**Catheter-Associated Urinary Infections and Consequences of Using Coated versus Non-Coated Urethral Catheters—Outcomes of a Systematic Review and Meta-Analysis of Randomized Trials**

**Vineet Gauhar 1, Daniele Castellani 2,\*, Jeremy Yuen-Chun Teoh 3, Carlotta Nedbal 2, Giuseppe Chiacchio 2, Andrew T. Gabrielson 4, Flavio Lobo Heldwein 5, Marcelo Langer Wroclawski 6,7, Jean de la Rosette 8, Rodrigo Donalisio da Silva 9, Andrea Benedetto Galosi <sup>2</sup> and Bhaskar Kumar Somani <sup>10</sup>**


**Abstract:** Coated urethral catheters were introduced in clinical practice to reduce the risk of catheteracquired urinary tract infection (CAUTI). We aimed to systematically review the incidence of CAUTI and adverse effects in randomized clinical trials of patients requiring indwelling bladder catheterization by comparing coated vs. non-coated catheters. This review was performed according to the 2020 PRISMA framework. The incidence of CAUTI and catheter-related adverse events was evaluated using the Cochran–Mantel–Haenszel method with a random-effects model and reported as the risk ratio (RR), 95% CI, and *p*-values. Significance was set at *p* < 0.05 and a 95% CI. Twelve studies including 36,783 patients were included for meta-analysis. There was no significant difference in the CAUTI rate between coated and non-coated catheters (RR 0.87 95% CI 0.75–1.00, *p* = 0.06). Subgroup analysis demonstrated that the risk of CAUTI was significantly lower in the coated group compared with the non-coated group among patients requiring long-term catheterization (>14 days) (RR 0.82 95% CI 0.68–0.99, *p* = 0.04). There was no difference between the two groups in the incidence of the need for catheter exchange or the incidence of lower urinary tract symptoms after catheter removal. The benefit of coated catheters in reducing CAUTI risk among patients requiring long-term catheterization should be balanced against the increased direct costs to health care systems when compared to non-coated catheters.

**Keywords:** urinary catheters; catheters; indwelling; catheter-related infections

**Citation:** Gauhar, V.; Castellani, D.; Teoh, J.Y.-C.; Nedbal, C.; Chiacchio, G.; Gabrielson, A.T.; Heldwein, F.L.; Wroclawski, M.L.; de la Rosette, J.; Donalisio da Silva, R.; et al. Catheter-Associated Urinary Infections and Consequences of Using Coated versus Non-Coated Urethral Catheters—Outcomes of a Systematic Review and Meta-Analysis of Randomized Trials. *J. Clin. Med.* **2022**, *11*, 4463. https://doi.org/10.3390/

jcm11154463

Academic Editor: Javier C. Angulo

Received: 3 July 2022 Accepted: 28 July 2022 Published: 30 July 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/).

#### **1. Introduction**

The word catheter is derived from the ancient Greek *kathiénai*, literally meaning "to thrust into" or "to send down" [1]. In use for more than 3500 years, urethral catheters are a bane and boon for patients and urologists alike as they may pose a risk to patients requiring long-term catheterization. The most common problems include hematuria, catheter encrustation requiring frequent catheter exchange, and catheter-acquired urinary tract infection (CAUTI).

With technical advancements in bioengineering and materials science, several types of indwelling catheters were developed to prevent CAUTI and improve patient tolerance. Coating agents were added to catheter surfaces to improve antimicrobial proprieties and to provide robust antibiofilm/antimicrobial activity, without causing an increase in patient discomfort [2,3]. Coated catheters can be divided into two types: those coated in antifouling materials, and those impregnated with bactericidal molecules.

Antifouling substances do not kill the bacteria but rather perturb their ability to colonize surfaces, preventing the formation of biofilms in the bladder or on the catheter surface. The most common antifouling materials are hydrogel and polytetrafluoroethylene (PTFE). Hydrogel catheters may reduce encrustation via forming hydration layers on the catheter surface; however, studies have demonstrated a similar incidence of nosocomial CAUTI and a higher rate of blockage when compared to standard silicone catheters [4]. PTFE-coated catheters seem to be more suitable candidates to inhibit biofilm formation because of their low coefficient of friction. Unfortunately, studies have demonstrated that PTFE-coated catheters are not superior to hydrogel or standard silicone catheters in preventing CAUTI [2].

Catheters can also be coated with antimicrobial agents such as metal ions (i.e., silver, gold, and/or palladium), antibiotics, and nitrofurazone. Among bactericidal-coated catheters, silver-coated catheters are the most popular and widely tested catheters. The release of silver ions into the bladder induces oxidative stress and disrupts bacteria membrane and proteins, but antimicrobial efficacy may vary with the silver-coated substance used. Although in vitro and in vivo studies have shown great efficacy in preventing infections [5], these have not necessarily translated to clear benefits in clinical trials [6].

Antibiotic-coated catheters are less frequently used, especially with the increased frequency of having multi-drug-resistant bacteria [2]. Nitrofurazone was a promising coating agent in in vivo and in vitro studies, but it was not efficient in preventing infections in clinical studies and caused patient discomfort [7].

This study aimed to systematically review the incidence of CAUTI and its adverse effects in randomized clinical trials of patients requiring indwelling bladder catheterization (transurethral or suprapubic) by comparing coated vs. non-coated catheters.

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

#### *2.1. Aim of This Review*

The present study aims to systematically review the incidence of CAUTI in patients requiring indwelling bladder catheterization by comparing coated vs. non-coated catheters. The primary outcome was the CAUTI rate between the two types of catheters. The secondary outcomes were the CAUTI rate according to catheterization time (cut-off: 14 days) and the rate of catheter-related adverse events (i.e., hematuria, need for catheter exchange or catheter removal, urinary symptoms after catheter removal). Additionally a cost-effectiveness analysis was performed.

#### *2.2. Literature Search*

This review was performed according to the 2020 Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) framework. A broad literature search was performed on 1 May 2022, using MEDLINE, EMBASE, and Cochrane Central Register of Controlled Trials. Medical Subject Heading (MeSH) terms and keywords such as (urinary tract infection OR infections OR sepsis) AND (short term OR long OR indwelling) AND (standard urethral catheter OR impregnated urethral catheter OR silicone OR hydrogel OR antibiotic coated OR silver-impregnated) were used. The search was restricted to English papers only. No date limits were imposed. Pediatric and animal studies were excluded. The review protocol was submitted for registration in PROSPERO (receipt #332889).

#### *2.3. Selection Criteria*

The Patient Intervention Comparison Outcome Study (PICOS) model was used to frame and answer the clinical question. P: adults requiring bladder catheterization; Intervention: coated catheters; Comparison: non-coated catheters; Outcome: CAUTI and catheter-related adverse effects; Study type: prospective and randomized studies. Patients were assigned to two groups according to the type of catheter (coated vs. non-coated catheters).

#### *2.4. Study Screening and Selection*

Two independent authors screened all retrieved records through Covidence Systematic Review Management® (Veritas Health Innovation, Melbourne, Australia). Discrepancies were solved by a third author. Studies were included based on PICOS eligibility criteria. Only prospective and randomized studies were accepted. Meeting abstracts, retrospective, and prospective nonrandomized studies were excluded. Case reports, reviews, letters to the editor, and editorials were excluded. The full text of the screened papers was selected if found relevant to the purpose of this study.

#### *2.5. Statistical Analysis*

The incidence of CAUTI and catheter-related adverse effects was evaluated using the Cochran–Mantel–Haenszel method with a random-effects model and reported as the risk ratio (RR), 95% CI, and *p*-values. For studies with 3 groups of patients, intervention groups were combined to create a single pair-wise comparison [8]. Analyses were two tailed and significance was set at *p* < 0.05 and a 95% CI. Study heterogeneity was assessed utilizing the I2 value. Substantial heterogeneity was defined as an I<sup>2</sup> value > 50%. Meta-analysis was performed using Review Manager (RevMan) 5.4 software by Cochrane Collaboration. The quality assessment of the included studies was performed using RoB 2 [9].

#### **3. Results**

The literature search retrieved 2689 studies. After eliminating 297 duplicates, 2392 studies were left for screening. Another 2326 papers were further excluded against the title and abstract screening because they were unrelated to the purpose of this review. The full texts of the remaining 66 studies were screened and 54 papers were further excluded. Finally, 12 studies were accepted and included for meta-analysis. Figure 1 shows the PRISMA flow diagram.

