has demonstrated a method that reliably reduces residual meniscus extrusion [88].

Regardless of the surgical method chosen for the repair of MMRTs, it is of utmost importance to ensure that the repair occurs within the anatomical footprint of the MMPR. Unlike anatomical repair, non-anatomical repair cannot restore the loading profile of the knee joint, including the tibiofemoral contact area and pressure [28]. Additionally, complex suture configurations, such as the modified Kessler stitch or modified Mason-Allen stitch, have been reported to show better biomechanical properties than those with simple suture configurations [72,73]. Accordingly, efforts to perform surgical repairs using complex suture configurations at the anatomical footprint with a more familiar surgical technique would be paramount.

#### 5.3.3. Postoperative Rehabilitation

Almost all surgeons apply a rehabilitation program to their patients after the surgical repair of MMRTs. Biomechanically, the load applied to the meniscus generated by weight bearing can affect the repair site [3]. In addition, displacement of the repair construct has been reported after cyclic loading [89]. This displacement may lead to repair loosening and surgical failure. Indeed, the successful healing rate after the surgical repair of MMRTs is approximately 70% [70]. Therefore, well organized postoperative rehabilitation programs are required to provide an optimal healing environment for the repaired MMRT by minimizing the potential effects of harmful stimuli. There are no standardized rehabilitation protocols yet, but most consist of protective strategies, such as limited weight-bearing, range of motion exercises, and brace application [76,90]. The details of the protocols varied greatly among studies. According to a systematic study by Kim et al., range of motion exercises were initiated immediately after surgery or two to three weeks after surgery, and partial weight-bearing was started from one to six weeks after surgery [76]. A knee brace or splint was applied in a fully extended position two to six weeks after surgery [76]. This was followed by progressive muscle-strengthening exercises, and a return to sports is usually allowed six months after surgery [76,90]. Because meniscus-to-bone healing after surgical repair is reported to progress for up to approximately 12 weeks [91], a certain period of protection for the repair construct is necessary to promote healing. However, prolonged protection also carries the risk of developing complications, such as muscle weakness and joint stiffness. Therefore, systematic rehabilitation should be applied under close monitoring, depending on the patient. In addition, accompanying lesions or concomitant surgical procedures should be considered.

#### 5.3.4. Surgical Outcomes

Surgical repair of MMRTs provides both subjective and objective clinical benefits. The literature has consistently reported significant functional improvement, which could persist over the mid- to long-term after surgical repair of MMRTs [83,92,93]. Although surgical repair cannot prevent degenerative changes in the knee joint, the progression of osteoarthritis is less severe compared to that in non-operative treatment or meniscectomy [57,58,94,95]. Additionally, a paper reported that surgical repair is an economically superior treatment

approach compared with other treatment modalities [58]. Therefore, although further multidisciplinary studies are required, surgical repair should be recommended for MMRTs, rather than other treatment methods, unless contraindicated [95].

#### *5.4. High Tibial Osteotomy*

High tibial osteotomy may be a treatment option for MMRTs. Because most MMRTs are degenerative in nature, it is not uncommon for them to be accompanied by cartilage lesions or varus deformities of the lower extremities [96]. As high-grade cartilage lesions in the medial tibiofemoral joint and severe varus malalignment are poor prognostic factors for the surgical repair of MMRTs [66–69], high tibial osteotomy can be an alternative to the surgical repair of MMRTs for patients with medial compartment osteoarthritis in a varus knee by redistributing the load applied to the knee. High tibial osteotomy has been reported to show favorable clinical outcomes for patients regardless of the healing of MMRTs [97,98]. Furthermore, although it would theoretically be ideal to perform concurrent surgical repair of MMRTs during high tibial osteotomy, it has been reported that there are no clinical benefits in combined surgical procedures compared to isolated high tibial osteotomy [99–101].

#### **6. Author's Treatment Strategies for MMRTs**

For patients diagnosed with MMRTs, we attempted to perform surgical repair as much as possible, avoiding meniscectomy. If patients with MMRTs do not meet the indications for surgical repair, non-operative treatment or high tibial osteotomy can be suggested by comprehensively considering not only the underlying knee pathology, but also the patient's characteristics, comorbidities, and willingness for treatment. Our indications for surgical repair of MMRTs are when all of the following are satisfied: (1) not older patients (usually by age 65 years); (2) radiographic osteoarthritis grade ≤2, according to the Kellgren-Lawrence grading system; (3) no pathologic varus alignment of the lower extremity (hip-knee-ankle angle ≤10◦ and <5◦ greater compared to the hip-knee-ankle angle of the contralateral lower extremity); and (4) willingness to comply with a strict postoperative rehabilitation program (Figure 3). For the lower limb alignment, we aimed to avoid excluding patients with MMRTs from surgical candidates simply because the hipknee-ankle angle is >5◦, given that constitutional varus is present in a significant portion of the normal population [102,103]. Indeed, we recently found that the short-term clinical outcomes of surgical repair of MMRTs in patients with mild-to-moderate varus alignment were comparable to those in patients with neutral alignment [74]. Treatment plans for cartilage lesions in the medial tibiofemoral joint, which frequently accompany MMRTs, were determined using MRI and arthroscopic findings during surgery. The severity of cartilage lesions was assessed according to the International Cartilage Repair Society (ICRS) grading system [104]. In general, cartilage restoration procedures were not performed when cartilage lesions corresponded to the ICRS grade ≤3a, but were performed for cartilage lesions of the ICRS grade 3b to 3d (if the calcified cartilage layer was exposed). For patients with ICRS grade 4 cartilage lesions, surgical repair of MMRTs, combined with cartilage restoration procedures, was conducted only if they did not want alternative treatment strategies.

The surgical repair of MMRTs is performed through arthroscopic transtibial pull-out repair using a modified reverse Mason-Allen stitch (Figure 4) [38,74]. First, diagnostic arthroscopy is performed via the parapatellar high anterolateral portal, and the accompanying lesions are thoroughly evaluated. When an MMRT is identified, the frayed portion of the torn edge of the MMRT is gently debrided using an arthroscopic shaver, followed by surgical repair. Before the repair process, the medial joint space width is measured using a 5-mm hook on an arthroscopic probe. If the medial gap is considered narrow enough to perform meniscal procedures, a percutaneous pie-crusting release of the superficial medial collateral ligament is performed using a 19-gauge intravenous catheterization needle. This additional procedure reduces the risk of iatrogenic articular cartilage damage and facilitates the surgical procedure without affecting the surgical outcomes and residual valgus laxity

of the knee [105]. Subsequently, a crescent-shaped suture hook (Conmed Linvatec) or Knee Scorpion Suture Passer (Arthrex) is inserted into the joint through the anteromedial working portal, followed by making the stitch at approximately 3–5 mm medial to the posterior portion of the torn edge of the MMPR. This process is repeated on the anterior portion of the first stitch, and then a modified reverse Mason-Allen stitch with an ultra-high molecular weight polyethylene (UHMWPE) suture is made (Figure 5) [38,74,106]. We expect this horizontal loop stitch to increase the contact area of the meniscus-to-bone interface by pulling the meniscus, including its peripheral margin, from above. Additionally, using a No. 1 polydioxanone suture (PDS; Ethicon), an overlaid vertical stitch crossing the horizontal loop is created to produce a locking effect [107]. After suturing, an additional anteromedial portal is created. Subsequently, a posteromedial portal is created using the field of view provided by the arthroscope inserted through an additional anteromedial portal. The anatomical footprint of the MMPR is identified under observation through the posteromedial portal, and the bone bed of the footprint is decorticated using an arthroscopic rasp or curette to promote meniscus-to-bone healing. A tibial tunnel guide (Commed Linvatec) is then inserted into the joint through an additional anteromedial portal and placed over the footprint of the MMPR with reference to the remnant stump and posterior cruciate ligament. A tibial transosseous tunnel is created, and a wire loop is inserted into the joint through the tibial tunnel. Next, the sutures stitched to the meniscus are passed through the wire loop and pulled out through the tibial tunnel by pulling the wire loop. Finally, sutures are tied over the EndoButton (Smith & Nephew), which is placed on the anterior tibial cortex just above the pes anserinus. With the knee fully extended, the PDS suture is tied first using a sliding knot while maintaining adequate tension by pulling the UHMWPE suture, and the UHMWPE suture is tied.

**Figure 3.** Flowchart of authors' treatment strategies for medial meniscus posterior root tear. K-L: Kellgren-Lawrence.

**Figure 4.** The process of arthroscopic transtibial pull-out repair of a medial meniscus posterior root tear (MMRT). (**A**,**B**) MMRT is confirmed, and a modified reverse Mason-Allen stitch is made (view from the parapatellar high anterolateral portal). (**C**,**D**) After identification of the anatomical footprint of the medial meniscus posterior root and decorticating the bone bed, a transosseous tibial tunnel is created, and a wire loop is inserted into the joint (view from the posteromedial portal). (**E**) Suture strands are pulled out through a tibial tunnel (view from the posteromedial portal) and (**F**) tied (view from the parapatellar high anterolateral portal). Reprinted with permission from Ref. [74]. Copyright 2021, copyright with permission from Moon and Kim.

Postoperatively, the patients are instructed to adhere to a strict rehabilitation program to ensure optimal conditions for MMRT healing. With the use of crutches and a hinged knee brace in the fully extended position, patients are restricted to toe-touch weight-bearing for four weeks postoperatively, followed by partial weight-bearing (<50% of their body weight) for six weeks. The crutches and brace are discontinued 10 weeks after surgery, and the patients are allowed to have a full weight-bearing gait. Passive range of motion exercises are initiated two weeks postoperatively and gradually increased. Similar to weight bearing, active range of motion exercises are allowed from 10 weeks postoperatively. For

muscle strengthening, closed kinetic chain exercises are initiated 10 weeks postoperatively, gradually increasing the exercise intensity [108]. Deep knee flexion should be avoided during daily activities and exercise.

**Figure 5.** Pictorial illustration of the transtibial pull-out repair of a medial meniscus posterior root tear using a modified reverse Mason-Allen stitch. The stitch consists of the horizontal loop stitch (red thread; ultra-high molecular weight polyethylene suture [Tigerwire suture]) and the overlaid simple vertical stitch (blue thread; No. 1 PDS). Reprinted with permission from Ref. [38]. Copyright 2020, copyright with permission from Moon and Kim.

#### **7. Summary and Future Direction**

MMPR plays a vital biomechanical role in transmitting and distributing the load applied to the meniscus, thereby reducing the impact on the adjacent articular cartilage and preserving the knee joint. Damage to this structure can lead to functional meniscal deficiency, followed by a poor clinical course and rapid progression of osteoarthritis. Most MMRTs are degenerative in nature and usually occur during light daily activities with clinical symptoms, such as painful popping. MMRTs are usually diagnosed based on several characteristic MRI findings. Although there are several treatment options for MMRTs, surgical repair should be prioritized as much as possible to restore the biomechanical function of MMPR, which is supported by the favorable clinical outcomes reported in several clinical studies. However, consensus has yet to be reached regarding the optimal surgical indications, surgical techniques, and rehabilitation protocols for MMRTs. Addressing these issues necessitates further multidisciplinary research. Furthermore, unresolved challenges persist, such as residual meniscus extrusion, the limited healing rate of MMRTs, and the progression of osteoarthritis [38,70,109,110]. Tissue engineering and regenerative medicine may offer potential solutions for these matters. In particular, cell-based tissue regeneration, utilizing various cell sources, is particularly promising as a novel strategy to improve meniscus repair [111,112]. Several aspects in this field also need to be addressed, such as the recruitment and adhesion of cells to the site of meniscus tears, their differentiation into the appropriate cell phenotype, proliferation, and manipulation methods for clinical application [111]. Nonetheless, considering their reparative potential and mounting preclinical evidence, it is expected that most of these issues will be resolved in the near future. Collaboration and joint research with these innovative fields are anticipated to potentially delay the progression of osteoarthritis by overcoming several existing limitations and optimizing the treatment strategies for MMRTs.

**Author Contributions:** Conceptualization, H.-S.M. and S.-H.K.; resources, H.-S.M., C.-H.C., M.J. and S.-H.K.; data curation, H.-S.M., K.C., S.-H.J. and Y.-H.K.; writing—original draft preparation, H.-S.M.; writing—review and editing, C.-H.C., M.J., K.C., S.-H.J. and S.-H.K.; supervision, S.-H.K.; project administration, S.-H.K.; funding acquisition, S.-H.K. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was supported by a faculty research grant from the Yonsei University College of Medicine (6-2021-0230).

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** Not applicable.

**Acknowledgments:** MID (Medical Illustration & Design), as a member of the Medical Research Support Services of Yonsei University College of Medicine, providing excellent support with medical illustration.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


**Disclaimer/Publisher's Note:** The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

**Tae Woo Kim \* and June Seok Won**

Department of Orthopedic Surgery, Seoul National University College of Medicine, Seoul National University Boramae Medical Center, Seoul 07061, Republic of Korea; chris9906@naver.com

**\*** Correspondence: orthopassion@naver.com

**Abstract:** *Background:* Accurate pre-operative planning is essential for successful high tibial osteotomy (HTO). The lateral tibial spine is a commonly used anatomical landmark for weight-bearing line assessment. However, studies on the mediolateral (M-L) position of the lateral tibial spine on the tibial plateau and its variability are limited. *Purpose:* This study aimed to (1) analyze the M-L position of the lateral tibial spine on the tibial plateau and its variability, (2) investigate radiologic parameters that affect the position of the lateral tibial spine, and (3) determine whether the lateral tibial spine can be a useful anatomical landmark for weight-bearing line assessment during HTO. *Materials and Methods:* Radiological evaluation was performed on 200 participants (64% female, mean age 42.3 ± 13.2 years) who had standing anterior–posterior plain knee radiographs with a patellar facing forward orientation. The distances from the medial border of the tibial plateau to the lateral spine peak (dLSP) and lateral spine inflection point (dLSI) were measured using a picture archiving and communication system. The medial–lateral inter-spine distance (dISP) was also measured. All parameters were presented as percentages of the entire tibial plateau width. The relationships between the parameters were also investigated. *Results:* The mean value of dLSP was 56.9 ± 2.5 (52.4–64.5)%, which was 5% lower than the Fujisawa point (62%). The mean value of dLSI was 67.9 ± 2.2 (63.4–75.8)%, which was approximately 5% higher than the Fujisawa point. The values of the dLSP and dLSI were variable among patients, and the upper and lower 10% groups showed significantly higher and lower dLSP and dLSI, respectively, than the middle 10% group. The mean value of dISP was 16.5 ± 2.4%, and it was positively correlated with dLSP and dLSI. *Conclusions:* On average, the dLSP and dLSI were located −5% and +5% laterally from the conventional Fujisawa point, and they may be useful landmarks for correction amount adjustment during HTO. However, it should be noted that correction based on the lateral tibial spine can be affected by anatomical variations, especially in patients with small or large inter-spine distances.

**Keywords:** high tibial osteotomy; planning; landmark; lateral tibial spine; anatomy

#### **1. Introduction**

High tibial osteotomy (HTO) is an efficient surgical option for treating middle-aged patients with medial knee osteoarthritis (OA) and varus alignment [1–5]. Accurate alignment correction is essential for successful HTO in patients with OA [6–11]. Under-correction can result in residual knee pain and shorten the longevity of the effects of this procedure [12]. Over-correction can induce lateral and patellofemoral arthritis and cause dissatisfaction associated with excessive valgus deformity [13–15].

To reduce the correction error during HTO, pre-operative planning based on weightbearing line assessment plays an important role, and the first step in HTO pre-operative planning is the determination of the appropriate target point on the tibial plateau [16–21]. Since it was first documented in 1979, the Fujisawa point (62–63%) has been commonly used as an ideal realignment target during HTO, and the outcome of HTO has also been evaluated

**Citation:** Kim, T.W.; Won, J.S. Anatomical Study of the Lateral Tibial Spine as a Landmark for Weight Bearing Line Assessment during High Tibial Osteotomy. *Medicina* **2023**, *59*, 1571. https:// doi.org/10.3390/medicina59091571

Academic Editors: Yong In and In Jun Koh

Received: 21 July 2023 Revised: 25 August 2023 Accepted: 26 August 2023 Published: 29 August 2023

**Copyright:** © 2023 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

using this reference point [22–24]. However, some authors recommend adjusting the correction target depending on the severity of osteoarthritis and the reason for surgery [2,25].

The lateral tibial spine can also be used as an anatomical landmark for weight-bearing line assessment during HTO. Lee et al. reported that medial opening-wedge HTOs aimed at the lateral tibial spine showed comparable clinical outcomes to those of HTOs targeting the conventional Fujisawa point [26]. The lateral tibial spine is a bony eminence on the tibial plateau to which the anterior cruciate ligament attaches, which varies anatomically between individuals [27,28]. However, it is unclear whether the lateral tibial spine can be a constant anatomical landmark for HTO in the mediolateral dimension of the tibial plateau.

Therefore, this study aimed to (1) analyze the M-L position of the lateral tibial spine on the tibial plateau and its variability, (2) investigate the radiological parameters that affect the position of the lateral tibial spine, and (3) determine whether the lateral tibial spine can be a useful anatomical landmark for weight-bearing line assessment during HTO. We hypothesized that the lateral tibial spine can be a useful anatomical landmark for HTO and that it can also show anatomical variation correlated with specific radiologic parameters.

#### **2. Materials and Methods**

#### *2.1. Study Population*

A total of 517 consecutive standing knee anteroposterior (AP) radiographs of 517 patients diagnosed with medial compartment knee osteoarthritis at Seoul National University Boramae Medical Center between January 2019 and March 2020 were retrospectively reviewed. Of these, 317 radiographs were excluded, and the remaining 200 knee AP radiographs were examined. The exclusion criteria were as follows: (1) Kellgren–Lawrence grade IV with a large osteophyte; (2) bone deformity of the knee due to previous surgery or trauma; (3) rheumatoid arthritis; (4) infection; and (5) absence of a standing knee AP radiograph with proper lower extremity rotation (i.e., with the patella facing forward). In patients with bilateral knee OA, only the side with the more progressed knee OA was evaluated. Patient demographic data, including age and sex, and data on laterality were also collected. This study was approved by the Institutional Review Board (IRB No.20-2021-22), and the requirement for informed consent was waived due to the retrospective study design.

#### *2.2. Radiologic Assessment*

Standing knee AP radiographs were obtained using a UT 2000 X-ray machine (Philips Research, Eindhoven, The Netherlands) set at 90 Kv and 50 mA. Knee rotation was controlled using our standard protocol to locate the patella at the center of the femoral condyle during the examination. All radiographs were digitally acquired using a picture archiving and communication system (PACS), and radiographic assessments were performed using the PACS (INFINITT, Seoul, Republic of Korea). The standard knee AP view was marked with a digital ruler, and each value was calibrated automatically on the PACS system.

The peak and inflection points of the lateral tibial spine were determined as the most prominent points and the intersection point between the tibial spine slope and the tibial plateau, respectively. The distance between the medial tibial border and lateral spine peak point (dLSP), the distance between the medial tibial border and lateral spine inflection point (dLSI), and the mediolateral inter-spine distance (dISP) were measured and presented as a percentage of the entire tibial plateau width (Figure 1).

For the interobserver reliability test, two orthopedic surgeons independently measured the radiographic parameters. Each observer was blinded to the measurements of the other observer. For the intra-observer reliability test, each observer measured the radiological parameters twice at 8-week intervals.

**Figure 1.** Radiographic measurements of the lateral tibial spine. Peak point: the most prominent point of the lateral tibial spine, inflection point: the intersection point between the tibial spine slope and tibial plateau, dLSP: distance between the tibial medial border and lateral spine peak point, dLSI: distance between the tibial medial border and lateral spine inflection point; dISP: medio-lateral inter-spine distance.

#### *2.3. Analysis of Radiologic Parameters*

To quantitatively analyze the location of the lateral tibial spine on the tibial plateau, the dLSP and dLSI were measured. To evaluate the variability in each parameter statistically, the upper 10% and lower 10% groups were compared with the middle 10% group, respectively. To determine whether other radiologic parameters could predict the amount of dLSP or dLSI, correlations between dLSP, dLSI, tibial plateau M-L width, and dISP were analyzed.

#### *2.4. Statistical Analysis*

All statistical analyses were performed using SPSS version 25.0 (IBM, Armonk, NY, USA). An a priori power analysis was performed with the assumption that a 2.0% M-L dimension was radiologically significant, with a standard deviation of 2.4% mm. It was found that 20 knees in each group were sufficient for all parameters (*α* = 0.05, *β* = 0.8). *p*-values of <0.05 were considered statistically significant. All data are presented as mean and standard deviation. Student's *t*-test was used for intergroup comparisons. Correlation analyses between radiological parameters were performed using Pearson's correlation coefficients. Inter- and intra-observer reliabilities were evaluated using the intra-class correlation coefficient (ICC).

#### **3. Results**

In total, 200 standing AP knee radiographs (115 right and 85 left) from 200 patients (92 males, 108 females) were evaluated. The mean age of the patients was 56.2 ± 7.2 years, and the severity of knee OA was K-L grades I, II, and III in 14, 76, and 110 patients, respectively (Table 1). For all radiological parameters, the inter- and intra-observer ICC were >0.8 (Table 2).


**Table 1.** Patient demographics and measurements of radiologic parameters.

K-L, Kellgren–Lawrence; dLSP, the distance between the tibial medial border and lateral spine peak point; dLSI, the distance between the tibial medial border and lateral spine inflection point; dISP, medio-lateral inter-spine distance.

**Table 2.** Intra-observer and inter-observer reliability of radiographic measurements.


dLSP, the distance between the tibial medial border and lateral spine peak point; dLSI, the distance between the tibial medial border and lateral spine inflection point; dISP, medio-lateral inter-spine distance.

In the quantitative analysis of the lateral tibial spine position, the mean values of dLSP, dLSI, and dISP were 56.9 ± 2.5%, 67.9 ± 2.2%, and 16.5 ± 2.4%, respectively. The peak point and inflection point of the lateral tibial spine were located approximately 5% lower and 5% higher, respectively, than the conventional Fujisawa point (62–63%). However, both dLSP and dLSI varied broadly between individuals. The maximum and minimum values of dLSP were 64.5% and 52.4%, respectively, and the maximum and minimum values of dLSI were 75.8% and 63.4%, respectively. The upper 10% group showed significantly higher dLSP and dLSI than the middle 10% group (*p* = 0.001, *p* = 0.002); the lower 10% group also showed significantly lower dLSP and dLSI than the middle 10% group (*p* = 0.001, *p* = 0.001) (Figures 2 and 3).

In the correlation analysis, the dISP showed significant positive correlations with the dLSP (r = 0.787, *p* = 0.001) and dLSI (r = 0.756, *p* = 0.001) (Figure 4). In patients with larger inter-spine distances, the peak and inflection points of the lateral tibial spine were located more laterally. Similarly, in patients with a narrow inter-spine distance, the peak and inflection points of the lateral tibial spine were located more medially. However, the tibial plateau M-L width, sex, and laterality did not affect the dLSP or dLSI (*p* > 0.05) (Table 3).

**Figure 2.** The mean and maximum/minimum values for dLSP and its variation. Red dot: peak point of lateral tibial spine, red dotted line and red font: mean value of dLSP, red frame: lower, middle, and upper 10% groups.

The mean value of dLSP was 56.9 ± 2.5%, and the maximum and minimum values of dLSP were 64.5% and 52.4%, respectively. The upper 10% and lower 10% groups showed significantly higher and lower dLSP than the middle 10% group, respectively.

The mean value of dLSI was 67.9 ± 2.2%, and the maximum and minimum values of dLSI were 75.8% and 63.4%, respectively. The upper 10% and lower 10% groups showed significantly higher and lower dLSP than the middle 10% group, respectively.

In the correlation analysis, dISP showed a significantly positive correlation with dLSP (r = 0.787, *p* < 0.05) and dLSI (r = 0.756, *p* < 0.05). In patients with large and small dISP, dLSP and dLSI also showed increasing and decreasing tendency, respectively.

**Figure 3.** The mean and maximum/minimum values of dLSI and its variation. Yellow dot: inflection point of lateral tibial spine, red dotted line and red font: mean value of dLSI, red frame: lower, middle, and upper 10% groups.

**Table 3.** Comparison of radiologic parameters depending on gender and laterality.


**Figure 4.** Correlation analysis between radiologic parameters. Yellow dot: inflection point of lateral tibial spine, red dot: peak point of lateral tibial spine, yellow arrow: dISP, shorter white arrow: dLSP, longer white arrow: dLSI.

#### **4. Discussion**

The principal finding of this study was that the lateral tibial spine could be a useful anatomical landmark for weight-bearing line assessment during HTO. The mean values of the peak and inflection points of the lateral tibial spine were located approximately −5% and +5% laterally from the conventional Fujisawa point, and it could be easily used for adjustment of the amount of correction. However, the values of the dLSP and dLSI varied among patients, and dISP correlated with dLSP and dLSI. Therefore, the hypothesis of this study was fully supported.

Previous studies on HTO that used the lateral tibial spine as a landmark for alignment correction reported results similar to those of the present study. Lee et al.'s study that compared HTOs aiming at either the peak of the lateral tibial spine or the Fujisawa point showed that the lateral spine peak point was located at 57.4 ± 1.6% from the medial border of the tibia [26]. Jiang et al.'s study also showed that the mean WBL percentage of the top of the tibial spine was 57.7 ± 2.1% [29]. However, the inflection point of the lateral tibial spine in our study was quite different from that of Jiang et al. (67.9 ± 2.2% vs. 74.6 ± 3.3%). In this study, the lateral end of the lateral tibial spine was defined as the inflection point where the tibial spine slope and tibial plateau intersect. On the other hand, in Jiang et al.'s study, the bottom point of the lateral slope was defined as the intersection point between the lateral slope of the lateral tibial spine and the joint line of the tibia. Since the lateral tibial slope and tibial plateau intersect in a round shape rather than an acute angle, the bottom point is located further to the lateral side than the inflection point. This discrepancy can be attributed to the different measurement methods of the intersection point between the tibial spine slope and the tibial plateau.

Quantitative analysis of the lateral tibial spine can be helpful in adjusting the amount of correction during HTO. Jakob et al. suggested that correction of the mechanical axis should differ based on the thickness of cartilage loss in the medial compartment (1/3 cartilage loss: WBL 55–57.5%, 2/3 cartilage loss: WBL 60–62.5%, full thickness cartilage loss: WBL 65–67.5%) [25]. Based on our results, the location of the peak, mid, and inflection points of the lateral tibial spine correspond well with Jakob et al.'s suggestion, depending on different arthroses of the medial compartment. Recent studies have also shown that HTO targeting the peak point of the lateral tibial spine showed clinical outcomes comparable to those of HTO targeting the conventional Fujisawa point [26]. Adjustment of the correction angle may also be necessary depending on the purpose of HTO. When anterior cruciate ligament (ACL) or posterior cruciate ligament (PCL) reconstruction fails in patients with varus deformity, neutral or slight varus alignment is usually targeted during HTO. Even during protective realignment surgery combined with cartilage regeneration procedures, the WBL target is a little more medial than the conventional Fujisawa point. In these cases, the peak point of the lateral tibial spine can be a more useful anatomical landmark than the conventional Fujisawa point.

Another remarkable finding of our study was that the dLSP and dLSI varied among patients. The discrepancy between maximum and minimum values in dLSP, and dLSI were more than 10%, respectively. Also, the upper 10% group showed significantly higher dLSP and dLSI than the middle 10% group; the lower 10% group also showed significantly lower dLSP and dLSI than the middle 10% group. These results can be explained by anatomical variation in the location and size of the lateral tibial spine. Anatomical studies that indicate variable tibial footprint sizes of the ACL among individuals may also indirectly support the variability in the inter-spine distance [27,28].

In the correlation analysis, dLSP and dLSI were significantly correlated with interspine distance in this study. Although the dLSP and dLSI were located −5%, and +5% laterally from the conventional Fujisawa point on average, anatomical variation can be an obstacle that reduces the accuracy and efficiency of dLSP or dLSI as an anatomical landmark for HTO. The inter-spine distance can be easily used to predict large or small dLSP or dLSI and to avoid over- or under-correction without direct measurement of each point. To the best of our knowledge, this is the first study to document anatomical variation in the lateral tibial spine and the radiologic predictor for anatomical variance in the lateral tibial spine.

This study had some limitations. First, this study investigated osteoarthritic knees in an Asian population, and the results may be different from those of a Western population. However, the location of the peak and inflection points in the lateral tibial spine was presented as a percentage of the entire tibial plateau width in this study, which may remove the effect of individual size differences in the tibial plateau and tibial spine. Further anatomical studies in different races are necessary to generalize the results of this study. Second, in contrast to the peak point method, the method used to define the intersection point between the lateral tibial spine and the tibial plateau can differ between examiners. The different values of the intersection point in the lateral tibial spine and tibial plateau between Jiang et al.'s study (bottom point) and this study (inflection point) may be related to different measurement methods, and more effort to reach a consensus is required. Third, in this study, the correlation between dLSP and dLSI and diverse radiologic parameters was not investigated. The femoral notch width, which is known to be related to the size of the ACL footprint, can also be a predictor for dLSP and dLSI. Thus, it is an interesting topic for future study. Fourth, this study was conducted using a standing knee AP image, which may be different from the actual situation in which HTO planning is performed in the long leg view. However, in our hospital, a patient's posture is identical between the standing knee AP view and the long leg view, and we believe that radiologic parameters related to the lateral tibial spine may not be different between the standing knee AP view and the long leg view.

#### **5. Conclusions**

On average, the dLSP and dLSI were located −5%, and +5% laterally from the conventional Fujisawa point, and they may be useful landmarks for correction amount adjustment during HTO. However, it should be noted that correction based on the lateral tibial spine can be affected by anatomical variations, especially in patients with small or large inter-spine distances.

**Author Contributions:** Conceptualization, methodology, writing-review and editing, supervision, funding acquisition: T.W.K., Data collection, data analysis, writing-original draft preparation, revision: J.S.W. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by a grant No. HI22C1879 from the Korea Health Industry Development Institution.

**Institutional Review Board Statement:** This study was conducted in accordance with the Declaration of Helsinki, and approved by the Institutional Review Board of Seoul National University Boramae Medical Center (IRB No.20-2021-22).

**Informed Consent Statement:** Informed consent was waived due to the retrospective study design.

**Data Availability Statement:** The data presented in this study are available on request from the corresponding author. The data are not publicly available due to privacy.

**Conflicts of Interest:** The authors declare that they have no conflict of interest.

#### **References**


**Disclaimer/Publisher's Note:** The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

### *Article* **Comparison of Anatomical Conformity between TomoFix Anatomical Plate and TomoFix Conventional Plate in Open-Wedge High Tibial Osteotomy**

**Sung-Sahn Lee 1, Jaesung Park <sup>2</sup> and Dae-Hee Lee 2,\***


**Abstract:** *Background and Objectives*: The TomoFix anatomical plate was developed to improve plate position, proximal screw direction, and post-correction tibial contouring. The purpose of this study was to compare postoperative configurations between the TomoFix anatomical plate and the TomoFix conventional plate. It was hypothesized that the new modified plate provides a better fixative coaptation than the conventional plate. *Materials and Methods*: A total of 116 cases (112 patients) were enrolled in this study from March 2015 to February 2021. Among them, 63 patients underwent surgery using the TomoFix conventional plate, and 53 underwent surgery using the TomoFix anatomical plate. The radiographic outcomes, including the hip–knee–ankle (HKA) angle, medial proximal tibial angle (MPTA), tibial slope, plate angle, proximal screw angles, and plate-to-cortex distance at #1 hole (just below the osteotomy site) were compared between the two groups. *Results*: Patients with the TomoFix anatomical plate showed similar results in terms of the pre- and postoperative HKA angle, MPTA, and tibial slope. The TomoFix anatomical group showed a significantly greater plate angle (39.2◦ ± 8.1◦ vs. 31.7◦ ± 7.0◦, *p* < 0.001) and less screw angles, indicating that the TomoFix anatomical plates allowed a more posterior plate position than the conventional plate. The plate-to-cortex distance was significantly less in the TomoFix anatomical group than in the TomoFix conventional group (*p* < 0.001). *Conclusion*: The TomoFix anatomical plate showed a more posteromedial plating position, better proximal screw direction to the lateral hinge, and improved post-correction tibial contour compared to the TomoFix conventional plate.

**Keywords:** high tibial osteotomy; TomoFix; plate position; anatomical conformity

#### **1. Introduction**

High tibial osteotomy is a well-established surgical procedure for the management of medial unicompartmental arthritis of the knee joint [1–4]. A lateral closed-wedge high tibial osteotomy has potential disadvantages, including difficulty in intraoperatively controlling the amount of correction and concerns of neurologic deficits [5,6]. Due to these disadvantages, open-wedge high tibial osteotomy (OWHTO) is becoming popular [7].

The mechanical stability of the osteotomy site is a major issue following OWHTO. Correction loss and non-union of the osteotomy site have been observed after surgery [8–10]. However, with the introduction of the TomoFix plate (Synthes GmbH, Oberdorf, Switzerland), which is characterized by a T-shaped, long, rigid-angle stable implant, the incidence of non-union following OWHTO has declined significantly [8,9].

Plate position and proximal screw angles are important for fixation stability in OWHTO [11]. Posteromedial plating provides better fixation stability than anteromedial plating [11,12]. The proximal screw in the direction of the lateral hinge rather than the posterior tibial cortex is associated with increased stability and less neurovascular

**Citation:** Lee, S.-S.; Park, J.; Lee, D.-H. Comparison of Anatomical Conformity between TomoFix Anatomical Plate and TomoFix Conventional Plate in Open-Wedge High Tibial Osteotomy. *Medicina* **2022**, *58*, 1045. https://doi.org/ 10.3390/medicina58081045

Academic Editor: Erich Rutz

Received: 24 June 2022 Accepted: 1 August 2022 Published: 3 August 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

injury [12]. Anatomical contouring to the post-correction medial tibial geometry is also important for decreasing the distance between the plate and cortex [13,14]. The TomoFix anatomical plate was developed to improve plate position, proximal screw direction, and post-correction tibial contouring (Figure 1).

**Figure 1.** (**A**) Schematic diagram of the new plate lateral view. While the previous plate was a 90◦ T-shaped plate with no left and right distinction, the TomoFix anatomical plate has a 5◦ posterior slope, allowing the plate to be positioned relatively backward. As a result, the new plate system (**B**) enables parallel screw fixation than the previous plate (**C**).

Therefore, the purpose of the current study is to compare postoperative configurations between the new modified plate and the conventional plate (the TomoFix Anatomical plate versus the TomoFix conventional plate). It was hypothesized that the new modified plate would provide a better anatomical fitness than the conventional plate.

#### **2. Materials and Methods**

This study enrolled patients with primary medial osteoarthritis who underwent primary OWHTO between March 2015 and February 2021. The included subjects were patients aged <65 years who had osteoarthritis with persistent pain despite conservative management for more than three months. Patients were excluded if they had severe osteoarthritis of the patellofemoral joint or lateral compartment, ligament laxity, rheumatoid arthritis, or flexion contracture > 15◦, and required an angle of correction > 20◦. A total of 116 cases (112 patients) were enrolled in the current study. Among them, 63 patients underwent surgery using the conventional TomoFix plate and 53 using the TomoFix Anatomical plate. For patients who underwent surgery between March 2015 and July 2018, OWHTO was performed using the TomoFix conventional plate, after which surgery was performed using the TomoFix anatomical plate. Both plates were made of titanium. The patient demographics and preoperative data are summarized in Table 1. This study was approved by the Ethics Committee of our institution (SMC2021-05-051 at 26 May 2021), and written informed consent was obtained from all patients.

**Table 1.** Demographic and preoperative data.


BMI, body mass index; HKA, hip-knee-ankle; MPTA, medial proximal tibial angle.

#### *2.1. Surgical Technique*

All surgeries were performed by the senior author. The correction angle was measured using Miniaci's method [15,16]. The alignment of the lower limb was adjusted by passing the point located at 62.5% of the tibial plateau width when measured from the edge of the proximal tibial medial plateau [17]. Biplanar osteotomy was performed in all knees. After skin incision, a guide wire was inserted with visual assistance by an image intensifier. The oblique part of the osteotomy was performed with a saw for a distance of up to 1 cm from the lateral cortex. The osteotomy of the vertical part was performed at the posterior aspect of the tibial tubercle, thus not violating the bony portion to which the patellar tendon was attached, leaving most of the tendon attached to the distal tibial fragment. The anterior gap of the osteotomy sited behind the tibial tuberosity was intended to be approximately 1/2 to 2/3 of the posterior opening gap at the posteromedial corner of the proximal tibia to maintain tibial slope. Target alignment was achieved under intraoperative fluoroscopy, and the osteotomy was stabilized using the TomoFix conventional plate or the TomoFix anatomical plate. A cortical screw was used on the most proximal hole of the distal holes (the so-called #1 hole), if the bone-to-plate distance was more than 5 mm, to avoid fixative failure [18]. An allogenic bone graft (Junyoung Medical, Seoul, South Korea) was inserted into the osteotomy gap to minimize postoperative loss of correction.

The patients were started on isometric quadriceps, active ankle, and straight legraising exercises on the day after surgery. Knee motion exercises were initiated on the second postoperative day. Patients were restricted toe toe-touch weight bearing for the first 2 postoperative weeks, followed by partial weight bearing for next 4–6 weeks. Full weight-bearing was permitted by 6–8 weeks.

#### *2.2. Radiographic Evaluation*

Plain radiographs were obtained before surgery and three months postoperatively. The radiographs included whole leg standing radiographs (patella facing forward and full knee extension) and lateral views with 30◦ flexion. The change in limb alignment from before to after surgery was determined by measuring the hip–knee–ankle (HKA) angle on full-length standing radiographs, using a picture archiving and communication system (Centricity; General Electric, Chicago, IL, USA) [19,20]. The HKA angle was measured by the angle subtended by a line drawn from the center of the femoral head to the center of the knee and a line drawn from the center of the knee to the center of the talus. The medial proximal tibial angle (MPTA) was measured as the angle between the proximal anatomical axis of the tibia and the tangent along the articular surface of the tibial plateau on full-length standing radiographs [21]. The angle of the tibial slope was defined using the proximal tibial anatomical axis method [19]. The MPTA and tibial slope were also compared before and after surgery and between both plate systems.

Computed tomography (CT) scans (MDCT; Brilliance 64, Phillips, Cleveland, OH, USA) were performed on the fifth postoperative day. The proximal plate position was evaluated by the proximal plate angle on the axial CT images [14]. The proximal plate angle was defined as the angle between the posterior cortex line and the line connecting the front and back of the plate proximal area (Figure 2A). The most proximal screw (A screw, B screw, and C screw, each from front to back, respectively, Figure 2B) angles were also measured on axial CT images. Each screw angle was defined as the angle between the posterior cortex and the line drawn along the screw. The plate-to-cortex (just below the osteotomy site) distance at #1 hole was measured, and the usage of cortical screw on #1 hole was counted to evaluate the post-correction tibial contouring of the plate (Figure 3).

**Figure 2.** Measurement of the plate angle and screw angles on the CT axial scan. (**A**) The plate angle was defined as the angle between the anteroposterior line connecting the proximal part of the plate and the posterior tibial condylar line. (**B**) The A, B, and C screw angle was defined as the angle between the line drawn along the screw and the posterior tibial condylar line. (A, B, and C screws from anterior to posterior, respectively).

**Figure 3.** Measurement of the plate-to-cortex distance at #1 hole. It was defined as the distance between the inner surface of the plate and the outer surface of the tibial cortex at #1 hole.

All measurements were performed by two orthopedic surgeons with significant experience in OWHTO, but who were not associated with the subject of this study to assess interobserver reliability, and were repeated six weeks later to assess intra-observer reliability.

#### *2.3. Statistical Analysis*

The *Shapiro–Wilk* test was used to investigate the normality of distribution. The *Student's t*-test was used for continuous variables, whereas the *chi-squared* test was used for categorical variables to compare the demographic data and radiographic data. Statistical significance was set at *p* < 0.05. All data were analyzed using SPSS software (version 22, IBM Corp, Armonk, NY, USA). To have a 90% probability of detecting a 1 mm difference in the plate-to-cortex distance, we needed to enroll 72 patients, assuming an overall standard deviation in 2 mm and a 2-tailed alpha-level of 5%.

#### **3. Results**

All inter- and intra-observer intraclass correlation coefficients showed good agreement for all radiographic measurement reliabilities (>0.80).

Compared to the TomoFix conventional group, the TomoFix anatomical group showed similar results in terms of sex distribution, mean age, body mass index, and preoperative and postoperative simple radiographic measurements (Table 2).

**Table 2.** Comparison of the demographic and simple radiographic data between the TomoFix conventional plate and the TomoFix anatomical plate groups.


BMI, body mass index; HKA, hip–knee–ankle; MPTA, medial proximal tibial angle.

With respect to CT scan measurements, the TomoFix anatomical group showed a greater plate angle than the TomoFix conventional group (39.2◦ ± 8.1◦ vs. 31.7◦ ± 7.0◦, *p* < 0.001, Table 3). This result indicates that the TomoFix anatomical plates allowed a more posterior plate position than the conventional plate. Moreover, all screw angles were significantly lower in the TomoFix anatomical group. These results were also caused by positioning the plate more posteriorly than in the TomoFix conventional group.

**Table 3.** Comparison of the fixative device position on CT scan between the two groups.


In the TomoFix conventional group, a cortical screw on #1 hole (the most proximal hole of distal 4 holes) was used in 16 cases (25.4%). However, cortical screws were used in eight cases (15.1%) in the TomoFix anatomical group. The anatomical plate group used fewer cortical screws on #1 hole, but the difference was not statistically significant (*p* = 0.18). The plate-to-cortex distance was significantly shorter in the TomoFix anatomical group than in the TomoFix conventional group regardless of the use of a cortical screw on the #1 hole. (*p* < 0.001).

#### **4. Discussion**

The principal finding of the current study was that the TomoFix anatomical plate represented a more posteromedial plating position, proximal screw toward the lateral hinge rather than posterior cortex, and improved post-correction tibial contour compared to the TomoFix conventional plate.

Numerous studies have demonstrated that posteromedial plating provides better fixative stability than anteromedial plating in OWHTO [11,12,22]. Takeuchi et al. [11]. compared the stress on the TomoFix plate between the anteromedial and medial plating positions. They reported that the medial plate position provided significantly less stress on the plate than the anteromedial plate position. Moreover, previous studies reported that the anterior plate position was associated with an increased posterior tibial slope, which might lead to anterior cruciate ligament damage after OWHTO [23,24]. The TomoFix conventional plate was a 90◦ T-shaped plate with no left and right distinction, while the TomoFix anatomical plate had a 5◦ posterior slope, allowing the plate to be positioned relatively backward. In the current study, the patients using the TomoFix anatomical plate showed a significantly posterior position compared to those using the conventional plate among single-surgeon cases. These results are consistent with the intent of the TomoFix anatomical plate design.

It is well known that an increased plate–bone distance might reduce the fixative stability when using a locking compression plate and screws. Ahmad et al. [18]. demonstrated that, 5 mm from the bone, the locking plate exhibited increased plastic deformation during cyclical compression and required lower loads to induce mechanical failure when compared to a lesser distance between the bone and plate. Therefore, post-correction anatomical contouring is important for reducing the plate–bone distance in OWHTO. The plate-fitting technique using cortical lag screws was widely performed [25]. However, it might induce a change in the opening gap or lateral hinge fracture [26]. In the current study, in the patients with the TomoFix anatomical plate, fewer cortical screws were used on #1 hole (15.1% vs. 25.4%, *p* = 0.128). Moreover, the plate-to-cortex distances were significantly shorter in the TomoFix anatomical group regardless of the use of a cortical screw. These results demonstrated that the TomoFix anatomical plate showed better anatomical contouring to the post-correction tibia.

OWHTO is a promising surgery for medial compartmental knee joint osteoarthritis. However, adverse events were reported following OWHTO in 37% to 55% of cases [25,27,28]. Sidhu et al. [29] demonstrated that a low rate of serious complications (6.5%) requiring unplanned additional surgery was noted after OWHTO using a TomoFix locking plate. However, 52% of the knees required elective hardware removal due to soft tissue irritation. The TomoFix anatomical plate is expected to have a lower hardware removal rate as it has a more posterior position and is more suitable for the post-correction anatomical structure of the tibia.

This study has some limitations. First, although the TomoFix anatomical plate provided a better configuration than the previous version plate system, the superiority of the actual mechanical stability was not proven. An investigation including a mechanical test is needed to identify the association between fixative configuration and mechanical stability in the future. Second, the current study did not include clinical outcomes, including patient-reported outcome measurements, union rate, union period, and correction loss rate. Therefore, this study was not able to identify the effects of plate design changes on clinical outcomes. Third, the two groups had OWHTO performed at different times (TomoFix conventional plate was used between March 2015 and July 2018, TomoFix anatomical plate was used between August 2018 to February 2021), which may have affected the outcome. However, this study was based on data from a single surgeon, which would have had little effect on the results.

#### **5. Conclusions**

The TomoFix anatomical plate showed a more posteromedial plating position, proximal screw direction toward the lateral hinge rather than posterior cortex, and improved post-correction tibial contour compared to the TomoFix conventional plate.

**Author Contributions:** Conceptualization, S.-S.L. and D.-H.L.; methodology, S.-S.L. and D.-H.L.; validation, S.-S.L. and J.P.; formal analysis, S.-S.L. and J.P.; investigation, J.P.; data curation, S.-S.L. and J.P.; writing—original draft preparation, S.-S.L. and D.-H.L.; writing—review and editing, S.-S.L. and D.-H.L.; visualization, S.-S.L. and J.P.; supervision, D.-H.L.; project administration, D.-H.L. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Institutional Review Board Statement:** The protocol used to evaluate radiographic findings and intraoperative navigation data was approved by our institution's investigational review board. (SMC2021-05-051 at May 26 2021).

**Informed Consent Statement:** Informed consent was obtained from all individual participants included in the study.

**Data Availability Statement:** Not applicable.

**Conflicts of Interest:** The authors declare that they have no conflict of interest.

#### **References**


### *Article* **Central Sensitization Is Associated with Inferior Patient-Reported Outcomes and Increased Osteotomy Site Pain in Patients Undergoing Medial Opening-Wedge High Tibial Osteotomy**

**Jae-Jung Kim 1, In-Jun Koh 2, Man-Soo Kim 1, Keun-Young Choi 2, Ki-Ho Kang <sup>1</sup> and Yong In 1,\***


**Abstract:** *Background and Objectives*: Studies have shown that centrally sensitized patients have worse clinical outcomes following total knee arthroplasty (TKA) than non-centrally sensitized patients. It is unclear whether central sensitization (CS) affects patient-reported outcomes (PROs) and/or level of osteotomy site pain in patients undergoing medial opening-wedge high tibial osteotomy (MOWHTO). The purpose of this study was to determine whether CS is associated with PROs and osteotomy site pain following MOWHTO. *Materials and Methods*: A retrospective evaluation was conducted on 140 patients with varus knee osteoarthritis (OA) who were treated with MOWHTO and monitored for two years. Before surgery, the Central Sensitization Inventory (CSI) was used to assess CS status, and a CSI of 40 or higher was considered indicative of CS. The Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) and pain visual analogue scale (VAS) were used to assess PROs. The minimal clinically important difference (MCID) for the WOMAC was set as 4.2 for the pain subscore, 1.9 for the stiffness subscore, 10.1 for the function subscore, and 16.1 for the total based on the results of a previous study. The WOMAC score, pain VAS score of the osteotomy site, and the achievement rates of WOMAC MCID were compared between the CS and non-CS groups. *Results*: Thirty-seven patients were assigned to the CS group, whereas 84 were assigned to the non-CS group. Before surgery, the CS group showed a higher WOMAC score than the non-CS group (58.7 vs. 49.4, *p* < 0.05). While there was a statistically significant improvement in WOMAC subscores (pain, stiffness, function, and total) for both groups at two years after surgery (all *p* < 0.05), the CS group had a higher WOMAC score than the non-CS group (37.1 vs. 21.8, *p* < 0.05). The CS group showed significantly inferior results in pre- and postoperative changes of WOMAC subscores (pain, function, and total) relative to the non-CS group (all *p* < 0.05). In addition, pain at the osteotomy site was more severe in the CS group than in the non-CS group at two years after surgery (4.8 vs. 2.2, *p* < 0.05). Patients with CS had worse MCID achievement rates across the board for WOMAC pain, function, and total scores (all *p* < 0.05) compared to the non-CS group. *Conclusions*: Centrally sensitized patients following MOWHTO had worse PROs and more severe osteotomy site pain compared to non-centrally sensitized patients. Furthermore, the WOMAC MCID achievement rate of patients with CS was lower than that of patients without CS. Therefore, appropriate preoperative counseling and perioperative pain management are necessary for patients with CS undergoing MOWHTO. *Level of Evidence*: Level III, case-control study.

**Keywords:** medial opening-wedge high tibial osteotomy; central sensitization; patient-reported outcomes; osteotomy site pain; minimal clinically important difference

#### **1. Introduction**

Medial compartment osteoarthritis (OA) with varus deformity [1,2] is the most common indication for medial opening-wedge high tibial osteotomy (MOWHTO). Chondral

**Citation:** Kim, J.-J.; Koh, I.-J.; Kim, M.-S.; Choi, K.-Y.; Kang, K.-H.; In, Y. Central Sensitization Is Associated with Inferior Patient-Reported Outcomes and Increased Osteotomy Site Pain in Patients Undergoing Medial Opening-Wedge High Tibial Osteotomy. *Medicina* **2022**, *58*, 1752. https://doi.org/10.3390/ medicina58121752

Academic Editor: Federico Canavese

Received: 14 September 2022 Accepted: 29 November 2022 Published: 29 November 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

lesions in the medial compartment [3], osteonecrosis of the medial femoral condyle [4], and varus thrust instability [3] are all reasons to consider MOWHTO if they are accompanied by varus deformity. By moving the mechanical axis to the lateral compartment, it is possible to reduce load on the medial compartment and postpone the progression of knee OA [5,6].

While total knee arthroplasty (TKA) is effective in eliminating pain by cutting off the source at the joint periphery, about 20–40% of patients are dissatisfied with the results and complain of persistent pain [7]. Due to these results, research into the mechanisms of central pain has increased [8]. Studies have shown that centrally sensitized patients had worse clinical outcomes and pain following TKA than non-centrally sensitized patients [9–11]. Central sensitization (CS) is considered an etiology of persistent pain and dissatisfaction following TKA [9–12]. Hyperalgesia and allodynia are hallmarks of CS, which is caused by dysfunction in the central nervous system [10,11]. However, there is a lack of studies on patient-reported outcomes (PROs) in CS patients following MOWHTO. In addition, unlike TKA, in which the painful lesion is surgically removed, MOWHTO is performed on the bone around the lesion. Currently, no studies have examined the PROs and degree of osteotomy site pain after MOWHTO according to the presence or absence of CS.

PROs have gained traction in recent years, and patient-reported outcome measures (PROMs) are widely used to assess the success of orthopedic procedures [13]. One of the most common and reliable PROMs used following knee surgery is the Western Ontario and McMaster Universities Arthritis Index (WOMAC) [14,15]. The pain visual analogue scale (VAS) score is also extensively used for evaluating outcomes following orthopedic surgery [16]. Understanding the minimal clinically important difference (MCID) and factors that may be used to forecast whether the MCID will be exceeded might have a major bearing on the definition of surgical outcomes and on the development of patient-centered decision-making aids [17,18]. Due to its emphasis on clinical relevance over statistical significance, it is recommended to use the MCID in PROMs [18,19].

It is currently unclear whether CS affects PROs and/or osteotomy site pain in patients undergoing MOWHTO. Therefore, the purpose of this study was to determine whether CS is associated with PROs and osteotomy site pain following MOWHTO. We predicted that centrally sensitized patients would have worse PROs and increased osteotomy site pain than those who were not centrally sensitized.

#### **2. Materials and Methods**

As part of a retrospective record review research, we included 140 patients who received MOWHTO between May 2015 and April 2019 at a single hospital. The criteria for inclusion were age < 70 years and isolated medial compartment OA with varus deformity. Patients with lateral compartment and patellofemoral OA, osteonecrosis, inflammatory or traumatic OA, missing data, and/or loss to follow-up within two years were excluded from the study to reduce the potential impact of preoperative variables on MOWHTO outcomes. The final study contained 121 patients, after excluding 19 for various reasons (Figure 1). Informed consent was acquired from all patients, and the study was approved by the Institutional Review Board of Seoul St. Mary's Hospital (KC22RASI0419 and 15 June 2022).

#### *Surgical procedure*

A tourniquet was placed in the proximal femur and maintained during surgery with a pressure of 300 mmHg. Arthroscopy was performed prior to osteotomy in all cases. If the meniscus tear was present including medial meniscal root tear, partial meniscectomy was performed. Multiple drilling was also performed for cartilage defects. A 7 cm long skin incision was made in the middle between the tibial tuberosity and the posteromedial border of the tibia, beginning 1 cm below the knee joint. Same skin incision was used in all patients. The Dugdale method [20] was used to determine the correction angle, and the Fujisawa point [21] was chosen as the surgical target. The Dugdale method requires the use of two lines to calculate the correction angle. Two lines are drawn: one from the center of the femoral head to 62.5% of the tibial width and the other from the center of the tibiotalar joint to the same percentage of the tibial width. The angle between these two lines is known as the correction angle [22]. Biplanar osteotomy was used in the surgical operation. A locking plating system (TomoFix®, DePuy Synthes, Oberdorf, Switzerland) was used to secure the osteotomy site. No bone grafts or artificial materials were used to fill the opening gap space.

#### *Postoperative rehabilitation*

Following surgical procedures, all patients participated in the same rehabilitation regimen. To improve knee flexion range, exercise on a continuous passive motion machine was started on postoperative day one at 60 degrees and progressed by 5 degrees daily up to 130 degrees of flexion. Within four weeks, patients were allowed to walk with crutches, and after six weeks, they were permitted full weight bearing.

#### *Clinical assessment and Radiographic evaluation*

The Central Sensitization Inventory (CSI) was used to assess the status of CS, and patients were given the questionnaire the day before the surgery [23]. The CSI is a valid and reliable self-report assessment of central sensitivity [23,24]. The purpose of this 25-item survey is to evaluate both emotional state and pain perception in the course of daily living. Each survey uses a Likert scale ranging from 0 to 4 to collect data. Based on the findings of prior research, a CSI of 40 or higher is considered indicative of CS, and a CSI score less than 40 indicates non-CS [24]. In this study, 37 patients were assigned to the CS group, whereas 84 were assigned to the non-CS group (Figure 1).

Preoperatively and at two years postoperatively, the WOMAC scores were used to assess knee PROs [25]. The 24-item disease-specific questionnaire (WOMAC) is a popular tool for assessing patient progress following knee surgery. It is comprised of three subscores evaluating pain (five items), stiffness (two items), and function (17 items). Each of the items has five possible responses, with scores ranging from 0 to 4. Scores range from 0 to 96, with 20 points assigned to pain, 8 points to stiffness, and 68 points to function. A lower score indicates a more favorable clinical result. In addition, pain VAS scores were used to assess pain at the osteotomy site at two years after surgery. The MCID for the WOMAC was set as 4.2 for the pain subscore, 1.9 for the stiffness subscore, 10.1 for the function subscore, and 16.1 for the total score based on the results of a previous study [26] that refers to the smallest improvement that a patient might perceive as beneficial. The achievement rates of WOMAC MCID were also measured. Between the CS and non-CS groups, the WOMAC scores, pain VAS scores of the osteotomy site, and the achievement rates of WOMAC MCID were compared.

Age, sex, operation side, body mass index (BMI), American Society of Anesthesiologists (ASA) score, comorbidities (diabetes and hypertension), alcohol intake, smoking, and preoperative OA severity were also recorded. The Kellgren and Lawrence (K-L) grade was used to quantify the severity of OA [27]. Hip–knee–ankle (HKA) angle and weight-bearing line (WBL) ratio were assessed as radiological variables preoperatively and after 2 years postoperatively. HKA angle is a measure of lower limb alignment, defined as the angle between the mechanical axes of the femur and the tibia. The WBL ratio was estimated by measuring from the medial border of the proximal tibia to the spot where the WBL meets the proximal tibia and dividing that number by the total width of the tibia. HKA angle and WBL ratio were measured on weight-bearing bilateral standing long-leg radiographs. In addition, correction angles were collected for surgical factors.

Preoperative and postoperative HKA angle and WBL ratio were analyzed twice by two orthopedic surgeons. They were orthopedic surgeons with more than five years of experience. Each analyzer was blinded to the results of the other and to all patient information. The average of the two analyzers' results was used. Using the intraclass correlation coefficient (ICC), intraobserver and interobserver reliability were evaluated to determine the reliability of the measurement. In this study, both intraobserver and interobserver ICC values were greater than 0.8.

#### *Statistical Analysis*

The mean and standard deviation are presented as descriptive statistics for all continuous variables, and frequency statistics are supplied for all noncontinuous variables. We compared preoperative PROMs, two-year postoperative PROMs, and osteotomy site pain between the CS and non-CS groups using Student's *t*-test. The rate of WOMAC MCID achievement was compared between the two groups using the chi-square test. A post hoc power analysis was conducted to evaluate whether our findings had sufficient statistical power. SPSS was used conduct statistical analyses (version 21.0; IBM). A *p*-value less than 0.05 was considered statistically significant.

#### **3. Results**

The 121 patients included 37 (30.5%) patients with CS and 84 (69.4%) patients without CS. The mean follow-up period was 2.0 ± 0.1 years. Table 1 shows the demographic data comparing patients with and without CS.

Before surgery, the CS group showed a significantly higher mean WOMAC score than the non-CS group (*p* < 0.05, Table 2). While there was a statistically significant improvement in WOMAC subscores (pain, stiffness, function, and total) for both groups (all *p* < 0.05), the CS group had a significantly lower two-year postoperative score and lower WOMAC subscores (pain, function, and total) than in the non-CS group (all *p* < 0.05, Table 2). In addition, pain VAS at the osteotomy site was more severe in the CS group at two years after surgery (*p* < 0.05, Table 2). With an alpha value of 0.05, a post hoc power analysis revealed

greater than 90% power to detect a difference in postoperative two-year WOMAC (pain, function, and total) and postoperative two-year osteotomy site pain VAS.


**Table 1.** Patient characteristics.

Data are provided as mean ± SD or *n* (%). CS, central sensitization; CSI, Central Sensitization Inventory; K–L, Kellgren–Lawrence; ASA, American Society of Anesthesiologists.



Data are provided as mean ± SD or *n* (%). WOMAC, Western Ontario and McMaster Universities arthritis index score; VAS, visual analogue scale.

MCID achievement was examined using the results of a previous study [26]. Patients with CS had worse MCID achievement rates across the board for WOMAC pain, function, and total scores (all *p* < 0.05) compared with the non-CS group (Table 3).

**Table 3.** MCID achievement.


Data are provided as mean ± SD or *n* (%). MCID, minimal clinically important difference.

During the follow-up, there were no major complications requiring additional surgery. Minor complications occurred in four knees. Three patients had superficial wound infections that were treated with intravenous antibiotics. Routine DVT evaluation was not performed. However, one symptomatic patient was confirmed as DVT under CT venography and was treated with oral anticoagulants. No neurovascular injury, pulmonary embolism, or delayed union case was observed during the follow-up period.

#### **4. Discussion**

The most important finding of this study was that patients with CS showed inferior outcomes in terms of PROs and greater osteotomy site pain compared to those without CS despite significant improvement in WOMAC subscores.

It has been established that CS is a major contributor to persistent pain following TKA [9–12]. It is believed that severe chronic pain is induced by reduced activation of the descending antinociceptive pathway, perhaps as a result of inactivation by norepinephrine/serotonin [10,11]. Unlike neurogenic or inflammatory pain, central pain is defined as dysfunctional [28,29]. The Central Sensitization Inventory (CSI) is frequently used to evaluate CS [23,24]; in addition, a whole-body pain diagram or quantitative sensory testing (QST) can be used for assessment [30–33]. In our study, the CSI was used for diagnosis of CS.

There have been many studies that focused on CS in relation to the spine and joints, especially in the field of TKA. It has been reported that CS patients showed inferior PROs to non-CS patients in the short term (three months to two years) after TKA [9–11].

Although MOWHTO is a surgical procedure that also involves bone cutting, there has been no research on the relationships between clinical outcomes and CS. As such, this study investigated the relationships between CS and clinical outcomes following MOWHTO.

Even though the PROs of the CS group showed significant improvement after surgery, scores for the preoperative, postoperative, and amount of change in WOMAC in patients with CS were worse than those for patients without CS. Two years after surgery, pain at the osteotomy site of patients with CS was more severe than in non-CS patients. Therefore, surgery outcome, including osteotomy site pain, may be inferior for CS patients who undergo MOWHTO.

Most recently, the MCID (minimum gain judged clinically important vs. statistically significant) has been given more weight [34,35]. Due to the lower improvement in WOMAC scores of patients with CS compared to those without CS, obtaining a minimally acceptable improvement in clinical outcomes is more difficult. In our study, the WOMAC scores of CS patients improved statistically significantly after surgery, but the WOMAC MCID achievement rate of the CS group was only 56.8%. This means that it is difficult to expect clinical improvement in nearly half of patients when performing MOWHTO in the CS group, possibly explaining why CS patients who undergo MOWHTO report lower levels of satisfaction.

Although there was no statistically significant difference in sex between the two groups, the proportion of females in the CS group tended to be higher than in the non-CS group (*p* = 0.064). Women have been shown in several studies to be more sensitive to centrally mediated pain than males [36–38]. This is due to their decreased modulation of conditioned pain [36], increased referred pain following intramuscular experimental pain [37], and increased temporal summation to thermal pain [38]. However, there has been a lack of studies that clearly examine sex differences in pain sensitivity in patients with knee OA, which may be attributable to the confounding factor of disease.

The reason for the differences in PROs and osteotomy site pain between patients with and without CS is not well understood. It is considered that individuals with and without CS have different clinical results after surgery [9–11]. Even before surgery, patients with CS had worse pain and function than patients without CS. Despite the clinical improvement following surgery for both groups, patients with CS might experience more pain in response to the same stimulus (hyperalgesia) and could interpret normal stimulation as painful (allodynia) [9,10]. As a result, it is recommended that patients with CS be given different perioperative pain control when assessing the PROs and osteotomy site pain in patients undergoing MOWHTO. With an incidence of 30.5% of patients with CS in our study, active pain control regimens should be incorporated focusing on preemptive, multimodal pain management in patients with CS undergoing MOWHTO. Studies exist that have demonstrated the efficacy of duloxetine in TKA patients [11,39,40]. Additionally, a recent study suggested that perioperative duloxetine could reduce pain after HTO [41]. Although we think duloxetine can help control pain in MOWHTO patients, further randomized study is warranted. A better understanding of differences in PROs and osteotomy site pain between patients with and without CS for MOWHTO would lead to better overall patient perioperative pain management.

There were limitations in this study. First, as seen in previous research with Koreans who have been diagnosed with OA, patients in this study were all Korean, with females as the majority of patients, which limits the application of the study to other environments [42,43]. In Korea, women have been shown to account for 80–90% of cases of degenerative arthritis and TKA [44,45], indicating an acceptable degree of bias in our study. Second, as a retrospective analysis, there is a chance of selection bias, even though CSI and WOMAC scores were prospectively assessed in all patients. Additional prospective studies are required to verify these findings. Third, due to the nature of the study, only 37 patients were included in the CS group, even though 121 patients who had MOWHTO were studied to compare the PROs and osteotomy site pain according to CS [46,47]. However, our preoperative prevalence of CS was consistent with other studies that found central sensitization in 20–40% of patients before surgery [46–49], which was 30.5% in our study. Fourth, there is a variety of questionnaires that can be used to evaluate the degree of CS [23,24,30–33], which means that study results may vary depending on which questionnaire is used. Unfortunately, there is no agreed-upon approach for assessing CS, and it is unclear whether previous evaluations were validated for use in knee OA patients undergoing MOWHTO. In this study, CS was evaluated using the CSI, which should be considered before applying our results to other studies. Finally, to determine the degree of WOMAC score improvement, we depended on the previously reported MCID of 16.1 points [26]. It has been reported that there are several variables related to MCID, and these differences are clinically significant [34,35]. Despite these limitations, the findings in this study are consistent with those of previous studies, indicating some degree of generalizability [11,46,47]. Therefore, this study contributes important knowledge on CS and its impact on PROs following MOWHTO.

#### **5. Conclusions**

The PROs and severity of osteotomy site pain of patients with CS following MOWHTO were worse than those of patients without CS. Moreover, the MCID achievement rate of patients with CS was lower than that of patients without CS. Therefore, appropriate preoperative counseling and perioperative pain management should be considered for patients with CS undergoing MOWHTO.

**Author Contributions:** Conceptualization, J.-J.K. and Y.I.; methodology, J.-J.K., I.-J.K., M.-S.K. and Y.I.; software, J.-J.K., K.-Y.C. and K.-H.K.; validation, J.-J.K., I.-J.K., M.-S.K. and Y.I.; formal analysis, J.-J.K., K.-Y.C. and K.-H.K.; investigation, J.-J.K., I.-J.K. and M.-S.K.; resources, J.-J.K., K.- Y.C. and K.-H.K.; data curation, K.-Y.C. and K.-H.K.; writing—original draft preparation, J.-J.K.; writing—review and editing, J.-J.K., I.-J.K., M.-S.K., K.-Y.C., K.-H.K. and Y.I.; visualization, K.-Y.C. and K.-H.K.; supervision, I.-J.K., M.-S.K. and Y.I.; project administration, J.-J.K. and Y.I. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Institutional Review Board Statement:** The study was conducted in accordance with the Declaration of Helsinki and approved by the Institutional Review Board of Seoul St. Mary's Hospital (KC22RASI0419, 15 June 2022).

**Informed Consent Statement:** Informed consent was obtained from all subjects involved in the study.

**Data Availability Statement:** The data published in this research are available on request from the first author (J.-J.K).

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


### *Article* **Automated Detection of Surgical Implants on Plain Knee Radiographs Using a Deep Learning Algorithm**

**Back Kim 1,†, Do Weon Lee 2,†, Sanggyu Lee 3, Sunho Ko 4, Changwung Jo 5, Jaeseok Park 1, Byung Sun Choi 4, Aaron John Krych 6, Ayoosh Pareek 6, Hyuk-Soo Han 1,4 and Du Hyun Ro 1,4,7,\***

	- <sup>5</sup> Seoul National University Hospital, Seoul 03080, Republic of Korea
	- <sup>6</sup> Department of Orthopedic Surgery, Mayo Clinic, Rochester, MN 55902, USA
	- <sup>7</sup> CONNECTEVE Co., Ltd., Seoul 03080, Republic of Korea
	- **\*** Correspondence: duhyunro@gmail.com; Tel.: +82-2-2072-4060
	- † These authors contributed equally to this paper.

**Abstract:** *Background and Objectives*: The number of patients who undergo multiple operations on a knee is increasing. The objective of this study was to develop a deep learning algorithm that could detect 17 different surgical implants on plain knee radiographs. *Materials and Methods*: An internal dataset consisted of 5206 plain knee antero-posterior X-rays from a single, tertiary institute for model development. An external set contained 238 X-rays from another tertiary institute. A total of 17 different types of implants including total knee arthroplasty, unicompartmental knee arthroplasty, plate, and screw were labeled. The internal dataset was approximately split into a train set, a validation set, and an internal test set at a ratio of 7:1:2. You Only look Once (YOLO) was selected as the detection network. Model performances with the validation set, internal test set, and external test set were compared. *Results*: Total accuracy, total sensitivity, total specificity value of the validation set, internal test set, and external test set were (0.978, 0.768, 0.999), (0.953, 0.810, 0.990), and (0.956, 0.493, 0.975), respectively. Means ± standard deviations (SDs) of diagonal components of confusion matrix for these three subsets were 0.858 ± 0.242, 0.852 ± 0.182, and 0.576 ± 0.312, respectively. True positive rate of total knee arthroplasty, the most dominant class of the dataset, was higher than 0.99 with internal subsets and 0.96 with an external test set. *Conclusion*: Implant identification on plain knee radiographs could be automated using a deep learning technique. The detection algorithm dealt with overlapping cases while maintaining high accuracy on total knee arthroplasty. This could be applied in future research that analyzes X-ray images with deep learning, which would help prompt decision-making in clinics.

**Keywords:** automated detection; detection algorithm; deep learning

### **1. Introduction**

As knee operations are increasing in number, the number of patients who undergo multiple knee surgeries is increasing [1,2]. Surgical history must be identified to check whether there is an existing implant and what type of implant it is [3]. However, some patients undergo surgeries at multiple institutions, which hinders obtaining accurate surgical history [4]. According to a 2012 survey on American Association of Hip and Knee Surgeons (AAHKS) members, 10% of implants before operations and 2% of implants during surgery could not be identified [5]. Moreover, the median time taken for implant identification by surgeons was 20 min [5].

Automated identification using artificial intelligence (AI) is an efficient way to overcome this situation. Machine learning (ML) methods have been demonstrated to be fast

**Citation:** Kim, B.; Lee, D.W.; Lee, S.; Ko, S.; Jo, C.; Park, J.; Choi, B.S.; Krych, A.J.; Pareek, A.; Han, H.-S.; et al. Automated Detection of Surgical Implants on Plain Knee Radiographs Using a Deep Learning Algorithm. *Medicina* **2022**, *58*, 1677. https://doi.org/10.3390/ medicina58111677

Academic Editors: Yong In and In Jun Koh

Received: 7 November 2022 Accepted: 17 November 2022 Published: 19 November 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

and accurate in processing medical images such as X-ray, MRI, and pathology slides [6,7]. Among diverse ML approaches, deep learning (DL) especially stands out because it can find distinctive features in given data without direct human supervision, unlike classic ML models [8]. DL model could significantly reduce time and cost spent in identifying implants by examining X-ray images directly and automatically. Tiwari et al. [9] and Patel et al. [10] have directly compared DL models with human experts in examining X-ray images and proven that DL models are better in both speed and accuracy.

Ren et al. [11] have reviewed studies on identification of orthopedic implants using AI. Their systematic review covered several different anatomic sites such as hip, knee, and shoulder. It is reasonable because basic concepts are similar regardless of anatomical locations. Among the 11 studies they included, those conducted by Yi et al. [12] and Karnuta et al. [13] dealt with knee X-rays.

Yi et al. [12] were the first who tried to identify implants on knee plain radiographs using AI. They developed three different DL models: (1) a model that judges whether there is total knee arthroplasty (TKA) on the image, (2) a model that classifies TKA and unicompartmental knee arthroplasty (UKA), and (3) a model that distinguishes between two TKA models. They utilized 237 to 274 images, which varied by each model. All three models with binary classifications achieved a perfect AUC of 1.0.

Karnuta et al. [13] have expanded the task into multi-class classification. Their DL system classified nine different implant models with AUC of 0.99. They also increased the size of the dataset into 682 images.

Contrary to these previous studies [12,13], this study employed a detection network because there were 'overlapping' cases in the dataset. A majority of previous studies focused on classifying companies and models within the scope of TKA, whereas this study aimed to figure out the type and location of every implant on an X-ray image. For example, some images might include TKA and three screws simultaneously. Such images could be easily identified with detection networks.

Covering a variety of implants other than TKA is important because TKA is not the only surgical intervention that a patient may have undergone. For instance, if a patient received high tibial osteotomy (HTO), a plate and screws would appear on the X-ray, but there would be no TKA. Another example is anterior cruciate ligament (ACL) reconstruction. Usually, only screws would appear in this case. Most of the previous studies could not deal with these surgeries because their models classified TKA only. Thus, the main purpose of this study was to develop a DL model to identify 17 different types of objects and determine their locations in the form of bounding boxes.