#### *3.1. Study Characteristics and Quality Assessment*

Twelve prospective, randomized studies compared coated vs. non-coated catheters in patients requiring an indwelling catheter [7,10–20]. No study with a suprapubic catheter was retrieved. Study characteristics are summarized in Table 1. Only one study had catheters with antibacterial/antifouling coating (i.e., hydrogel) [16] and the other 11 had catheters coated with bactericidal molecules, i.e., pure silver ions [7,10,12,13,18], noble ions (silver, gold, palladium) [14], or silver ions mixed with hydrogel [19], nitrofurazone [7,11,15,17], and a polymer of zinc oxide bonded carbon nanotube [20]. There were 36,783 patients included in 12 studies: 19,404 patients in the coated catheter group and 17,379 in the non-coated catheter group.

**Figure 1.** PRISMA diagram of this study.

Table 2 shows data on pathogen species isolated in urine culture. The most common detected pathogens were *Escheria coli, Enterococcus*, *Pseudomonas* spp., *Klebsiella* spp., Gram-positive cocci, including *Staphylococcus aureus*, followed by *Candida* spp. and *Yeasts*. Polymicrobial infections were uncommon.

*J. Clin. Med.* **2022**, *11*, 4463

**Inclusion Criteria Exclusion Criteria Length of Follow Up Type of Coated Catheter Number of Patients Included in Coated Catheter Mean Age (Standard Deviation) in Coated Group Type of Non-Coated Catheter Number of Patients Included in Non-Coated Catheter Mean Age (Standard Deviation) in Non-Coated Group Catheter Duration (Days)** Akcam 2019 [10] Patients admitted to the intensive care unit and anticipated to require long-term urinary catheterization Patients with any infectious disease on admission or with pyuria/bacteriuria in the first urine specimen collected following catheter placement Until discharge of patients Silver-coated silicone catheters 28 70.61 (NA) silicone catheters 26 69.23 (NA) NA Bonfill 2017 [18] Patients with traumatic or medical SCI requiring an indwelling urinary catheter for at least 7 days Patients with demonstrable UTI at the time of inclusion; taking antibiotic treatment at the time of inclusion for any infectious condition or within 7 days before inclusion 12 months Silver alloy catheters 243 55.30 (16.35) silicone catheters 246 57.25 (16.32) 27 in coated 28 in non-coated Erickson 2008 [16] Men undergoing urethral reconstruction None 20 months Hydrogel-coated latex foley 42 40 (NA) silicone catheters 43 43 (NA) 14–21 Johnson 1990 [13] Patients with a steady catheter that was expected to remain indwelling for at least 24 h UTI at enrollment 16 months Silicone catheter coated with a layer of silicone elastomer containing micronized silver oxide 207 50 (NA) silicone catheters 208 47 (NA) 3 in coated 4 in non-coated Karchmer 2000 [12] Hospitalized patients with vesical catheters Pediatric, obstetrics, gynecology, and psychiatry wards excluded 12 months Silver-alloy, hydrogel-coated latex catheters 13,945 NA silicone catheters 13,933 NA >7 days Lee 2004 [17] Patients who were catheterized for more than 24 h conditions such as silicone sensitivity, nitrofurazone or nitrofurantoin sensitivity, pregnancy, lactating, hospitalization for more than 7 days, and having urinary diseases; positive urine culture result before catheter insertion or when the catheter was removed within 24 h of insertion 7 days Release nitrofurazone foley catheter 92 55.3 silicone catheters 85 54.1 3.9–4.4 Menezes 2019 [11] urethral catheterization for kidney transplantation with a livingdonorasymptomatic bacteriuria or urinary tract infection at baseline, deceased kidney transplant donors, hypersensitivity to 22 months Nitrofuralimpregnated silicone catheter 88 38.4 (NA) silicone catheters 88 39.6 (NA) 5.1

 spinal cord injury.

nitrofurantoin,

 pregnancy







Figure 2 shows the details of quality assessment in the included studies. Six studies showed a low overall risk of bias and the remaining six demonstrated some concerns. The most common reason for bias arose from the randomization process, followed by bias due to missing outcome data.

**Figure 2.** Risk of bias of the included study (Rob2): (**a**) review authors' judgments about each risk of bias item presented as percentages across all included studies; (**b**) review authors' judgments about each risk of bias item for each included study.

#### *3.2. Meta-Analysis of CAUTI*

Meta-analysis from 12 studies (19,328 cases in the coated and 17,287 cases in the noncoated group) showed that the risk of CAUTI did not differ significantly between the groups (RR 0.87 95% CI 0.75–1.00, *p* = 0.06) (Figure 3). There was no significant heterogeneity among the studies (I<sup>2</sup> = 22%). Subgroup analysis for catheter dwelling time demonstrated that the risk of CAUTI was significantly lower in the coated group compared with the non-coated group (RR 0.82 95% CI 0.68–0.99, *p* = 0.04). Only one study reported the rate of sepsis and another the rate of cystitis, making meta-analysis not feasible.

**Figure 3.** Meta-analysis of CAUTI incidence.

#### *3.3. Meta-Analysis of Need for Catheter Removal or Catheter Exchange*

Meta-analysis from three studies (499 cases in the coated and 502 cases in the noncoated group) showed no significant risk in the need for catheter removal or exchange (OR 0.93 95% CI 0.52–1.65, *p* = 0.80) (Figure 4). There was no significant heterogeneity among the studies (I<sup>2</sup> = 0%).

**Figure 4.** Meta-analysis of need for removal/change of catheter.

#### *3.4. Meta-Analysis of Lower Urinary Tract Symptoms at Follow-Up after Removal of Catheter*

Meta-analysis from four studies (4245 cases in the coated and 2419 cases in the noncoated group) showed that the number of patients complaining of lower urinary tract symptoms after catheter removal did not differ between the groups (OR 1.05 95% CI 0.87–1.17, *p* = 0.58) (Figure 5). There was no significant heterogeneity among the studies (I2 = 14%).


**Figure 5.** Meta-analysis of the number of patients reporting lower urinary tract symptoms at followup after removal of catheter.

#### *3.5. Meta-Analysis of Hematuria Incidence*

There was only one study reporting hematuria, making meta-analysis not feasible.

#### **4. Discussion**

In our meta-analysis, we found no difference in the incidence of CAUTI in patients with coated and non-coated catheters even though subgroup analysis regarding dwelling time (short- vs. long-term catheterization) showed a significantly lower risk for CAUTI in patients using coated catheters (*p* = 0.04). The interest in developing catheters that can decrease the risk of CAUTI started in 1979 with Akyama and Okamoto, who were the first to describe a decreased risk for bacteria associated with coated urinary catheters [21]. Other studies reported only a "protective effect" of coated urinary tract catheters but these trials were performed with a small number of patients [19,22,23]. Thibon et al. evaluated the effects of coated catheters with hydrogel and silver salts on the incidence of hospitalacquired urinary tract infection and showed no protective effect of coated catheters [19]. With regard to studies that reported a significant reduction in CAUTI in patients on silveralloy catheters [12,22,23], some methodological critiques were made to these studies as they were performed by randomizing the hospital unit instead of the individual patients, which could lead to bias since hospital units can differ significantly in terms of catheter placement technique, indwelling time, and patient comorbidities.

Another confounding factor in considering indwelling catheters and CAUTI risk is the surgical procedure performed. Ideally, catheters should be removed at the earliest possible time. The misconception that the use of antibiotic- or silver-coated catheters has better outcomes in patients undergoing urological procedures needing a short duration of catheterization was refuted in a study by Pickard et al. [7]. Likewise, Erickson et al. compared silicone- and hydrogel-coated latex catheters in men needing short-term postoperative bladder drainage after urethral surgeries and showed no absolute advantage for either type [16]. Menzies et al. compared nitrofurazone-coated and non-coated urinary catheters in kidney transplant recipients and did not find any difference in the rate of urinary tract infection (8% and 6.8%, *p* = 0.99) among the two groups [11]. Instead, the incidence of adverse events was more frequent in the nitrofurazone-impregnated silicone urinary catheter group (46.6% and 26.1%, *p* = 0.007) [11]. Tae et al. studied the incidence of CAUTI in patients who underwent radical cystectomy with an orthotopic neobladder for bladder cancer and received either a coated or conventional non-coated catheter for 2 weeks [20]. The incidence of CAUTI 2 weeks after radical cystectomy and orthotopic neobladder was 21.95% (case) and 27.27% (control), with no significant difference between the two groups. However, asymptomatic bacteriuria was significantly lower in the antibiotic-coated catheter group [20]. The authors concluded that the prevention of biofilm formation on coated catheters has the potential to prevent CAUTI. One explanation for why the CAUTI rate was similar between the groups is that the duration of catheterization was short for this cohort (2 weeks); as we demonstrated in our meta-analysis, coated catheters may only be of benefit during longer catheterization durations. When taken together, the results of the present meta-analysis (Figure 3) support the safety of using non-coated catheters in patients undergoing surgical procedures in which catheter duration is expected to be less than 14 days. For patients requiring long-term catheters, the use of coated catheters may lower the risk of CAUTI together with routine catheter and/or drainage bag changes [24].

In a randomized trial of 17 patients, Priefer et al. observed that the practice of monthly catheter exchange resulted in fewer symptomatic urinary tract infections when compared to patients in whom catheters were exchanged at the time of either obstruction or infection [25]. In contrast, White et al. found that when patients were divided into short- versus long-term catheter exchange intervals, the incidence of infection was greater in those whose catheters were changed in 2 weeks or less [26]. Only 15.4% remained free of infection after one month in this group, whereas 80% of those whose catheters were changed between 4 and 6 weeks remained free of infection after 6 weeks. The number of exchange and the number of nurses who performed the catheter exchange might have influenced the CAUTI risk. Indeed, there is insufficient evidence to assess the value of different policies for replacing longterm urinary catheters on patient outcomes [24]. We found that the incidence of CAUTI was decreased when maintained well even for a long duration (RR 0.82 95% CI 0.68–0.99, *p* = 0.04). Thus, maybe the implementation of protocols using coated catheters could be of interest to prevent encrustation, obstruction, and infection, and increase the intervals between changes.

Adverse events related to catheter use, such as hematuria, irritative lower urinary tract symptoms, or the need for catheter exchange or removal, were investigated as secondary endpoints in our study. Only one article classified the infections by differentiating into cystitis or urinary sepsis, preventing our analysis from evaluating these secondary outcomes. Furthermore, no studies comparing coated versus non-coated catheters evaluated rates of pyelonephritis. There were insufficient data to determine the relative influence of coated urinary catheters on hematuria. Hematuria, which was reported in only a single study, occurred in 18/243 (7.4%) patients in the silver alloy-coated catheter group and 20/246 (8.1%) patients using conventional catheters and this was not significantly different between groups [18]. Three studies involving a total of 1001 patients reported on the need for catheter removal or exchange. Overall, the need for urinary catheter exchange or removal was similar between non-coated and coated catheters [14,18,20]. In our analysis, four studies, which included 6664 patients, provided information on lower urinary tract

symptoms (LUTS) after catheter removal [7,14,17,18]. LUTS ranged from 1.2% to 22% in the coated group and from 0.4% to 22.6% in the control group. Compared to standard urinary catheters, we found that the use of coated catheters did not significantly increase the risk of LUTS.

Salient to the discussion of comparing antibiotic- or alloy-coated catheters to conventional silicone/latex catheters is cost-effectiveness. Overall, four studies incorporate cost-effectiveness analyses [12,27–29]. Cost analyses can be further stratified into comparisons of cost among different catheters and their associated components as well as analyses incorporating both catheter costs as well as the estimated cost of consequent nosocomial urinary tract infections. The latter cost assessment can be challenging to perform as it may be difficult to delineate how much a CAUTI contributes to the length of hospital stay or utilization of hospital resources. Nonetheless, several studies have provided estimates of these costs.

In a 12-month randomized crossover trial comparing CAUTI rates in patients with silver alloy-coated versus non-coated catheters, the use of silver alloy-coated catheters was associated with a 2.5-fold higher direct material cost when compared to non-coated catheters [12]. However, when taking into account the estimated costs associated with CAUTI and associated sequela (i.e., bloodstream infection, upper tract involvement, need for intensive care unit stay) within their study population, the use of silver alloy-coated catheters yielded significant aggregate savings due to a reduction in CAUTI rates. The lower and higher estimate of cost savings were USD 14,000 and 500,000, respectively [12]. This finding was similarly demonstrated by Bologna et al., where the use of silver alloy-coated catheters was predicted to lead to superior cost savings over standard latex catheters [27]. However, this cost analysis was limited to a single institution, whose differential CAUTI rate between silver alloy-coated and standard silicone catheters significantly differed from that of the other four institutions included in the analysis. The authors also relied on estimates of cost savings by attributing CAUTI as a major driver of hospital and intensive care unit length of stay [27]. Importantly, a recent prospective crossover study comparing silver alloy-coated to standard silicone catheters demonstrated a 12% risk reduction against CAUTI with the use of silver alloy-coated catheters. This is contrary to a prior study that assumed a 30–40% relative reduction in the CAUTI rate with the use of silver alloy-coated catheters in their cost-effectiveness analyses [29]. Therefore, if the difference in the CAUTI rate between catheter types is modest, the cost savings with the use of silver alloy-coated catheters may be negated and may not outweigh the increased direct costs associated with these catheters [29].

In another large study involving 7102 patients admitted to NHS England hospitals, cost-effectiveness analysis demonstrated that nitrofurazone-coated catheters were the least costly [30]. When compared to nitrofurazone-coated catheters, PTFE and silver alloycoated catheters cost on average USD 11 and 19 more, respectively. Based on their analysis, nitrofurazone-coated catheters had an approximately 70% chance of being a cost-saving and had an 84% chance of having an incremental cost per quality-adjusted life year [incremental cost-effectiveness ratio of < GBP 300,000 (USD 47,500), the willingness-to-pay threshold suggested by the UK National Institute of Health and Clinical Excellence] [30]. Conversely, silver alloy-coated catheters had a 0% chance of being cost-effective at all threshold values between GBP 0 and 50,000. Nonetheless, nitrofurazone-coated catheters were associated with greater patient discomfort and the cost-saving estimates were based on assumptions of large attribution of CAUTI as the main driver of the length of hospital stay. These results, therefore, do not provide robust evidence of cost-effectiveness for one catheter over another within a universal health care system [30].

When taken together, the use of metal alloy-coated or antibiotic-coated catheters may increase direct costs to health care systems when compared to standard silicone or latex catheters; however, it is unclear whether the risk reduction in the CAUTI rate (and associated health care utilization) outweighs this cost.

Our study has some limitations. This study precludes us from making absolute deductions on which coated catheters are better for minimizing CAUTI, and better clinical trials should address this in the future. We could deduce that patients with long-term indwelling catheters could be the ideal candidates for coated catheters and it is necessary to provide proper training to patients and caregivers for catheter maintenance. This could help optimize the cost-effectiveness for the patients as, from our results, due to paucity of information and likely variability in health care systems, it was difficult to make concrete conclusions on cost-effectiveness. Finally, there was no randomized clinical trial comparing coated vs. non-coated suprapubic catheters, considering that UTI incidence is not significantly different between urethral and suprapubic catheters in spinal cord injury and neurogenic bladder [31].

#### **5. Conclusions**

In this systematic review of randomized trials, we found that the use of indwelling coated catheters was not associated with a lower incidence of CAUTI and the need for removal/change of catheter compared to non-coated catheters. In addition, we also found no difference in lower urinary tract symptoms after catheter removal. However, the incidence of CUATI was significantly lower using silver alloy-coated catheters in patients who require more than 14 days of dwelling time. The utility of coated catheters to reduce CAUTI risk versus standard catheters must be balanced with differences in direct costs to patients and health care systems.

**Author Contributions:** Conceptualization, V.G., B.K.S. and J.d.l.R.; methodology, V.G. and D.C.; data gathering, V.G., D.C., C.N., G.C., A.T.G., R.D.d.S., F.L.H., J.Y.-C.T. and M.L.W.; validation, B.K.S., J.Y.-C.T., A.B.G. and J.d.l.R.; formal analysis, D.C.; writing—original draft preparation, V.G., D.C., C.N., G.C., A.T.G., R.D.d.S., F.L.H. and M.L.W.; writing—review and editing, D.C., V.G., B.K.S., A.B.G., J.Y.-C.T. and J.d.l.R. All authors have read and agreed to the published version of the manuscript.