#### **2. Materials and Methods**

This study was approved by the Institutional Review Board (No. H-1801-061-915). The overall process of this study, from organizing dataset to evaluating the trained DL model, is summarized in Figure 1.

#### *2.1. Dataset and Pre-Processing*

The internal dataset consisted of 5206 X-rays taken at a single, tertiary hospital from 4435 patients. The external test contained 238 X-rays from another tertiary institution. X-rays from each hospital were retrospectively collected from 2009-01-01 to 2019-12-31.

All images were knee anteroposterior (AP) view images. These images were first converted from Digital Imaging and Communications in Medicine (DICOM) format to Portable Network Graphics (PNG) format. They were subsequently converted into Joint Photographic Expert Group (JPG) format. JPG quality was 70 for internal data and 95 for external data.

The external dataset consisted of 71 images including both legs (29.8%), 73 images including only the left leg (30.7%), and 94 images including only the right leg (39.5%). Those images were flipped horizontally and concatenated with the original ones. If the

original image contained the right leg, the flipped one was put left to it. If the original one included the left leg, the flipped one was placed right to it.

**Figure 1.** A schematic process of the deep learning system development in this study. (TP = true positive, TN = true negative, FP = false positive, FN = false negative, mAP = mean average precision).

A total of 17 different types of objects were labeled: TKA\_femur, TKA\_tibia, UKA, staple, wire, screw, plate, IM nail, washer, smooth pin, metal button, bone cement, tumor prosthesis, ruler, External fixator, Patellofemoral arthroplasty, and TKA\_stem (TKA = total knee arthroplasty, UKA = unicompartmental knee arthroplasty, IM nail = intramedullary nail). The type and location of each object were labeled so that both classification and detection were possible. Table 1 shows the number of images and instances that belong to each class in internal and external datasets.

**Table 1.** The number of images and instances by class in internal and external dataset.



**Table 1.** *Cont.*

TKA, total knee arthroplasty; UKA, unicompartmental knee arthroplasty; IM, intramedullary.

The internal dataset was split into train set, validation set, and internal test set. The split ratio was train: validation: test = (about) 7:1:2. Multiple X-rays from a single patient were assigned to the same subset. In addition, the class imbalance was adjusted by applying a ratio of 7:1:2 in a per-class manner. Except for this principle, the splitting process was done in a random manner. Finally, 3701, 517, and 1042 images were assigned to the train set, validation set, and internal test set, respectively. The external dataset was solely used as the external test set. These 238 images were neither used in the training nor in the validation of DL process.

#### *2.2. Deep Learning Model Training*

You only look once (YOLO) was utilized as the DL algorithm. The detection approach before YOLO was divided into two steps. For example, R-CNN could "first generate potential bounding boxes in an image and then run a classifier on these proposed boxes" [14]. Using YOLO, the detection process became a single-step, "straight from image pixels to bounding box coordinates and class probabilities" [14]. In other words, it can judge the type and location of an object simultaneously. YOLO is fast and light because of this feature. Yolov5, the latest version of YOLO, was employed in this research. Yolov5 offers multiple options such as size and complexity, whose basic approaches are identical. Yolov5s has the most simple and lightest architecture among them. Thus, it was selected in this study.

The programming language and the deep learning framework used for model training were Python 3.8.5 and PyTorch, respectively. Graphics processing unit (GPU) NVIDIA GeForce RTX 3090 was utilized for fast training. Transfer learning was conducted. That is, Yolov5s trained with Common Objects in Context (COCO) dataset was fine-tuned to adjust to our dataset. COCO is a large-scale dataset frequently used in detection and segmentation tasks like ImageNet in classification tasks. It consists of 330K images with 1.5 million object instances labeled into 80 object categories [15]. Detection models pretrained with COCO is expected to perform well even on novel tasks because it has already learned how to find meaningful features on various kinds of objects (e.g., finding contours or rectangles). The model was trained and validated in 50 epochs. The batch size was 16. Online imagespace and colorspace augmentations were applied to the train set [16]. Three random images were presented with the original image each time when an image was loaded for training [17].

#### *2.3. Evaluation Metrics*

Train losses and validation losses of each epoch were examined to verify that no overfitting occurred. In that case, the model 'memorizes the answers' of the training set instead of recognizing genuine features of each image. This phenomenon can be sensed by tracking validation loss because an overfitted model cannot perform well on novel data. Thus, validation loss increases with each epoch in the overfitted model.

Accuracy, sensitivity, specificity, and confusion matrix were used to evaluate the model's performance. These metrics were each measured with the validation set, internal test set, and external test set separately. Accuracy, sensitivity, and specificity of each class were first calculated. The total score was defined as their weighted average, whose weight was the number of instances that belonged to each class. Accuracy was defined as (TP+TN)/(TP+TN+FP+FN). Sensitivity was defined as TP/(TP+FN) and specificity was defined as TN/(TN+FP). The confusion matrix is a figure that shows TP, TN, FP, and FN rates of each class. TP, TN, FP, and FN mean true positive, true negative, false positive, and false negative, respectively. The metrics of each class could be figured out by treating the task like a binary classification problem. For instance, the sensitivity value of TKA\_stem was figured out by defining TKA\_stem as 'True' and all other classes as 'False'. The judgment of true and false is intuitive in classification. For example, if the ground truth is '0' and the model prediction is '0', it is true. If the model prediction is '1', it is false. However, in the case of detection, the location of the bounding box was also taken into consideration.

Intersection over union (IoU) was used as a criterion for detection. IoU was defined as (area of overlap intersection)/(area of union). It was determined by overlap intersection and union between two boxes: ground truth and prediction. IoU became higher when the two boxes overlapped more, i.e., the area of intersection of the two boxes was larger while the area of symmetric difference was smaller. It is because the area of union is the sum of the area of intersection and symmetric difference. The threshold of minimum IoU can be selected by users. In this study, a threshold of 0.5 was used for accuracy, sensitivity, and specificity, while a threshold of 0.45 was for confusion matrices. It means that if the threshold is 0.7, predictions with IoU lower than 0.7 are false predictions.

#### **3. Results**

Figure 2 shows no overfitting in the training process. There was no abnormal increase in validation loss during the task.

**Figure 2.** Train (**A**) & validation (**B**) losses of the Yolov5s model in the study. As the number of epochs increases, train losses always decrease. Validation losses decrease if the model is robust on new data. If the model is overfitted to the train set, validation losses would increase.

Tables 2–4 present accuracy, sensitivity, and specificity tested with the validation set, internal test set, and external test set, respectively. Total accuracy, total sensitivity, total specificity value of each subset were (0.978, 0.768, 0.999), (0.953, 0.810, 0.990), and (0.956, 0.493, 0.975), respectively.


**Table 2.** Evaluation metrics on the validation set.

TKA, total knee arthroplasty; UKA, unicompartmental knee arthroplasty; IM, intramedullary.

**Table 3.** Evaluation metrics on the internal test set.


TKA, total knee arthroplasty; UKA, unicompartmental knee arthroplasty; IM, intramedullary.

**Table 4.** Evaluation metrics on the external test set.



**Table 4.** *Cont.*

TKA, total knee arthroplasty; UKA, unicompartmental knee arthroplasty; IM, intramedullary.

As for the internal validation set, the accuracy ranged from 0.703 to 1. Accuracy values for TKA\_tibia, UKA, bone cement, and tumor prosthesis were 1. The value for Patellofemoral arthroplasty was 0.703. Sensitivity values for TKA\_tibia, UKA, metal button, bone cement, tumor prosthesis, and TKA\_stem were 1. Specificity values for TKA\_femur, TKA\_tibia, UKA, wire, plate, bone cement, tumor prosthesis, External fixator, and Patellofemoral arthroplasty were also 1.

As for the internal test set, the accuracy ranged from 0.402 to 0.999. Accuracy values for UKA, bone cement, tumor prosthesis, ruler, and External fixator were 0.999. This value for Patellofemoral arthroplasty was 0.402. Sensitivity values for UKA, bone cement, and tumor prosthesis were 1. The specificity value for External fixator was also 1.

As for the external test set, the accuracy ranged from 0.670 to 0.998. The accuracy value for External fixator was 0.998. This value for screw was 0.670. Sensitivity values for TKA\_femur and UKA were 1. Specificity values for UKA, smooth pin, and External fixator were also 1.

Confusion matrices of the validation set, internal test set, and external test set are displayed in Figure 3. Means ± SDs of diagonal components of the validation set, internal test set, and external test set in the confusion matrices were 0.858 ± 0.242, 0.852 ± 0.182, and 0.576 ± 0.312, respectively.

**Figure 3.** *Cont*.

**Figure 3.** Confusion matrices on validation set (**A**), internal test set (**B**), and external test set (**C**). Among 18 labels on each axis, only 'background' is not an object class. If the 'True label' is background and the 'Predicted label' belongs to one of the object classes, it means that the model made a false positive judgment on the presence of an implant. If the 'Predicted label' is background and the 'True label' belongs to one of the object classes, it means that the model made a false negative judgment on the presence of an implant.

#### **4. Discussion**

The most dominant orthopedic implant of our dataset was TKA. This is because the number of arthritis patients is increasing due to the aging population. In addition, TKA has been acknowledged as a highly effective treatment, especially for senior patients. Nevertheless, there are other types of implants for other diseases. The number of these cases was not negligible. Thus, the authors tried to define as diverse classes as possible to cover a wide range of knee implants.

Table 5 compares the approach and the result of this research with previous studies. Among previous studies that automatically identified implants on X-rays, seven studies that dealt with knee radiographs were included. In comparison, our study dealt with large amount of dataset with comparable accuracy to other DL models while adapting a different kind of approach (detection method) to identify multiple different kinds of implants.


**Table 5.** Comparison with previous studies.

AUC = Area Under the Curve.

Belete et al. [18] have developed a DL model with eight different classes, and they considered calculating the probability with the softmax function in the evaluation process. Sharma et al. [19] have classified six implant types using 1078 images and compared four different networks with an external test set that consists of 162 images. Tiwari et al. [9] have classified six implant types using 521 images, and they have shown the superiority of ML models over human experts regarding the average accuracy. Patel et al. [10] have used both knee and hip radiographs, identifying 12 implant types using 427 knee and 922 hip unilateral images. Klemt et al. [20] have also classified both knee and hip arthroplasties, identifying 14 TKA designs and 24 total hip arthroplasty (THA) designs using a total of 11,204 knee X-rays.

Gurung et al. [21] have reviewed nine studies using AI for the identification of orthopedic implants. They also analyzed three articles that focused on predicting the risk of implant surgery failure. Ren et al. [11] and Gurung et al. [21] both pointed out that most of the existing studies didn't use external datasets, which means they only used radiographs from one institution. Our model was tested on a balanced external test set to resolve this problem.

Among studies that only handled locations other than the knee, such as the hip and shoulder, the one conducted by Kang et al. [22] was notable because they automatically cropped images with YOLOv3, a detection network. The final prediction was done with a simple CNN classification network. The research by Urban et al. [23] was also noticeable because they compared the performances of DL networks with other ML classifiers (e.g., Gradient Boosting).

The current study employed YOLO, a detection network, unlike majority of previous studies that utilized classification networks such as ResNet. These previous studies have focused on identifying different companies or models of arthroplasty implants like TKA because different models of arthroplasties are usually incompatible. In contrast, this study aims to identify implants that can appear simultaneously in the same image. In other words, images that include each type of object are not disjoint in our dataset. For example, both UKA and screw can appear in some images. Therefore, when applying classification networks, this task becomes not only a multi-class classification, but also a multi-label classification. To handle this problem, either the dataset or model algorithm should be transformed. That was why this study employed detection approach instead. Unlike classification, detection networks can inherently handle overlapping cases. The model output includes every object in an image. For example, the model can detect eight smooth pins and two wires in an image and figure out their locations.

As to the pre-processing of images, conversion into low-quality JPGs can reduce not only computational resources required for DL training and inference, but also model training and predicting time. Another way to save time and resources is by cropping images to only include the region of interest (ROI), i.e., the area around the knee joint. However, cropped dataset limits types and locations of implants that can be detected. For example, detection of TKA\_stem and differentiation between IM\_nail and TKA\_stem becomes easier with the dataset that is uncropped. Thus, the whole image was used in this study.

In terms of defining object classes, the reason why TKA was split into TKA\_femur and TKA\_tibia and why there were classes that were not surgical implants (i.e., ruler and external fixator) was related to the performance of the detection model. TKA was split into TKA\_femur and TKA\_tibia because unless it was separated, bounding boxes became too large and the performance of the detection model would decline. The reason why there were 'ruler' and 'external fixator' classes was because unless they were assigned to some classes, the detection model could mistake them for some other implants like screws or staples.

As for model evaluation, the metrics were measured with three subsets for comparison. Test sets were more novel than the validation set to the model. The external test set was from different institution. Thus, how the model would respond in a new situation can be predicted by comparing its performances with these three subsets. Moreover, the reason why the 'weighted' average of accuracy, sensitivity, and specificity was calculated was because it could help deal with the class imbalance problem by reflecting the number of

instances. Despite covering diverse cases, the proportion of TKA instances was high, which was more than half of the entire dataset.

Confusion matrices in Figure 3 showed that the model was more than 99% accurate on TKA in the case of internal subsets. This score was almost the same as the binary TKA classifier described by Yi et al. [12]. TP rate of TKA on external subset was also high, which was about 0.96. The fact TKA was the most dominant class might explain this result. However, this result might also be due to fact that TKA implant is relatively big and easy to identify owing to its distinct shape. UKA shares these characteristics with TKA. Its TP rates were 1.00 in all subsets. Interestingly, TP rates of screw, the second most dominant class after TKA, were less than 0.9 in all three subsets. It might be because the model confused screws with other small implants like staples.

External fixator was not detected successfully because the number of their instances was too small. In the case of tumor prosthesis whose number of instances was also small, the model showed perfect TP rates with internal subsets, but a poor TP rate of 0.3 with the external dataset. It means the model is not robust enough for detecting tumor prosthesis. Notably, the score on patellofemoral arthroplasty was fine, although its number of instances was similar to that of tumor prosthesis. The fact that its shape and location are very distinct from other implants might explain this outcome.

The metrics in Tables 2–4 display similar tendencies with the confusion matrices. Performances for TKA and UKA were high in all three subsets. The model could successfully identify the presence of TKA and UKA. Overall scores with the external dataset show that the robustness of the model needs to be improved. However, the evenness of the external test set should be considered, which can be verified through the number of images by class such as those shown in Table 1.

This research would be helpful not only in clinical fields including medical centers performing orthopedic surgeries, but also in future DL studies. Inclusion or exclusion criteria for DL research on X-rays may contain whether a patient has undergone knee surgeries. Automated identification of implants would help us quickly screen images for implants.

This study has several limitations. First, there was a class imbalance problem in the internal dataset. The proportion of TKA instances was very high, which was more than half of the entire dataset. It can make an objective evaluation of the DL models difficult. This problem could be partly resolved by using a relatively balanced external dataset. Second, the sensitivity of the model for certain kinds of implants such as staples was low. Since this was the first study, to our knowledge, that dealt with these low profile implants, further study is needed to improve the sensitivity of the model. Third, default hyperparameters were used for model training. The model performance could become better if hyperparameter tuning was applied.

#### **5. Conclusions**

In conclusion, this study explored automation of implant identification process on plain knee radiographs using DL techniques. True positive rates of TKA, the most dominant class, for the validation set, internal test set, and external test set were 1.00, 0.99, and 0.95, respectively. The total accuracy value of the internal test set was 0.953, meaning that the model classified other classes with a fine performance as well. Unlike the models presented in previous studies, the one in this study can identify various cases other than TKA in clinical fields by detecting multiple implants that can be overlapped. Approaches of this study can also be applied in future research that analyzes X-ray images with DL techniques.

**Author Contributions:** Conceptualization, D.H.R. and B.K.; methodology, B.K.; software, S.K.; validation, S.L., J.P. and D.W.L.; formal analysis, B.K.; investigation, J.P.; resources, S.K., A.J.K., A.P. and H.-S.H.; data curation, S.L.; writing—original draft preparation, B.K.; writing—review and editing, D.W.L. and B.S.C.; visualization, C.J.; supervision, D.H.R.; project administration, D.H.R. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Institutional Review Board Statement:** The study was conducted in accordance with the Declaration of Helsinki, and approved by the Institutional Review Board of Seoul National University Hospital. (No. H-1801-061-915, 16 September 2022).

**Informed Consent Statement:** Patient consent was waived due to anonymity of the data used in the study. It was approved by the Institutional Review Board of Seoul National University Hospital.

**Data Availability Statement:** The data presented in this study are available on request from the corresponding author. The data are not publicly available due to privacy.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


### *Article* **Dual-Energy CT-Based Bone Mineral Density Has Practical Value for Osteoporosis Screening around the Knee**

**Keun Young Choi 1,2,†, Sheen-Woo Lee 3,†, Yong In 2,4, Man Soo Kim 2,4, Yong Deok Kim 1,2, Seung-yeol Lee 1,2, Jin-Woo Lee 1,2 and In Jun Koh 1,2,\***


**Abstract:** *Introduction*: Adequate bone quality is essential for long term biologic fixation of cementless total knee arthroplasty (TKA). Recently, vertebral bone quality evaluation using dual-energy computed tomography (DECT) has been introduced. However, the DECT bone mineral density (BMD) in peripheral skeleton has not been correlated with Hounsfield units (HU) or central dual-energy X-ray absorptiometry (DXA), and the accuracy remains unclear. *Materials and methods*: Medical records of 117 patients who underwent TKA were reviewed. DXA was completed within three months before surgery. DECT was performed with third-generation dual source CT in dual-energy mode. Correlations between DXA, DECT BMD and HU for central and periarticular regions were analyzed. Receiver operating characteristic (ROC) curves were plotted and area under the curve (AUC), optimal threshold, and sensitivity and specificity of each region of interest (ROI) were calculated. *Results:* Central DXA BMD was correlated with DECT BMD and HU in ROIs both centrally and around the knee (all *p* < 0.01). The diagnostic accuracy of DECT BMD was higher than that of DECT HU and was also higher when the T-score for second lumbar vertebra (L2), rather than for the femur neck, was used as the reference standard (all AUC values: L2 > femur neck; DECT BMD > DECT HU, respectively). Using the DXA T-score at L2 as the reference standard, the optimal DECT BMD cut-off values for osteoporosis were 89.2 mg/cm<sup>3</sup> in the distal femur and 78.3 mg/cm<sup>3</sup> in the proximal tibia. *Conclusion*: Opportunistic volumetric BMD assessment using DECT is accurate and relatively simple, and does not require extra equipment. DECT BMD and HU are useful for osteoporosis screening before cementless TKA.

**Keywords:** dual-energy CT; Hounsfield unit; bone mineral density; volumetric phantomless BMD; opportunistic CT

#### **1. Introduction**

The number of younger and more active patients treated with total knee arthroplasty (TKA) continues to increase rapidly [1–5], but TKA in younger patients has exhibited a lower 10-year survivorship with higher demand for revision [6–9]. Thus, long-term and biologic fixation after TKA has been revisited, and interest in cementless TKA, which theoretically enables physiologic fixation, has increased [10,11]. The earlier cementless TKA prostheses had unacceptably high failure rates and poor clinical outcomes [12,13], but major advances in design and materials in recent years have greatly improved the newer generation of cementless prostheses [14–16]. Stereo imaging analysis has shown that the new cementless prosthesis is superior to traditional press fit design and cemented prostheses in reducing micromotion, and it is anticipated that the risk of aseptic loosening will be greatly reduced [17].

**Citation:** Choi, K.Y.; Lee, S.-W.; In, Y.; Kim, M.S.; Kim, Y.D.; Lee, S.-y.; Lee, J.-W.; Koh, I.J. Dual-Energy CT-Based Bone Mineral Density Has Practical Value for Osteoporosis Screening around the Knee. *Medicina* **2022**, *58*, 1085. https://doi.org/10.3390/ medicina58081085

Academic Editor: Massimo De Martinis

Received: 29 June 2022 Accepted: 8 August 2022 Published: 11 August 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

Preoperative bone quality and bone mineral density (BMD) around the knee joint are strongly associated with adequate bone ingrowth and initial prosthesis stability after cementless TKA [16,18,19]. However, there is no gold standard method as yet to evaluate BMD around the knee joint. Previous study has shown that BMD around the knee can be estimated by CT attenuation in Hounsfield units (HU), which are a measure of the standardized linear attenuation coefficient of CT [20]. However, the HU-based BMD is limited due to distortion by changes in marrow composition, and diagnostic value around the knee joint remains unclear [20–23]. Dual-energy CT (DECT), which acquires CT attenuation data at two different energy levels, can generate images of soft tissue or bone marrow by decomposing different tissue characteristics [24]. It has recently been proposed as a tool to screen for osteoporosis with greater diagnostic accuracy than traditional quantitative CT (QCT) [22,25]. However, the DECT BMD for peripheral bone assessment has not been studied yet.

Therefore, the purposes of this study were (1) to assess the strength of correlation between central dual-energy X-ray absorptiometry (DXA) BMD, DECT BMD and HU at the lumbar spine, femur neck, distal femur, and proximal tibia; and (2) to calculate the diagnostic accuracy of DECT BMD and HU for bone quality assessment around the knee joint when the T-score of central DXA is used as the reference standard. We hypothesized that the volumetric BMD and HU from DECT would be well correlated with DXA BMD. We also hypothesized that the diagnostic accuracy of DECT BMD and HU in bone quality assessment around the knee joint would be high.

#### **2. Material and Methods**

A total of 125 patients who underwent TKA between November 2021 and January 2022 and had both preoperative third-generation dual-source DECT and DXA within three months before surgery were considered for study inclusion. The study was approved by the Institutional Review Board of our institute (PC22RISI0049) and it was exempted from informed consent because it is a retrospective medical and radiological record review. Patients with metal prosthesis, fracture or infection in central (spine or hip) or peripheral (distal femur or proximal tibia) regions were excluded (Figure 1). DXA at the lumbar spine and left pelvic area had been taken within 3 months before TKA (Figure 2A), and had identified osteoporosis in 11% of L2 vertebral bodies and in 14.5% of femur necks (Table 1). Lower extremity CT was performed for the purpose of preoperative planning for TKA with a measured resection technique. All CT studies were performed on a third-generation dualsource CT system in dual-energy mode (SOMATOM Force; Siemens Healthineers, Erlangen, Germany), with tube A at 90 kVp and 180 mAs and tube B at Sn150 kVp [0.64-mm tin filter] and 180 mAs. Image series were collected in a craniocaudal direction with the patient in a supine position without administration of a contrast agent. Three image sets, 90 kVp, Sn150 kVp, and weighted average (ratio 0.5:0.5), were acquired in each CT examination to resemble the contrast properties of single-energy bone CT images. The images were reconstructed with a dual-energy bone kernel (B69f), transferred to the image archiving system (SyngoVia, Siemens Healthineers, Erlangen, Germany), and then transferred to a personal computer carrying the analysis software for postprocessing.

Phantomless volumetric BMD assessment of L2 and the left femur neck with DECT requires manual delineation of trabecular volumes of interest (VOI) in the L2 vertebra and the left pelvis, which was carried out by one of the authors with dedicated software (Examine, Siemens Healthcare, Erlangen, Germany). Regions of interest (ROIs) defined to best include trabecular bone and exclude any cortical bone were also drawn manually on the images loaded into the software (Figure 2B). The software then performed calculations according to a dedicated algorithm and the resulting output included the volumetric BMD values.

For DECT HU analysis, one of the authors, working at a conventional PACS workstation, manually defined polygonal ROIs on standard bone reconstructions in sagittal or axial image series (Figure 2C). The ROIs were positioned in the anterior trabecular bone

space of the L2 vertebral body, as proposed by several studies [26–28]. Thereby, the reader was instructed to avoid attenuation heterogeneity by placing the ROIs in areas of spinal hemangiomas or other causes of attenuation heterogeneity. All HU values were obtained as averages of three serial polygonal ROIs.

**Figure 1.** Standards for Reporting of Diagnostic Accuracy Studies (STARD) flow chart of patient inclusion.


**Table 1.** Patient demographics and preoperative characteristics.

† Data are presented as numbers (percentage) of patients. \* Data are presented as the means ± standard deviations (range). DXA, dual x-ray absorptiometry; BMD, bone mineral density.

**Figure 2.** (**A**) Assessment of dual x-ray absorptiometry (DXA) in the lumbar spine and femur neck; (**B**) Manual definition of the region of interest (ROI) and assessment of bone mineral density (BMD) derived from dual-energy computed tomography (DECT) in the distal femur and proximal tibia using dedicated DECT postprocessing software; (**C**) Manual definition of the region of interest (ROI) and assessment of Hounsfield unit (HU) derived from dual-energy computed tomography (DECT) in the distal femur and proximal tibia using dedicated DECT postprocessing software.

Correlation between DXA BMD or T-scores for L2 or the left femur neck and DECT BMD or HU of L2, left femur neck, distal femur, and proximal tibia was analyzed and receiver-operating characteristic (ROC) curves were plotted to evaluate diagnostic accuracy. In addition, the area under the curve (AUC), optimal threshold, and sensitivity and specificity of DECT BMD and HU for distal femur and proximal tibia were determined by using the DXA T-scores for L2 and femur neck as the reference standards.

The values assessed by the researchers were obtained as the average of two evaluators. Each researcher assessed every radiological variable two times with the interval at least two weeks. Intra- and inter-observer reliability for each measurement were expressed as intraclass correlation coefficients (ICCs).

#### *Statistical Analysis*

All computations were performed with Statistical Package for Social Sciences (SPSS) version 21 (IBM Corp., Armonk, NY, USA), with significance set at *p* < 0.05. Variables are presented as mean ± standard deviation. The correlation between DXA BMD and DECT BMD or HU was analyzed by using the Pearson product moment correlation. ROC curve analysis and calculation of the AUC were performed to evaluate optimal cut-off values for distinguishing osteoporosis from normal BMD. Osteoporosis [T < −2.5], osteopenia [−1.0 ≤ T ≤ −2.5], and normal BMD [T > −1.0] were classified according to the DXA T-score. Sensitivity, specificity, positive and negative predictive values (PPV and NPV), and accuracy were computed from these cut-off values.

#### **3. Results**

Of a total of 125 patients, three patients with a history of metal implants in spine or hip, tumor or previous fracture at spine or hip, previous vertebroplasty or kyphoplasty at spine, or previous infection at spine or hip were excluded. Five patients with metal implant, previous fracture, or history of previous osteotomy around a knee were excluded. Thus, a total of 117 patients were finally included and their medical and radiographic data were reviewed (Figure 1). The mean age was 70.6 years and the mean body mass index (BMI) was 26.6 kg/m2 (Table 1).

DXA BMD of L2 and femur neck were strongly correlated with DECT BMD and HU at their own region (all *r* > 0.5, all *p* < 0.01). In addition, the DXA BMD of L2 and the femur neck were significantly correlated with DECT BMD and HU in the other ROIs, surpassing moderate degree (all *r >* 0.3), except for the DXA BMD of the femur neck and HU of the proximal tibia (*r =* 0.286) (all *p* < 0.01) (Table 2). In addition, the correlation value (*r*) was higher for L2 than for the femur neck for all ROIs except for their own region.

The DECT BMD showed a stratified result from normal bone quality for osteoporosis measured by the DXA T-score in every ROI (Figure 3A–D). When the DXA T-score at L2 was used as the reference standard, the optimal cut-off values of DECT BMD for diagnosing osteoporosis were calculated as 89.2 mg/cm<sup>3</sup> in the distal femur and as 78.3 mg/cm<sup>3</sup> in the proximal tibia. In terms of DECT HU, the optimal cut-off values were 104.5 in the distal femur and 66.5 in the proximal tibia. When the DXA T-score at the left femur neck was used as the reference standard, the optimal DECT BMD cut-off values for diagnosing osteoporosis were 96.9 mg/cm<sup>3</sup> in the distal femur and 80.9 mg/cm<sup>3</sup> in the proximal tibia and the optimal cut-off values of DECT HU were 117.4 in the distal femur and 66.8 in the proximal tibia (Table 3). In addition, all AUC values of L2 surpassed femur neck and AUC values of DECT BMD surpassed DECT HU, respectively (Table 3). Thus, the diagnostic accuracy of both DECT BMD and HU was better when the DXA T-score for L2, rather than the femur neck, was used as the reference standard (Figure 4A,B).


**Table 2.** Correlation analysis between DECT BMD or HU and central DXA BMD or T-score.

Pearson correlation analysis demonstrated significant correlation of dual-energy computed tomography (DECT) bone mineral density (BMD) and Hounsfield unit (HU) values with DXA (dual x-ray absorptiometry)-based BMD and T-score values.

**Table 3.** Diagnostic accuracy of DECT BMD and HU for osteoporosis diagnosis in distal femur and proximal tibia using DXA as the reference standard.


AUC: area under the curve. CI: confidence intervals. BMD: bone mineral density. HU: Hounsfield unit. DXA: dual-energy X-ray absorptiometry. Min: Minimum. Max: Maximum. L2: second lumbar spine.

**Figure 3.** *Cont*.

**Figure 3.** (**A**) Box plots of volumetric bone mineral density (BMD) and Hounsfield unit (HU) values in the distal femur derived from dual-energy computed tomography (DECT). The plots show the distribution of values that were categorized as normal BMD, osteopenia, and osteoporosis according to the dual x-ray absorptiometry (DXA)-derived T-score in the lumbar spine, which served as the reference standard; (**B**) Box plots of volumetric bone mineral density (BMD) and Hounsfield unit (HU) values in the proximal tibia derived from dual-energy computed tomography (DECT). The plots show the distribution of values that were categorized as normal BMD, osteopenia, and osteoporosis according to the dual x-ray absorptiometry (DXA)-derived T-score in the lumbar spine, which served as the reference standard; (**C**) Box plots of volumetric bone mineral density (BMD) and Hounsfield unit (HU) values in the distal femur derived from dual-energy computed tomography (DECT). The plots show the distribution of values that were categorized as normal BMD, osteopenia, and osteoporosis according to the dual x-ray absorptiometry (DXA)-derived T-score in the femur neck, which served as the reference standard; (**D**) Box plots of volumetric bone mineral density (BMD) and Hounsfield unit (HU) values in the proximal tibia derived from dual-energy computed tomography (DECT). The plots show the distribution of values that were categorized as normal BMD, osteopenia, and osteoporosis according to the dual x-ray absorptiometry (DXA)-derived T-score in the femur neck, which served as the reference standard.

**Figure 4.** (**A**) Representative receiver operating characteristic (ROC) curves of phantomless volumetric bone mineral density (BMD) (blue line) values and Hounsfield unit (HU) measurements (green line) derived from dual-energy computed tomography (DECT) for the detection of osteoporosis using the dual x-ray absorptiometry (DXA) derived T-score of the lumbar spine as the reference standard. (Left: values in distal femur; Right: values in proximal tibia); (**B**) Representative receiver operating characteristic (ROC) curves of phantomless volumetric bone mineral density (BMD) (blue line) values and Hounsfield unit (HU) measurements (green line) derived from dual-energy computed tomography (DECT) for the detection of osteoporosis using the dual x-ray absorptiometry (DXA) derived T-score of the femur neck as the reference standard. (Left: values in distal femur; Right: values in proximal tibia).

Intra- and inter-observer reliability for all radiographic measurements was considered acceptable, ranging from 0.81 to 0.99 and 0.81 to 0.96, respectively.

#### **4. Discussion**

Aseptic loosening is a major cause of failure after TKA in young and active patients, and the need for optimal biological fixation of cementless prostheses is growing [1,6–9,29,30]. Preoperative poor bone quality is associated with a higher rate of micromotion and migration in cementless prostheses. Meanwhile, there is no gold standard method for evaluating bone quality in peripheral regions [19,31]. We investigated the diagnostic accuracy of DECT BMD and HU for osteoporosis around the knee and compared the DECT values with the central DXA. In addition, we calculated the optimal cut-off values of DECT BMD and HU for diagnosing osteoporosis around the knee.

We found that BMD of central DXA was significantly correlated with DECT BMD and HU in both central and peripheral ROIs. The DXA BMD and T-scores for L2 and the femur neck were strongly correlated with DECT BMD and HU in their own region (all correlation coefficients > 0.5). In addition, the DXA BMD of L2 and the femur neck were moderately correlated with DECT BMD and HU in all other ROIs, except for a weak correlation of the DXA BMD at the femur neck with DECT HU at the proximal tibia. Our results concur with previous studies reporting the correlation between DXA BMD and radiologic values obtained from opportunistic CT scans [20,26,27]. However, while a substantial number of studies have found correlations between central DXA BMD and DECT BMD or HU [20–22,26,27], it is difficult to compare those studies with ours because all of the previous studies were confined to ROIs other than the knee joint. According to our study, DECT BMD and HU were more strongly correlated with DXA BMD at L2 than at the femur neck. In addition, there was stronger correlation between DECT BMD than DECT HU with DXA BMD. These results concur with a previous study reporting a stronger correlation between DXA BMD and DECT BMD than DECT HU [21]. Our study results demonstrate that DECT can yield more accurate and precise volumetric BMD than HU, permitting opportunistic BMD measurements in routine CT scans without the need for calibration phantoms. Our results indicate that DECT for the routine preoperative evaluation of lower extremity axial alignment can also be a useful method of evaluating preoperative bone quality in patients who are about to undergo cementless TKA.

The result of this study, using DXA T-score as the reference standard, also supports the hypothesis that DECT BMD and HU can provide reliable diagnostic accuracy in assessing bone quality around the knee joint. DECT BMD provided the highest diagnostic accuracy with DXA BMD at L2 as the reference standard (all AUC values: L2 > femur neck; DECT BMD > DECT HU, respectively) (Table 3). Moreover, the reference of L2 provided more precise diagnostic value in the tibia than in the femur. This result concurs with previous studies finding that phantomless volumetric DECT BMD offered significantly more accurate BMD assessment and superior diagnostic accuracy for osteoporosis than DECT HU [21]. Our results suggest that DECT BMD has the potential to be a gold standard method for evaluation of BMD around the knee joint. However, because this finding is based on the central DXA T-score as the reference standard, further biomechanical research regarding the correlation between DECT BMD or HU and the true bone strength around the knee is required.

Interestingly, BMD evaluation using DECT may have additional benefits and possibilities compared to DXA. Although DXA has been regarded as a gold standard preferred by the International Society for Clinical Densitometry (ISCD), it has certain limitations in clinical application, and variations in body composition can lead to up to 20% error [32]. DXA cannot be used in patients with scoliosis or calcifications from chronic disease, metal implants in both hips or at multiple levels of the spine, or cement in a vertebral body. In addition, DXA is based in two dimensions, meaning it cannot distinguish cancellous bone from cortical bone quality. On the other hand, the three-dimensional DECT analysis can discriminate cancellous bone, which may comprise most of the bone-prosthesis interface between cementless TKA and cortical bone. There are already studies reporting a higher rate of successful fusion and lower rates of adjacent spine fracture following spine surgery in patients with higher HU [33–36]. Moreover, there also are studies reporting no significant difference between HU measurements on sagittal and axial CT images [27]. These previous studies may further support the possibility of utilizing DECT BMD and HU as a tool for evaluation of preoperative bone quality around the knee in patients undergoing TKA. This study assessing BMD around the knee joint proposes a novel method in evaluating bone quality around peripheral regions. Moreover, there is a possibility to expand this method to other peripheral joints. If surgeons can get information about a patient's bone quality around the knee or other peripheral joints from a preoperative CT scan, they may be able to screen for the suitability of a cementless implant and modify their technique or type of implant according to the findings. Routine preoperative DECT may be helpful for detecting unexpected poor bone quality that could provoke early prosthesis failure.