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

**Data Availability Statement:** Data will be provide by the corresponding author upon a reasonable request.

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

#### **References**


## *Opinion* **Low- vs. High-Power Laser for Holmium Laser Enucleation of Prostate**

**Vasileios Gkolezakis 1, Bhaskar Kumar Somani <sup>2</sup> and Theodoros Tokas 3,4,\***


**Abstract:** Holmium laser enucleation of the prostate (HoLEP) constitutes an established technique for treating patients with symptomatic bladder outlet obstruction. Most surgeons perform surgeries using high-power (HP) settings. Nevertheless, HP laser machines are costly, require high-power sockets, and may be linked with increased postoperative dysuria. Low-power (LP) lasers could overcome these drawbacks without compromising postoperative outcomes. Nevertheless, there is a paucity of data regarding LP laser settings during HoLEP, as most endourologists are hesitant to apply them in their clinical practice. We aimed to provide an up-to-date narrative looking at the impact of LP settings in HoLEP and comparing LP with HP HoLEP. According to current evidence, intra- and post-operative outcomes as well as complication rates are independent of the laser power level. LP HoLEP is feasible, safe, and effective and may improve postoperative irritative and storage symptoms.

**Keywords:** prostate enucleation; laser power; holmium; HoLEP

#### **1. Introduction**

Benign prostate hyperplasia (BPH) with consecutive lower urinary tract symptoms (LUTS) constitutes a significant health issue of the aging male. Traditional transurethral resection of the prostate (TURP) remains the standard treatment for small- and mediumsized prostate glands and patients who fail medical therapy and may have complications of outlet obstruction such as bladder stones, urinary retention, or renal insufficiency [1]. Nonetheless, between several surgical treatment modalities, transurethral holmium laser enucleation of the prostate (HoLEP) has emerged as a distinctive technique that can be applied to prostates of all sizes [2–4]. Compared to standard TURP, HoLEP offers better hemostasis, shorter catheterization and hospitalization times, and nullifies the rate of TURP syndrome [3,5]. The holmium laser technology enables the prostatic tissue to be enucleated from the capsule while simultaneously coagulating the capsular surface. HoLEP, which offers long-term functional results superior to TURP and comparable to open simple prostatectomy but with lower treatment morbidity and complication rates, is therefore regarded as a procedure of reference for the surgical treatment of large prostate glands [2–4]. Surgeons usually apply power settings of 80–100 W with 2 J energy and 40–50 Hz frequency and an occasional power reduction for coagulation (75 W, 1.5 J, and 50 Hz) and apical preparation (30 W, 0.6 J, and 50 Hz) [6,7]. These settings provide the ability to adjust pulse duration to energy and frequency but also necessitate more expensive equipment with numerous high-power plugs, which are generally considered limitations of the widespread adoption of HoLEP.

Low-power (LP) devices are also available on the market, functioning at powers of 20, 30, and 50 W with lower startup costs and no demand for specialized plugs. The same equipment can be used successfully for lithotripsy and BPH surgery. Comparing these

**Citation:** Gkolezakis, V.; Somani, B.K.; Tokas, T. Low- vs. High-Power Laser for Holmium Laser Enucleation of Prostate. *J. Clin. Med.* **2023**, *12*, 2084. https://doi.org/10.3390/ jcm12052084

Academic Editor: Richard Naspro

Received: 14 December 2022 Revised: 27 February 2023 Accepted: 3 March 2023 Published: 6 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/).

qualities to high-power (HP) units could be advantageous. Rassweiler et al. were the first to use LP settings (24 W, 2 J, and 12 Hz, or 39.6 W, 2.2 J, and 18 Hz) to treat 129 patients, proving the treatment's viability, safety, and effectiveness [7]. These findings implied a significant reduction in the initial capital equipment cost, which may make adopting this technique more bearable if the method's effectiveness is preserved. Despite this, the LP method has not gained much support, and there are still few reports on LP HoLEP in the era of rising HP machine output. On the other hand, the higher price tag that comes with these sophisticated devices is a significant disadvantage, and many endourologists are looking for less expensive options. Furthermore, while HP and unique technologies like MOSES may only be available in referral centers, LP holmium laser machines are universally available, as they are frequently adopted for other endourological procedures (i.e., lithotripsy). Hence, this work aimed to compare LP to standard HP HoLEP regarding perioperative parameters, complications, and functional outcomes.

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

#### *2.1. Literature Search*

We performed a literature search in PubMed and used the following keywords: 'prostat\* hypertrophy', 'prostat\* hyperplasia', 'BPH', 'BPO', 'HoLEP', and 'holmium laser'. We limited our search to papers written in English.

#### *2.2. Selection Criteria*

The PICOS (patient, intervention, comparison, outcome, study type) model was employed. Patient: adults undergoing HoLEP for BPH; intervention: LP HoLEP; comparison: HP HoLEP; outcome: surgical time, operative efficiency, postoperative catheterization time, length of hospital stay, re-catheterization, blood transfusion, incontinence rate, international prostate symptom score (IPSS), maximum peak flow (Qmax), and post-void residual urine (PVR) at last follow-up; study type: randomized, prospective non-randomized, and retrospective studies. HP HoLEP was conducted at 100 W, whereas LP was performed at 30 W, 40 W, and 50 W. Due to study heterogeneity and the non-standardized quality appraisal, we performed a narrative synthesis. The limitations of using a single database for a review are also taken into account [8]. Furthermore, our results might be constrained by study heterogeneity and selection bias. We included randomized, prospective non-randomized, and retrospective studies and excluded case studies, reviews, and editorials.

#### **3. Results**

The literature search identified 969 records (Figure 1). Additionally, we included four studies from other sources. Following title, abstract, and full-text screening, we selected and included 11 studies in the review (Tables 1 and 2). Five studies described the functions and presented outcomes of LP HoLEP and six compared LP with HP HoLEP in terms of procedure times, peri- and postoperative outcomes, and complications. We included four meeting abstracts [9–12], one prospective comparative study [13], one prospective randomized trial [14], one prospective case series [15], three retrospective case series [16–18] and one ex vivo porcine study [19].

#### *3.1. Efficiency and Speed of LP HoLEP*

Operative time (OT), enucleation time (ET), operative efficiency (OE; defined as resected prostate weight divided by operative time in g/min), enucleation efficiency (EE; defined as resected prostate weight divided by enucleation time in g/min), laser/prostate ratio (defined as laser energy consumed divided by resected prostate weight, in KJ/g), and laser rate (defined as the laser energy consumed) were among the outcome measures evaluated. In the first randomized controlled trial comparing LP (50 W, 2 J, and 25 Hz) versus traditional HoLEP (100 W, 2 J, and 50 Hz), the authors found comparable outcomes in terms of EE (1.42 ± 0.6 g/min vs. 1.47 ± 0.6 g/min), OE (1.01 ± 0.4 g/min vs. 1.09 ± 0.4 g/min), and OT (81 min vs. 75.5 min) regardless of the surgeon's experience [14]. Two prospective case-control studies from the same group presenting the records of 316 patients with any prostate volume (range 10–200 g), normal PSA, Qmax < 15 mL/s, IPSS > 10, and PVR < 300 cc and comparing the efficacy of en-bloc no-touch LP HoLEP (40 W, 2 J, and 20 Hz) with HP HoLEP (100 W, 2 J, and 50 Hz) revealed identical results regarding mean ET (27.5 min vs. 31 min) and EE (1.7 g/min vs. 1.64 g/min, respectively) in the hands of an experienced surgeon [10,11]. The authors reported a reduction in energy consumption of nearly one-third.

Gazel et al. compared the impact of two different LP settings on enucleation and hemostasis in 160 patients and recorded increased EE (1.2 vs. 0.78 g/min, *p* = 0.001) while administering 37.5 W (1.5 J and 25 Hz) as opposed to 20 W (1 J and 20 Hz) [17]. In addition, the mean enucleation rate (0.64 vs. 0.88%, *p* = 0.001) and laser efficiency (2.07 vs. 2.12 joule/g, *p* = 0.003) were significantly higher with 37.5 W. The enucleation time was significantly shorter (54 vs. 75.5 min, *p* = 0.002). The authors concluded that using 37.5 W, both enucleation and hemostasis could be performed successfully, while using 100 W in the bladder neck shortens the duration of the procedure. Furthermore, in an experimental ex vivo study, Yilmaz et al. demonstrated that the HP–Ho:YAG's efficiency (evaluated by a numerical measurement of the "tissue pocket" created by separating the fascial layers of a porcine belly, measured in cm2/min) was reduced by 50% in an LP (3 J and 10 Hz) compared to a medium-power (3 J, 25 Hz) laser setting. Additionally, the authors demonstrated more favorable dissection results with HP systems applying high single-pulse energy, short pulses, and medium frequency [19].