This study had several limitations. First, it was confined to an Asian population, and most of the patients undergoing TKA who were included in this study are women

(102/117, 87%), and it may be difficult to generalize our results to other ethnicities. It is not yet clear why the majority of patients with arthritis in Korea are females [37]. Second, the study was confined to middle-aged and older patients (70.6 ± 6.5, 54 to 88) because only patients who underwent TKA were included. Third, although there was significant correlation between DXA BMD and DECT BMD, the clinical meaning of DECT BMD values remains unclear. As a way to overcome this limitation, further study examining the direct correlation between DECT BMD values and the properties of actual bone is required. Fourth, there are concerns of increased radiation exposure by utilizing DECT. However, DECT has been reported not to increase radiation exposure compared with single energy CT [38]. In addition, preoperative lower extremity CT is already part of the routine preoperative planning protocol for TKA by measured resection technique. Thus, the opportunistic measurement of BMD around the knee using DECT does not seem to increase radiation exposure compared with standard clinical practice. Fifth, because of the small study population, it is possible that the study was underpowered and subject to type-II error with respect to detecting all relevant outcomes. Sixth, analysis on factors possibly affecting BMD is not done in this study. Further study focused on the possible factors is required to expand the clinical relevance and would provide more valuable information. Finally, there is a possibility of limited reproducibility because of discrepancies in HU and BMD values among institutions with different CT scanners, DXA devices, and software [20]. Nonetheless, it is still possible to use DECT BMD as a tool to evaluate BMD around the knee, and despite these limitations, our study appears to provide valuable information on the practical use of DECT for osteoporosis screening around the knee in patients scheduled for cementless TKA.

#### **5. Conclusions**

Volumetric BMD assessment using DECT is accurate, relatively simple, and does not require further equipment. DECT BMD and HU are useful tools for osteoporosis screening before cementless TKA in clinical practice. Further research focused on the correspondence between the actual bone quality of the distal femur and proximal tibia and DECT BMD and HU is required to expand the clinical relevance.

**Author Contributions:** Data curation, I.J.K. and S.-W.L.; Software, S.-W.L.; Investigation, K.Y.C., Y.I., M.S.K., Y.D.K., S.-y.L. and J.-W.L.; Supervision, I.J.K.; Visualization, K.Y.C.; Writing—original draft, K.Y.C.; Writing—review and editing, K.Y.C., S.-W.L. and I.J.K. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was supported by a grant funded by The Catholic University of Korea, Eunpyeong St. Mary's Hospital, Research Institute of Medical Science in program year 2022. (EPSMH-C-2022-01).

**Institutional Review Board Statement:** The study was conducted according to the guidelines of the Declaration of Helsinki and approved by the Institutional Review Board of Eunpyeong St. Mary's Hospital (PC22RISI0049, 2018-05-04).

**Informed Consent Statement:** Patient consent was waived because this study is a retrospective medical and radiologic record review.

**Data Availability Statement:** Data collected for this study, including individual patient data, will not be made available.

**Acknowledgments:** We thank Seongyong Pak, Siemens Healthcare, for technical consultation.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


### *Article* **Guided-Motion Bicruciate-Stabilized Total Knee Arthroplasty Reproduces Native Medial Collateral Ligament Strain**

**Dai-Soon Kwak 1, Yong Deok Kim 2,3, Nicole Cho 4, Ho-Jung Cho 1, Jaeryong Ko 2,3, Minji Kim 5, Jae Hyuk Choi 5, Dohyung Lim <sup>5</sup> and In Jun Koh 2,3,\***


**Abstract:** *Background and Objectives*: Guided-motion bicruciate-stabilized (BCS) total knee arthroplasty (TKA) includes a dual cam-post mechanism with an asymmetric bearing geometry that promotes normal knee kinematics and enhances anterior-posterior stability. However, it is unclear whether the improved biomechanics after guided-motion BCS TKA reproduce soft tissue strain similar to the strain generated by native knees. The purpose of this cadaveric study was to compare medial collateral ligament (MCL) strain between native and guided-motion BCS TKA knees using a video extensometer. *Materials and Methods*: Eight cadaver knees were mounted onto a customized knee squatting simulator to measure MCL strain during flexion in both native and guided-motion BCS TKA knees (Journey II-BCS; Smith & Nephew, Memphis, TN, USA). MCL strain was measured using a video extensometer (Mercury® RT RealTime tracking system, Sobriety s.r.o, Kuˇrim, Czech Republic). MCL strain level and strain distribution during knee flexion were compared between the native and guided-motion BCS TKA conditions. *Results*: The mean and peak MCL strain were similar between native and guided-motion BCS TKA knees at all flexion angles (*p* > 0.1). MCL strain distribution was similar between native and BCS TKA knees at 8 of 9 regions of interest (ROIs), while higher MCL strain was observed after BCS TKA than in the native knee at 1 ROI in the mid portion of the MCL at early flexion angles (*p* < 0.05 at ≤30◦ of flexion). *Conclusions:* Guided-motion BCS TKA restored the amount and distribution of MCL strain to the values observed on native knees.

**Keywords:** knee; medial collateral ligament; strain; video extensometer; total knee arthroplasty

#### **1. Introduction**

Despite advancements in total knee arthroplasty (TKA), many patients have residual symptoms or unsatisfactory outcomes after TKA [1–6]. Recently, guided-motion bicruciatestabilized (BCS) TKA, which involves asymmetric bearing geometry and dual substitution for the anterior cruciate ligament (ACL) and posterior cruciate ligament, has been introduced [7–9]. Guided-motion BCS TKA has an asymmetric femoral component, a polyethylene insert with 3◦ of tibial varus, a medially concave and laterally convex shape, and a dual cam-post mechanism (Figure 1). The goal of guided-motion BCS TKA is to facilitate guided motion that is closest to normal knee kinematics. A growing body of evidence supports that guided-motion BCS TKA can mimic native knee kinematics, improve recovery and activity, and provide more natural knee sensations compared to conventional TKA [5,6,8–11]. However, the biomechanical mechanisms that underlie these improved results remain unclear.

**Citation:** Kwak, D.-S.; Kim, Y.D.; Cho, N.; Cho, H.-J.; Ko, J.; Kim, M.; Choi, J.H.; Lim, D.; Koh, I.J. Guided-Motion Bicruciate-Stabilized Total Knee Arthroplasty Reproduces Native Medial Collateral Ligament Strain. *Medicina* **2022**, *58*, 1751. https://doi.org/10.3390/ medicina58121751

Academic Editor: Jose Antonio de Paz

Received: 25 October 2022 Accepted: 26 November 2022 Published: 29 November 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

**Figure 1.** The JOURNEYTM II Bi-cruciate Stabilized Total Knee System (Smith & Nephew, Memphis, TN, USA) has an asymmetric femoral component, a polyethylene insert replicating 3◦ of the tibial varus, a medially concave and laterally convex shape, and a dual cam-post mechanism.

As the medial collateral ligament (MCL) plays critical roles in primary restraint against mechanical stresses and neurosensory feedback, changes in MCL tension after TKA affect postoperative kinematics and outcomes. Therefore, restoration of appropriate MCL strain is essential for optimal performance after TKA. Despite the significant influence of the ACL on restoring native knee kinematics, most TKA designs sacrifice the ACL without providing a substitute for its function. In the conventional ACL-deficient TKA design, the MCL and remaining capsular structures provide restraint against anterior tibial translation. Previous studies have reported that MCL strain, compared to native knees, worsened after conventional ACL-deficient TKA [12–18]. Theoretically, BCS TKA may restore post-TKA MCL strain to levels observed in native knees by enhancing anterior-posterior stability. However, no previous studies have investigated the changes in MCL strain after BCS TKA, and it is unclear whether BCS TKA restores native strain of the MCL.

The purpose of this study was to compare MCL strain between native knees and guided-motion BCS TKA knees using a video extensometer, a highly accurate tool for measuring surface strain in human cadavers [19,20]. We hypothesized that guided-motion BCS TKA would restore the strain level and distribution of the MCL to the values observed in native healthy knees.

#### **2. Materials and Methods**

Eight fresh-frozen knees from eight men with a mean age of 79 years (range: 49–96 years) were used. The specimens were macroscopically intact and did not exhibit any gross pathology. The specimens were frozen at −20 ◦C until the evening before dissection, when they were thawed at room temperature. The lower extremity specimens were prepared by disarticulating the hip joint; high-resolution photographs were obtained to measure the preoperative hip-knee-ankle axis. The hip-knee-ankle axis showed a varus angle of 1.3 ± 3.2◦ (range: varus 4◦ to valgus 6◦). The skin and subcutaneous tissue were dissected without damage to the extensor mechanism, retinaculum, knee capsule, or periarticular soft tissues. The quadriceps femoris was separated into the vastus medialis, rectus femoris/vastus intermedius, and vastus lateralis. The hamstring muscles were separated into the biceps femoris and semimembranosus/semitendinosus. The separated muscle branches were sutured using wire to connect the muscles and allow transmission of force. The femur was cut 30 cm proximal to the joint line; the tibia was cut 25 cm distal to the joint line. The ends of the femur and tibia were anatomically positioned and a cylindrical resin mold (Z-Grip, Evercoat, OH, USA) was created to mount the knee squatting simulator. Each specimen was securely mounted in its original axial alignment on a customized knee squatting simulator system (RNX & Corentec, Seoul, Republic of Korea) based on the original Oxford rig (Figure 2) [21]. This system, which is an opened loop control system with a motor type actuator with 7.5º/sec angular velocity, produced continuous flexion-extension motion while permitting physiological muscle loading and 6◦ of freedom positioning. The multiplane loading of the quadriceps and hamstring muscles was used to simulate physiological loading of the knee joint (vastus medialis 51 N, rectus femoris/ vastus intermedius 87 N, and vastus lateralis 77 N, biceps femoris 31 N, semimembranosus/ semitendinosus 54 N). The total loading was 300 N [17]. These loading parameters were based on the ratios of the physiological cross-sectional area of the muscles as described in a previous anatomical study [22]. The muscles were loaded in a multiplanar fashion as described in previous studies [23,24].

**Figure 2.** Schematic drawing of knee squatting simulator with six degrees-of-freedom.

A single experienced surgeon (one of authors) performed the arthroplasties using the JOURNEYTM II Bi-cruciate Stabilized Total Knee System (Smith & Nephew, Memphis, TN, USA). A subvastus approach was used to expose the knee joint, whereas the patellae were left un-resurfaced. TKA was performed using the conventional measured resection technique. The distal femur was resected using individualized intramedullary instrumentation based on the difference between the mechanical and anatomical axes of the specimen. The trans-epicondylar axis served as a reference to determine the extent of external rotation of the femoral component. Coronal and sagittal resection of the proximal tibia was performed using extramedullary instrumentation at a cutting angle of 90◦ to the tibial axis. Finally, flexion gaps at 0 and 90◦ were measured using a tensor device (B Braun-Aesculap, Tuttlingen, Germany) with a 200-N distraction force [25].

The superficial MCL was identified as previously described, then stained with multiple random speckles prior to measurement of MCL strain [18,26]. Real-time changes in MCL strain during flexion were analyzed using a non-contact video extensometer that consisted of a high-resolution digital camera (ISG, MONET 3D, Sobriety s.r.o, Czech Republic) and real-time image processing software (ISG, Mercury RT x64 2.7, Sobriety s.r.o). The camera was installed 1 m from the specimen to produce a field of view of 485 (w) × 383 mm and resolution of 1.87 μm. Illumination was minimized by performing the tests in darkness under two 36-watt light-emitting diode lights (Figure 3). MCL strain was measured at knee flexion angles of 0–120◦, at intervals of 15◦. Repeatability was checked in real-time to raise reliability. Each measurement was performed in three flexion-extension cycles and all of graphs following each of the three cycles were plotted at the same time. The patterns of all three cycles were compared and repeatability was analyzed. After repeatability was confirmed, data from the 3rd cycle were used for statistical analyses in all specimens. The native and post-TKA MCL strains were compared in terms of quantity and distribution. The mean MCL strain was determined by measuring the strain over the entire MCL and the peak MCL strain at one-fourth of the entire MCL area. The distribution of MCL strain was evaluated by measuring MCL strain at 9 regions of interest (ROIs), drawn by dividing the ligament vertically (front to back) into equal anterior (A), middle (M), and posterior (P) regions, and horizontally (top to bottom) into equal regions numbered 1–3. MCL strain was measured in all specimens before and after arthroplasty.

**Figure 3.** Experimental setup for biomechanical testing. A high-resolution digital camera was installed 1 m from the specimen, which was mounted onto a customized knee squatting simulator system.

#### *Statistical Analysis*

Data are presented as means ± standard deviations. Paired t-tests were used to determine whether the mean and peak MCL strains differed between native and post-TKA knee specimens. The Shapiro–Wilk test confirmed that this data set was normally distributed. The variables subjected to multiple between-group comparisons included the mean, peak, and distribution of MCL strain, which were analyzed using repeated-measures ANOVA, followed by the Bonferroni corrected post hoc test in order to protect against multiple comparison bias. In addition, paired t-tests were used to determine differences in the distribution of MCL strain at 9 ROIs between native and BCS TKA knee specimens. Data analysis was performed using SPSS software for Windows (ver. 26.0; IBM Corp., Armonk, NY, USA). *p* < 0.05 was considered to indicate statistical significance. A priori power analysis based on the results of our previous study of changes in the MCL strain in native knees was performed to determine the necessary sample size to achieve sufficient statistical power. Using a two-sided hypothesis test at an alpha level of 0.05 and power

of 80%, 7 knees were required to detect a 5% difference. A 5% change was considered biomechanically meaningful since ligament damage occurs at 5% strain [27].

#### **3. Results**

Guided-motion BCS TKA restored the mean and peak MCL strain to the levels present in native knees during knee flexion. Guided-motion BCS TKA provided mean strain measurements similar to the levels in native knees at all flexion angles [Mean (SD) native vs. J2 BCS mean MCL strain (%) at 15◦, 2.4 (1.5) vs. 3.5 (1.1), *p* = 0.96; at 30◦, 3.7 (2.7) vs. 5.2 (2.4), *p* = 0.43; at 45◦, 4.5 (3.0) vs. 6.0 (3.0), *p* = 0.48; at 60◦, 5.1 (3.1) vs. 6.6 (3.7), *p* = 0.56; at 75◦, 5.4 (3.4) vs. 7.1 (4.4), *p* = 0.64; at 90◦, 5.6 (3.5) vs. 7.6 (5.2), *p* = 0.72; at 105◦, 5.9 (3.7) vs. 8.0 (5.7), *p* = 0.88; at 120◦, 5.9 (3.9) vs. 8.2 (5.9), *p* = 0.80] (Figure 4). The peak strains at one-fourth of the MCL area were similar between guided-motion BCS TKA and native knees [Mean (SD) native vs. J2 BCS peak MCL strain (%) at 15◦, 5.3 (4.3) vs. 7.7 (3.4), *p* = 0.32; at 30◦, 11.5 (10.9) vs. 14.3 (13.9), *p* = 0.87; at 45◦, 14.6 (11.1) vs. 17.2 (16.7), *p* = 0.92; at 60◦, 17.3 (11.4) vs. 19.1 (17.3), *p* = 0.96; at 75◦, 18.6 (12.0) vs. 20.6 (17.5), *p* = 0.99; at 90◦, 19.4 (12.5) vs. 21.3 (17.4), *p* = 0.93; at 105◦, 20.3 (12.3) vs. 21.4 (16.4), *p* = 0.70; at 120◦, 20.6 (12.0) vs. 20.8 (14.2), *p* = 0.94] (Figure 5).

**Figure 4.** Area for mean MCL strain measurements (**A**) and comparisons of mean MCL strain between native and BCS TKA knees (**B**). The mean strain after BCS TKA was similar to the mean strain in native knees at all flexion angles. Error bars indicate standard deviations. The correspondence between color and strain is shown in the color bar.

Guided-motion BCS TKA restored the native knee MCL strain distribution. Significant strain differences between guided-motion BCS TKA and native knees were observed in 1 of the 9 ROIs (i.e., at M2). In this ROI, strain was higher after guided-motion BCS TKA than in native knees only at ≤30◦ of flexion [Mean (SD) native vs. J2 BCS MCL strain at M2 (%) at 15◦, 0.69 (0.21) vs. 2.4 (0.4), *p* = 0.048; at 30◦, 0.74 (0.30) vs. 2.9 (0.6), *p* = 0.024; at 45◦, 1.3 (0.4) vs. 3.2 (0.7), *p* = 0.12; at 60◦, 1.7 (0.5) vs. 3.4 (0.9), *p* = 0.32; at 75◦, 1.9 (0.6) vs. 3.7 (1.2), *p* = 0.72; at 90◦, 2.0 (0.7) vs. 3.9 (1.5), *p* = 0.93; at 105◦, 2.0 (0.7) vs. 4.3 (1.9), *p* = 0.99; at 120◦, 1.9 (0.7) vs. 4.6 (2.0), *p* = 0.94] (Figure 6).

**Figure 5.** Area for peak MCL strain measurements (**A**) and comparisons of peak MCL strain between native and BCS TKA knees (**B**). The peak strain was measured at one-fourth of the entire MCL area, including the highest MCL strain portion. The peak MCL strain after BCS TKA was similar to the peak MCL strain of native knees at all flexion angles. Error bars indicate standard deviations. The correspondence between color and strain is shown in the color bar.

#### **4. Discussion**

Despite improvements in TKA design, most TKA systems sacrifice the ACL, which may lead to suboptimal satisfaction and function. Guided-motion BCS TKA replicates the normal knee motion by promoting kinematics and anterior-posterior stability. Enhanced stability after guided-motion BCS TKA may normalize the soft tissue tension that contributes the perception of a normal knee. In this cadaveric study, we compared MCL strain

between native and post-BCS TKA knees using a video extensometer to determine whether guided-motion BCS TKA restores MCL strain to the level found in native knees.

Our findings suggest that guided-motion BCS TKA normalizes MCL strain. In this study, no significant differences were observed in the mean and peak strain measurements between native and guided-motion BCS-TKA knees at any flexion angle. The MCL provides restraint against anterior tibial translation in ACL-deficient knees [28]. Multiple previous studies have reported that MCL strain was significantly higher in post-TKA knees than in native knees [14,15,18] and postoperative MCL laxity was also higher in post-TKA knees than in native knees [16,17]. Our findings, when taken into account with previous studies, suggest that guided-motion BCS TKA successfully restored MCL strain to the level found in native knees. Additionally, free nerve endings that serve as a nociceptive system were reported to be the most commonly observed mechanoreceptors in the MCL [29]. Our findings, when taken into account with this anatomical detail, suggest that guided-motion BCS TKA may provide more normal feelings of the knee, which is strongly associated with patient satisfaction [11,12,30]. However, future studies that evaluate the MCL strain thresholds necessary to perceive the differences between normal and prosthetic knees are needed. Our findings also indicated that the standard deviations of the peak MCL strain were much higher than those of the mean MCL strain. One plausible explanation is that the difference in measuring area may affect MCL strain. In this study, the mean MCL strain was determined by measuring the strain over the whole MCL area and the peak MCL strain was found at one-fourth of the entire MCL area, and the distribution of MCL strain was evaluated at each of ROIs. Therefore, the surface strains measured at a smaller area, such as the peak MCL strain and strain at ROIs, are more susceptible to specimen-specific anatomical conditions, such as the bone contour underneath the measured MCL area and the soft tissues connected soft tissues to the MCL, than those measured at the entire MCL.

The results of the present study suggest that guided-motion BCS TKA can normalize the MCL strain distribution pattern. We found that guided-motion BCS TKA restored the strain to the level generated in native knee at 8 of 9 ROIs. Higher MCL strain was found at 1 ROI in the mid-portion of the MCL during flexion at ≤30◦. A previous cadaveric study discovered significantly higher strain after conventional ACL-deficient mechanically aligned TKA, compared to native knees, at all flexion angles in the proximal mid to posterior portion of the MCL [18]. Similar to the findings of previous studies, the current results suggest that guided-motion BCS TKA was able to better restore the native MCL strain pattern than conventional TKA. Considering the role of the MCL in neurosensory feedback, patients undergoing BCS TKA may experience greater perception of a normal knee, compared to patients who undergo conventional TKA. Yet, further studies evaluating the relationship between MCL strain and neurosensory feedback changes are required.

There were several noteworthy limitations in this cadaveric study. First, the specimen preparation and testing environment may not have replicated natural conditions and normal squatting loads and patterns. Properties of the MCL may have been changed following rigorous soft tissue balancing, which could have affected the measured surface strain. Additionally, we used muscle loads from previous cadaveric studies investigating knee squatting motions, to simulated physiological loads [23,24], but these loading parameters were too small to simulate in a normal-weight person. This should be considered before extrapolating our cadaveric test findings to real clinical scenarios. Second, we only tested guided-motion BCS TKA implants without comparing them to conventional ACL-deficient TKA implants. Thus, we compared our data with previous studies that investigated conventional mechanically aligned TKA. Third, it is difficult to determine the clinical relevance of our findings because both the threshold of MCL strain that triggers nociception and the cut-off strain value for mechanical failure in the human knee are unknown. Finally, video extensometer analyses have some inherent limitations in terms of image processing, such as poor resolution and quality, as well as distortion of the digital image. Nevertheless, this novel, non-contact analysis technique eliminates the need for strain gauge implantation, thereby reducing the risk of changes to ligament properties. Despite its limitations, this

cadaveric study is the first to report MCL strain measurements using a video extensometer after guided-motion BCS TKA. Therefore, our results provide valuable information regarding MCL strain patterns following guided-motion BCS TKA.

#### **5. Conclusions**

Our study demonstrates that guided-motion BCS TKA restores the amount and distribution of MCL strain to the levels found in native knees during knee flexion. These findings may explain the perception of a normal knee after guided-motion BCS TKA, rather than conventional ACL-deficient TKA.

**Author Contributions:** Conceptualization, I.J.K. and D.L.; Data curation, I.J.K., D.-S.K., H.-J.C., M.K. and D.L.; Formal analysis, I.J.K., Y.D.K., J.K. and N.C.; Funding acquisition, I.J.K. and D.L.; Resources, D.-S.K. and H.-J.C.; Software, M.K., J.H.C. and D.L.; Investigation, D.-S.K., H.-J.C., M.K., J.H.C., D.L. and I.J.K.; Supervision, D.L. and I.J.K.; Validation, I.J.K.; Visualization, I.J.K., D.-S.K. and D.L.; Writing—original draft, D.-S.K. and I.J.K.; Writing—review and editing, N.C., D.-S.K., D.L., J.K. and I.J.K. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work was supported by the National Research Foundation of Korea (NRF) grant funded by the Korea government (MSIT) (No. 2019 R1F1A 1057842 and No. 2017M3A9E9073545).

**Institutional Review Board Statement:** This cadaveric study was exempt from the institutional review board of our institution because it did not involve human subjects.

**Informed Consent Statement:** Written informed consent for use of the cadavers and consent for use of future research on the related materials were provided by all donors or authorized representatives.

**Data Availability Statement:** All data are presented in the article. Instrumental readings are available upon request from the corresponding author.

**Acknowledgments:** We thank Smith & Nephew, Korea for providing the surgical instruments used in this study. We also thank Su Gu Chai, B.S. and Seung Jun Lee, P.R.S., at Smith & Nephew, Korea, Ki Joon Yoo, B.A., Jae Young Sung, B.S., and Hyunggu Han, B.A., of Daon HealthCare, Korea, for their assistance in the test. We also thank the cadaver donors and their families.

**Conflicts of Interest:** The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript; or in the decision to publish the results.

#### **References**


### *Article* **Accuracy of the Tibial Component Alignment by Extramedullary System Using Simple Radiographic References in Total Knee Arthroplasty**

**Jin-Ho Cho, Jun Young Choi and Sung-Sahn Lee \***

Department of Orthopedic Surgery, Ilsan Paik Hospital, Inje University School of Medicine, Goyang-si 10380, Korea

**\*** Correspondence: sungsahnlee@gmail.com; Tel.: +82-31-910-7301

**Abstract:** *Background and Objectives*: The tibial component alignment is an important issue for the longevity of total knee arthroplasty (TKA). The purpose of our study was to investigate the usefulness of proximal tibial references determined by pre-operative radiography and intraoperative C-arm-guided hip and ankle center marking for the extramedullary guided tibial cut in mild (<10◦) and severe (≥10◦) varus knee TKA. *Materials and Methods*: A total of 150 consecutive patients (220 cases) who underwent total knee arthroplasty who were recruited from July 2011 to April 2017 were reviewed retrospectively. Before surgery, the proximal tibial reference point and medio-lateral cut thickness difference were identified. Then, hip and ankle centers were checked using a C-arm intensifier intraoperatively. The hip–knee–ankle (HKA) alignment and medial proximal tibial angle (MPTA) were assessed pre-operatively and post-operatively. More than 3◦ varus or valgus of HKA alignment or tibial component angle was defined as an outlier. *Results*: Mean follow-up duration was 26.9 months. Among 220 cases, 111 cases are classified as mild varus group and 109 cases are classified as severe varus group. The HKA alignment is significantly improved (*p* < 0.001). The average tibial component angle after surgery is 90.1◦. A total of 21 cases (9.5%) and 3 cases (1.4%) are classified as outliers of HKA alignment and MPTA, respectively. Among MPTA outliers, one case is in the mild varus group and two cases are in the in severe varus group (*p* = 0.62). *Conclusion*: Measurement of proximal tibial radiographic references and checking the C-arm-guided intraoperative hip and ankle center could be helpful to obtain the favorable coronal position of the tibial component in the extramedullary guided tibial cut.

**Keywords:** total knee arthroplasty; tibial component alignment; radiographic references; extramedullary system

#### **1. Introduction**

Kinematic alignment was recently investigated as an alternative to mechanically aligned total knee arthroplasty (TKA) [1]. Several studies demonstrate that kinematically aligned TKA shows similar or better clinical outcomes due to less disruption of the native soft tissue envelope [2,3]. As a similar theoretical background, some studies show that slight under-corrected TKA results in better clinical outcomes than neutrally aligned TKA [4]. However, many studies suggest that post-operative coronal alignment is associated with implant survival; in particular, tibial component varus position is strongly correlated to implant loosening [5–7].

To enhance post-operative alignment, the computer navigation instruments or patientspecific implants (PSI) can be used [8–10]. However, these devices are not always available and have additional costs, therefore, many surgeons perform conventional jig-based TKA—intramedullary guided distal femoral cut and extramedullary guided proximal tibia cut. Various anatomical landmarks (extensor hallucis longus, dorsal pedis artery,

**Citation:** Cho, J.-H.; Choi, J.Y.; Lee, S.-S. Accuracy of the Tibial Component Alignment by Extramedullary System Using Simple Radiographic References in Total Knee Arthroplasty. *Medicina* **2022**, *58*, 1212. https://doi.org/10.3390/ medicina58091212

Academic Editors: Yong In and In Jun Koh

Received: 9 August 2022 Accepted: 1 September 2022 Published: 2 September 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

intermalleolar point, anterior tibial border, intercondylar eminence) are used as a reference for extramedullary alignment to enhance tibial component coronal alignment [11–13]. Thippana et al. [14] reported that using the line that connects the proximal tibial reference point defined by pre-operative radiography and ankle center was helpful to set the extramedullary guide. In addition to this landmark, we thought pre-determined proximal tibial medio-lateral (ML) cut thickness difference by radiograph (Figure 1) and the intraoperative C-arm-intensifier-guided hip and ankle center marking method are helpful to enhance tibial component position.

**Figure 1.** Proximal tibial medio-lateral (ML) cut thickness difference. The lines were drawn (1) along the anatomical axis of the tibia and (2) perpendicular to first line (starting from medial condylar edge). The *white arrow line* indicates proximal tibial 'ML cut thickness difference'.

The purpose of this study was (1) to investigate the usefulness of proximal tibial references determined by pre-operative radiography and intraoperative C-arm-guided hip and ankle center marking, and (2) to compare radiographic measurements between the patients with pre-operative mild varus (<10◦) and severe varus (≥10◦) deformity. It was hypothesized that favorable coronal alignment of the tibial component might be shown by this method, regardless of the severity of pre-operative varus deformity.

#### **2. Materials and Methods**

#### *2.1. Patients*

This study is a retrospectively designed study. From July 2011 to April 2017, the patients who underwent primary TKA surgery with the same method for degenerative osteoarthritis with varus deformity were reviewed. The exclusion criteria were as follows: patients who (1) followed up for less than 1 year, (2) had undergone other previous bony procedure, such as osteotomy, and (3) diagnosed systemic arthritis such as rheumatoid arthritis. The enrolled patients were divided into mild varus group (<10◦) and severe varus group (≥10◦). based on pre-operative hip–knee–ankle (HKA) alignment as the definition of previous studies [15,16]. Written informed consents were obtained from all patients before enrolling them in the study. This study was approved by the IRB of the authors' affiliated institutions (ISPAIK 2019-02-014 at 20 February 2019).

#### *2.2. Radiographic and Clinical Assessment*

We analyzed the HKA alignment taken from the pre-operative and 1 year follow-up low extremity whole radiography [17,18]. As in previous studies, more than 3◦ varus or valgus of HKA alignment was defined as an outlier [4,19]. The medial proximal tibial angle (MPTA) is the medial angle of intersection between the anatomical axis of the tibia and the horizontal axis of the proximal tibia (pre-operative measurement) or tibial component (post-operative measurement). Refs. [20,21] are same as HKA alignment, more than 3◦ varus or valgus of tibial component angle was defined as an outlier [4,19]. The tibial slope was also measured as previously described [22,23] (Figure 2).

**Figure 2.** Measurement of the pre- and post-operative hip–knee–ankle (HKA) alignment, medial proximal tibial angle (MPTA, (**A**,**B**)) and tibial slope (**C**,**D**).

All radiographs were measured using Marosis software (INFINITT Healthcare, Seoul, Korea). The radiographs were evaluated by two independent orthopedic surgeons specializing in knee arthroplasty, who did not participate in the current study, to verify inter-observer reliability. The intra-observer reliability was checked by having the observers repeat the same measurements 6 weeks later. Intraclass correlation coefficients (ICCs) were used for intra-observer and inter-observer reliabilities.

The clinical parameters of pre-operation and last follow-up were evaluated by the following clinical scores: Hospital for Special Surgery (HSS) score [24] and Western Ontario Mac-Master University (WOMAC) Index [25,26]. In addition, the operative time and incision size were analyzed.

#### *2.3. Pre-Operative Planning*

Pre-operative tibial anteroposterior (AP) view was used to verify the proximal tibial reference point. The proximal tibial reference point is determined as the meeting point between the tibial anatomical axis and the proximal tibial joint line. Of the enrolled cases, all proximal tibial reference points existed between the center of intercondylar eminence and the lateral tibial spine. These points were classified as 3 zones—the center of intercondylar eminence, lateral tibial spine, and in-between (Figure 3).

The perpendicular line to the tibial anatomical axis was drawn from the edge of medial tibial condyle. The ML cut thickness difference of proximal tibia was measured (Figure 1).

#### *2.4. Intraoperative Planning*

After induction of anesthesia to the patients, the ankle center and hip center should be assessed under a C-arm intensifier to apply the pre-operative planning before draping. A long rod (High tibia osteotomy alignment rod, DePuy Synthes, Raynham, MA, USA) was used to confirm the hip center and ankle center. The patient was placed in a metal pelvic stabilizer with the mobile peg, so that the measured position did not change during

surgery [27]. The hip center was marked with a mobile peg, and the end of the long rod was marked in the patient's ankle using a marking pen (Figure 4).

**Figure 3.** Proximal tibial reference point. These points were classified as 3 zones—(**A**) the center of intercondylar eminence, (**B**) in-between, and (**C**) lateral tibial spine.

**Figure 4.** Marking hip and ankle centers. (**A**,**B**) Check hip and ankle centers using a metal rod under C-arm intensifier. (**C**) A hip center was marked using mobile peg in a pelvic stabilizer and an ankle center was marked with a marking pen.

#### *2.5. Operative Procedure*

All surgeries were performed by the senior author. In all cases, posterior, stabilized TKA implant was used (LEGION, Smith & Nephew, London, UK). Skin incision and medial parapatellar arthrotomy were performed. Both cruciate ligaments were removed from their femoral and tibial attachment sites and osteophytes were removed. Then, the intramedullary guided femoral cut was performed.

Extramedullary guided tibia cut was performed using pre- and intra-operatively checked references. The cartilages of lateral tibial condyle were removed. Extramedullary tibial cutting apparatus was installed using the proximal tibial reference point and marked ankle center. After that, we checked the ML thickness difference was similar between radiographic and intraoperative measurements using a cut thickness gauge.

After tibial cut and trial component insertion, marked ankle center and hip center should be connected by flexible cable (we used the cable-connecting electrocautery device) to check the tibial component coronal position and lower limb alignment. A proximal tibial cut was re-performed if the tibial component position was not perpendicular to cable (Figure 5).

**Figure 5.** Operative procedure. (**A**) Gentle peeling of cartilage of the lateral condyle for precise intraoperative measurement. (**B**,**C**) Application of tibial cutting apparatus according to the 'proximal tibial reference point', 'ankle center', and pre-operatively measured medio-lateral cut thickness difference. (**D**) Check the limb and component coronal alignment using cable method.

After the surgery, full weight-bearing was allowed. Then, after the removal of drains, a range of motion exercises and quadriceps strengthening exercises were started.

#### *2.6. Statistical Analyses*

The intraclass correlation coefficient was used to quantify both inter-observer and intra-observer reliabilities for radiographic assessment. A descriptive analysis was performed for all the variables, including calculation of the mean and standard deviation. Data normality was tested using a *Kolmogorov–Smirnov test*. The *paired T-test* was used to

compare between the pre-operative and post-operative parameters. The *Student's T-test* for continuous variables and chi-square tests for categorical variables were used to compare the parameters between mild varus group and severe varus group.

Statistical analyses were performed using SPSS software, version 18 (IBM Corp., Armonk, NY, USA). Statistical significance was determined as a *p*-value less than 0.05 for all analyses.

#### **3. Results**

A total 266 consecutive patients who underwent primary TKA surgery with the same method for degenerative osteoarthritis with varus deformity were reviewed. Among them, 80 patients were excluded who were followed up for less than 1 year. A total of 26 patients were excluded who previously underwent high tibial osteotomy or surgery for fracture. Ten patients who were diagnosed with rheumatoid arthritis were excluded. Finally, 150 patients (220 cases) were enrolled in the present study (Table 1). Among 220 cases, 111 cases are classified as mild varus group (<10◦) and 109 cases are classified as severe varus group (≥10◦).

**Table 1.** Demographic and preoperative radiographic parameters in 220 cases.


<sup>a</sup> Values are presented as mean ± standard deviation (range).

Mean follow-up duration was 26.9 months. Inter-observer and intra-observer ICCs show good agreement regarding radiographic measurement reliability (>0.80). The HKA alignment is significantly decreased (*p* < 0.001). The average tibial component angle after surgery is 90.1◦. A total of 21 cases (9.5%) and 3 cases (1.4%) are classified as an outlier of HKA alignment and tibial component angle, respectively. The average tibial slope before surgery is 9.3◦ and after surgery is 3.2◦ (*p* < 0.001). The average incision size is 12.3 cm and the average operative time is 75.1 min. The post-operative clinical outcomes including HSS score and WOMAC index are significantly improved compared to pre-operative values (*p* < 0.05) (Table 2).

**Table 2.** Comparison of clinical and radiographic data between pre- and post-operation.


HKA: hip–knee–ankle, MPTA: medial proximal tibial angle, HSS: Hospital for Special Surgery, WOMAC: Western Ontario Mac-Master University.

The severe varus group shows significantly more post-operative HKA alignment outliers than the mild varus group (mild varus group—6 cases (5.4%), severe varus group—15 cases (13.8%), *p* = 0.04, Table 3). The pre-operative MPTA is significantly greater in the mild varus group (*p* < 0.001). However, the post-operative MPTA outliers did not show statistical significance (mild varus group—one case (0.9%), severe varus group—two cases (1.5%), *p* = 0.62).


**Table 3.** Comparison of measurements between mild varus group and severe varus group.

HKA: hip–knee–ankle, MPTA: medial proximal tibial angle.