**Figure 1.** Flow chart.


**Table 1.** Characteristics of low-power HoLEP studies with initial efficacy parameter.



**Table 2.** Characteristics of included studies comparing low-power vs. high-power holmium laser enucleation of the prostate with initial efficacy parameter.


#### *3.2. Functional Outcomes of LP HoLEP*

Two prospective studies showed significant improvement in IPSS scores and Qmax at three months (24 vs. 5, *p* < 0.001 and 7.8 mL/s vs. 28 mL/s, *p* < 0.001, respectively) [13] and 12 months follow-up (22 vs. 6, *p* < 0.001 and 12 mL/s vs. 29.3 mL/s, *p* < 0.001, respectively) [14] compared to the preoperative assessment. The authors also observed significant improvements regarding PVR at three months (100 mL vs. 30 mL, *p* < 0.001) and 12 months follow-up (135 mL vs. 11.15 mL, *p* < 0.001) [15]. Gilling et al. prospectively compared LP to HP HoLEP and observed a considerable and persistent improvement in these parameters for both energies at up to 12 months follow-up compared to baseline (30.9 mL/s vs. 7.7 mL/s for 50 W setting, 19 mL/s vs. 7.4 mL/s for 100 W setting, statistical evaluation not reported) [9]. In addition, one prospective comparative study [10] and one randomized trial [14] showed similar results with no difference among LP and HP HoLEP regarding IPSS scores at three months follow-up (6.5 ± 5 d.s. vs. 7.8 ± 5 d.s.) [10] and IPSS scores and Qmax at 12 months follow-up (3 vs. 4, *p* = 0.4, 21.1 mL/s vs. 21.8 mL/s, *p* = 0.7) [14].

#### *3.3. Postoperative Stress Urinary Incontinence (SIU) and Dysuria after LP HoLEP*

Becker et al. reported postoperative SIU rates of 16.7% after one month, declining to 0% after six months [15], whereas Gazel et al. showed similar postoperative SIU rates with both 20 W and 37.5 W energy settings (3.7 vs. 2.5% at three months and 1.2 vs. 0% at 12 months follow-up) [17]. Minagawa et al. assessed SIU retrospectively and found that in 55 patients without preoperative SIU, postoperative SIU was observed in seven patients (12.7%) at one month postoperatively and in three patients (5.5%) at three months postoperatively [18]. Scoffone et al. [10] and Elshal et al. [14] demonstrated similar long-lasting incontinence rates at three months (1.6 vs. 1.4%) [10] and four months (1.6 vs. 1.7%) [14] in patients undergoing LP HoLEP and HP HoLEP.

#### *3.4. Safety of LP HoLEP*

Reported complication rates of LP HoLEP ranged from 7 to 24% [12]. Of them, 3.7% were Clavien grade 3a, and 5.5% were Clavien 3b [15]. Transfusion rates varied from 0%, with only one case among 74 patients requiring hemostasis under anesthesia (1.3% Clavien grade 3a) [18], to 5% in large adenomas >80 g [12]. The hemoglobin decrease typically varies from 0.5 g/dL [17] to 1.5 g/dL [15]. Tokatli et al. found that patients who had undergone prostate biopsy before HoLEP treatment had a significant hemoglobin drop (*p* = 0.002) regardless of the type of laser device used [13]. The excellent coagulation effect obtained with LP HoLEP was confirmed in the randomized controlled trial by Elshal et al.; the authors found no statistically significant difference between LP and HP HoLEP in median perioperative hemoglobin deficit (0.9 vs. 0.7, *p* = 0.6), blood transfusion rate (0% vs. 0%), median hospital stay (1 day vs. 1 day, *p* = 0.052) and time to catheter removal (1 day vs. 1 day, *p* = 0.7) [14]. Occasionally, LP HoLEP devices produced results that were marginally preferable, such as shorter median catheter times (17.5 vs. 25.1 h) and recovery times (26.6 vs. 32.5 h), although statistical significance was not reached in these cases [9]. When comparing the mean catheterization time (42 h for the 20 W setting vs. 27 h for the 37.5 W setting, *p* = 0.008), Gazel et al. recorded a significant improvement with 37.5 W, whereas no significant difference was found in terms of mean hospitalization time (28 vs. 33 h, *p* = 0.16) [17]. Furthermore, in two prospective and one retrospective LP case series, the median time to catheter removal was 2 [13,15] and 2.6 days [18]. The median hospital stay ranged from 2 [15] to 5.3 days [18]. A statistically significant shorter length of stay was observed in patients with a previous transperineal biopsy (1.3 vs. 3 days, *p* < 0.001) [16].

#### **4. Opinion**

Since the initial groundbreaking research [20,21], the HoLEP treatment has advanced alongside other developments in urological technology [2–4]. According to the most recent research, one pedal should deliver a high laser intensity (>80 W) throughout the entire procedure [18,22,23]. While a quick enucleation can be achieved this way, irritative symptoms frequently remain even a year later [18]. In a recent editorial, Scoffone et al. [23] reported that they retained enucleation effectiveness and efficiency while reducing the laser photothermic effect on the capsule by decreasing the power output from 50 to 20 W. Several authors backed this finding by showing that LP is just as effective as HP HoLEP [12,15,18]. Cecchetti et al.'s "in-vitro" research [24] convincingly showed how different holmium laser settings interact with shockwaves and produce temperatures on soft tissues. The researchers demonstrated that the lowest threshold for plasma bubble generation and shockwave noise for soft tissue ablation was detected at an energy level of 1.4 J and a frequency of 10 Hz. With a particular quantity of joules provided at a lower frequency and an additional longer pulse duration, thermal relaxation time is significantly increased, fewer photothermic side effects are created, and the photomechanical effects are softened while also maintaining laser effectiveness [24,25]. However, it is crucial to remember that while the frequency can be dropped somewhat proportionately, the energy should not be decreased significantly [26].

It is challenging to predict how a laser will affect a particular type of tissue because of the complex interactions between the laser (wavelength, absorption coefficient, power, and pulse), tissue (water concentration, hardness, and absorption coefficient), environment (air and liquid), and the distance and inclination angle between the fibre tip and tissue [27]. A vaporization zone (vaporization volume), incision depth, width, coagulation zone, carbonization zone, and thermo-mechanical or laser damage zone are the parameters that define laser incisions, which are created by explosive tissue water vaporization [28]. Protein denaturation and pyrolysis induce thermal coagulation to develop between 60 and 100 ◦C. The release of carbon atoms after the vaporization of water molecules causes the adjacent tissue to become carbonized [29]. Perfusion simplifies the process of transferring heat from the laser incision into healthy tissue below, reducing heat damage. However, perfused and non-perfused porcine kidneys show similar laser damage zones [30]. The type of laser used during endoscopic prostate enucleation can affect the type of incision and the power settings used [31]. Using HP lasers, faster procedure times and more significant hemostasis can be achieved, along with broader and deeper tissue incisions [32]. However, when operating near the prostate pseudo-capsule, deeper incisions, especially with a more expansive thermo-mechanical damage zone, may result in collateral damage, such as a neurovascular bundle injury [4]. In contrast, the minimal carbonization zone associated with LP lasers might reduce postoperative urge symptoms [33] and improve histological findings [28].

The most crucial distinction between LP and HP HoLEP is operational effectiveness. The primary evidence for the efficiency of LP HoLEP was provided by the single-series retrospective investigation by Minagawa et al. [18]. In this study, HoLEP procedures were carried out using an 80 W device with a 30 W power setting by surgeons with various surgical skills. HoLEP was successfully treated on every occasion, regardless of the LP setting, without increasing the laser's output, and no patient required a blood transfusion. Furthermore, the authors assessed the outcomes while considering the surgeon's level of expertise and concluded that the enucleation time was significantly reduced when an experienced surgeon carried out the HoLEP operation. Moreover, the EE results aligned with other publications that used an HP laser.