#### **4. Discussion**

The principal finding of our study is that a favorable coronal position of the tibial component could be obtained by using preplanned proximal tibial radiographic references and intraoperative synchronizing. The severe varus group shows similar outliers of tibial component coronal position, but significantly more outliers of HKA alignment compared to the mild varus group.

Computer navigation instruments or patient-specific instruments (PSI) are helpful in post-operative limb and implant alignment [8,9]. Cheng et al. investigated the efficacy of computer-navigation-assisted surgery by meta-analysis of 41 randomized controlled trials [8]. They report that malalignment of >3◦ from neutral alignment in the HKA alignment occurs in fewer patients in the computer-navigation-assisted group than in the conventional group (12.2% vs. 28.3%, respectively). Schotanus et al. conduced a meta-analysis of comparison between PSI-assisted and conventional TKA [9]. They demonstrate that MRI-based PSIs show a 19% decline in outliers compared to conventional TKAs. However, these devices are not always available and have additional cost, therefore, many surgeons perform conventional jigbased TKA. Both extramedullary and intramedullary guided cut can be used, and show similar results [28,29]. However, the intramedullary system shows less accuracy in patients with tibial bowing or post-traumatic deformities [30]. The installation of extramedullary tibia cutting instruments depends on anatomical landmarks, which include intercondylar eminence, center of tibial plateau, posterior cruciate ligament, and tibial tuberosity for the proximal side, and anterior tibialis tendon, extensor hallucis longus, dorsal pedis, and intermalleolar center for the distal side [11–13,31]. These landmarks may not be palpable, and the surgeon's experience is important to obtain an accurate tibial component position. We suggest that additional radiographic references (proximal tibial reference point, ML cut thickness difference) and intraoperative hip and ankle center marking by C-arm intensifier could enhance the tibial component coronal position. These additional references and center markings provide an accurate location of the extramedullary instrument and allow a second chance to check limb and component alignment after trial insertion. It is well-known that high-grade pre-operative varus deformity is associated with residual post-operative varus alignment [32]. In the current study, the severe varus group shows more HKA alignment outliers than the mild varus group, which is concurrent with previous studies. However, MPTA outliers are not significantly different between both groups. Previous studies suggest that the increasing severity of preoperative varus deformity is associated with complex bone cuts to restore a neutral mechanical alignment [1,33]. According to our findings, we think our methods aid in obtaining tibial component neutral alignment, regardless of the severity of pre-operative varus deformity.

Proximal tibial reference points of all cases exist between the center of intercondylar eminence and the lateral tibial spine in the current study. Kim et al. [34] suggest that lateral intercondylar spine rather than the center of intercondylar eminence should be used as a reference for the proximal tibial side in extramedullary guided tibial cut, in order to obtain a neutral component position. We agreed that routine use of the center of intercondylar eminence as a proximal reference point has a chance to highlight the varus position of the tibial component. However, we think an individualized proximal tibial reference point is better than routine use of lateral intercondylar spine. A further comparative study is needed to find the most reliable reference point of the proximal tibia.

This study has several limitations. First, it was not a comparative study with other methods. Second, the follow-up period was relatively short, and further studies are needed to evaluate implant survival. Third, there is not enough discussion on femur cutting with emphasis on increasing the accuracy of tibia cutting. Forth, we only focused on correcting coronal malalignment, and we need to discuss the correction technique for the sagittal plane as well. Fifth, in addition to the bone cutting, the collateral ligament and soft tissue were involved in the gap-balancing, but there was not sufficient information regarding this.

#### **5. Conclusions**

The measurement of proximal tibial radiographic references and checking C-armguided intraoperative hip and ankle centers could be helpful in obtaining a favorable coronal position of the tibial component for the extramedullary guided tibial cut in TKA.

**Author Contributions:** Conceptualization, J.-H.C. and S.-S.L.; methodology, J.-H.C. and S.-S.L.; validation, J.-H.C. and J.Y.C.; formal analysis, J.Y.C. and S.-S.L.; investigation, J.Y.C. and S.-S.L.; data curation, J.Y.C. and S.-S.L.; writing—original draft preparation, J.-H.C. and S.-S.L.; writing—review and editing, J.-H.C. and S.-S.L.; visualization, J.-H.C. and J.Y.C.; supervision, J.-H.C. and S.-S.L.; project administration, J.-H.C. and S.-S.L. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Institutional Review Board Statement:** The protocol used to evaluate radiographic findings and intraoperative navigation data was approved by our institution's investigational review board. (ISPAIK 2019-02-014 at 20 February 2019).

**Informed Consent Statement:** Informed consent was obtained from all individual participants included in the study.

**Data Availability Statement:** Not applicable.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


### *Article* **Topical Tranexamic Acid Can Be Used Safely Even in High Risk Patients: Deep Vein Thrombosis Examination Using Routine Ultrasonography of 510 Patients**

**Yong Bum Joo 1, Young Mo Kim 1, Byung Kuk An 1, Cheol Won Lee 1, Soon Tae Kwon <sup>2</sup> and Ju-Ho Song 3,\***


**Abstract:** *Background and Objectives*: Previous studies regarding tranexamic acid (TXA) in total knee arthroplasty (TKA) investigated only symptomatic deep vein thrombosis (DVT), or did not include high risk patients. The incidence of DVT including both symptomatic and asymptomatic complications after applying topical TXA has not been evaluated using ultrasonography. *Materials and Methods*: The medical records of 510 patients who underwent primary unilateral TKA between July 2014 and December 2017 were retrospectively reviewed. Because TXA was routinely applied through the topical route, those who had a history of venous thromboembolism, myocardial infarction, or cerebral vascular occlusive disease, were not excluded. Regardless of symptom manifestation, DVT was examined at 1 week postoperatively for all patients using ultrasonography, and the postoperative transfusion rate was investigated. The study population was divided according to the use of topical TXA. After the two groups were matched based on the propensity scores, the incidence of DVT and the transfusion rate were compared between the groups. *Results:* Of the 510 patients, comprising 298 patients in the TXA group and 212 patients in the control group, DVT was noted in 22 (4.3%) patients. Two patients had DVT proximal to the popliteal vein. After propensity score matching (PSM), 168 patients were allocated to each group. In all, 11 patients in the TXA group and seven patients in the control group were diagnosed with DVT, which did not show a significant difference (*p* = 0.721). However, the two groups differ significantly in the transfusion rate (*p* < 0.001, 50.0% in the TXA group, 91.7% in the control group). *Conclusions:* The incidence of DVT, whether symptomatic or asymptomatic, was not affected by the use of topical TXA. The postoperative transfusion rate was reduced in the TXA group. Topical TXA could be applied safely even in patients who had been known to be at high risk.

**Keywords:** tranexamic acid; venous thromboembolism; total knee arthroplasty; transfusion

#### **1. Introduction**

Total knee arthroplasty (TKA) leads to substantial blood loss and poses a risk of transfusion [1,2]. Because allogenic blood transfusion is associated with several complications, such as allergic reactions and metabolic imbalances [3,4], surgeons have made an effort to reduce postoperative blood loss. The use of tranexamic acid (TXA), a synthetic amino acid derivative of lysine which prevents fibrin degradation [5–8], has been proven efficacious in reducing blood loss after TKA, and thus the rate of allogenic transfusion [9–11].

A substantial body of research has endorsed the efficacy and safety of TXA [2,8,12,13]. However, previous studies often excluded patients with medical comorbidities that were associated with postoperative thromboembolism. There remains a concern that TXA

**Citation:** Joo, Y.B.; Kim, Y.M.; An, B.K.; Lee, C.W.; Kwon, S.T.; Song, J.-H. Topical Tranexamic Acid Can Be Used Safely Even in High Risk Patients: Deep Vein Thrombosis Examination Using Routine Ultrasonography of 510 Patients. *Medicina* **2022**, *58*, 1750. https:// doi.org/10.3390/medicina58121750

Academic Editors: Yong In and In Jun Koh

Received: 11 October 2022 Accepted: 26 November 2022 Published: 29 November 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

may increase the risk of thromboembolic complications [14–17]. In particular, patients undergoing TKA are prone to deep vein thrombosis (DVT) because surgical trauma and tourniquet application accelerate local fibrinolytic activity [18]. The incidence of DVT after arthroplasty has been investigated based on manifested symptoms in most studies [19–21]; thus, the incidence would only indicate the tip of the iceberg [22,23].

The topical use of TXA has an equivalent effect in reducing the transfusion rate after TKA, compared to intravenous (IV) use. Recent guidelines published by the American Association of Hip and Knee Surgeon (AAHKS) stated that the strong supporting evidence for use of TXA in high risk patients was lacking [24]. At our institution, topical TXA was applied without restrictions in high risk patients, such as those who had history of venous thromboembolism, myocardial infarction, or cerebral vascular occlusive disease. This study aimed to determine whether the use of topical TXA affected the true incidence of DVT using ultrasonography, regardless of symptom manifestation.

#### **2. Materials and Methods**

Medical records of 510 patients who underwent primary unilateral TKA between July 2014 and December 2017 were retrospectively reviewed after approval was obtained from our institutional review board. Because TXA was routinely applied through the topical route after September 2015, those who had history of venous thromboembolism, myocardial infarction, cerebral vascular occlusive disease, or cancer [25], were not excluded. The study population was divided according to the use of topical TXA.

#### *2.1. Study Intervention*

All TKAs were performed by two senior surgeons with the same surgical principle and prosthesis (Scorpio NRG, Stryker, Mahwah, NJ, USA). In the TXA group, 3.0 g of TXA in 100 mL of saline solution was applied directly into the knee joint cavity while suturing. An intra-articular drain was left for 48 h postoperatively. After the removal of drains, a range of motion exercise and walker-aided ambulation were encouraged. For DVT prophylaxis, intermittent pneumatic compression device was routinely used and a low molecular weight heparin was injected for 1 week after surgery [26,27]. There was no long-term anticoagulant therapy, regardless of the groups.

#### *2.2. Study Design and Propensity Score Matching*

DVT was examined using ultrasonography (Philips HD15, Bothwell, WA, USA) by two experienced radiologists. Lower extremity ultrasonography was performed at 1 week postoperatively for all patients, regardless of symptom manifestation. Proximal DVT was defined as the thrombosis that occurred proximal to the popliteal vein, and distal DVT was defined as the thrombosis of the anterior and tibial, peroneal, gastrocnemial, and soleal veins.

Postoperative transfusion records were reviewed, and the transfusion rate was compared according to the use of TXA. The necessity for transfusion was determined based on the guidelines by the National Institutes of Health Consensus Conference: hemoglobin level < 8.0 g/dL, or hemoglobin level < 10.0 g/dL with intolerable anemic symptoms or any anemia-related organ dysfunctions.

The two groups were matched at a 1:1 ratio based on propensity score [28]. The relevant variables were applied in a logistic regression analysis to calculate propensity scores. The incidence of DVT and the transfusion rate were compared between the groups after propensity score matching (PSM).

#### *2.3. Statistical Analysis*

The sample size of each group (168 patients) was confirmed by post hoc analysis to achieve a power of 96% to reject the null hypothesis with regard to the incidence of DVT, with a significance level of 0.05. Post hoc power analysis was performed using G\*Power (Version 3.1.7, Franz Faul, Christian-Albrechts-Universitätzu Kiel, Kiel, Germany). The propensity score variables included age, sex, body mass index (BMI), American Society of Anesthesiologists (ASA) score, smoking, hypertension, diabetes mellitus, chronic kidney disease, arrhythmia, blood profiles (platelet count, prothrombin time (PT), and activated partial thromboplastin time (aPTT)), and medical comorbidities (cerebrovascular accident, myocardial infarction, other thromboembolism, and cancer). Propensity scores were matched using one-to-one nearest neighbor matching with no replacement and no caliper width. The matched patients were selected randomly to avoid a potential bias that came from an imbalance in the number of patients between the groups. After PSM, intergroup comparison of each variable was performed to confirm the validity of the matching. Categorical variables including the incidence of DVT and the transfusion rate were analyzed by Chi-square test when the expected value of the cell was 5 or more in at least 80% of the cells; otherwise, Fisher exact test was used. Continuous variables were analyzed by *t* test. All statical analyses were performed using the R software version 4.1.1 (R foundation for Statistical Computing, Vienna, Austria), with a *p* value < 0.05 considered statistically significant.

#### **3. Results**

A total of 510 patients with a mean age of 69.8 ± 7.6 years (range, 53–86 years) were followed up for 39.6 ± 23.5 months (range, 12–86 months). There were 298 patients in the TXA group and 212 patients in the control group (Table 1). Overall, DVT was noted in 22 (4.3%) patients, with two patients having DVT proximal to the popliteal vein. Computed tomography pulmonary angiography was performed in those patients, and none of them showed pulmonary embolism. No patients exhibited symptoms for DVT. The odds ratio of topical TXA was 1.03 (95% confidence interval 0.43–2.45).

**Table 1.** Patient characteristics between the TXA and the control groups before PSM.


TXA, tranexamic acid; PSM, propensity score matching; BMI, body mass index; ASA, American Society of Anesthesiologists; PT, prothrombin time; INR, international normalized ratio; aPTT, activated partial thromboplastin time.

After PSM, 168 patients were allocated to each group (Table 2). In all, 11 patients in the TXA group and seven patients in the control group were diagnosed with DVT. One patient in each group had DVT proximal to the popliteal vein. There was no significant difference in the incidence of DVT between the groups (*p* = 0.721). A total of 84 (50.0%) patients in the TXA group and 154 (91.7%) patients received allogenic transfusion, which did not show a significant difference between the groups (*p* < 0.001; Table 3).


**Table 2.** Patient characteristics between the TXA and the control groups after PSM.

TXA, tranexamic acid; PSM, propensity score matching; BMI, body mass index; ASA, American Society of Anesthesiologists; PT, prothrombin time; INR, international normalized ratio; aPTT, activated partial thromboplastin time.


**Table 3.** Comparison of outcomes between the TXA group and the control group.

TXA, tranexamic acid; DVT, deep vein thrombosis.

#### **4. Discussion**

The primary finding of the present study was that topical TXA did not increase the incidence of DVT that was evaluated using ultrasonography. In this study, postoperative ultrasonography was performed in all patients to find both symptomatic and asymptomatic thrombosis, which could be the source of other serious complications such as pulmonary thromboembolism and cerebral infraction. The incidence of DVT represented the potential risk of TXA that could trigger systemic coagulation. This study proved that topical TXA reduced postoperative transfusion rate in TKA, which is consistent with previous studies [12,18,29].

It is surprising that concerns over the safety of TXA has not been fully addressed, given that TXA is a widely used modality to reduce the risk of transfusion after TKA. Based on recent guidelines by AAHKS, the recommendation for use of TXA in high risk patients is limited due to a lack of strong evidence [24]. There have been several randomized controlled trials (RCT) or level 1 studies investigating the efficacy and safety of TXA; however, those studies excluded high risk patients [13–15,29–31]. In their notable study, Whiting et al. performed a retrospective review of 1002 total joint arthroplasty patients with ASA score ≥ 3 to evaluate the outcomes of high risk patients who received TXA [19]. They found no differences in 30-day postoperative symptomatic thromboembolic events and postoperative transfusion rate. However, a concern regarding the potential procoagulant

effect of TXA still remained because only symptomatic events were assessed. Besides, the indication of TXA was different among surgeons in their study, which would cause selection bias. Sabbag et al. also reviewed 1262 primary total hip or knee arthroplasty patients with a history of DVT [21]. There was no difference in the incidence of recurrent DVT between patients who received TXA and those who did not receive TXA. However, asymptomatic DVT and medical comorbidities could not be investigated in detail because the study was based on total joint registry data.

Another remarkable study by Jules-Elysee et al. compared local and systemic levels of thrombogenic markers and TXA between IV TXA group and topical TXA group [32]. They collected peripheral and wound blood samples, measuring levels of plasmin-anti-plasmin (PAP, a measure of fibrinolysis), prothrombin fragment 1.2 (PF1.2, a marker of thrombin generation), and TXA. The authors recommended a single dose of IV TXA because no major difference was found in the mechanism of action, coagulation, and fibrinolytic profile between topical TXA and a single dose of IV TXA. However, postoperative (1 and 4 h after tourniquet release) systemic PF1.2 levels were higher than intraoperative levels in both IV TXA group and topical TXA group. The procoagulant effect of TXA cannot be ruled out although TXA was not the only reason of increased PF1.2 levels. In the present study, the incidence of DVT was investigated as an indicator of thromboembolic complications by topical TXA.

This study proved the efficacy of topical TXA in reducing the transfusion rate after TKA. The AAHKS guidelines stated that a superior method of administration could not be identified [24]. The aforementioned study by Jules-Elysee et al. concluded that a single dose of IV TXA was preferrable to topical TXA because of its convenience without making a significant difference in coagulation profile, compared to the latter. However, the safety of IV TXA in high risk patients has not been strictly tested. Based on the present study, topical TXA could be considered in high risk patients undergoing TKA.

Several limitations should be noted. First, the retrospective nature of this study brought potential bias. The ideal design in examining the effects of TXA would be RCT. However, as in the previous RCT studies [18,29,32], including high risk patients in the study population would face an ethical issue. Although this study was based on retrospective data, PSM was applied to minimize the possible confounding effects. Second, this study showed relatively high transfusion rates, compared to other studies [19,29]. Because our hospital is the only tertiary referral center in the region, TKA patients often have multiple comorbidities that necessitate low threshold of transfusion. This can be the reason of the high transfusion rates. Third, DVT might have occurred after ultrasonographic examination that was performed at 1 week postoperatively. However, the delayed thrombosis can hardly be attributed to TXA, considering the short half-life of it.

#### **5. Conclusions**

The incidence of DVT, whether symptomatic or asymptomatic, was not affected by the use of topical TXA. Postoperative transfusion rate was reduced in the TXA group. Topical TXA could be applied safely even in patients who had been known to be at high risk.

**Author Contributions:** Conceptualization, J.-H.S.; data curation, B.K.A., C.W.L. and S.T.K.; writing—original draft, Y.B.J.; Writing—review and editing, Y.M.K.; supervision, Y.M.K.; statistical analysis, J.-H.S. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work was supported by research fund of Chungnam National University.

**Institutional Review Board Statement:** The study was conducted in accordance with the Declaration of Helsinki, and approved by the Institutional Review Board of Chungnam National University Hospital (No. 2021-09-084, 5 February 2021).

**Informed Consent Statement:** Patient consent was waived due to the retrospective nature of the study.

**Data Availability Statement:** The manuscript has associated data, which will be deposited in repositories if necessary.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


### *Article* **Continuous Cold Flow Device Following Total Knee Arthroplasty: Myths and Reality**

**Michele Coviello 1,\*,†, Antonella Abate 2,†, Francesco Ippolito 2, Vittorio Nappi 2, Roberto Maddalena 2, Giuseppe Maccagnano 3,\*, Giovanni Noia <sup>3</sup> and Vincenzo Caiaffa <sup>2</sup>**


**Abstract:** *Background and Objectives*: To assess the effect of continuous cold flow (CCF) therapy on pain reduction, opioid consumption, fast recovery, less perioperative bleeding and patient satisfaction in patients undergoing a total knee arthroplasty. *Materials and Methods*: Patients affected by knee osteoarthritis between September 2020 and February 2022 were enrolled in this case-control study. Patients were randomly divided into two groups (*n* = 50, each): the study group received postoperative CCF therapy while the control group was treated by cold pack (gel ice). The CCF device is a computer-assisted therapy with continuous cold fluid, allowing a selective distribution, constant and uniform, of cold or hot on the areas to be treated. In both groups, pre- and postoperative evaluations at 6, 24, 72 h and at the fifth day were conducted using Visual Analogic Scale (VAS), opioid consumption, passive range of motion, preoperative hematocrit, total blood loss by Gross formula, transfusion requirement and patient satisfaction questionnaire. *Results*: One hundred patients, 52 women (52%), were included in the study. Reduction of pain, opioid consumption and increase in passive range of movement were statistically significantly demonstrated in the study group on the first and third days. Patients were satisfied with adequate postoperative pain management due to CCF therapy (*p* = 0.01) and they would recommend this treatment to others (*p* = 0.01). *Conclusions*: A continuous cold flow device in the acute postoperative setting after total knee arthroplasty is associated with pain reduction and improving early movement. Patients were almost satisfied with the procedure. The management of perioperative pain control could improve participation in the early rehabilitation program as demonstrated by the increase in ROM, psychological satisfaction and reduction in opioid use.

**Keywords:** continuous cold flow therapy; cryotherapy; total knee arthroplasty; pain; opioids consumption; patient satisfaction

### **1. Introduction**

Total knee arthroplasty (TKA) is a widely accepted and successful procedure for endstage arthritis. However, being a major orthopedic intervention, it is accompanied by tissue damage and an inflammatory response, manifesting as local swelling and edema [1,2], reduced range of motion (ROM), stiffness and reduced quadriceps strength [3,4]. Acute postoperative period after TKA may be quite challenging; patients may experience pain, swelling, restricted knee joint excursion, nausea or vomiting, and potential blood loss. A painful acute postoperative period may affect the so called fast track recovery [3], with patient discomfort and long-term effects on knee function [5,6].

**Citation:** Coviello, M.; Abate, A.; Ippolito, F.; Nappi, V.; Maddalena, R.; Maccagnano, G.; Noia, G.; Caiaffa, V. Continuous Cold Flow Device Following Total Knee Arthroplasty: Myths and Reality. *Medicina* **2022**, *58*, 1537. https://doi.org/10.3390/ medicina58111537

Academic Editors: Yong In and In Jun Koh

Received: 6 September 2022 Accepted: 24 October 2022 Published: 27 October 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

Pain is often severe in the early postoperative period, impeding rehabilitation [7]. Multimodal approaches to obtain pain control, throughout advances in anesthetic technique and narcotics administration, is not always sufficient to promote early rehabilitation and is often associated with side effects, such as nausea, vomiting, sedation, pruritus, hypotension, and respiratory depression, which can limit activity, result in longer hospital stays, increase morbidity, and reduce patients' satisfaction [8,9].

Cold therapy for pain reduction is an accepted and frequently used treatment in daily practice after trauma or surgery. It involves the application of cold to the skin surrounding the injured soft tissues and in joint surgery is supposed to reduce the intraarticular temperature [10] (Figure 1). It is supposed to reduce the local blood flow by vasoconstriction, swelling, and pain experience by slowing the conduction of nerve signals [11,12].

**Figure 1.** Application of cryotherapy pad on a patient.

Several cold therapeutic options are available, including first generation cold therapy, such as crushed ice in a plastic bag; second generation cold therapy with circulating ice water with or without compression; and third generation advanced computer-assisted devices with continuous controlled cold therapy. The advantage of these latter devices is the possibility of temperature modulation with cooling at a specific and continuous temperature (generally 5–11◦) for a prolonged time. These devices also allow a progressive increase in the temperature before stopping the treatment to avoid secondary cold-induced vasodilatation, and they are not needed to be periodically filled with cold water or ice [13].

Even though some studies show excellent results regarding computer-assisted continuous cold flow therapy, the quality of the available literature is not convincing and level I evidence is still missing [14].

The main aim of this study is to assess the effect of continuous cold flow therapy on pain reduction, opioid consumption, ROM recovery, less perioperative bleeding and patient satisfaction in patients undergoing a TKA.

#### **2. Materials and Methods**

This is a prospective, case-control, monocenter study, validated by the Ethics Committee (protocol number: 10/CE/2020—1 June 2020) and performed in accordance with the ethical standards laid down in the 1964 Declaration of Helsinki. All the patients involved in the study gave their informed consent prior to their inclusion in the study.

The patients who were scheduled to undergo primary TKA replacement for grade IV gonarthrosis due to a proven painful knee osteoarthritis, who need to obtain pain relief and improve function, who were able and willing to follow instructions, were enrolled and treated at the Di Venere Hospital of Bari between September 2020 and February 2022.

The patients were divided into two groups using a predefined program (http://www. randomization.com, accessed on 1 September 2020). After surgery, the circulating nurse reviewed the random numbers list. The study group was represented by patients who were treated with postoperative continuous cold flow (CCF) therapy; the control group included patients treated with a cold pack (gel ice) postoperatively.

The inclusion criteria were primary knee osteoarthritis, age between 40 and 81, body mass index (BMI) between 20 and 29.9 and chronic history (for at least 4 months) of knee joint pain.

The exclusion criteria were inflammatory diseases, infection, coagulopathy, previous knee surgery, history of deep vein thrombosis or pulmonary embolism, rheumatoid arthritis, pregnancy and patients who were not able to understand and complete the procedure due to cognitive dysfunction or language barrier.

The sample size as a prospective study was calculated using a pain score tested by visual analogic scale (VAS), referring to previous paper values on the same treatment protocol [15] as primary endpoint. Using a standard deviation of 0.6-point [16,17], we estimated that we would need 23 participants in each group to detect a statistically significant differences at an α level of 0.05 and power level of 80%.

Between September 2020 and February 2022, one hundred and twelve patients with grade IV gonarthrosis were evaluated for eligibility for the present study. Four patients had previous surgery of the knee, four patients participated in another study, two were living in another country, one declined to participate, and one other patient had reprieved the operation. Therefore, they were excluded from the study. Subsequently, 100 patients were included. An overview of the number of patients recruited, enrolled and analyzed in this work is presented in Figure 2 in the manner recommended by consolidated standards of the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines [18].

**Figure 2.** Diagram of the number of patients enrolled and analyzed in this study using STROBE guidelines.

For each patient, the following data were recorded: age; sex; BMI; side of surgery; American Society of Anesthesiologists score (ASA) [19], surgical time; VAS (visual analogic scale) (X), opioid consumption; passive range of motion (ROM); preoperative hematocrit; total blood loss (using the formula described by Gross [7], transfusion requirement and patient satisfaction questionnaire.

Data were recorded at the following times: T0 (before the surgical procedure); T1 (six-hours post-surgery); T2 (24 h post-surgery); T3 (72 h post-surgery) and T4 (fifth day post-surgery).

Opioid consumption was measured as number of tablets, tramadol (50 mg), used in the postoperative days. All patients received standard pain therapy with acetaminophen 1 g every 12 h, ketorolac tromethamine 30 mg and daily low molecular heparin injections (Fondaparinux, Arixtra, Glaxo-SmithKline, Brentford, Middlesex, UK) as thrombosis prophylaxis up to 4 weeks after surgery.

Passive range of motion as functional recovery was measured by the same operator for each patient using an orthopedic protractor.

Blood loss was evaluated only at T2 using the formula described by Gross.


Transfusion requirement was evaluated as the number of blood transfusions during hospitalization (transfusion trigger was set at 8.0 g/dL of hemoglobin).

Patient satisfaction was assessed via questionnaire. Participants reported their overall satisfaction by placing an X on a 10-cm line (from 0 (extremely unsatisfied) to 10 (extremely satisfied)). The distance was measured in centimeters from the beginning of the line to the center of the patient's X, using a ruler with 1-mm increments. Participants were also asked to recommend the method of cooling, answering "yes" or "not".

According to a standardized protocol, patients received antibiotic prophylaxis with cefazolin 2 g and premedication 2 h before operation with acetaminophen 1 g. Patients were operated under spinal anesthetic treatment by one experienced knee surgeon (VC), performing a minimum of 150 TKA procedures annually. Patients were operated on using a midline anterior incision with a medial parapatellar arthrotomy and with the use of dedicated instruments for the implantation of a cemented Postero-Stabilized (PS) TKA [20] (Zimmer Biomet, Warsaw, IN, USA). No tourniquet was used during surgery. Before skin closure, hemostasis was controlled using diathermia. The wound margins were infiltrated with 140 cc ropivacaine (0.2%) as local infiltration anesthesia (LIA). No adrenaline was used during LIA, since it was shown that adrenaline could be omitted from the LIA-mixture [21]. A wound drain was used. All wounds were closed with a barbed suture wire. The surgical wound was covered with a hydrocolloid dressing (Aquacel® Surgical, ConvaTec Inc., Reading, UK) for 7 days and a compressive dressing was applied in the operating room. A strict urinary nurse-led bladder scan management protocol was used for the prevention of postoperative urinary retention [22]. All patients were planned to be discharged on the fifth day after surgery.

The study group received advanced cooling devices. The machine used (Zamar Z-one MG465A, Croatia, Balkans) was dedicated to each patient during the entire period of hospital stay. It is a computer-assisted continuous cold flow therapy allowing a selective distribution, constant and uniform of cold or hot, on the areas to be treated. Patients cooled the affected knee according to a fixed protocol.

The application started the day before surgery for 3 h and repeated within 6 h postoperation for 3 h. Then it was applied for3hrespectively in the morning and in the afternoon, during the five postoperative days.

Temperature was set at 5◦ as efficient CCF therapy requires cooling the skin and achieving reduction in pain according to Chesterton et al. [1]. Compression was granted at the minimum level to avoid discomfort.

In the control group patients received 15 min cold pack (conserved at −17◦) treatment with two cold packs anterior and one posterior to the knee within 6 h post-operation and then repeated at 2 and 4 h after surgery. The following days patients received the same cold pack 15 min after each physiotherapy session (10 am and 3 pm). Cold packs were also permitted as requested during the night. No compression was applied.

Thereafter the same rehabilitation program was employed for all patients, consisting of full weight bearing and active range of motion exercises at day of surgery. A continuous passive motion device was used twice a day.

As a primary endpoint, pain was quantified using the VAS scale with scores ranging from 0 (no pain) to 10 (worst imaginable pain).

The functional recovery, pain management, blood loss and patient satisfaction were assessed as a secondary endpoint.

A prospective clinical study was conducted. The data were collected and analyzed using SPSS (v 23; IBM® Inc., Armonk, NY, USA). Descriptive statistics were calculated for the overall sample and for follow-up. Categorical variables were presented as numbers or percentages. Continuous variables were presented as mean and standard deviation. Due to the non-homogeneous distribution of the values using the Kolmogorov–Smirnov test (*p* > 0.05), non-parametric tests were considered. To compare the average values between the groups at the same times, the U Mann–Whitney test or Fischer's test were used, when appropriate. To compare the value within the same group at different times, the Wilcoxon test or Related-Samples Friedman's test Two-Way Analysis of Variance were used. To demonstrate the correlation between the CCF device and variables, a multiple regression model was then fitted. For all the tests, a *p*-value of less than 0.05 was considered to be statistically significant.

The data presented in this study are available on request from the corresponding author.

#### **3. Results**

One hundred consecutive patients treated with TKA were enrolled in this study and allocated into two groups, with 50 cases in each group. We compared the study and control group at recruitment (Table 1). No statistical differences emerged between groups.


**Table 1.** Main data of the study.

(One hundred patients; U Mann–Whitney and Fischer's test; data are presented as mean ± standard deviation or number and percentage; BMI: Body Mass Index; ASA: American Society of Anesthesiologists score).

None of the patients experienced any skin complications due to the CCF therapy or gel ice. None of patients needed revision surgery due to infection and mechanical relaxation in the early period. Neither intraoperative nor postoperative complication was recorded.

The VAS scores before and after surgery were noted for each time. No difference was shown preoperatively. The study group recorded lower values if compared with control, but only at one day postoperatively was the difference statistically significant (*p* = 0.01, Table 2 and Figure 3).


**Table 2.** Differences in pain and opioid treatment between groups.

(One hundred patients; U Mann–Whitney test for VAS and Fischer's test for opioid consumption; data are presented as mean ± standard deviation or number of tablets tramadol and equivalent mg; VAS: visual analogic scale).

**Figure 3.** Visual analogic scale within each group (\* *p* = 0.01).

In the study group a reduction in opioid consumption was demonstrably statistically significant in the first- and third-days postoperative. Overall, there was a lower consumption in the study group than in the control (*p* = 0.02, Table 2).

Descriptive values and comparison results of passive knee range of motion values are given in Table 3. It was found that the difference between preoperative and postoperative knee flexion measurements was significantly improved in both groups (*p* = 0.01 in both groups using Two-Way Analysis of Variance) and the differences between the groups were changed one- and three-day postoperative.


**Table 3.** Differences in passive range of motion, total blood loss using Gross formula and patient satisfaction between groups.

(One hundred patients; U Mann–Whitney test and Fischer's test; data are presented as mean ± standard deviation or number and percentage; ROM: range of motion).

No statistically significant difference was found between the groups regarding blood loss and transfusion number (Table 3).

Patients treated with CCF therapy were more satisfied than the control group (Table 3). The multiple linear regression models showed postoperative VAS score and opioid consumption were influenced by CCF therapy rather than other preoperative variables (Table 4).

**Table 4.** Multiple linear regression models for postoperative VAS and opioid consumption.


(BMI: Body Mass Index).

#### **4. Discussion**

After a major orthopedic surgical procedure, such as TKA, pain control is imperative to promote rapid recovery and collaboration to the rehabilitation program. Nowadays, specific pathways of care, including enhanced recovery after surgery (ERAS), have been introduced. It has been proposed these pathways, along with multimodal analgesia, have reduced the length of hospital stay (LOS) for patients undergoing TKA [23]. Analgesic options for TKA include pre-emptive analgesia, local infiltration, systemic analgesics (opioids, non-opioids, and patient-controlled analgesia), epidural analgesia and, more recently, femoral nerve block [24]. It has been postulated that peripheral nerve blocks provide intense, site-specific analgesia and are associated with a lower incidence of side effects when compared with many other modalities of postoperative analgesia [25]. Another approach to reduce pain and promote recovery consisted of cooling obtained by a different kind of CCF therapy approach [26,27]. However, consensus has not successfully been reached on the better modality, its efficiency and scheme of application.

The most important finding in our study is that postoperative computer assisted CCF therapy is effective in pain control after TKA and associated with reduction in analgesic use during the early postoperative period. It is also associated with a major passive ROM during the first two days and with a great grade of patient satisfaction. We enrolled a larger study population if compared to power analysis due to the possibility of patient refusal and to strengthen the results.

Significant reduction in VAS was observed only at T2 (24 h), however there was a reduction in VAS after every cooling session in the study group. Similar to these findings, pain reductions in the acute postoperative phase were reported [7,28,29]. Inflammatory conditions secondary to surgery cause most of the perceived pain in the immediate postoperative period after knee replacement. Additionally, patients undergoing knee surgery have already suffered pain for a long period secondary to established central nervous system sensitization. It seems to be intuitive that any approach aimed to reduce pain should be pursued to avoid risk of development of chronic pain [30]. Bech et al. found no superiority in reducing pain compared with traditional icing. Cooling reduces the tissue metabolic rate and relieves inflammation by suppressing enzymatic activity and preventing secondary tissue damage, reducing muscle tissue spasm [31].

Consumption of analgesic was also investigated in this study. In our paper the consumption of additional opioids as rescue medication was up to two times greater in the control group than in the study group. However, this trend was significant only in the first two postoperative days, with these days being the worst period for pain management.

The incremented use of opioids can slow a patient's rehabilitation program due to side effects (nausea, vomiting, headache, constipation). A slower rehabilitation can be associated with retarded discharge and increased costs [32].

Su et al. [33] showed a lower consumption of opioids in a multicenter RCT. In another recent systematic review and meta-analysis of 11 RCT, Addie et al. [28] showed no benefits for analgesic use. Thienpont et al. [14], on the contrary, found no difference in VAS and morphine use in their RCT. They shared Algafly's conviction that a decreased nerve conduction velocity resulting in better pain tolerance can explain the local anesthetic effect of intense cooling [11]. However, this is merely a temporary measure that disappears after ice removal.

The CCF therapy is crucial and ensures that the knee ROM movements start in the early postoperative period. There were studies reporting that the CCF therapy affects the knee movement gap in a positive way in the early postoperative period [34–36]. In our study passive ROM at first and second postoperative day was better in the cold compression group. Su et al. found no differences for ROM or functional testing between a cryopneumatic device and ice packs [33]. Kullemberg et al. [12] and Holmstrom et al. [37] reported that cold compression was more efficient than epidural analgesia with better ROM, reducing hospital stay.

Patients reported high satisfaction rate with postoperative cooling and were likely to recommend icing device. Less pain could substantiate this validation. The high level of satisfaction may also be associated to greater patient contact with the staff conduction device function and periodic skin and dressing check. The psychological aspect plays a key role in rehabilitation [38]. Only two patients reported that the device was uncomfortable and that some nurses appeared not to be sure how to apply it.

Blood loss evaluation is a key point to justify routine use of a cold flow device, considering its additional costs. Reduction in blood loss and blood transfusion may reduce

costs associated with TKA. Additionally, blood transfusion is associated with an increased rate of periprosthetic infection. In our study we found that the total blood loss measured by Gross formula in the postoperative period in the study group was lower than in the control group; however, non-statistically significant values were observed in total blood loss. Kuyucu et al. [26] also did not find significant difference in blood loss between their study groups. Ruffilli et al. [7] carefully evaluated blood loss throughout drain output, blood loss with Gross formula, transfusion requirements, and did not find any difference between their study groups. Contradictory results have been found by several authors regarding visible and real blood loss [37,39]. This could be explained by the fact that CCF therapy may cause an immediate reduction of bleeding by vasoconstriction and compression but later lead to a cold-induced vasodilatation and interfere with secondary hemostasis [40]. Hemostasis and coagulation are biochemical reactions that are most efficient at basal body temperature. Computed control of temperature progressive increase before cryotherapy removal should be considered to reduce secondary vasodilatation and blood loss [41]. Karaduman et al. found the Hb levels measured in the ice pack group were lower than those in experimental groups treated with CCF therapy in the postoperative period. They also sustained that postoperative CCF therapy was effective in reducing blood loss. However, preoperative CCF therapy application had no significant effect [15].

The evaluation method of blood loss could explain the difference in results in the literature. We decided to use the formula described by Gross et al. [42] for blood loss calculation after arthroplasty surgery, which is one of the most popular among surgeons. However, the Gross equation does not involve Hb-related factors. Actual blood loss and anemia are revealed by the calculated perioperative volume and the changing in hematocrit [42]. This could consist of a limitation of this method and additional evaluation of Hb values and drainage blood loss should be considered.

The novelty of this study is the careful analysis of the patient's outcomes combined with their satisfaction. Psychological well-being improves clinical outcomes and makes the patient prone to better physiotherapy [38].

Our study has several strengths. This was performed in one single center, one single experienced surgeon performed all surgeries using the same surgical approach, one type of anesthesia, implant, pain and rehabilitation program. The self-reported patient satisfaction questionnaire is an additional element considering the importance of a fast-track recovery plan in replacement surgery.

However, some limitations must be considered. The sample size could be increased. Better evaluation of blood loss throughout alternative formula, including Hb values, should be adopted. Another limitation is that patients were evaluated only during the hospital stay so it cannot draw definitive long-term conclusions about CCF therapy efficacy.

#### **5. Conclusions**

Continuous cold flow devices in the acute postoperative setting after TKA are associated with pain reduction and improving early ROM. Although reduction in blood loss throughout the Gross formula was observed, no significant differences were found. Patients were almost satisfied with the procedure and should recommend it to improve perioperative pain control and participation to the early rehabilitation program due to lower opioid consumption and higher passive ROM.

**Author Contributions:** Conceptualization, M.C.; data curation, M.C.; writing—original draft preparation, A.A., V.N. and F.I.; writing—review and editing, G.M. and G.N.; supervision, R.M. and V.C. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Institutional Review Board Statement:** The study was conducted according to the guidelines of the Declaration of Helsinki and approved by the Ethics Committee (protocol number: 10/CE/2020—1 June 2020).

**Informed Consent Statement:** Informed consent was obtained from all subjects involved in the study. Written informed consent has been obtained from the patient(s) to publish this paper if applicable.

**Data Availability Statement:** The data presented in this study are available on request from the corresponding author.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


### *Article* **Optimal Release Timing of Drain Clamping to Reduce Postoperative Bleeding after Total Knee Arthroplasty with Intraarticular Injection of Tranexamic Acid**

**Myung-Ku Kim, Sang-Hyun Ko, Yoon-Cheol Nam, Yoon-Sang Jeon, Dae-Gyu Kwon and Dong-Jin Ryu \***

Department of Orthopaedic Surgery, Inha University Hospital, Incheon 22332, Korea **\*** Correspondence: mdryu24@naver.com; Tel.: +82-10-7161-3684; Fax: +82-32-890-2387

**Abstract:** *Background and Objectives*: Intraarticular injection of tranexamic acid (IA-TXA) plus drainclamping is a preferred method of reducing bleeding after total knee arthroplasty (TKA). However, no consensus has been reached regarding the timing of the clamping. The purpose of this study was to determine the optimum duration of drain-clamping after TKA with IA-TXA. *Materials and Methods*: We retrospectively reviewed 151 patients that underwent unilateral TKA with IA-TXA plus drain-clamping for 30 min, 2 h, or 3 h. The total drained volume was reviewed as the primary outcome, and hematocrit (Hct) reductions, estimated blood loss (EBL), transfusion rates, and wound complications were reviewed as secondary outcomes. *Results*: The mean total drained volume, Hct reduction, and EBL were significantly less in the 3 h group than in the 30 min group. Between the 2 h and 3 h groups, there was no statistical difference in the mean total drained volume, Hct reduction, or EBL. The proportion of patients who drained lesser than 300 mL was high in the 3 h group. No significant intergroup difference was observed for transfusion volume, transfusion rate, and wound related complications. *Conclusions*: In comparison of the IA-TXA plus drain-clamping after TKA, there was no difference in EBL between the 2 h group and the 3 h group, but the amount of drainage volume was small in the 3 h group.

**Keywords:** total knee arthroplasty; tranexamic acid; clamping time; transfusion; estimated blood loss

#### **1. Introduction**

Total knee arthroplasty (TKA) is a major orthopedic surgical procedure used to treat end-stage osteoarthritic knees, and it has good clinical and functional outcomes [1]. However, TKA is associated with significant blood loss, and 18% to 67% of patients require a blood transfusion, which is associated with poor outcomes such as allergic reaction, extended hospitalization, thromboembolic events, and mortality [2–4]. For these reasons, various methods such as a tourniquet application, drain clamping, epinephrine or tranexamic acid (TXA), and fibrin sealant have been proposed to reduce blood loss [5–11].

TXA is a hemostatic substance that inhibits fibrinolysis, providing a pharmacological option to reduce blood loss. Previous studies have reported that intraarticular TXA (IA-TXA) combined with drain clamping is a more effective mean of preventing blood loss than drain clamping alone [12–16]. Liao et al. conducted a meta-analysis of the results of seven different studies and confirmed the efficacy of IA-TXA plus drain-clamping [16]. However, in this meta-analysis, little data were available for clamping times between 1 and 4 h, and it remains controversial (1 h: Onodera et al. [17], Mutsuzaki et al. [18]; 2 h: Sa-Ngasoongsong et al. [11,19]; 3 h: Chareancholvanich et al. [20]; 4 h: Wang et al. [21], Wu et al. [22]).

To the best of our knowledge, no study has compared blood loss with respect to drainclamping time after TKA with IA-TXA. Accordingly, the present study was performed to determine an optimum drain-clamping time after TKA with IA-TXA by comparing blood loss and complication rates for different drain-clamping times. In this study, total drained volume

**Citation:** Kim, M.-K.; Ko, S.-H.; Nam, Y.-C.; Jeon, Y.-S.; Kwon, D.-G.; Ryu, D.-J. Optimal Release Timing of Drain Clamping to Reduce Postoperative Bleeding after Total Knee Arthroplasty with Intraarticular Injection of Tranexamic Acid. *Medicina* **2022**, *58*, 1226. https:// doi.org/10.3390/medicina58091226

Academic Editors: In Jun Koh and Yong In

Received: 18 August 2022 Accepted: 1 September 2022 Published: 5 September 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

was reviewed as the primary outcome and hematocrit (Hct) reductions, estimated blood loss (EBL), transfusion rates, and wound complications were reviewed as secondary outcomes.

#### **2. Materials and Methods**

This study was approved by our Institutional Review Board (IRB No. INHAUH 2020-03-035 at 20 April 2020). The requirement for informed consent was waived due to the retrospective nature of the study. The medical and surgical records of 151 patients that underwent TKA surgery at our hospital from January 2017 to December 2019 were retrospectively reviewed. According to our database, 328 patients underwent TKA surgery after being diagnosed with knee osteoarthritis. However, 177 patients were excluded after applying the following exclusion criteria: simultaneous bilateral TKA; concomitant operation; TKA with lateral retinacular release; TKA with patella resurfacing; use of an extended stem; a diagnosis of secondary osteoarthritis; a neurologic disorder; and the receipt of medications, such as antiplatelet or anticoagulant medications, likely to interfere with findings.

The 151 study subjects were allocated to 3 groups according to clamping time; that is, Group A (*n* = 60) had a clamping time of 30 min, Group B (*n* = 42) had a clamping time of 2 h, and Group C (*n* = 49) had a clamping time of 3 h. A schematic of the patient selection is presented in Figure 1.

**Figure 1.** Flowchart of patient selection for this study.

#### *2.1. Operation Procedures*

All patients underwent TKA by a single senior surgeon (MKK) and were provided with the same procedures and post-operative managements, except clamping time. Spinal anesthesia was used in most cases (93.4%), except for patients in whom spinal anesthesia was impossible due to severe degenerative deformation of the spine. In the supine position, a standard mid-line skin incision was made using a medial parapatellar approach after applying a pneumatic tourniquet at 350 mmHg. The same implant system (Persona-Zimmer Biomet, Warsaw, IN, USA) with cement fixation (Optipac 80, Biomet Orthopaedics GmbH, Dietikon, Switzerland) was applied to all patients, and a 3.2 mm drainage tube and a BAROVAC (400 mL, Sewoon Medical, Seoul, Republic of Korea, negative pressure 90 mmHg) comprised the drainage system. IA-TXA (3 g/30 cc + normal saline 70 cc) was administered immediately after joint capsule closure [23,24]; 1-0 Vicryl simple interrupted sutures were used for joint capsules and tendons, and 3-0 Vicryl simple running sutures were used for synovium. After subcutaneous and skin closures with 2-0 Vicryl and skin staples, respectively, we confirmed no leakage. An aseptic compression dressing was applied using an elastic bandage. Drains were clamped off in a timely manner.

#### *2.2. Postoperative Management*

All case-patients received the same perioperative management, including preemptive medications. Anti-embolism stockings and intermittent pneumatic compression were applied in all cases to prevent deep vein thrombosis (DVT) or pulmonary thromboembolism (PTE). In addition, anticoagulant (rivaroxaban, Xarelto®, 10 mg) was given from postoperative days (POD) 3 to 14. Patients with a hemoglobin level of <7 g/dL received a unit of packed red blood cells (RBCs), and patients that maintained a hemoglobin level between 7 and 9 g/dL received a unit of packed RBCs postoperatively [25]. All patients performed weight-bearing exercise (using a walker), active thigh lifting exercise, passive range of motion (ROM) exercise, and cryotherapy from POD 1 to 14 in consultant with the Department of Rehabilitation.

#### *2.3. Perioperative Laboratory Factors and Hemodynamic Factors*

Because of massive irrigation during TKA, blood loss under anesthesia cannot be measured appropriately. We used Nadler's formula adjusted for height and body weight using Mercuriali's formula to calculate blood volume from preoperative and POD 5 hematocrit values [26,27]. Mercuriali's and Nadler's formulae are as follows.

**EBL** = **Blood volume** × - *Hctpreop* <sup>−</sup> *Hct***<sup>5</sup>** *days postoperative* + **volume of transfused RBCs Blood volume in men** (**L**) = **Height** (**m**) <sup>3</sup> <sup>×</sup> 0.367 <sup>+</sup> **Body weight** (**kg**) <sup>×</sup> 0.032 <sup>+</sup> 0.604 **Blood volume in women** (**L**) = **Height** (**m**) <sup>3</sup> <sup>×</sup> 0.356 <sup>+</sup> **Body weight** (**kg**) <sup>×</sup> 0.033 <sup>+</sup> 0.183

#### *2.4. Outcomes Measurement*

Total drained volume was reviewed as the primary outcome. Drainage amounts were recorded at 24 and 48 h postoperatively, and all drains were removed at 48 h postoperatively. For secondary outcomes, we measured Hct reduction and EBL with Hct value at preoperation and POD 5. The transfusion rate was calculated by counting the number of patients who received a transfusion after surgery.

#### *2.5. Complications*

Complications such as DVT and PTE were evaluated closely because they can occur during clamping. From POD 2, surgical wounds were monitored to evaluate superficial infections and wound complications such as major bruises, oozing, hemarthrosis, subcutaneous hematoma, and blisters. Wounds were assessed up to POD 12 to evaluate possible superficial or deep wound infections. Major bruises were defined as bruises that extended >5 cm around wounds. Oozing was defined when three or more gauzes were soaked with blood.

#### *2.6. Statistical Analysis*

Results are presented as means ± standard deviations for continuous variables and as numbers and relative frequencies for categorical variables. Groups were compared using one-way analysis of variance (ANOVA) for quantitative data or Pearson's chi-squared test for qualitative data. The test of significance was conducted on IBM SPSS Statistics for Windows version 25.0 (IBM Corp., Armonk, NY, USA), and statistical significance was accepted for *p*-values < 0.05. A post hoc power analysis was performed among the three groups using the G power 3.1 software (JMP., Lane Cove, NSW, Australia). The statistical power was 0.9505, which means that the number of subjects and the results were significant.

#### **3. Results**

#### *3.1. Patient Demographics*

Age, gender, surgery side, height, weight, blood volume, anesthesia method, and preoperative Hct level before surgery are presented in Table 1. No significant difference was found between these variables in the three groups (all *p* > 0.05).


**Table 1.** The patients' preoperative characteristics data.

Hct: Hematocrit. <sup>a</sup> Data presented as mean ± standard deviation. <sup>b</sup> Data presented as number of patients having that condition (percentage of this group).

#### *3.2. Drainage Amount*

Mean total drainage amounts at 48 h postoperatively in groups A, B, and C were 332.3 ± 100.2, 286.4 ± 127.9, and 255.8 ± 84.5 mL, respectively (*p* = 0.001). Group C had a significantly lower amount than group A (*p* = 0.001), but no significant difference was observed between groups A and B (*p* = 0.09) or groups B and C (*p* = 0.495) (Table 2). The proportions of patients in the three groups with a drainage amount < 300 mL were 36.6%, 59.6%, and 79.6%, respectively. Notably, as clamping time increased, the percentage of patients with a drainage amount of <300 mL also increased (Figure 2).


**Table 2.** Blood loss and blood transfusion outcome in three groups.

<sup>a</sup> Data presented as mean ± standard deviation. <sup>b</sup> Data presented as number of patients having that condition (percentage of this group). Statistical significance was determined by one-way ANOVA followed by Scheffe's post hoc analysis.

#### *3.3. Total Blood Loss*

Mean EBL calculated using Mercuriali's and Nadler's formulae was higher in Group A than in the other groups (*p* ≤ 0.001) (Table 2). Mean EBL showed a decreasing tendency as clamping time increased.

**Figure 2.** Percentages of patients with drainage volumes of <300 mL. Percentages of patients with a drainage volume of <300 mL increased with drainage time.

#### *3.4. Need for Transfusion*

Regarding the need for transfusion, results showed a tendency similar to EBL and drainage amounts. Mean transfusion volume was highest in group A and tended to decrease with clamping time. Transfusion rates showed a similar tendency and were 16.7, 9.5, and 4.0% in groups A, B, and C, respectively.

#### *3.5. Complications*

No deep vein thrombosis or superficial infection occurred. Wound complications were categorized as major bruises, hemarthrosis, subcutaneous hematomas, and blisters, but no significant intergroup difference was observed (Table 3).


**Table 3.** Complications.

<sup>a</sup> Data presented as number of patients having that condition (percentage of this group). Data presented as number (%). Statistical significance was determined by Pearson's chi-squared test.

#### **4. Discussion**

Previous studies have reported that IA-TXA combined with drain clamping effectively prevents blood loss after TKA [12–16]. However, to our knowledge, no study has determined the optimum timing of drain clamp release. In the present study, TKA with IA-TXA plus drain clamping for 3 h resulted in a significant blood loss reduction compared

to clamping for 30 min. Between the 2 h and 3 h groups, although there were no statistical differences, the proportion of patients who drained a lower volume than 300 mL was high in the 3 h group. No significant intergroup difference was observed for complication rates. Blood loss is an important postoperative consideration that must be considered after TKA. Bleeding into soft tissues surrounding the knee increases pain, stiffness, and length of recovery following surgery [28]. TXA application has recently become one of the most popular methods for reducing blood loss and transfusion requirements. TXA is an antifibrinolytic agent and was discovered in 1962. It prevents the formation of plasmin, inhibiting the breakdown of fibrin clots and decreasing bleeding. Although TXA has been administered intramuscularly, intravenously, and intraarticularly, it is being increasingly administered locally due to theoretically lower rates of systemic effects, including those related to thromboembolic disease [13,23,24,29]. However, due to safety concerns regarding PTE, interest is growing in the use of TXA as an IA agent in TKA. In the present study, we decided to use IA-TXA to reduce blood loss after TKA.

IA-TXA with drain-clamping reduces blood loss in TKA compared to IA-TXA without clamping [16]. This method is considered effective for reducing bleeding by forming a tamponade before opening. Prior studies have examined many methods, such as the intermittent method and specific timed drain clamping after surgery to reduce blood loss postoperatively [18,20–22]. However, no study has determined the optimum timing of drain clamp release after TKA with TXA. Liao et al. [16] conducted a systematic review and meta-analysis on the efficacy of TXA plus drain-clamping in TKA and concluded that this technique reduced blood loss and the need for transfusion. However, the seven clinical studies [11,17–22] included in their meta-analysis [16] were conducted using different clamping times (1 h: Onodera et al. [17], Mutsuzaki et al. [18]; 2 h: Sa-Ngasoongsong et al. [11,19]; 3 h: Chareancholvanich et al. [20]; 4 h: Wang et al. [21], Wu et al. [22]) and TXA dosages (range 250 to 1000 mg). To avoid confusion, we tried to define an effective clamping time by injecting TXA at 3 g/30 cc + 70 cc of normal saline and found EBL decreased and the percentage of patients with a drainage amount of <300 mL increased as the clamping time increased from 30 min to 3 h.

When clamping is released early, effective bleeding control cannot be achieved, i.e., longer clamping times are required to form effective tamponades. Since the half-life of TXA is 3 h, we examined the effects of clamping times up to3h[29]. Furthermore, it should be noted that complications such as hematoma can occur when clamping times are excessive (ca. > 4 h), as accumulations of blood in knee joints can lead to swelling, delayed wound healing, and increased risk of infection [14]. We found no significant difference between the three groups in terms of complications such as DVT, superficial infections, and wound complications.

The cytotoxic effect of IA-TXA on cartilage should be considered when the surgical intention is to preserve native cartilage tissues; its cytotoxicity may not affect total joint arthroplasties involving the removal of entire articular cartilage. Effective dosing for topical TXA ranges from 15 to 100 mg/mL. Increased exposure time to TXA at high concentrations is cytotoxic to cartilage. Because patients included in this study did not undergo patella resurfacing, we needed to minimize TXA exposure time and concentration on the articular surface and, thus, decided to use TXA at a concentration of 30 mg/mL and to limit the maximum exposure time to 3 h.

This study has several limitations. First, this is a retrospective study. Although we excluded the confounding factors that could affect bleeding tendency, a randomized control trial is needed for exact evaluation. Second, a relatively small number of cases were included in each group because all operations were performed by one surgeon in a single center. Third, postoperative blood loss was low in some patients when the surgeon was able to well identify and cauterize bleeding vessels, though vessel bleeding was carefully cauterized in all cases. Fourth, individual bleeding tendencies differ, and numerous factors can affect blood loss. However, considering the size of our cohort, we excluded factors that may have confounded results, such as a history of anticoagulant/antiplatelet medication, abnormal coagulation factors, and patients at a high risk of blood loss. Fifth, as blood loss could not be accurately measured, EBL was calculated using Mercuriali's and Nadler's formulae. However, Nadler's formula calculates blood volume based on weight and height; thus, fluid-induced body changes pre-operation to POD 5 may have introduced errors. Sixth, postoperative pain is an important evaluation factor, and blood management can affect postoperative pain. However, in this study, we focused on blood loss rather than postoperative pain. Seventh, because of the limited sample size and low prevalence of perioperative complications, this study design was insufficient to draw a clear conclusion towards complications. Thus, we performed a power analysis among the three groups using the G power 3.1 software, and the power calculated was 97.9%. Despite these limitations, we believe our findings are meaningful as they provide evidence of the optimal duration of the drain-clamp application with IA-TXA after TKA. Further, our study provides orthopedic surgeons with a rationale for how to minimize bleeding after TKA.

#### **5. Conclusions**

Temporary drain clamping after TKA with an intraarticular injection of tranexamic acid can effectively reduce EBL. Although there were no statistical differences between the groups of 2 h and 3 h in terms of blood loss, the proportion of patients who drained lesser than 300 mL was notably higher in the 3 h group. In comparison to IA-TXA plus drain-clamping after TKA, there was no difference in EBL between the 2 h group and the 3 h group, but the amount of drainage was small in the 3 h group.

**Author Contributions:** Conceptualization, D.-J.R., Y.-S.J. and M.-K.K.; methodology, D.-J.R., Y.-S.J. and M.-K.K.; validation, D.-J.R. and D.-G.K.; formal analysis, Y.-C.N. and S.-H.K.; investigation, Y.-C.N. and S.-H.K.; data curation, Y.-C.N. and S.-H.K.; writing—original draft preparation, D.-J.R., D.-G.K., Y.-S.J. and M.-K.K.; writing—review and editing, D.-J.R., Y.-S.J. and M.-K.K.; visualization, D.-J.R., Y.-S.J. and M.-K.K.; supervision, M.-K.K.; project administration, M.-K.K. All authors have read and agreed to the published version of the manuscript.

**Funding:** This study was supported by Inha University Hospital research grant.

**Institutional Review Board Statement:** The design and protocol of this study were reviewed and approved by the institutional review board of Inha university hospital (IRB No. INHAUH 2020-03- 035). All the experiments were performed in accordance with the relevant guidelines and regulations.

**Informed Consent Statement:** An exemption from informed consent was obtained from the institutional review board of Inha university hospital due to its retrospective nature.

**Data Availability Statement:** Not applicable.

**Conflicts of Interest:** The authors declare that they have no conflict of interest.

#### **References**


### *Article* **Current Trends in Knee Arthroplasty: Are Italian Surgeons Doing What Is Expected?**

**Lorenzo Moretti 1,†, Michele Coviello 1,\*,†, Federica Rosso 2, Giuseppe Calafiore 3, Edoardo Monaco 4, Massimo Berruto <sup>5</sup> and Giuseppe Solarino <sup>1</sup>**


**Abstract:** Objectives: The purpose of this study is to evaluate Italian surgeons' behavior during knee arthroplasty. Materials and Methods: All orthopedic surgeons who specialized in knee replacement surgeries and were members of the Italian Society of Knee, Arthroscopy, Sport, Cartilage and Orthopedic Technologies (SIGASCOT) between January 2019 and August 2019 were asked to complete a survey on the management of knee arthroplasty. Data were collected, analyzed, and presented as frequencies and percentages. Results: One-hundred and seventy-seven surgeons completed the survey and were included in the study. Ninety-five (53.7%) surgeons were under 40 years of age. Eighty-five surgeons (48%) worked in public hospitals and 112 (63.3%) were considered "high volume surgeons", with more than 100 knee implants per year. Postero-stabilized total knee arthroplasty was the most commonly used, implanted with a fully cemented technique by 162 (91.5%) surgeons. Unicompartmental knee arthroplasty (UKA) was a rarer procedure compared to TKA, with 77% of surgeons performing less than 30% of UKAs. Most common TKA pre-operative radiological planning included complete antero-posterior (AP) weight-bearing lower limb radiographs, lateral view and patellofemoral view (used by 91%, 98.9% and 70.6% of surgeons, respectively). Pre-operative UKA radiological images included Rosenberg or Schuss views, patellofemoral view and magnetic resonance imaging (66.1%, 71.8% and 46.3% of surgeons, respectively). One hundred and thirtytwo surgeons (74.6%) included an AP weight-bearing lower limb X-ray one year after surgery in the post-operative radiological follow-up. Furthermore, 119 surgeons (67.2%) did not perform a post-operative patellofemoral view because it was not considered useful for radiological follow-up. There was no uniformity in the timing and features of post-operative follow-up, with 13 different combinations. Conclusions: Italian surgeons perform TKA more commonly than UKA. Pre-operative TKA planning is quite uniform rather than UKA planning. Despite literature evidence, there is no agreement on follow-up. It may be useful to create a uniform checklist, including correct timing and exams needed. This analysis is also part of a society surgical educational project for training doctor.

**Keywords:** orthopedic surgeon; planning; survey; total knee arthroplasty; unicompartmental knee arthroplasty

#### **1. Introduction**

Total knee arthroplasty (TKA) is a very common procedure in orthopedic surgery. It has increased by 162% over the past twenty years with approximately 250,000 primary and

**Citation:** Moretti, L.; Coviello, M.; Rosso, F.; Calafiore, G.; Monaco, E.; Berruto, M.; Solarino, G. Current Trends in Knee Arthroplasty: Are Italian Surgeons Doing What Is Expected? *Medicina* **2022**, *58*, 1164. https://doi.org/10.3390/ medicina58091164

Academic Editors: Yong In and In Jun Koh

Received: 26 July 2022 Accepted: 20 August 2022 Published: 26 August 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

revision arthroplasties performed each year [1]. Main indications for knee arthroplasty remain primary or secondary osteoarthritis, rheumatoid arthritis in association or not with limitation in range of movement (ROM) or deformities. Modifiable and nonmodifiable prognostic factors were associated with the rate of unsatisfied patients from 5 to 40% after TKA [2,3]. Nowadays there are several TKA designs, which can be chosen depending on the patients' age, expected activity level, pre-operative deformity and stability of the knee [4]. Particularly, different types of constraint can be used in primary TKA, from cruciate-retaining (CR) to postero-stabilized (PS) implants. There are also "high constrained" implants, such as condylar constrained, but they are normally reserved to revision TKA due to lower survivorship in primary TKA [5,6]. The first TKA designs did not include patellar implants, and they were characterized by high rate of post-operative anterior knee pain [7]. In the 1980s, patella-related complications accounted for up to 50% of complications following TKA [8]. Consequently, patellar resurfacing was introduced, but different complications were described. For these reasons, despite the number of studies, there is still some disagreement between surgeons who prefer patellar resurfacing, surgeons who never resurface the patella and surgeons who resurface it in selected cases [8–12]. In the case of degenerative arthritis involving only the medial or lateral compartment, unicompartmental knee arthroplasty (UKA) can be performed [13]. UKA needed the patient to have an intact anterior cruciate ligament and correctable knee alignment [14]. Similarly, isolated patellofemoral arthroplasty (PFA) can also be performed in selected cases with isolated patellofemoral osteoarthritis (PFOA) [15].

Both in TKA or UKA, correct pre-operative planning is mandatory to achieve a good outcome [16]. Different X-ray protocols have been described, but the most used are still the long-leg anteroposterior (AP) view for evaluation of the anatomical and mechanical axis [17–20], weight-bearing AP and lateral view of the knee and patellar view [21,22]. Magnetic resonance imaging (MRI) is usually not useful in TKA pre-operative planning, but it may be reasonable in UKA planning to evaluate cruciate ligaments status as well as cartilage status of the not-replaced compartment. Furthermore, it may also be useful in some revision surgery cases [23]. Careful pre-operative planning allowed the surgeon to predict possible difficulties or complications, need for higher constraint or specific instrumentation and it is mandatory to plan the surgeries [24]. Despite different studies on the correct pre-operative planning needed for both TKA and UKA, in practice there is still lack of uniformity. Current literature was investigated to compare our results with other large working groups, such as national registries, demonstrating similarities and differences. With regard to the Nordic Arthroplasty Register Association (NARA), that confirmed some Italian data, TKA was used as a primary implant (92%) in patients less than 65 y.o. (64%), while UKA and PFA were used less (3% and 1%, respectively) [25]. The experienced surgeons' behaviors together with new learning technologies, such as augmented reality or cadaver labs, represent the focus of young surgeons trained by orthopedic societies [26,27].

The purpose of this study is to evaluate how Italian surgeons specialized in knee replacement behave in pre-operative planning and surgery. The authors, as active members of Italian orthopedic society and as active knee surgeons, guided this analysis as a starting point for educational pathways for young surgeons.

#### **2. Materials and Methods**

All orthopedic surgeons who specialized in knee replacement surgeries and were members of the Italian Society of Knee, Arthroscopy, Sport, Cartilage and Orthopedic Technologies (SIGASCOT) were asked to complete a survey on the current management of TKA and UKA, between January 2019 and August 2019, including pre-operative planning, implants used and characteristics of follow-up. Surgeons interviewed belong to the "knee surgery" specialized group of the society with 415 members. This is represented by doctors who spend more than half of their surgery in knee procedures with both arthroplasty and arthroscopy and who have followed high specialization courses.

An online questionnaire was built using SurveyMonkey (Portland, OR, USA®), a free, open-source online survey tool. The 25 multiple choice questions were divided by subject into 5 parts: surgeons' characteristics (i.e., age, volume of surgeries), TKA data (i.e., type of insert, cementation technique), UKA data (numbers of surgeries, type of implants), data regarding the pre-operative planning and follow-up (Scheme 1). The survey required approximately 10–12 min to be completed. Results from the survey were collected electronically and anonymously using SurveyMonkey (Portland, OR, USA®). All data were analyzed using SPSS 25.0 (IBM, Armonk, NY, USA). Data were evaluated using descriptive analysis, and they were presented as frequencies and percentages. The data used to support the findings of this study are available from the corresponding author upon request.

**Scheme 1.** Flow diagram of questionnaire.

#### **3. Results**

One hundred and seventy-seven orthopedic surgeons (42.65%) completed the survey. Table 1 reports the survey in detail. Of the respondents, 46.9% worked in semiprivate hospitals, and 53.7% were under the age of 40. The demographic data, workplace and number of total annual implants included in the study are summarized in Table 2 and Figure 1.

**Table 1.** Questions asked and possible responses.


#### **Table 1.** *Cont.*


**Table 1.** *Cont.*


(y = year, n◦ = number, PS = postero stabilized, CR = cruciate retaining, m = month).

**Table 2.** Demographic data, workplace and number of total annual implants.


(y = year, n◦ = number).

**Figure 1.** Demographic data, workplace and number of total annual implants. (y = year, n◦ = number).

With regard to total knee arthroplasty, most of the surgeons preferred a posterostabilized (PS) (78%) cemented (91.5%) implant. Focusing on the type of cement preferred, most of the surgeons (45.2%) used antibiotic-loaded cement only in selected patients, while 35% of the surgeons always used antibiotic-loaded cement. The cruciate retaining (CR) insert was used by 20.3% of surgeons, while only 1.7% preferred the medial pivot insert. Nearly half of the surgeons preferred selected patellar resurfacing (46.3%). Data related to the TKA surgical technique are presented in Table 3.

**Table 3.** Total knee arthroplasty related data.


With regard to unicompartmental knee arthroplasty (UKA) and patellofemoral arthroplasty (PFA), UKA was chosen less compared to TKA, with 40% of surgeons performing it in less than 10% of the cases, while 37.3% of surgeons performed UKA in 30% of the cases. Surgeons did not recognize a correct indication towards UKA (46.4%) and they were unfamiliar with the surgical technique (35.7%). During their activity, surgeons performed patellofemoral arthroplasty (PFA) in 36.2% of cases. UKA and PFA related data are summarized in order of frequency in Figure 2.

**Figure 2.** UKA and PFA related data. (N = number of answers).

With regard to radiological planning, characteristics of pre-operative planning were shared by most of the surgeons. For example, patellar view X-ray was required both in the total (70.6%) and in the unicompartmental knee arthroplasty (71.8%). Other routinely used radiographs were lateral view in 98.9% of cases and complete lower limb AP weightbearing in 91% of cases. Some surgeons only used the latter for unicompartmental (1.7%) and others found it not useful (7.3%) due to its frequent execution errors (27.3%) or due to organization budget reasons (36.3%). However, in most of the cases the long-leg view was performed in bipodalic position (80.8%). Varus/valgus stress view was required only in 0.6% of cases, while Rosenberg or Schuss views in unicompartmental knee arthroplasty were requested in two thirds of cases (66.1%). The majority of surgeons recognized the importance of adequate pre-operative planning (80.2%). Regarding pre-operative MRI, 53.7% of surgeons did not require it and 37.3% required it only for UKA. Radiographic pre-operative data are summarized in order of frequency in Table 4.

**Table 4.** Radiographic pre-operative data.


**Table 4.** *Cont.*


(TKA = total knee arthroplasty, UKA = unicompartmental knee arthroplasty, AP = antero-posterior, MRI = magnetic resonance imaging, N = number of answers).

With regard to post-operative follow-up, 74.6% of surgeons requested complete lower limb AP weight-bearing X-rays one year after knee replacement surgery. Only 23.2% did not use this view, mainly because they did not consider it useful (53,7%). Patellar view X-ray, very important in the pre-operative planning, was not required in the postoperative evaluation by 67.2% of the surgeons. When it was required, in one third of cases (32.8%), different views were performed, with the most commonly requested being the Merchant view at 45◦ of knee flexion (71% of the cases). The last two questions of the survey evaluated the post-operative follow-up. Unfortunately, there was no uniformity in the management of post-operative follow-up, with thirteen different combinations of timing and exams required, with none exceeding 35%. Radiographic and follow-up post-operative data are summarized in order of frequency in Table 5.

**Table 5.** Radiographic and follow-up post-operative data.


**Table 5.** *Cont.*


(AP = antero-posterior, UKA = unicompartmental knee arthroplasty, N = number of answers).

#### **4. Discussion**

The present survey was carried out in collaboration with the Italian Society of Knee, Arthroscopy, Sport, Cartilage and Orthopedic Technologies (SIGASCOT) and members of the Italian Society of Orthopedics (SIOT). The purpose of this study was to summarize Italian surgeons' preferences in pre-operative planning, surgical technique and post-operative follow-up for TKA and UKA. The analysis of our data is the research subject of the Italian orthopedics society for educational pathway promotion for training doctors. The study was strongly supported by the society because it represents one of the starting points of training: the role of possible tutors and training centers. Every year the universities try to ensure proper education for their training doctors. The role of the societies in this educational program is to improve the choice of non-university training centers. The debate is still open regarding hospital type, the role of the tutor and the main features of the training center. An Italian survey example carried out by the same society with an education objective was published in 2017 [27].

Survey analysis revealed a population of young surgeons who worked mainly in semi-private hospitals and with "high volume" surgeries (more than 100 arthroplasties per year) (Table 2). Italian orthopedic training was made up of various resources, such as cad-lab, face-to-face and multimedia courses, and indexed journals [27]. This training leads young surgeons to being open to innovations and continuous updating.

TKA is still one of the most common orthopedic procedures, with good reported outcomes. However, almost 20% of the patients are still unsatisfied post-operatively and there is also a considerable complication rate, including periprosthetic joint infection [28,29]. Historically, UKA and PFA were considered "at risk" procedures, with high failure rates, mainly due to poor implants and surgical techniques [15,30]. However, recently, better outcomes were reported with new implants, with lower failure rate and higher patient satisfaction compared to TKA, especially for "high volume" surgeons [31]. However, despite improvement in UKA outcomes, Italian knee surgeons still performed more TKAs compared to UKAs or PFAs. The TKA planning was uniform among the surgeons, with a prevalence of the PS-TKA over the other types of implants. Instead, the planning of the UKA has proved uneven, with a major impact of surgeon volume on planning type. Finally, the post-operative follow-up was too different in terms of timing and type of radiological examination required. Emerged data respected the current literature review with lack of standardization of UKA. Many authors have recognized the importance of standing AP, lateral, Merchant and Rosenberg, stress views and MRI for UKA focusing attention on some radiographic prognostic values, but no recommendations have yet emerged from these results [31]. Furthermore, recent literature has also shown great interest in MRI in total knee arthroplasty. Some authors demonstrated that measuring the distances of Achilles tendon from the mechanical axis of lower limb in magnetic resonance imaging of the ankle helps towards correct alignment of the components in the coronal plane. The pre-operative planning role is fundamental for the success of the arthroplasty. Correct pre-operative studies, such as X-ray or MRI, are necessary for the evaluation of the axes of the knee [32]. Additionally, a new MRI-based approach for the analysis of thigh muscles was described to improve a patient-specific rehabilitation program [33]. Computerized tomography seems to be of great importance in the use of robot surgery, as demonstrated in a 2020 ESSKA review [34]. The robotic-assisted procedure had significantly lower postoperative pain score, significantly reduced time until hospital discharge and significantly better functional scores when compared with traditional surgery [34].