The level of surgical experience may be a significant confounder affecting the procedure results. Without utilizing a control group, a study looked at the EE of HoLEP surgery performed by two experienced surgeons using a 50 W device (2.2 J and 18 Hz) [15]. The authors found that their EE values were higher than those reached by HP laser devices after comparing their results to those of earlier HP HoLEP series. They also stressed that the surgeon's experience is more crucial than the device's power for acquiring high EE values. Elshal et al. [14] observed no statistically significant differences between the two groups for any operational parameters, including EE values, in the first randomized controlled experiment contrasting LP HoLEP vs. conventional HP HoLEP (50 W and 100 W energy

settings). When contrasting the findings of studies comparing the effects of various energy settings on the efficiency of enucleation during the HoLEP procedure, it is apparent that EE values were reported to be lower in studies conducted before 2013 (mean EE values ranged between 0.45 and 0.94 g/min) [7,34] compared with studies conducted in 2017 or later (mean EE values ranged between 1.1 and 1.7 g/min) [14,15]. These findings demonstrate the importance of gaining experience over time by showing that regardless of the device's power, the procedure's effectiveness improves as the surgeon's experience increases. This conclusion implies that starting HoLEP with large-volume prostates is not advisable for novice surgeons.

Endourologists commonly base their selections on the safety of the procedure and the potential for excessive intraoperative bleeding when selecting the laser settings for HoLEP. Using 25 W vs. 40 W power settings in the initial experiment, Rassweiler et al. [7] discovered an average hemoglobin reduction of 3.1 g/dL and an 8% transfusion rate in their patient groups. Despite the unexpectedly high results, several papers with patients receiving procedures with LP settings have reported acceptable values. In their prospective investigation, Becker et al. [15] used the 39.6 W energy setting on 50 W Ho:YAG laser equipment for HoLEP surgery and collected data on more than 50 patients. They noted a 1.9% transfusion rate and an average hemoglobin decrease of 1.5 g/dL. In a different study comparing 50 W and 100 W energy settings, the decrease in hemoglobin was 0.9 and 0.7 g/dL, respectively [18]. Results from a multiple regression analysis by Tokatli et al. [13] revealed that the sole independent predictor of hemoglobin decline was the existence of biopsy anamnesis. Some underlying factors for the increased bleeding risk include acute or chronic inflammatory reactions that cause granulation in the tissue. HoLEP and removing the adenomatous tissue from the prostatic capsule may also be more challenging to accomplish if there is an inflammatory reaction following the biopsy.

One of the problematic consequences of the HoLEP procedure is postoperative SIU, which can occur to an extent in between 2–15% of patients [35–37]. SIU is typically described as temporary, which is reassuring the patients and adds significantly to the preoperative counseling process [38]. The two leading causes of SIU are significant urethral sphincter traction during surgery and tissue damage caused by laser energy close to the prostate's apex. The likelihood of SIU can be decreased by the adenoma's low energy consumption close to the urethral sphincter. Prospectively, Becker et al. [15] found that the postoperative immediate SIU rate at 1-month follow-up was high (16.7%). However, the rate of SIU decreased to zero by the 6-month follow-up, in line with what is seen with other HP HoLEP series, TURP, and open prostatectomy [3,36,39–41].

Research teams using LP HoLEP report results equivalent to those seen when using HP settings. However, the fact that every surgeon who reports on LP HoLEP uses a different enucleation technique, adding the advantages of each to the LP settings, may be a source of bias. We could also converse concerning how to interpret each outcome measure. For example, the length of stay in the hospital is another indirect indicator that may be highly "environment-dependent" since a surgeon may be reluctant to remove the catheter too soon for fear that doing so will result in an immediate re-catheterization (e.g., hospital stay time is longer in Japan for this reason due to the insurance system). Even enucleation efficiency, which seems to be a very reliable indicator of the intraoperative outcome and reflects both the efficiency of the laser and the clarity of vision in a particular setting, may be influenced by several factors, such as the size of the adenoma (large adenomas significantly improve it, while smaller ones worsen it), so the range of adenoma volume within a case series may influence this factor. Also, the narrative structure of this work underscores the paucity of reliable evidence on this subject. The works included in this study are heterogeneous, particularly regarding the types of laser fibres and the laser and irrigation settings used. In addition, the descriptions of tissue effects during laser ablation differ between study groups in terms of definitions, units, and extra details like laser activation/deactivation intervals or laser tip/tissue distance. The critical endpoints and objectives of the included papers also vary. Further multicentric studies are needed to determine how variables deemed

necessary for enucleation, such as prostate size, surgical technique, and surgeon expertise, may alter the outcomes of problems when different holmium machines are used [42].

#### **5. Conclusions**

LP HoLEP is feasible, safe, and effective and may help lessen the frequency, severity, and duration of postoperative dysuria and storage symptoms. The laser power level does not significantly affect the intra- and postoperative variables and the complication rates. While more comparative studies are still required to confirm the efficacy of LP HoLEP with various enucleation techniques, the physical background for LP HoLEP is valid and supports its use, encouraging surgeons with access to LP machines to use this method.

**Author Contributions:** Concept and Study design: T.T.; Methods and experimental work: V.G. and T.T.; Results analysis and conclusions: V.G.; Manuscript preparation: V.G., B.K.S. and T.T. 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:** Not applicable.

**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* **Radiomics in Urolithiasis: Systematic Review of Current Applications, Limitations, and Future Directions**

**Ee Jean Lim 1,\*, Daniele Castellani 2, Wei Zheng So 3, Khi Yung Fong 3, Jing Qiu Li 1, Ho Yee Tiong 4, Nariman Gadzhiev 5, Chin Tiong Heng 6, Jeremy Yuen-Chun Teoh 7, Nithesh Naik 8, Khurshid Ghani 9, Kemal Sarica 10, Jean De La Rosette 11, Bhaskar Somani <sup>12</sup> and Vineet Gauhar <sup>6</sup>**


**Abstract:** Radiomics is increasingly applied to the diagnosis, management, and outcome prediction of various urological conditions. Urolithiasis is a common benign condition with a high incidence and recurrence rate. The purpose of this scoping review is to evaluate the current evidence of the application of radiomics in urolithiasis, especially its utility in diagnostics and therapeutics. An electronic literature search on radiomics in the setting of urolithiasis was conducted on PubMed, EMBASE, and Scopus from inception to 21 March 2022. A total of 7 studies were included. Radiomics has been successfully applied in the field of urolithiasis to differentiate phleboliths from calculi and classify stone types and composition pre-operatively. More importantly, it has also been utilized to predict outcomes and complications after endourological procedures. Although radiomics in urolithiasis is still in its infancy, it has the potential for large-scale implementation. Its greatest potential lies in the correlation with conventional established diagnostic and therapeutic factors.

**Keywords:** radiomics; urolithiasis; therapeutic applications

#### **1. Introduction**

The exponential growth of medical digitalization and data acquisition has led to the healthcare sector embracing artificial intelligence (AI) to manage and optimize data accruement and utilization [1]. The scope of analysis has correspondingly broadened and introduced a new scientific field collectively called "omics" [2]. The branches of science known informally as omics refers to a field of study in biological sciences that ends with -omics, such as genomics, transcriptomics, proteomics, or metabolomics. The application of AI capabilities within the context of medical imaging is known as radiomics. Radiomics is a quantitative method that primarily extracts extensive amounts of mineable data from medical imaging and radiographic images [3]. These features are subsequently input into statistical frameworks and evaluated. It quantifies textural information by using analysis

**Citation:** Lim, E.J.; Castellani, D.; So, W.Z.; Fong, K.Y.; Li, J.Q.; Tiong, H.Y.; Gadzhiev, N.; Heng, C.T.; Teoh, J.Y.-C.; Naik, N.; et al. Radiomics in Urolithiasis: Systematic Review of Current Applications, Limitations, and Future Directions. *J. Clin. Med.* **2022**, *11*, 5151. https://doi.org/ 10.3390/jcm11175151

Academic Editor: Kent Doi

Received: 13 August 2022 Accepted: 29 August 2022 Published: 31 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/).

methods from the field of artificial intelligence to analyse "big data" [4]. Big data is defined as "a term that describes large volumes of high velocity, complex and variable data that require advanced techniques and technologies to enable the capture, storage, distribution, management, and analysis of the information" [5]. AI is used for mathematical extraction of the spatial distribution of signal intensities and pixel interrelationships and quantifies textural information which is otherwise imperceptible to humans [6,7]. Radiomics aims at improving precision medicine by using AI to improve diagnostic and prognostic information [8,9]. It surpasses the human ability to identify key imaging characteristics imperceptible to the naked human eye, picking up hidden objective data that may influence subsequent treatment decisions [10].

Data-characterization algorithms such as machine learning (ML), deep learning (DL), and artificial neural networks (ANNs) have already been incorporated to generate radiomicsguided learning models that guide diagnosis, stratification, and treatment [11,12]. In recent years, radiomics has been increasingly applied to the diagnosis, management, and outcome prediction of several medical and urological conditions.

First utilized in oncology, radiomics has been successfully investigated to differentiate benign renal mass from malignancy and predict histopathology, survival, and outcome of various urologic cancers [13]. Radiomics aims to analyse and translate medical images into quantitative data and provide an image-based biomarker to aid clinical decisions and improve precision medicine [14]. Success in the oncologic field has drawn attention to the application of radiomics in benign urologic conditions, especially urolithiasis. An illustration of the workflow of radiomics in kidney stone disease is presented in Figure 1: (1) Image acquisition and pre-processing, (2) Validation and training dataset creation, (3) Extraction and feature segmentation, and (4) Model building, e.g., kidney stone analysis. Figure 2 illustrates the utility of radiomics when applied specifically to patients with kidney stone disease, with four potential key areas: (1) Diagnosis and prediction of pathological features in patients with kidney stone disease, (2) Risk stratification and prognosis of stone forming patients, (3) Categorisation and molecular profiling of high-risk stone formers; and (4) Implementation of personalized medicine in kidney stone formers.

**Figure 1.** Potential contributions of radiomics and radiogenomics to the management of a patient with urolithiasis.

**Figure 2.** Radiomics approach in the treatment of patients with kidney stone disease.

The aim of the scoping review is to evaluate if radiomics-based applications can help endourologist overcome some confounders in stone management such as preoperative identification of stone composition, identifying phleboliths, and predicting stone free rate after medical expulsion therapy.

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

#### *2.1. Literature Search*

A literature review of the usage of radiomics in the setting of urolithiasis was performed using the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) framework for scoping reviews and metanalysis guidelines. An electronic literature search was conducted on PubMed, EMBASE, and Scopus from inception to 21 March 2022 without language restrictions (Appendix A) [15]. The full search strategies are outlined in Appendix B. Abstracts and full texts retrieved were reviewed by two independent investigators; conflicts were resolved by a third author. The inclusion criteria were: (1) Use of radiomics in diagnostics, treatment prediction, or therapeutics; (2) Any type of urolithiasis, including nephrolithiasis, ureterolithiasis, and cystolithiasis. Case reports, abstracts, and reviews were excluded from the analysis. The data were extracted using a standardized data collection template with predefined data fields including study characteristics, objective of radiomics, study findings, and study conclusions.

#### *2.2. Study Selection*

The search strategy retrieved 1332 studies; after removal of 322 duplicates, the remaining 1010 studies were screened by title and abstract. Of the 108 studies shortlisted for full-text screening, 7 were eventually included in this review.

#### **3. Results**

The potential domains in which radiomics can contribute significantly are diagnostics, therapeutic, and interventional outcomes. A total of seven studies were included; one study

by Perrot et al. [16] examined the use of radiomics in differentiating between phleboliths and calculi. Four studies reviewed identified stone type and composition, with two studies that looked at interventional outcomes [17,18]. Table 1 shows the included studies.

**Table 1.** Summary of included studies.



**Table 1.** *Cont.*

NR: Not reported; AUC: Area under the curve; CT: Computed Tomography; PCNL: Percutaneous nephrolithotomy; RIRS: Retrograde intrarenal surgery.

#### **4. Discussion**

#### *4.1. Diagnostics*

#### 4.1.1. Differentiating Ureteric Calculi and Phleboliths

First described in the 19th century, phleboliths generally present as layers of calcified fibrous tissue covered by a layer of endothelium which is continuous with the intimal layer of vein wall [19]. Differentiating characteristics include a central lucency, comet tail sign, and anatomical distribution [20]. Although advances in radiology have improved the landscape of differentiating phleboliths from ureteral calculi, they still present a diagnostic challenge particularly in the emergency setting, leading to unnecessary intervention and associated financial and resource burdens. Perrot et al. [16] sought to utilize the capabilities of radiomics to improve the use of low-dose unenhanced computed tomography (LDCT) in distinguishing renal calculi from pelvic phleboliths. The study involved independent training (369 patients, 211 kidney stones, and 201 phleboliths) and a testing cohort (43 patients, 24 kidney stones, and 23 phleboliths) for training and experimentation of the machine-learning classifier, respectively. Both patient groups presented with acute renal colic and subsequently underwent LDCT for radiological assessment. A total of 147,029 radiomics features (first-order, shape, gray level co-occurrence matrix (GLCM), gray level size zone matrix (GLSZM), gray level run length matrix (GLRLM), neighboring gray-tone difference matrix (NGTDM), and gray level dependence matrix (GLDM)) extracted from LDCT images were used for prediction by the model, demonstrating an overall accuracy of 85.1%, 91.7% sensitivity, and 78.3% specificity with a ROC-AUC value of 0.902. This radiomics-reinforced machine-learning algorithm proves itself to be a highly objective method for discerning renal calculi and might be helpful in limiting unnecessary interventions.

#### 4.1.2. Pre-Operative Identification of Stone Type

Radiomics features have also been employed to guide the detection of calculi material, largely within the pre-operative context to guide downstream management. Cui et al. [21] developed a radiomics signature created with ensemble learning based on bagged trees and applied it to non-contrast CT images of 157 patients diagnosed with either infection stone (98 patients) or non-infection stone (59 patients). With the least absolute shrinkage and selection operator (LASSO) algorithm, 27 radiomics features with the highest predictabilities were selected. The model reported 90.7% accuracy, 85.8% sensitivity, and 94.0% specificity with a ROC value of 0.97 in determining the presence of infection kidney stones. In the same vein, Zheng et al. [22] established a radiomics-signature incorporated radiomics model after extraction of data from CT images of 1198 urolithiasis patients, with 24 best radiomics features finalized by LASSO from 1316 radiomics features. AUC values of 0.898 (95% CI 0.840–0.956), 0.832 (95% CI 0.742–0.923), 0.825 (95% CI 0.783–0.866), and 0.812 (95% CI 0.710–0.914) were attained with the model on training and validation cohorts. The model also performed significantly better (*p* < 0.001) than urine pH, urine white blood cell count, urine nitrite, and presence of urease-producing bacteria in determining the existence of infection renal stones.

Tang et al. [23] specifically looked at the prediction of the occurrence of calcium oxalate monohydrate (COM) stones, the most prevalent stone type in routine practice. A total of 1218 radiomics features were extracted from 337 COM and 107 non-COM calculi seen on pre-operative non-contrast CT images, and 8 with non-zero coefficients were selected for the model by LASSO. Incorporation into the AI model revealed an accuracy, sensitivity, and specificity of 88.5%, 90.5%, and 84.3%, respectively, with an AUC value of 0.935 (95% CI 0.907–0.962) in the training cohort and 0.933 (95% CI 0.893–0.973) in the testing set for preoperative prediction of COM vs. non-COM stones. Hameed et al. [24] applied deep learning convolutional neural network (DLCNN) guided by radiomics features demonstrating 87% accuracy of prediction of calculi type. Specificity of each type of calculi was 89% for COM stones, 85% for calcium oxalate dihydrate stones, 86% for struvite stones, 93% for uric acid stones, and 89% for calcium hydrogen phosphate stones. However, despite improvements with added anatomical location and the ability to aid in differentiating between pelvic phleboliths and ureteric calculi, there are still sizable inaccuracies if artificial intelligence is used alone. Like AI, radiomics too can efficiently process vast quantities of data. With the shift towards electronic patient records, increasingly more big data sets are created and this will allow AI and radiomics to analyse and detect novel diagnostic and treatment patterns in the future [25].

Summary: With increasing application and accuracy of radiomics in differentiating phleboliths from true calculi and stone type, this can potentially influence the choice of treatment modality and limit unnecessary surgical intervention with its associated financial burden and morbidity.

#### *4.2. Evaluating Treatment Outcomes*

Radiomics has been also applied in the field of urolithiasis to predict the complications and outcomes of endourological procedures. We review their role in the various treatment modalities in predicting treatment efficacy.