Similar society analyses have been conducted in the literature. Friederich et al. [35] investigated the computer-assisted use for total knee surgery among members of the European Society of Sports Traumatology Knee Surgery and Arthroscopy (ESSKA) and the Swiss Orthopedic Society (SGO-SSO). Authors described half of surgeons using this technology and the improvement in alignment of prosthesis was the most strongly cited reason for using a navigation system. Jaap et al. [36] studied realistic expectations for recovery one year after TKA among Dutch orthopedic surgeons using a hospital for special surgery score. They concluded that greatest improvement was predicted for the items "pain relief" and "walking short distances". The British Association for Surgery of the Knee in conjunction with the James Lind Alliance investigated assessment, management and rehabilitation of patients with persistent symptoms after knee arthroplasty including patients, surgeons, anesthetists, nurses, physiotherapists and researchers. They concluded that the top ten research priorities focused on pain, infection, stiffness, health service configuration, surgical and non-surgical management strategies and outcome measures [37]. Alexander et al. [38], instead, surveyed orthopedic surgeons affiliated with the American Association of Hip and Knee Surgeons to inquire into the global impact of the COVID-19 pandemic on patient care. They described that all respondents noted their practices had been reduced and 70% of the surgeons canceled elective procedures. Our study, unlike the previous ones, examined Italian knee surgeons' behaviors. The questionnaire is completed

with a pre- and post-operative analysis, including the main types of implants available. Furthermore, it is important to underline the careful selection of the interviewees, as members of a specialized group of society.

Considering the type of implants used, data from this study can be compared to different international arthroplasty registers. The Nordic Arthroplasty Register Association (NARA) confirmed some Italian data. TKA was used as a primary implant (92%) in patients younger than 65 y.o. (64%), while UKA and PFA were less used (3% and 1%, respectively). From our data, UKA was chosen in less than 10% of cases and PFA was implanted in 36.2%. NARA demonstrated that cemented fixation was used in 92% of all TKAs as in our survey (91.5%). The patella was resurfaced in 22% of cases, while our survey showed almost double the value. Hybrid fixation was used in 5% of all TKAs [25] as with our data (2.8%). Similar data emerged from the Italian register of arthroplasty (RIAP). TKA was chosen more often than UKA (83.6% and 16.4%, respectively), as emerged from our data. Cemented fixation was used in 66.9% of TKAs and 65.7% of all UKAs, slightly lower values than ours. Hybrid fixation was used in 3.4% and 6.9% of TKAs and UKAs, respectively, as with our data. Patellar resurfacing was only used in 12.1% of TKAs and 1.8% of UKAs [39]. These data seem lower than ours due to the "Selected patients" answers to question 10. This created a bias regarding the absolute value of resurfacing, masking data similar to the Italian register

In the United States in 2016, 50% of surgeons preferred postero stabilized (PS) implants, and cruciate retaining (CR) was the second design commonly used, with almost 42% of total procedures [40]. These data are in line with this study, in which PS implants are preferred by most of the surgeons. This is probably due to contraindication to CR implants, such as posterior cruciate ligament (PCL) insufficiency, significant coronal deformity, extensor mechanism deficiency, posterolateral instability and inflammatory arthritis. Moreover, PS-TKA is generally easier to perform in most surgical situations without concern for obtaining appropriate tension on the PCL [41].

Cemented TKA guarantees good clinical outcomes with a long-term survival rate of up to 99% in comparison to a survival rate of up to 97% documented in cementless TKA [6,42].

Initial total knee arthroplasty designs did not include patellar implants, in fact high rate of anterior knee pain was found following these operations [7]. Modern TKA designs have all-polyethylene patellar implants, as the older metal-backed implants had high rates of wear and loosening [43]. Despite this innovation there are still complications related to the patellar implant, so much so that patellar resurfacing has been subject to controversy for several years [44].

Different implant types were studied by researchers but not included in our survey due to their non-popularity. Sabatini et al. [45] reviewed the second generation of bicruciate-retaining TKA. They summarized that in cases of high demand, end-stage bi- or tricompartmental knee arthritis, coronal malalignment <15◦, ACL integrity and minimal ROM reduction (< 5/10◦), this procedure could be a valid alternative to TKA or UKA [45].

Custom TKA, as new implants, are useful in cases of anatomical and functional variability. They added asymmetry and sizes to the existing implants. Actually, our understanding of the relation between the dynamic aspects of gait and position and form of the knee implants is lacking. Custom arthroplasties could also address these conceptual and practical difficulties due to robotics and artificial intelligence [46].

New materials for use are currently being researched. Ultra-high molecular weight polyethylene (UHMWPE) is widely chosen for its biomechanical characteristics. For reducing the polyethylene wear, one of the efforts is to investigate the selection of metal materials. Jamari et al. [47] analyzed the relationship between the polyethylene and metals via finite element analysis. They described, in total hip arthroplasty, that titanium alloy is able to reduce cumulative contact pressure if compared to stainless steel and cobalt chromium molybdenum on UHMWPE [47].

As emerged from the survey, there is no uniformity in the management of postoperative follow-up, with thirteen different answers. Different studies confirmed the importance of a complete post-operative follow-up, but also in the literature there is no uniformity in timing of evaluation, with variability ranging from 3 to 6 months [15].

The novelty of this study is the careful analysis of the Italian surgical situation compared to other countries. With the high number of annual procedures, the state could play a key role in research and innovation.

This study had some limitations. The number of participants in the study was low compared to those registered with the society. The sample was not homogeneous in terms of age, location of work and number of prostheses, leading to some bias. The analysis does not consider the new technologies present in the literature such as robots, patientspecific instrument or fluoroscopy-guided surgery. On the other hand, thanks to a complete questionnaire, the study analyzes the behavior of Italian surgeons and provides society with the starting point for the educational analysis of young surgeons. Although the questionnaire was quite complete, further aspects could be investigated about the surgical technique, such as surgical time, surgical approach and others.

An analysis is currently underway with similar questions addressed to training doctors, with the aim of comparing the answers of this analysis. This second survey represents the end of the research project which will be followed by the conception of educational practical courses, together with other projects already in progress.

#### **5. Conclusions**

Despite the improvement in UKA and PFA, we conclude that TKA still remains the preferred surgical option for Italian surgeons. PS cemented implants are the most commonly used, and patellar resurfacing was selected by most of the surgeons. Pre-operative planning is consistent with those reported in the literature and there is some agreement between surgeons, especially for TKA planning. When evaluating data regarding UKA, there is less agreement in pre-operative X-ray evaluation, with some surgeons requesting MRI and some surgeons preferring stress X-rays. Similar to the literature, there is absolutely no agreement in post-operative follow-up, both in terms of timing or radiological evaluation for both TKA and UKA. Considering these data, it may also be useful to promote some educational programs for training doctors after knowing their starting point.

**Author Contributions:** Conceptualization, L.M.; data curation, M.C.; writing—original draft preparation, F.R., E.M. and G.C.; writing—review and editing, M.B. and M.C.; supervision, G.S. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Institutional Review Board Statement:** The study was conducted according to the guidelines of the Declaration of Helsinki and approved by the Ethics Committee of Policlinico di Bari Hospital (delib. 0207 approved on 15 September 2018).

**Informed Consent Statement:** Informed consent was obtained from all subjects involved in the study. Written informed consent has been obtained from the patient(s) to publish this paper if applicable.

**Data Availability Statement:** The data presented in this study are available on request from the corresponding author.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


### *Article* **Spacer Block Technique Was Superior to Intramedullary Guide Technique in Coronal Alignment of Femoral Component after Fixed-Bearing Medial Unicompartmental Knee Arthroplasty: A Case–Control Study**

**O-Sung Lee 1, Myung Chul Lee 2, Chung Yeob Shin <sup>2</sup> and Hyuk-Soo Han 2,\***


**Abstract:** *Backgrounds and Objectives*: The spacer block technique in unicompartmental knee arthroplasty (UKA) has still a concern related to the precise position of the component in the coronal and sagittal planes compared to intramedullary guide technique. The purposes of this study were to explore whether the spacer block technique would improve the radiological alignment of implants and clinical outcomes compared with the outcomes of the intramedullary guide technique in fixed-bearing medial UKA. *Materials and Methods*: In total, 115 patients who underwent unilateral, fixed-bearing medial UKA were retrospectively reviewed and divided into group IM (intramedullary guides; n = 39) and group SB (spacer blocks; n = 76). Clinical assessment included range-of-motion and patient-reported outcomes. Radiological assessment included the mechanical femorotibial angle, coronal and sagittal alignments of the femoral and tibial components, and coronal femorotibial congruence angle. *Results*: All clinical outcomes showed no significant differences between groups. The coronal femoral component angle was valgus 2.4◦ ± 4.9◦ in IM group and varus 1.1◦ ± 3.2◦ (*p* < 0.001). In group IM, the number of outlier in coronal femoral component angle (<−10◦ or 10◦<) was 3 cases, while in group SB, there was no outlier (*p* = 0.014). The coronal femorotibial congruence angle was significantly less in group SB (mean 1.9◦, range, −3.2◦~8.2◦) than in group IM (mean 3.4◦, range, −9.6◦~16.5◦) (*p* = 0.028). *Conclusions*: In the group SB, the coronal alignment of femoral component was closer to neutral, and outlier was less frequent than in the group IM. The spacer block technique was more beneficial in achieving proper coronal alignment of the femoral component and congruence of femorotibial components compared to the intramedullary guide technique in fixed-bearing medial UKAs.

**Keywords:** spacer block; intramedullary rod; femorotibial congruence; unicompartmental arthroplasty

#### **1. Introduction**

Unicompartmental knee arthroplasty (UKA) is a more attractive treatment option than total knee arthroplasty (TKA) for patients with unicompartmental knee osteoarthritis; it is associated with less postoperative morbidity, more rapid recovery, more physiological kinematics, and greater patient satisfaction [1–4]. However, the national registries of Europe and Oceania consistently report inferior long-term survivorship of UKAs compared to TKAs [5]. It has been suggested that both implantation and ligament balancing during UKA may be rather inaccurate. Inappropriate component alignment may trigger edge loading, early polyethylene wear, and a high revision rate [6,7]. Mobile-bearing UKA features a round-on-round interface and provided that the orientation of knee components is within a tolerable range, any effect on knee mechanics may be negligible [8,9]. However,

**Citation:** Lee, O.-S.; Lee, M.C.; Shin, C.Y.; Han, H.-S. Spacer Block Technique Was Superior to Intramedullary Guide Technique in Coronal Alignment of Femoral Component after Fixed-Bearing Medial Unicompartmental Knee Arthroplasty: A Case–Control Study. *Medicina* **2023**, *59*, 89. https:// doi.org/10.3390/medicina59010089

Academic Editor: Vassilios S. Nikolaou

Received: 28 November 2022 Revised: 28 December 2022 Accepted: 29 December 2022 Published: 31 December 2022

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

fixed-bearing UKA features a round-on-flat interface; thus, it exhibits relatively narrower tolerance in terms of component-to-component malpositioning [10]. These characteristics of fixed-bearing UKA are associated with risks of edge loading and rapid wear [11,12].

Although new patient-specific instrument and navigation techniques are under development, conventional UKA techniques using intramedullary guides or spacer blocks remain widely used. In the context of mobile-bearing UKA, it remains unclear whether new intramedullary rod guides improve radiological alignment and clinical outcomes [13,14]. During fixed-bearing UKA, intramedullary rods have been widely used to align the femoral component. However, given the unsatisfactory accuracy of intramedullary alignment, extramedullary alignment using a spacer block or a tensor device is becoming popular [15–18]. Gap creation using a spacer block aligns the distal femoral resection parallel to the tibial resection in extension; the component gap can be predicted prior to femoral osteotomy. However, it is difficult to define a gap that reflects the physiologically relevant soft tissue balance; a surgeon's technical proficiency and subjective assessment of gap balancing are contributing factors.

In contrast to a TKA, coronal alignment of the components of a UKA is independent of the axis of the lower leg. Rather, the femoral and tibial component have an effect on each other especially in their coronal positioning in knee extension [19]. Although the spacer block technique shows promise in terms of reconstructing physiological joint tension, limited data are available regarding optimal femoral component positions in the coronal and sagittal planes. This technique produces the alignment of the femoral component according to positioning of the spacer introduced after cutting the tibia. Surgeon connects the block to this spacer for cutting the distal femur parallel to the tibial cutting surface on the coronal plane.

The purposes of the present study were to explore whether the spacer block technique improved implant radiological alignment and clinical outcomes, compared to use of an intramedullary guide, in patients undergoing fixed-bearing medial UKAs. It was hypothesized that the use of a spacer block would improve postoperative femoral component positioning and the congruence of femorotibial (FT) components.

#### **2. Materials and Methods**

In total, 129 consecutive patients who underwent unilateral, fixed-bearing medial UKA from February 2010 to May 2018 were eligible for inclusion. Fourteen were excluded for the following reasons: fewer than 2 years of follow-up (n = 4), postoperative complications including an infection (n = 1) and loosening (n = 3), and missing clinical data or radiographs (n = 6). Finally, 115 patients who were available for clinical and radiographic evaluation with minimum 2-year follow-up were divided into two groups according to the surgical instrument and implant used: group IM (intramedullary rod guide; n = 39) (MIS Miller/Galante, Zimmer, Warsaw, IN, USA) and group SB (spacer block; n = 76) (Sigma High Performance Partial Knee, DePuy Synthes, Warsaw, IN, USA). The study was approved by the institutional review board (approval no. H-2005-180-1126). The indications for UKA were identical in both groups: A clinical and radiographic diagnosis of isolated medial compartmental osteoarthritis, passive knee flexion contracture <10◦, active maximal flexion >100◦, and varus deformity <10◦. Patient demographics were collected from medical records; these data are summarized in Table 1. There were no significant differences in demographic characteristics between the groups.


**Table 1.** Demographic and preoperative clinical and radiological data of patients in the intramedullary guide and spacer block groups.

BMI, body-mass index; WOMAC, Western Ontario and McMaster Universities Arthritis Index; n.s, not significant. Related to Mechanical femorotibial angle and Medial proximal tibial angle, varus alignment was designated as negative values, and valgus alignment was designated as positive values. \* Derived with Pearson chi-square test. † Derived with Student's *t*-test.

#### **3. Surgical Technique**

All patients were operated upon by two experienced surgeons (HSH and MCL). A medial mini-parapatellar arthrotomy was performed through a skin incision of approximately 10 cm in length. After osteophyte resection, medial soft tissue release was confined to the deep medial collateral ligament. No further release to correct varus alignment was performed. The tibia was first cut using extramedullary guides to ensure coronal alignment between the native medial proximal tibial angle (MPTA) and 0◦ varus/valgus to the mechanical tibial axis. The sagittal alignment was aimed to be similar to the native medial posterior tibial slope. For femoral preparation, in group IM, a hole was created for insertion of the intramedullary guide (using an 8 mm drill), 1 cm anterior to the origin of the posterior cruciate ligament and immediately anterior to the femoral intercondylar notch. The intramedullary rod was inserted, and a distal femoral resection guide was then attached with the valgus angle measured on preoperative standing whole lower extremity radiographs. In group SB, a spacer block of the appropriate size (thickness) was placed; the gap from full extension to full flexion was checked. The goal was to ensure that the flexion and extension gaps were balanced with the spacer blocks in various sizes. A distal femoral cutting guide of the same thickness was inserted and pinned in place with the knee in extension. In both groups, femoral component rotation was performed to create a rectangular flexion gap. For confirmation of proper gap, the flexion gap is measured in about 100◦ flexion using the gap checking device that matches the spacer block thickness used for cutting the distal femur. Additionally, then, the extension gap is measured with this device. The correct gap indicating natural tension was assessed subjectively by checking whether the gap was loose with a thinner gap checker and tight with a thicker one (Figure 1). All components were cemented in both groups. All patients followed the same postoperative rehabilitation protocol, and full weight-bearing (as tolerated) the day after surgery.

**Figure 1.** Schematic diagram of surgical procedure in UKA. MCL; medial collateral ligament, IM; intramedullary, SB; space block.

#### **4. Clinical and Radiological Assessment**

Clinical and radiological evaluations were performed preoperatively, at 6 weeks and 3 months postoperatively, and annually thereafter. Information was retrieved from the clinical data repository regarding the pre- and post-operative ranges of motion, as well as clinical scores including the Hospital for Special Surgery score, the Knee Society Knee and Function Scores, the pain visual analog scale score, and the Western Ontario McMaster Universities Osteoarthritis Index.

Standardized anteroposterior and lateral radiographs of the knee, whole lower extremity standing anteroposterior (AP) radiographs were obtained for radiologic assessment. All radiological parameters were measured twice at a 2-week interval using a Picture Archiving and Communication System by two orthopedic surgeons who were not members of the surgical team. The preoperative mechanical FT varus angle, MPTA, and posterior tibial slope angle were measured. Postoperatively, the mechanical FT varus angle, coronal femoral component (α) angle and coronal tibial component (β) were determined by measuring varus (−)/valgus (+) alignment of each component relative to the long axis of the tibia on AP radiograph [11,20]. Sagittal femoral component angle was determined as the angle between a line perpendicular to the component part placed on distal femur cut and a long axis of the femur [21]. Sagittal tibial component angle was determined as the angle between a line perpendicular to the long axis of the tibia and a line tangential to the tibial component on a lateral radiograph (Figure 2). The coronal FT component congruence angle was defined as the angle of tilt between the line perpendicular to the long axis of the femoral component surface and the clearly visible lower margin of the tibial

component in the AP view; lateral convergences between these two lines were assigned positive values. Clinical and radiological results achieved at 2 years postoperatively were used for comparison between the groups.

#### **5. Statistical Analysis**

Continuous variables are presented as means with standard deviations; categorical variables are presented as frequencies with percentages. The Kolmogorov–Smirnov test was used to explore the normality of data distribution. Comparisons were performed using Pearson chi-square test and Student's *t*-test. Intraclass correlation coefficients were calculated to assess the inter- and intraobserver reliabilities of radiological measurements. To reveal the sample size of this study sufficient for adequate power, the statistical software G\*Power (Erdfelder, Faul, Lang and Buchner, 2014) was used for power analysis. A post hoc power analysis was performed and showed a sufficient sample size for this retrospective study with an alpha of 0.05 and a power > 0.8.) All statistical analyses were performed using SPSS software (ver. 25.0.0). A *p*-value < 0.05 was considered to indicate statistical significance.

#### **6. Results**

Neither demographic nor preoperative clinical and radiological data differed significantly between the two groups, with the exception of the mechanical FT varus angle (group IM, 6.6◦ vs. group SB 5.3◦; *p* = 0.05) (Table 1). No intraoperative complications or instrument-related problems were encountered in either group. The postoperative clinical and radiological data of both groups are summarized in Table 2. The clinical outcomes (except the maximal knee flexion angle) improved significantly in both groups. No significant differences were found in any postoperative clinical parameter in either group.


**Table 2.** Comparison of postoperative clinical and radiological data between patients in the intramedullary guide and spacer block groups.

FT, femorotibial; WOMAC, Western Ontario and McMaster Universities Arthritis Index; DFBA, distal femoral bowing angle; MPTA, medial proximal tibial angle; n.s, not significant. In representing mechanical FT angle, coronal femoral.

However, several radiological parameters differed significantly between the groups. Although the extent of correction of the mechanical FT angle did not differ between the groups, the postoperative mechanical FT varus angle was smaller in group SB than in group IM (−1.8◦ ± 3.7◦ vs. −0.3◦ ± 3.2◦, *p* = 0.025), which was similar to the preoperative difference. The coronal femoral component angle (α) was valgus 2.4◦ ± 4.9◦ in group IM, and varus 1.1◦ ± 3.2◦ in group SB, respectively (*p* < 0.001) (Figure 3).

**Figure 3.** Boxplot of the distribution of the coronal femoral component angle in group SB and IM.

The number of outlier in α angle (<−10 or 10<) was 3 in group IM, and 0 in group SB, respectively (*p* = 0.014) [22]. The postoperative coronal tibial component angle (β) significantly differed between the two groups (group IM, varus 1◦ ± 4.3◦ vs. group SB, 3.0◦ ± 2.4◦; *p* = 0.001), and MPTA change also significantly differed between the groups (group IM, 2.4◦ ± 4.9◦ vs. group SB, 0.2◦ ± 2.8◦; *p* = 0.003). Sagittal femoral and tibial component angles showed no significant difference between the two groups. Finally, the coronal FT component congruence angle of group SB (mean, 1.9◦; range, −3.2 to 8.2◦) was less than that of group IM (mean, 3.4◦; range, −9.6 to 16.5◦) (*p* = 0.028) (Figure 4). All radiological measurements exhibited excellent intraclass correlation coefficients in terms of both intra- and interobserver reliabilities (0.81–0.92).

**Figure 4.** Radiographs showing that the coronal FT component congruence angle of group SB is smaller than that of group IM. (**A**). Showing the coronal FT component congruence angle of group IM, (**B**) Showing the coronal FT component congruence angle of group SB.

#### **7. Discussion**

The principle findings of the present study were as follows: (1) the femoral component implanted by the spacer block technique has more neutral alignment relative to the long axis of the tibia with less outliers (>±10◦ deviation from neutral) [22]. (2) the mean and variation of the coronal FT component congruence angle in patients for whom the spacer block technique was used were lower than the corresponding values in patients treated using the intramedullary guide technique after fixed-bearing medial UKA. However, there were no significant between-group differences in clinical outcomes on short-term follow-up.

The success of UKA relies on appropriate implant alignment and positioning, as well as correct soft tissue tensioning, to ensure a balanced flexion-extension gap and stability [1–4]. Good surgical technique and instruments that avoid edge loading attributable to component-to-component mal-positioning are essential. In contrast to TKA, coronal alignment of the femoral component during UKA does not affect the leg axis; however, it influences relative component positioning in extension [8]. During UKA, the appropriateness of intramedullary alignment of the femoral component has been questioned, given the limited precision of manipulation, as well as potential morbidity created by trochlear cartilage perforation and medullary canal opening [15–17]. The spacer block technique seeks to ensure a rectangular, medial extension gap, thereby achieving maximum possible contact between femoral and tibial components. In this technique, the femoral component is aligned with the tibial cut by introduction of a spacer in extension after cutting the tibia. The intraoperatively, symmetrically resected medial extension gap also ensures parallel positioning of implants in the coronal plane.

In terms of evaluation of postoperative implant alignment after TKA, many studies have indicated that proper alignment of implant has a beneficial impact on implant survival and functional outcomes after TKA [23–25]. However, in contrast of TKA, the impact of implant alignment on long-term survival and clinical outcomes after UKA is unclear. UKA is known to have a widely acceptable range of component alignment within ±10◦ of coronal and sagittal alignment for the femoral component and ±5◦ coronal and sagittal alignment for the tibial component [11,22]. Despite acceptable deviation of postoperative limb alignment from neutral alignment may not have an effect on outcomes, far outliers of deviation may lead to harmful effect on survival and clinical outcomes after UKA. In our study, although the mean values of the coronal femoral component angle in both groups were within varus/valgus 3 degrees, the femoral component in the SB group has more neutral alignment relative to the long axis of the tibia with less outliers (>±10◦ deviation from neutral). Additionally, the mean coronal FT component congruence angle in SB group was lower than that in IM group, which may be associated with a risk of accelerated wear in the components.

Studies of mobile-bearing UKA using spacer blocks demonstrated a range of femoral component coronal alignments (8–11◦ valgus and 8–13◦ varus) [8,11]. However, during mobile-bearing UKA, because of the spherical nature of the femoral component, femoral malalignment of 10◦ and tibial malalignment of 5◦ did not hinder good clinical outcomes [11]. One retrospective study evaluated 193 consecutive patients who had undergone fixed-bearing medial UKA using the spacer technique [19]. The clinical results were comparable to those afforded by the intramedullary guide technique. However, the precision was lower and outliers were more frequent. In the present study, femoral implants exhibited no differences in terms of coronal or sagittal plane accuracies between groups SB and IM; however, the component-to-component congruence was better in group SB. When using the spacer block technique, the position of the femoral component is determined in the planes of the tibial component. Because femoral alignment is thus dependent on tibial alignment, the femoral and tibial components lie parallel, with only 1.9 ± 2.6◦ of asymmetry. This alignment is better than that of the intramedullary guide technique, where the femoral component position is independent of tibial component position. Given the narrower tolerable congruence between the femoral and tibial components of fixed-bearing UKA, compared to mobile-bearing UKA, the spacer block technique may be superior to the intramedullary guide technique.

When using the spacer block technique, care must be taken to avoid transference of any tibial malalignment to the femoral component; tibial resection must be very precise. However, no consensus has emerged regarding ideal positioning of the tibial component during UKA. In the present study, we first cut the tibia using extramedullary guides to achieve coronal alignment between the native MPTA and the 0◦ varus/valgus mechanical tibial axis; we also achieved sagittal alignment similar to that of the native medial posterior tibial slope. The mean coronal tibial component angle differed by 2◦ between groups IM

and SB. Although the angle of group IM was more neutrally aligned to the mechanical tibial axis, compared to the angle of group SB (89◦ and 87◦, respectively), the short-term clinical outcomes did not differ between the two groups. The significance of coronal tibial or femoral component alignment, and congruence between femoral and tibial components, should be explored over a longer follow-up period. Importantly, there is a need to address wear, aseptic loosening, and implant survival.

The optimal knee alignment during UKA remains controversial; few reports have addressed this aspect. The ideal component alignments and their effects on long-term UKA outcomes remain unknown. Overcorrection of a tibiofemoral deformity should be avoided to reduce the risk of degenerative change in the contralateral compartment; many authors have recommended that knee alignment should be relatively undercorrected during medial UKA [17,26]. However, undercorrection would increase the load on the medial compartment, which may accelerate polyethylene wear. In a retrospective case series (471 UKAs), more than 50% of failures occurred within 5 years of implantation [26]. The major cause of failure was development of other-compartment arthritis (39.5%), followed by aseptic loosening (25.4%). The FT angles tended to be greater in patients with contralateral arthritis. However, pre-medial UKA radiographs were not available for most patients. The use of a spacer block to determine femoral component position during UKA is associated with a risk of overcorrection [27]. However, the intramedullary guide technique is associated with risks of overstuffing and overcorrection when determining final polyethylene thickness. In the present study, we achieved a mean of 5◦ of correction and slight varus alignment in both groups. The correction was similar to that other studies (4◦ to 5◦) [19,28].

The present study had several limitations. First, it was retrospective in nature and did not involve randomization; selection bias may have influenced the findings. The preoperative mechanical FT angles significantly differed between the two groups. Second, we compared two different kinds of implants; this might have caused other forms of undetected bias. However, in terms of evaluation of the coronal and sagittal alignments of the tibial component, it is reasonable that our measurement method provides constant results regardless of the design of the implant (Figure 1). Additionally, in terms of measurement of the sagittal alignment of femoral component, we used the distal femur cutting line in order to minimize the bias due to the difference of the implants. Additionally, in terms of the coronal alignment of femoral component, we tried to minimize the bias by using the measuring method introduced by Gulati et al. [11], which is based on long axis of the femoral component regardless of the type of implant and most widely used in the evaluation of component alignment after UKA. Third, postoperative coronal tibial component angle (β) significantly differed between the SB and IM groups, which could affect the coronal femoral component angle (α). However, the more varus the coronal tibial component angle was set, the larger the coronal FT component congruence angle would become in the IM group (Figure 5). Therefore, we could reduce the concern about errors caused by the difference in the coronal tibial component angle after surgery. Finally, all measurements were performed using simple radiographs alone. Accordingly, component rotational errors could not be assessed.

**Figure 5.** Figure that expresses the imaginary lines showing the coronal FT component congruence angle increases as the varus degree of coronal tibial component angle increases. The (**a**) indicates the coronal FT component congruence, and the (**b**) indicates the the varus degree of coronal tibial component angle.

#### **8. Conclusions**

The spacer block technique was more beneficial in achieving proper coronal alignment of the femoral component compared to the intramedullary guide technique in fixed-bearing medial UKAs. The coronal alignment of femoral component was closer to neutral, and outlier was less frequent. Moreover, the congruence of FT components afforded by the spacer block technique was superior to that of the intramedullary guide technique after fixed-bearing medial UKA. Bone resection using a spacer block ensured that the implants were parallel in the coronal plane and may reduce edge loading. However, there were no significant differences in clinical outcomes between the two techniques on short-term follow-up.

**Author Contributions:** Conceptualization, H.-S.H. and M.C.L.; Methodology, H.-S.H. and O.-S.L.; Formal Analysis, H.-S.H. and O.-S.L.; Investigation, O.-S.L.; Resources, H.-S.H. and M.C.L.; Data Curation, C.Y.S., H.-S.H. and O.-S.L.; Writing—Original Draft Preparation, O.-S.L.; Writing—Review and Editing, C.Y.S., H.-S.H. and O.-S.L.; Visualization, O.-S.L.; Supervision, H.-S.H. and M.C.L. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Institutional Review Board Statement:** This study was approved by Seoul National University College of Medicine/Seoul National University Hospital Institutional Review Board (IRB no.: H-2005-180-1126).

**Informed Consent Statement:** Patient consent was waived due to anonymity of the data used in the study. It was approved by the Institutional Review Board of Seoul National University Hospital.

**Data Availability Statement:** The data presented in this study are available on request from the corresponding author. The data are not publicly available due to privacy.

**Conflicts of Interest:** The authors declare that they have no competing interest.

#### **References**


**Disclaimer/Publisher's Note:** The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

### *Article* **Risk Factors and Preventive Strategies for Perioperative Distal Femoral Fracture in Patients Undergoing Total Knee Arthroplasty**

**Ki Ho Kang, Man Soo Kim, Jae Jung Kim and Yong In \***

Department of Orthopaedic Surgery, Seoul St. Mary's Hospital, College of Medicine, The Catholic University of Korea, 222, Banpo-daero, Seocho-gu, Seoul 06591, Republic of Korea

**\*** Correspondence: iy1000@catholic.ac.kr

**Abstract:** *Background and Objectives* Perioperative distal femoral fracture is rare in patients undergoing total knee arthroplasty (TKA). In such rare cases, additional fixation might be required, and recovery can be delayed. Several studies have focused on perioperative distal femoral fractures in TKA, but there remains a lack of information on risk factors. The purpose of this study was to investigate risk factors for perioperative distal femoral fractures in patients undergoing TKA and suggest preventive strategies. *Materials and Methods*: This retrospective study included a total of 5364 TKA cases in a single institution from 2011 to 2022. Twenty-four distal femoral fractures occurred during TKA or within one month postoperatively (0.45%). Patient demographics, intraoperative findings, and postoperative progress were obtained from patient medical records and radiographs. Risk factors for fractures were analyzed using multivariate Firth logistic regression analysis. *Results*: Although all 24 distal femoral fractures occurred in female patients (24 of 4819 patients, 0.50%), the incidence rate of fracture between male and female patients was not significantly different (*p* = 0.165). The presence of osteoporosis and insertion of a polyethylene (PE) insert with knee dislocation were statistically significant risk factors (*p* = 0.009 and *p* = 0.046, respectively). However, multivariate logistic regression analysis showed that only osteoporosis with bone mineral density (BMD) < −2.8 (odds ratio (2.30), 95% CI (1.03–5.54), *p* = 0.043) was an independent risk factor for perioperative distal femoral fracture in TKA patients. *Conclusions*: Our results suggest that osteoporosis with BMD < −2.8 is a risk factor for distal femoral fractures in patients undergoing TKA. In these patients, careful bone cutting, adequate gap balancing, and especially the use of the sliding method for insertion of a PE insert are recommended as preventive strategies.

**Keywords:** total knee arthroplasty; femur fracture; polyethylene insert; osteoporosis; multivariate logistic analysis

#### **1. Introduction**

Fracture of the distal femur during primary total knee arthroplasty (TKA) is a rare complication, with two previous studies reporting an overall incidence ranging from 0.39% to 2.2% [1,2]. These fractures can happen intraoperatively or be found during the postoperative rehabilitation period without definite fracture history [3]. Although the majority of studies report an incidence below 1%, if a distal femoral fracture does occur, it can affect postoperative recovery and outcomes [4]. In fractures with minimal displacement, conservative treatment without additional fixation can lead to good outcomes [2]. However, most fractures require additional fixation [5]. Some severe cases require revision TKA [1]. Clinical outcomes are generally good with proper management; however, non-weight bearing or partial weight bearing for a period of 6–8 weeks as well as limitation of range of motion (ROM) after TKA are common, even in minimally displaced cases [4,6].

There are several known risk factors in distal femoral fracture during TKA, including osteoporosis, rheumatoid arthritis (RA), advanced age, female gender, and posteriorstabilized (PS) implant design [1,5,7–11]. Studies have shown that the wide box cut is a risk

**Citation:** Kang, K.H.; Kim, M.S.; Kim, J.J.; In, Y. Risk Factors and Preventive Strategies for Perioperative Distal Femoral Fracture in Patients Undergoing Total Knee Arthroplasty. *Medicina* **2023**, *59*, 369. https:// doi.org/10.3390/medicina59020369

Academic Editor: Vassilios S. Nikolaou

Received: 27 January 2023 Revised: 10 February 2023 Accepted: 13 February 2023 Published: 14 February 2023

**Copyright:** © 2023 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

factor for femoral fracture in PS TKA [8,9]. Recent PS TKA systems apply minimal boxcutting designs, and most PS surgeons would not change their main implant to a cruciateretaining type in their practice. In other studies reporting on female sex, osteoporosis, and RA as risk factors for femoral fractures during TKA, fewer than 10 cases of femoral fracture were included in each study [5,10,11]. Recently, one study presented risk factors in the Asian population [6], including female sex, osteoporosis, and small femur size; however, objective evaluation of osteoporosis was inadequate because bone mineral density (BMD) measurement was not routinely conducted. There is a lack of research on risk factors for femoral fracture during TKA [1,7].

The purpose of this study was to investigate risk factors of perioperative distal femoral fracture in a TKA series using PS implants at a single institution. We hypothesized that certain patient demographics or surgical techniques might be associated with perioperative distal femoral fracture. We would like to suggest optimal surgical strategies to prevent distal femoral fracture based on risk factors during TKA.

#### **2. Materials and Methods**

#### *2.1. Patient Selection*

A total of 5402 primary TKAs was performed on 3922 patients at our institution between September 2011 and October 2022 by two surgeons. All patients were of Asian ethnicity. Patients who previously underwent unicompartmental knee arthroplasty (UKA) and osteotomy around the knee were excluded. A total of 31 fractures occurred during the TKA procedure or within one month after TKA. A patient was excluded if the fracture occurred anywhere other than the distal femur. Cases that had occurred more than one month prior or in which the fracture was caused by definite trauma were also excluded. The remaining 5340 primary TKAs and 24 distal femoral fracture cases were enrolled in this study (Figure 1). This study was approved by the institutional review board of our institution {KC22RASI0865, approved on 25 November 2022}.

#### *2.2. Group Assessment*

Patients were divided into two groups with or without perioperative distal femoral fracture. The following demographic and surgical factors were compared between the two groups: age, sex, side, body mass index (BMI), preoperative hip-knee-ankle angle (HKAA), preoperative femorotibial angle (FTA), bone mineral density (BMD), surgical interval, type of implant, and mode of polyethylene (PE) insert insertion. All patients underwent BMD evaluation using dual-energy X-ray absorptiometry (DXA, Hologic Inc., Marlborough, MA, USA) prior to surgery. The BMD value was determined as the lowest T-score value among the femur head, total femur, and average lumbar spine, excluding outliers. The surgical interval of TKA was divided into three categories: unilateral TKA, same-day bilateral TKA, and staggered bilateral TKA with a one-week interval. PS-type implants of various knee systems were used, including 2244 TKAs using Lospa (Corentec, Seoul, Republic of Korea), 964 TKAs using Vanguard (Zimmer Biomet, Warsaw, IN, USA), 664 TKAs using advanced coated system (ACS) (Implantcast GmbH, Buxtehude, Germany), 351 TKAs using Exult (Corentec, Seoul, Republic of Korea), 261 TKAs using PFC (DePuy Synthes, Warsaw, IN, USA), 150 TKAs using Persona (Zimmer Biomet, Warsaw, IN, USA), 116 TKAs using Journey II (Smith & Nephew, Memphis, TN, USA), 56 TKAs using Legion (Smith & Nephew, Memphis, TN, USA), and 34 TKAs using Attune (DePuy Synthes, Warsaw, IN, USA). A total of 14 TKAs with fewer than 10 of a particular knee system were not described separately.

**Figure 1.** Participant flow diagram.

#### *2.3. Surgical Procedure*

All operations were performed by two surgeons in a standard fashion under general anesthesia with tourniquet inflation to 300 mmHg. A PS knee system with the subvastus approach was used in all cases. A measured resection technique was used for bone cutting. Proper gap balancing was applied after bone cutting. Extension and 90◦ flexion gaps were measured using a tensor device (B.Braun-Aesculap, Tuttlingen, Germany) and scaled forceps (B.Braun-Aesculap) with the application of a 200-N distraction force. In the case of a tight medial gap, multiple needling puncturing was performed with a standard 18-gauge needle based on digital palpation of taut medial collateral ligament (MCL) fibers [12]. All components were fixed using bone cement. Two PE insert insertion methods were used; the knee dislocation method and the sliding method (Figure 2). The PE insert insertion method was determined according to the design of the locking mechanism of the knee system or the status of soft tissue balance. Some knee systems required knee dislocation for secure PE insert insertion, and some allowed sliding insertion of PE insert without knee dislocation.

**Figure 2.** Modes of polyethylene (PE) inert insertion. (**a**) Insertion of the PE insert with knee dislocation. (**b**) Insertion of the PE insert with the sliding method.

#### *2.4. Statistical Methods*

Descriptive analyses were based on means and standard deviations for continuous variables and frequencies and percentages for categorical variables. The chi-square or Fisher exact test for categorical variables was used to compare two groups according to demographic characteristics. In this study, Firth logistic regression analysis was used to estimate the odds ratio (OR) with a 95% confidence interval (CI) due to the small number of fracture cases (24 cases) [13]. Multivariate logistic regression analysis was carried out to identify risk factors significantly related to perioperative distal femoral fracture (*p* < 0.1). All statistical analyses were performed using the SPSS ver. 29.0 program (SPSS Inc., Chicago, IL, USA). A *p*-value < 0.05 was considered statistically significant. Data are expressed as mean values with standard error of the mean (SEM) unless otherwise stated.

#### **3. Results**

Perioperative distal femoral fractures occurred with an incidence of 0.45% (24 cases) during TKA. Table 1 shows the demographics between patients that underwent TKA without fracture and those with perioperative distal femoral fracture. A total of 17 fractures occurred in 2701 Lospa (Corentec, Seoul, Republic of Korea) TKAs (0.63%) and 7 fractures in 717 ACS (Implantcast GmbH, Buxtehude, Germany) TKAs (0.98%). There were no significant differences between the without fracture group and with fracture group except for BMD value (*p* = 0.009), and mode of PE insert insertion (*p* = 0.046). Although all patients with fractures were female (0.50%), female sex was not a statistically significant factor (*p* = 0.165).

Among the 24 fractures, 22 (91.7%) had a varus knee deformity with an average HKAA of varus of 9.7◦. Seventeen fractures (70.8%) happened on the medial femoral condyle of varus deformed knee with an average HKAA of 9.4◦ (Table 2). The 21 cases (87.5%) that occurred intraoperatively were performed with screw fixation alone. All intraoperative fracture cases but one case achieved bony union, which required revision TKA because subsidence had occurred one month after fixation (Patient B; Table 2). All three postoperative fracture cases occurred 10 to 14 days after the operation without definite trauma history and had a fracture in the lateral femoral condyle that required revision TKA. Among the total 24 fracture patients, 19 (79.2%) were diagnosed with osteoporosis with a

BMD value of −2.5 or less, of which 16 cases (66.7%) were identified as osteoporosis with a BMD value less than −2.8.


**Table 1.** Demographics of patients with a perioperative distal femoral fracture with TKA.

HKAA; hip-knee-ankle angle; FTA; femorotibial angle.

Based on univariate factor analysis, variables that were statistically different and may affect fracture as potential predictors were used for additional univariate logistic regression analysis. Among them, BMD less than −2.8 (odds ratio [2.30], 95% CI [1.03–5.54], *p* = 0.042) was the only factor that correlated with a risk of perioperative distal femoral fracture in TKA. The BMD value of −2.8 was arbitrarily set as the cut-off value because it was the most statistically significant BMD value among values less than −2.5. BMD value of −2.5 corresponding to osteoporosis was not statistically significant. The parameters included in the multivariate logistic regression model were variables with a *p*-value less than 0.1 and were the mode of PE insert insertion method and BMD less than −2.8. Only one risk factor remained statistically significant. Patients with a BMD value less than −2.8 increased the risk for distal femoral fracture (odds ratio [2.30], 95% CI [1.03–5.54], *p* = 0.043) (Table 3).




**Table 3.** Multivariate logistic regression analysis of perioperative distal femoral fracture risk factors with TKA.

HKAA; hip-knee-ankle angle.

#### **4. Discussion**

This study evaluated the risk factors associated with perioperative distal femoral fractures in primary TKA. Among 5364 TKAs, 24 fractures occurred. The BMD value and mode of PE insert insertion (dislocation method) were the significant risk factors. In multivariate logistic regression analysis of risk factors, only BMD less than −2.8 showed a 2.30-fold increase in fracture risk.

Perioperative distal femoral fracture in TKA is uncommon, and there is a paucity of studies on this topic. Because some fractures that are not clinically important may be missed, the reported incidence might be underestimated [8]. In our study, there was a 0.45% incidence of perioperative femoral fracture with 5364 TKAs, similar to the incidence observed in previous studies (0.39% to 2.2%) [1,2]. According to a recently published systematic review, the overall incidence was from 0.2% to 4.4% [4]. Thus far, prospective studies have not been possible due to the small sample size and lack of predictability of fractures.

The risk factors for distal femoral fracture have been addressed in several previous studies. Osteoporosis is a well-known risk factor, but there is a lack of studies suggesting the cut-off value of osteoporosis. In our institution, DXA was included as a part of the routine preoperative workup for TKA candidates. Most of the patients with fractures were shown to have a low BMD value. In our study, BMD less than −2.8 was the only significant risk factor in multivariate logistic regression analysis. These patients should be informed about their low bone quality and associated risks preoperatively [14]. Then, the surgery should be performed with consent. McClung et al. reported that BMD value is important but emphasized that it was not the only determinant for fracture risk and assessment with individual patient risk [15]. However, according to our study, special attention should be given to patients with a BMD value less than −2.8 during surgery based on the 2.30-fold increase in fracture risk. In our practice, even in patients with osteoporosis, the same rehabilitation program was applied if there was no intraoperative fracture. However, medical treatment for osteoporosis was started or continued depending on the patient's situation [16].

Alden et al. reported that fracture occurs more commonly in female patients [1]. In our results, the fracture group was all female, which is consistent with previous studies. However, being female was not a significant risk factor in our analysis. According to large population-based cohort studies, an increased risk of symptomatic knee OA was associated with being female, particularly in the Asian population [17]. Although no fracture cases

occurred in male patients, since there were more female patients undergoing TKA, the risk of fracture in females was not significantly different in our study.

PS design of TKA has been associated with fracture. Lombardi et al. reported 40 distal femoral intercondylar fractures in 898 PS TKAs (4.4%) [9]. Alden et al. also reported a relative risk of 4.74 for femoral fracture with PS femoral components. In our study, all patients underwent the PS type TKA. Therefore, it was not possible to compare the sample with other types. The wider box cut in the PS type and inappropriate medial or lateral placement to a thin column of bone might result in a high incidence of fracture, which can be reduced by box design improvement and careful resection technique [1,9]. It was expected that advanced age would be a risk factor for fracture, but our results showed no relationship with age. A previous study showed that same-day bilateral TKA was a risk factor, but there was no such correlation in our study based on univariate logistic regression analysis [6]. Other known risk factors, such as chronic use of corticosteroids, RA, or neurological abnormalities, were not included in this analysis because there were few or no corresponding patients [18].

Perioperative distal femoral fractures can occur at various stages of the TKA procedure. It was reported that most fractures occurred during exposure and bone preparation of the femur and impaction of the femoral implants [1,2]. During the removal of a large osteophyte, inadequate box cut, thin column of the medial and lateral condyle, wrong hammering direction, and/or excessive force could cause fracture at this stage [6,10]. In our study, we found that distal femoral fracture, especially medial femoral condylar fracture, was more likely to occur at a certain stage of TKA. In patients with osteoporosis, we technically focused on soft tissue balancing, especially medial gap, and the PE insert insertion method. A tight medial gap in varus-deformed osteoporotic patients was the most common scenario for distal femoral fracture during TKA at the reduction stage of the knee joint after inserting PE insert with the knee dislocation. Depending on the knee systems, due to the design of the dovetail mechanism, knee dislocation is required to secure the insertion of the PE insert. Greater tension is applied to the MCL with the varus-deformed knee. If the medial gap balancing in the flexion position is not appropriate in these patients, the medial femoral condylar fracture risk is elevated during the knee reduction stage. There were only three cases (12.5%) of fractures among the implants using the sliding method in our study. Although this factor was not significant in our study, it reduced the risk of fracture by 2.73-fold in univariate logistic regression analysis. Alden et al. reported that performing a proper medial release prior to flexing the knee decreased tension on the MCL in the varus knee [1]. Extra attention should be paid to achieving the adequate medial gap, especially when inserting PE insert insertion using the knee dislocation method. After experiencing the fractures, the pie-crusting technique using multiple needle puncturing was applied to obtain the appropriate medial gap for the varus knee in our practice (Figure 3). The pie-crusting technique was reported to be effective and safe for both valgus and varus deformities during TKA [19,20]. In addition, multiple needle puncturing of the MCL is a more effective and safer technique relative to the blade knife technique [12,21].

Management options for perioperative distal femoral fracture are based on stable fixation and on surgeon preference. Huang et al. reported that screw fixation is more appropriate than plate fixation because the reduction of fracture fragments is easy, the plate could have affected the femoral component position, and the cement provides an additional stable force for the fracture fragment [11]. In our study, all fractures found intraoperatively were fixed with a screw alone. Among these cases, only one case of subsidence occurred after one month, and revision TKA was performed in that case. In the three cases of postoperative fractures, revision TKA was also required. These three cases occurred without any definite trauma and were different from previously reported periarticular fractures. The previously published periarticular fracture occurred on average two to four years after TKA with low-energy trauma [22]. It is thought that the perioperative fracture occurred during surgery, but that could not be verified. Shahi et al. reported 15 cases similar to our postoperative fracture cases in fracture patterns, and all of them underwent revision TKA [23]. With the recent development of three-dimensional technology, a clinical workflow for personalized surgical treatment can be developed in postoperative fracture cases [24]. After screw fixation, non-weight bearing or partial weight bearing was recommended to patients for 4–6 weeks, depending on the status of stability. Full weight bearing and full ROM exercise immediately after surgery were allowed in cases of revision TKA.

**Figure 3.** Ease of tension on the medial collateral ligament. (**a**) Pie-crusting with multiple needle puncturing. (**b**) Spreading medial flexion gap using a tensor device.

There exist limitations in our study. First, this study is a retrospective analysis. As mentioned earlier, the low incidence of perioperative fractures with TKA precludes a prospective study. Second, this study was carried out in a single hospital. If the study had included patients from several hospitals, the results would be more relevant, at least at the national level. Third, other known risk factors for distal femoral fracture, such as chronic use of corticosteroids, RA, and neurological abnormalities, could not be included in this analysis because there were few or no corresponding patients. In addition, clinical results confirmed by imaging studies were good in patients with appropriate treatment after a fracture. However, the clinical outcomes were not compared using patient-reported outcome measures (PROMs).

#### **5. Conclusions**

Perioperative distal femoral fractures during primary TKA are rare, but patients with osteoporotic bone and BMD values less than −2.8 need extra care during the TKA procedure. When inserting the PE insert, the sliding method can reduce the risk of distal femoral fracture. When using the knee dislocation method for insertion of PE insert, the medial and lateral gaps should be sufficiently balanced.

**Author Contributions:** Conceptualization, K.H.K. and Y.I.; methodology, K.H.K., M.S.K. and Y.I.; software, K.H.K. and J.J.K.; validation, K.H.K., M.S.K. and Y.I.; formal analysis, K.H.K. and J.J.K.; investigation, K.H.K. and M.S.K.; resources, K.H.K. and J.J.K.; data curation, J.J.K.; writing—original draft preparation, K.H.K.; writing—review and editing, K.H.K., M.S.K., J.J.K. and Y.I.; visualization, J.J.K.; supervision, M.S.K. and Y.I.; project administration, K.H.K. and Y.I. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Institutional Review Board Statement:** The study was conducted in accordance with the Declaration of Helsinki and approved by the Institutional Review Board of Seoul St. Mary's Hospital (KC22RASI0865 and 25 November 2022).

**Informed Consent Statement:** Informed consent was obtained from all subjects involved in the study.

**Data Availability Statement:** The data published in this research are available on request from the first author (K.H.K).

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


**Disclaimer/Publisher's Note:** The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

### *Review* **Detection of Prosthetic Loosening in Hip and Knee Arthroplasty Using Machine Learning: A Systematic Review and Meta-Analysis**

**Man-Soo Kim, Jae-Jung Kim, Ki-Ho Kang, Jeong-Han Lee and Yong In \***

Department of Orthopaedic Surgery, Seoul St. Mary's Hospital, College of Medicine, The Catholic University of Korea, 222, Banpo-daero, Seocho-gu, Seoul 06591, Republic of Korea **\*** Correspondence: iy1000@catholic.ac.kr; Tel.: +82-2-2258-2838

**Abstract:** *Background*: prosthetic loosening after hip and knee arthroplasty is one of the most common causes of joint arthroplasty failure and revision surgery. Diagnosis of prosthetic loosening is a difficult problem and, in many cases, loosening is not clearly diagnosed until accurately confirmed during surgery. The purpose of this study is to conduct a systematic review and meta-analysis to demonstrate the analysis and performance of machine learning in diagnosing prosthetic loosening after total hip arthroplasty (THA) and total knee arthroplasty (TKA). *Materials and Methods*: three comprehensive databases, including MEDLINE, EMBASE, and the Cochrane Library, were searched for studies that evaluated the detection accuracy of loosening around arthroplasty implants using machine learning. Data extraction, risk of bias assessment, and meta-analysis were performed. *Results*: five studies were included in the meta-analysis. All studies were retrospective studies. In total, data from 2013 patients with 3236 images were assessed; these data involved 2442 cases (75.5%) with THAs and 794 cases (24.5%) with TKAs. The most common and best-performing machine learning algorithm was DenseNet. In one study, a novel stacking approach using a random forest showed similar performance to DenseNet. The pooled sensitivity across studies was 0.92 (95% CI 0.84–0.97), the pooled specificity was 0.95 (95% CI 0.93–0.96), and the pooled diagnostic odds ratio was 194.09 (95% CI 61.60–611.57). The I2 statistics for sensitivity and specificity were 96% and 62%, respectively, showing that there was significant heterogeneity. The summary receiver operating characteristics curve indicated the sensitivity and specificity, as did the prediction regions, with an AUC of 0.9853. *Conclusions*: the performance of machine learning using plain radiography showed promising results with good accuracy, sensitivity, and specificity in the detection of loosening around THAs and TKAs. Machine learning can be incorporated into prosthetic loosening screening programs.

**Keywords:** loosening; arthroplasty; machine learning; transfer learning; review; prosthesis

#### **1. Introduction**

Total hip arthroplasty (THA) and total knee arthroplasty (TKA) are effective procedures that significantly improve quality of life and functional recovery in orthopedic surgery [1–8]. THA and TKA show sufficient survival rates [9,10], with 95% survivorship over 15 years in the case of TKA [10] and over 90% survivorship over 10 years in the case of THA, with continuous technological advancement [9]. Due to these successful results, the use of THAs and TKAs is continuously increasing. However, the ratio of revision THA and TKA is also expected to continue increasing because of continuous increases in total joint arthroplasty surgery and average life expectancy [11–15]. In the United States, revision TKA is expected to increase by 601% between 2005 and 2030; revision THA is expected to increase by 137% [13].

As revision THAs and TKAs increase, the associated economic burden is expected to gradually increase [16,17]. Therefore, understanding the causes and risk factors of

**Citation:** Kim, M.-S.; Kim, J.-J.; Kang, K.-H.; Lee, J.-H.; In, Y. Detection of Prosthetic Loosening in Hip and Knee Arthroplasty Using Machine Learning: A Systematic Review and Meta-Analysis. *Medicina* **2023**, *59*, 782. https://doi.org/10.3390/ medicina59040782

Academic Editor: Vassilios S. Nikolaou

Received: 18 February 2023 Revised: 2 April 2023 Accepted: 11 April 2023 Published: 17 April 2023

**Copyright:** © 2023 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

revision THA and TKA to improve the durability of revision surgery is increasingly important [16,17]. The most common causes of revision TKA are infection and aseptic loosening; the rate of aseptic loosening has increased recently [16–19]. In revision THA, infection and aseptic loosening also account for the largest proportion of revision surgeries [20–22]. As the diagnosis of prosthetic loosening is still challenging, various imaging tools are used for diagnosis, including plain radiographs, scintigraphy, arthrograms, fluorodeoxyglucosepositron emission tomography (FDG-PET) scans, and MRI [23,24]. However, except for plain radiography, these tools are invasive and expensive; therefore, plain radiography is the most cost effective method [25]. In addition, there is diversity in terms of concordance rates among the expert physicians assessing the cases [26,27].

Due to recent developments, machine learning has begun to be used widely in orthopedic surgery [28,29]. In particular, machine learning shows excellent performance in the field of diagnosis through images [28,29]. In fact, various techniques using machine learning are being used to diagnose lung disease [30] and breast cancer using radiographs [31]; these techniques are actually used as auxiliary diagnostic devices to assist doctors in diagnosis in hospitals [30,31]. Studies using machine learning to detect prosthetic loosening after arthroplasty surgery in orthopedic surgery are also continuously increasing in number [32–36]. Therefore, the purpose of this study is to conduct a systematic review and meta-analysis to demonstrate the analysis and performance of machine learning in diagnosing prosthetic loosening after total hip arthroplasty (THA) and total knee arthroplasty (TKA). We hypothesized that the model using machine learning is useful and can be incorporated into screening programs in diagnosing prosthetic loosening of THAs and TKAs.

#### **2. Materials and Methods**

This study was performed following the guidelines of the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) statement (S1 PRISMA Checklist) [37].

#### *2.1. Data and Literature Sources*

The study design was performed according to the Cochrane Review Methods. Multiple comprehensive databases, including MEDLINE, EMBASE, and the Cochrane Library, were searched in September 2022 for studies in English that detected the loosening of implants using machine learning (S1 Search Strategy). There were no restrictions on publication year. Search terms included mesh "Arthroplasty" and key words "replacement" "joint replacement" "alloarthroplasty", mesh "machine learning" and key words "transfer learning", "artificial intelligence", "deep learning", "neural network", "decision trees", and mesh "osteolysis" and key words "bone loss", "bone resorption", "loosening", "failure". After the initial electronic search, manual searches of the reference lists and the bibliographies of identified articles, including relevant reviews and meta-analyses, were conducted to identify trials that the electronic search may have missed. Identified articles were individually assessed for inclusion.

#### *2.2. Study Selection*

Two reviewers independently determined study inclusion according to the pre-defined selection criteria. Titles and abstracts were screened for relevance. In cases of uncertainty, the full article was evaluated to determine eligibility. Discrepancies were resolved through discussion. Studies included met these criteria: (1) used a machine learning algorithm as an index for the diagnosis of prosthetic loosening or osteolysis; and (2) the integrated data (true positive, false negative, false positive, and true negative) were provided directly or indirectly. The exclusion criteria included: (1) animal studies; (2) studies with incomplete data; and (3) reviews, comments, letters, and research for which full text cannot be obtained.

#### *2.3. Data Extraction*

Two reviewers independently extracted data from each study using a standardized data extraction form. Disagreements were resolved by discussion; those unresolved through discussion were reviewed by a third reviewer. The following variables were included: first author, publication year, country, study type, index test, eligibility criteria, reference standard, type of arthroplasty, sample size, machine learning algorithm, preprocessing, augmentation, model structure, the calculated area under the curve (AUC), accuracy, sensitivity, and specificity. We attempted to contact the study authors for supplementary information when there were insufficient or missing data in the articles. The third senior investigator was consulted to resolve any disagreement during data extraction.

The quality of all literature was evaluated by two researchers using the Quality Assessment of Diagnostic Accuracy Studies (QUADAS-2) [38], which is composed of patient selection, index test, reference standard, and flow and timing. If there is any disagreement in this process, the third author was responsible for making the decision.

#### *2.4. Statistical Analyses*

All extracted data analyses and picture production were performed with the R language using R studio. A bivariate random effect model was selected to analyze the true positive, false negative, false positive, and true negative values of 2 × 2 tables recorded in the sheet and test the heterogeneity. The sensitivity, specificity, positive likelihood ratio (PLR), negative likelihood ratio (NLR), diagnostic score, and diagnostic odds ratio (DOR) were calculated after integration. In addition, by drawing the summary receiver operating characteristics (SROC) through the Midas command, AUC discriminated the diagnostic ability of machine learning [39].

Heterogeneity was determined using the I2 statistic, with values of 25%, 50%, and 75% considered as indicating low, moderate and high heterogeneity, respectively. Due to the high levels of heterogeneity, a random-effects model was used to combine available data by meta-analysis. Random-effect DerSimonian and Laird models were used to calculate weighted averages of the transformed values, which were then back-transformed to produce final pooled rates [40]. Pre-planned sub-groups were designed according to the type of arthroplasty. Publication bias determination was usually only performed when the number of included articles was 10 or more; however, we used a funnel plot to evaluate publication bias.

#### **3. Results**

#### *3.1. Identification of Studies*

A study-flow diagram shows the process for study identification, inclusion, and exclusion (Figure 1). An initial electronic search yielded 1306 studies. Three additional publications were obtained through manual searching. In total, 82 potentially eligible studies were assessed for inclusion after screening titles and abstracts. After we reviewed the full texts, an additional 77 studies were excluded. Finally, five studies were included in the meta-analysis.

#### *3.2. Study Characteristics and Quality of Included Studies*

The study characteristics are summarized in Table 1. All studies were published from 2019 to 2022. All studies were retrospective studies. In total, 2013 patients with 3236 images were assessed, involving 2442 cases (75.5%) with THAs and 794 cases (24.5%) with TKAs. Among THAs, loosening occurred in 1136 cases (46.5%). Loosening occurred in 343 TKA cases (43.2%). Three studies included only THAs, one study included only TKAs, and one study included both THAs and TKAs. All five studies of these studies were retrospective studies and used the transfer learning method. The most common and best-performing machine learning algorithm was DenseNet; in one study, a novel stacking approach using a random forest showed similar performance to DenseNet. The characteristics of the included studies are summarized in Tables 1 and 2.

**Figure 1.** PRISMA flow diagram for systematic review.

#### *3.3. Descriptive Statistics*

The study with the highest accuracy was performed by Loppini et al. [34] in 2022 (96.8%), while the study carried out by Borjali et al. [32] in 2019 had the lowest accuracy (77.0%). The specificity and sensitivity values of each machine learning model are presented in Figure 2. Sensitivity values for machine learning models in this study ranged between 0.71 (95% CI: 0.60–0.80) and 0.97 (95% CI: 0.95–0.98) in THA. The pooled sensitivity across studies was 0.93 (95% CI: 0.81–0.98) in THA. Sensitivity values for machine learning models ranged between 0.70 (95% CI: 0.61–0.78) and 0.96 (95% CI: 0.93–0.98) in TKA. The pooled sensitivity across studies was 0.88 (95% CI: 0.81–0.98) in TKA. Specificity values for machine learning models in this study ranged between 0.95 (95% CI: 0.91–0.97) and 0.97 (95% CI: 0.95–0.98) in THA. The pooled sensitivity across studies was 0.96 (95% CI: 0.95–0.97) in THA. Specificity values for machine learning models ranged between 0.91 (95% CI: 0.87–0.94) and 0.95 (95% CI: 0.91–0.97) in TKA. The pooled specificity across studies was 0.93 (95% CI: 0.90–0.95) in TKA. The pooled sensitivity across studies was 0.92 (95% CI: 0.84–0.97), the pooled specificity was 0.95 (95% CI: 0.93–0.96) (Figure 2).






AI: artificial intelligence, AUC: area under the curve, DL: deep learning, CNN:

convolutional

 neural networks, THA: total hip arthroplasty,

 TKA: total knee arthroplasty.

**Figure 2.** Forest plots for sensitivity (**a**) and specificity (**b**). GLMM: generalized linear mixed model, CI: confidence interval, THA: total hip arthroplasty, TKA: total knee arthroplasty [32–36].

The pooled positive LR was 16.63 (95% CI: 10.55–26.20), while the pooled negative LR was 0.09 (95% CI: 0.03–0.27). The DOR values in this study ranged between 41.83 (95% CI: 20.51–85.30) and 863.23 (95% CI: 518.14–1438.14) in THA. The pooled DOR across studies was 282.93 (95% CI: 59.31–1349.64) in THA. The DOR values for machine learning models ranged between 43.85 (95% CI: 21.60–89.03) and 251.04 (95% CI: 108.70–579.74) in TKA. The pooled DOR across studies was 103.37 (95% CI: 18.70–571.35) in TKA. The pooled DOR was 194.09 (95% CI: 61.60–611.57) (Figure 3). The I2 statistics for sensitivity and specificity were 96.0% and 62.0%, showing that there was significant heterogeneity. The SROC curve indicated the sensitivity and specificity, as well as the prediction regions, with an AUC of 0.9853 (Figure 4).

**Figure 3.** Forest plots diagnostic odds ratio. IV: interval variance; CI: confidence interval, THA: total hip arthroplasty; TKA: total knee arthroplasty; DOR: diagnostic odds ratio [32–36].

**Figure 4.** Summary receiver operating characteristics (sROC) curve, the calculated area under the curve (AUC) = 0.985. SROC: summary receiver operating characteristics; SE: standard error; AUC: Area under the curve.

#### *3.4. Quality Assessment and Publication Biases*

The quality assessment results of five studies performed using the QUADAS-2 scale are indicated in Figure 2. The figure shows that the overall quality of the included studies was good; all studies were "unclear" or "low risk" with no study including "high risks". Although the sample size of the included literature was relatively small, the quality of the research was persuasive. (Figure 5) In addition, the funnel plot was asymmetrical, indicating the tendency of publication bias in this meta-analysis. (Figure 6).

**Figure 5.** Methodological assessment by QUADAS-2 [32–36].

**Figure 6.** Funnel plot.

#### **4. Discussion**

As a result of this systematic review, the diagnostic accuracy of the simple radiographic image-based artificial intelligence (AI) model for discriminating relaxation around implants after THA and TKA was over 0.9 in combined sensitivity, specificity, and AUC. As the machine learning for this purpose is in its immature stage, sufficient research results have not been secured. AI model studies using this imaging have not provided clear conclusions about clinical implementation and widespread use.

In the field of orthopedic surgery, THA and TKA are well known as the most effective and satisfactory surgical treatments [7,8]. In addition, long-term follow-up results have shown excellent survival rates [9,10]. The development of materials and technologies related to THA and TKA has contributed to the increase in the number of THA and TKA worldwide [41,42]. However, artificial joints inevitably have the risk of revision operation; thus, the most common cause of revision operation is loosening due to osteolysis [16,17]. Therefore, there has been a continuous demand for the early detection of prosthetic loosening around the implant. However, the accuracy of detection of prosthetic loosening in practice has not been as high as expected [43,44]. Temmerman et al. [44] analyzed the accuracy using radiography to diagnose cementless femoral component loosening; the sensitivity was 50% and the specificity was 89.5%. Cheung et al. [43] reported sensitivity and specificity of 83% and 82%, respectively. In addition to this, the inter-observer agreement between observers evaluating prosthetic loosening was also surprisingly poor [44]. Therefore, there has been a growing interest in tools that can have constant and high sensitivity, specificity, and accuracy in evaluating prosthetic loosening in THA and TKA [32,33]. Machine learning (ML) is on the rise as an alternative [32,33].

ML is a branch of AI that uses data to learn and improve tasks using various systems and algorithms [45,46]. Deep learning (DL) is a type of ML that allows complex tasks to be learned using large amounts of training information [47]. DL uses artificial neural networks (ANNs) made up of neurons arranged in a hierarchical structure. Convolutional neural networks (CNNs) are a subtype of DL that are effective and excellent at image processing. [48] A CNN uses a complex set of layers through which data is passed with a filter that can be trained to create a final or output layer [48]. The studies investigated in this review all have a common feature of using one of the various CNN algorithms [32–36]. This study is significant in its being the first to review studies that differentiate loosening around artificial joints using a CNN model. That model demonstrates excellent ability for image discrimination through plain radiographic images in the field of orthopedics [32–36].

Studies using AI in relation to arthroplasty in orthopedic surgery are gradually increasing in number [28,29]. These studies involve prediction of arthroplasty component size [49], length of stay and costs before primary arthroplasty [50], transfusion after arthroplasty [51], patient dissatisfaction following primary arthroplasty [52], and automated detection and classification of arthroplasty implant from knee radiograph [53]. In studies related to images, CNN models are generally most common; all the studies in this review detected prosthetic loosening using widely used CNN models learned through Imagenet. A further two studies used DenseNet [32,34] and one study used Exception [33]. The other two studies analyzed various CNN models and selected the model with the highest accuracy, which was DenseNet [35,36]. One study had limitations in accuracy with only images [36] while increasing the accuracy to more than 90% by providing additional patient data [36]. The remaining studies achieved satisfactory accuracy without additional patient data [32–35]. CNN models have been and are continuing to be developed. Additional research using the new CNN models will be needed in the future.

All five studies showed satisfactory results in terms of accuracy, sensitivity and specificity [32–36]. Accuracy, sensitivity, and specificity all showed high values of 0.9 or more [32–36]. The accuracy shown using simple plain radiographs, rather than more advanced imaging, shows similar or superior results compared to studies conducted only with imaging in other medical fields [28]. In a study by Adams et al., deep CNN was used for diagnosing femur neck fractures. The studies in this review showed excellent results with accuracy, specificity, and sensitivity of about 0.9 [54]. In a study by Urakawa et al., hip intertrochanteric fracture was diagnosed using a deep CNN model; the results of that study also showed a similar 95% accuracy [55]. In fact, two studies in this review compared diagnosis rates with orthopedic surgeons and showed higher accuracy than orthopedic surgeons [32,33]. Lau et al. reported that the accuracy of the machine learning model was 96.3%, while the average of the two orthopedic surgeons was 92% [33]. Borjali et al. demonstrated that the orthopedic surgeons' accuracy were 77%, which was lower than that of the machine learning model [32]. In particular, the specificity was similar; however, the sensitivity was poor in that study, indicating that loosening was not well differentiated [32].

Even when examining THA and TKA separately by sub-group analysis, the machine learning model showed high accuracy, sensitivity, and specificity in discriminating loosening around THAs and TKAs [32–36]. In the case of THA, four studies were included and showed excellent results, with accuracy, sensitivity, and specificity above 90% [32,34–36]. In the case of TKA, there are limitations as only data from two studies were included; however, this result also showed over 90% accuracy, sensitivity, and specificity [33,36]. Since the number of TKA samples was too small to perform sub-group analysis according to the type of joint arthroplasty, clear results require additional future TKA studies [33,36].

In the results of the meta-analysis, the heterogeneity among studies was particularly high at over 90% in sensitivity and, except in some cases, high in specificity [32–36]. Although difficulty arises in clearly explaining such high heterogeneity, potential reasons for heterogeneity include the diversity of CNN models used in this study, differences in patients' demographic data, differences in processes in creating CNN models, and the number of images used in the study [32–36]. This can be inferred from differences in quality. In this study, a random effect model was used due to high heterogeneity; this model was high in all values, including sensitivity and specificity.

There were some limitations in our research. Firstly, this study only included five articles; thus, the sample size was relatively small with only 1000 cases in the loosening group and 1500 cases in the non-loosening group [32–36]. Secondly, though four of the five studies contained data collected by the researchers at the hospital [32–34,36], one study contained image data collected from published papers available on the Internet [35]. Therefore, future research should focus on using high-quality data collected from hospitals or research centers in a broader population of patients with periprosthetic osteolysis or loosening with various clinical symptoms. Furthermore, future studies should explicitly describe the role of machine learning tools being developed as screening, diagnostic, or prognostic tools. A third limitation is that publication bias may exist. One possible cause is that there may be unpublished studies using underperforming models. In addition, the overall model of this study was highly heterogeneous and included a significant number of studies that could contribute to both the occurrence of publication bias and the high statistical power of the asymmetric test. Future meta-analyses should ensure conduct that includes pre-print articles that can reduce publication bias. Fourthly, the language restrictions in the included studies may have increased the risk of bias in the study results. Fifthly, there are many different machine learning models available with different variants and parameters. However, we were unable to compare each variant of the model due to the insufficient sample size of individual models. Lastly, most studies developed new models for TKA loosening detection but did not perform external validation using other data sources. One common concern is that the local data sets used for validation may not be representative of the target population on a global scale. In future studies, external validation requires improvement.

This meta-analysis is of substantial significance in that this is the first method to quantitatively combine and interpret data examining prosthetic loosening from different studies, potentially providing key clues for clinical application and further research. This study provided an opportunity to examine the accuracy of the method using ML in the detection of prosthetic loosening in THA and TKA. In addition, it was confirmed that the model using ML could be used as an auxiliary aid device in detecting prosthetic loosening in clinical practice. However, additional large-scale analysis studies and more research would be needed.

#### **5. Conclusions**

The performance of machine learning on plain radiographs showed promising results with good accuracy, sensitivity, and specificity in the detection of loosening around THAs and TKAs. Machine learning can be incorporated into prosthetic loosening screening programs. However, more research results are needed to clearly judge the ability of the machine learning model to discriminate loosening around the joint arthroplasty.

**Author Contributions:** Y.I. had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Concept and design: M.-S.K. and Y.I.; Acquisition, analysis, or interpretation of data: M.-S.K., J.-J.K., K.-H.K. and J.-H.L.; Drafting of the manuscript: M.-S.K. and Y.I.; Critical revision of the manuscript for important intellectual content: All authors; Administrative, technical, or material support: M.-S.K., J.-J.K., K.-H.K. and J.-H.L.; Supervision: Y.I. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was supported by Basic Science Research Program through the National Research Foundation of Korea (NRF) and funded by the Ministry of Education (2021R1I1A1A01059558).

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** The data presented in this study are available in the main article.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


**Disclaimer/Publisher's Note:** The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

MDPI St. Alban-Anlage 66 4052 Basel Switzerland www.mdpi.com

*Medicina* Editorial Office E-mail: medicina@mdpi.com www.mdpi.com/journal/medicina

Disclaimer/Publisher's Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Academic Open Access Publishing

mdpi.com ISBN 978-3-0365-9609-9