#### 4.2.1. Prediction of Spontaneous Stone Passage

Radiomics has also been applied to predict spontaneous stone passage rate in symptomatic patients. Mohammadinejad et al. compared the ability of a semi-automated radiomics analysis software in predicting the likelihood of spontaneous stone passage with manual measurements. Stone characteristics including length, width, height, maximal diameter, volume, the mean and standard deviation of the Hounsfield units, and morphologic features were extracted from CT images using automated radiomics analysis software [26]. Univariate analysis and multivariate analysis showed AUC of 0.82 and 0.83, respectively, for maximum stone diameter measured manually. The AUC for a model including automatic measurement of maximum height and diameter of the stone was

0.82. Hence, the authors concluded that semi-automated radiomics analysis shows similar accuracy compared with manual measurements in predicting spontaneous stone passage.

#### 4.2.2. Therapeutic Utility in ESWL

Despite numerous AI-based platforms exploring the utility of decision algorithms in ESWL, there were no articles that focused specifically on radiomics devised applications for ESWL, proving it to be uncharted therapy. It will be interesting to continue to monitor if refined technologies such as burst wave lithotripsy will fuel renewed interest in the application of radiomics in this modality of treatment [27].

#### 4.2.3. Predicting Stone Burden Affecting RIRS/PCNL Stone-Free Rates Outcomes

The application of radiomics in endourology is relatively novel, and only two reports have been so far published, one in PCNL [17] and the second one in RIRS [18]. Homayounieh et al. analysed 202 kidney stone adult patients who underwent CT scan for evaluation of renal colic or stones in three different CT machines [2]. The purpose of this study was to assess if an automatically segmented whole renal radiomics was able to estimate the stone burden and predict hydronephrosis and treatment strategies from CT images. All stone images were evaluated by a single experienced radiologist who assessed manually the stone location and burden (stone density, stone size, and stone contours) and the presence or not of hydronephrosis for each patient. A physician expert in image processing processed all CT examinations from a standalone radiomics prototype that automatically recognized and segmented the entire kidney volume, including all stones included within the segmentation contours. After confirmation of the contours, the radiomics prototype estimated 1690 first-, shape, and higher-order radiomics for each kidney. Among the 202 patients, the radiomics prototype was able to discriminate between patients with and without renal stones (AUC 0.84, 95% CI 0.78–0.89, *p* < 0.003). Radiomics was also able to accurately detect hydronephrosis (AUC 0.89, 95% CI 0.8–0.89, *p* < 0.003). In addition, radiomics was able to predict patients managed with PCNL. Stone burden in these patients was significantly larger than those managed conservatively (641 ± 1090 vs. 53 ± 8 mm3, *p* < 0.0001). Interestingly, there was no difference in radiomics vendors performance between the three CT machines across all study outcomes. The automatic segmentation and inclusion of the entire kidney volume enabled the authors to apply radiomics not only to the stone but to the whole renal volume to obtain a consistent and generalizable prediction of stone burden and the need for PCNL treatment.

Factors such as location [28], size and volume of stone burden [29,30], and Hounsfield units (HU) [31] are key determinants and predictors of stone-free rates in both normal and anomalous kidneys alike [32,33]. Stone size limits the use of HU for the prediction of stone composition, especially calcium oxalate stones, and is a known limitation for predicting successful outcomes in ESWL and PCNL [34,35] Xun et al. retrospectively assessed 264 patients with a solitary kidney stone who underwent RIRS [18]. Among these, 142 patients had a lower calix stone. Preoperative assessment was made with an unenhanced 64-slice CT scan. Stones were manually segmented on each transverse slice CT image. Radiomics feature extraction was accomplished operating an in-house texture analysis software, including a total of 604 radiomics features (first-order statistics, shape- and size-based features, textural features, and wavelet features) generated from each original CT image. A radiomics signature was generated by a linear combination of selected features weighted by their respective coefficients, and a radiomics score (Rad-score) was estimated for each patient. Finally, the authors developed a visual nomogram incorporating clinical and radiomics parameters to predict SFR, defined as residual fragments less than 2 mm. Interestingly, radiomics score significantly differed between SFR and non-SFR patients both in the test and validation group with higher scores observed in patients with higher SFR. The prediction nomogram was very accurate (AUC 0.94, 95% CI, 0.910–0.989) and its predictive efficacy was confirmed by the validation group (AUC 0.947, 95% CI 0.883–1). The inclusion of radiomics in this model demonstrated to be an effective preoperative prediction method for clinical decision-making in patients undergoing RIRS. The main advantage of using radiomics in this context relies mostly on the speed of the procedure in a more quantitative and reproducible manner as compared with the manual assessment which can be time-consuming and prone to intra and interobserver variations, particularly for complex renal stones requiring a PCNL treatment (i.e., staghorn and multiple stones) [36].

Summary: This adds a significant research potential wherein using the radiomics signatures comparisons can be made between the efficacy of single emission CT scans (SECT) vis a vis dual emission CT scans to accurately determine stone composition [37]. This information can enable endourologists to better choose the right intervention for their patient and potentially overcome limitations and act as adjuncts of various scoring systems used as surrogate tools for predicting success in endourology interventions [38].

#### *4.3. Current Limitations and Future Directions*

One limitation of deep learning-based radiomics is the dependent correlation between the features and the input data, as the features are generated from that very dataset. Therefore, in contrast to feature-based radiomics, large datasets are necessary to accurately identify the relevant and robust feature subsets. This limitation can be partially overcome by utilizing a machine learning technique called transfer learning, by using a pre-trained neural network on a different but similarly related task, e.g., Neural data that was trained to predict renal stones can also be used and trained on how to measure and classify residual fragments after a procedure [39]. Another limitation is the reproducibility and transferability of radiomics features as it is heavily dependent on size, quality, sequence, modality, resolution, and motion artifacts of image transfer; Traverso et al. performed a recent review and identified radiomics features that were reproducible and repeatable [40]. Moving forward, the Image Biomarker Standardization Initiative (IBSI) has been established to provide standardized image biomarker nomenclature and definition, as well as to aid in formulating reporting guidelines to regulate effective communication and verification within study groups in the field of radiomics [41]. These principles when applied to the field of radiomics for urolithiasis could help standardize and refine accessibility, facilitating a widespread acceptance of the same. Xun et al. developed and validated a clinical-radiomics nomogram model for pre-operatively predicting the stone free rate of flexible ureteroscopy. They demonstrated that when applied, radiomics scores from their nomogram had satisfactory predictive accuracy in clinical application [18]. Radiomics may be used in the future to generate or validate nomograms that aid in accessing or predicting stone-free rates based on the modality of intervention.

#### *4.4. Take Home Messages*

In summary, potential applications of radiomics in urolithiasis are:


#### **5. Conclusions**

Our review shows that radiomics in urolithiasis is still in infancy. Its best potential lies in identifying infectious stones preoperatively; whether this application can extend to all stone types remains undetermined. Future applications in ESWL and predicting stone free rates for different compositions are the next frontiers for research and development. It is hoped that with further correlation of radiomics with conventional established sources of diagnostic subsets such as clinical, molecular, and imaging can optimize disease management in urolithiasis and improve patient prognosis.

**Author Contributions:** Conceptualization, E.J.L., D.C., C.T.H., B.S. and V.G.; Data curation, N.N. and K.S.; Formal analysis, E.J.L. and K.S.; Funding acquisition, H.Y.T.; Investigation, W.Z.S. and J.Y.-C.T.; Methodology, E.J.L. and K.Y.F.; Project administration, H.Y.T., N.G., J.D.L.R., B.S. and V.G.; Resources, C.T.H., K.G. and V.G.; Software, K.Y.F. and K.G.; Supervision, J.Y.-C.T., N.N. and K.S.; Visualization, J.Q.L.; Writing—original draft, E.J.L., D.C., W.Z.S. and J.Q.L.; Writing—review & editing, H.Y.T., N.G., J.D.L.R., B.S. and V.G. 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:** Not applicable.

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

#### **Appendix A**

**Figure A1.** PRISMA 2020 Flow Diagram.

#### **Appendix B**

**Table A1.** Full search phrases used for the respective databases.


Date searched: 21 March 2022. Pre-deduplication 1332. Post-deduplication 1010.

#### **References**


MDPI St. Alban-Anlage 66 4052 Basel Switzerland Tel. +41 61 683 77 34 Fax +41 61 302 89 18 www.mdpi.com

*Journal of Clinical Medicine* Editorial Office E-mail: jcm@mdpi.com www.mdpi.com/journal/jcm

Academic Open Access Publishing

www.mdpi.com ISBN 978-3-0365-7767-8