**Cartilage Repair and Regeneration: Focus on Multi-Disciplinary Strategies**

Editors

**Marta Anna Szychlinska Giuseppe Musumeci**

MDPI • Basel • Beijing • Wuhan • Barcelona • Belgrade • Manchester • Tokyo • Cluj • Tianjin

*Editors* Marta Anna Szychlinska University of Palermo Italy

Giuseppe Musumeci University of Catania Italy

*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 *Applied Sciences* (ISSN 2076-3417) (available at: https://www.mdpi.com/journal/applsci/special issues/cartilage repair regeneration).

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

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

**ISBN 978-3-0365-3939-3 (Hbk) ISBN 978-3-0365-3940-9 (PDF)**

© 2022 by the authors. Articles in this book are Open Access and distributed under the Creative Commons Attribution (CC BY) license, which allows users to download, copy and build upon published articles, as long as the author and publisher are properly credited, which ensures maximum dissemination and a wider impact of our publications.

The book as a whole is distributed by MDPI under the terms and conditions of the Creative Commons license CC BY-NC-ND.

## **Contents**


## **About the Editors**

**Marta Anna Szychlinska** entered the international doctoral course in "Basic and Applied Biomedical Sciences" and she obtained her PhD in 2017 at the University of Catania, Italy. During her PhD studies, she was visiting scientist at the AO Research Institute Davos (Switzerland) in the Musculoskeletal Regeneration Program. She was University Researcher and Assistant Professor in Human Anatomy at the Department of Biomedical and Biotechnological Sciences at University of Catania, Italy, until 2021. She is currently a University Researcher and Assistant Professor in Human Anatomy at the Department of Biomedicine, Neuroscience and Advanced Diagnostics at the University of Palermo, Italy. She is a member of the editorial boards of international journals and national scientific Councils. Her leading research line regards musculoskeletal disorder studies with particular regard to tissue engineering applied to osteochondral regeneration. She is particularly interested in development of 3D in vitro models as technological platforms for the study and screening of degenerative diseases such as osteoarthritis. In this field she is author of more than 60 peer-reviewed papers on international journals.

**Giuseppe Musumeci** carried out his research activity at the Department of Biomedical Engineering and Orthopedic Surgery, Johns Hopkins University, Baltimore (Maryland, USA), the AO Research Institute, Davos (Switzerland) and the Department of Orthopedic Surgery, Medical University of Graz (Austria). Currently, he works as Full Professor of Exercise and Sport Sciences at the University of Catania, Italy. He is also Adjunct Professor at the Temple University's College of Science and Technology, Philadelphia, Pennsylvania, USA and the Vice Director of the Faculty of Sport Sciences of the Fujan Normal University, China. Prof. Musumeci is the Director of the Research Center on Motor Activities (CRAM), the Director of the master's Program, School of Posturology and Physical Exercise Sciences, the Dean of the Exercise and Sport Sciences School, the Head of the Movement Innovation PosturaLab and the Director of the Anatomy Museum at the University of Catania. He is a member of different Italian and international association of Anatomy, Histology, Kinesiology, Sports Medicine and Exercise. Since 2002 until today, he has been a Peer Reviewer in national and international scientific projects from 15 different countries and the European Community and in several international scientific journals. Moreover, he is an Editor in different international peer-reviewed journals, Founder and Editor-in-Chief of the Journal of Functional Morphology and Kinesiology (MDPI—Q2 journal ranking) and Section Editor-in-Chief in Heliyon Clinical research (Cellpress—Q1 journal ranking). Giuseppe Musumeci research interests are on morphological, molecular, and biochemical aspects of osteoarthritis and musculoskeletal disorders and the relative effects of tissue engineering, mechanobiology, diet, ageing and physical activity.

## **Preface to "Cartilage Repair and Regeneration: Focus on Multi-Disciplinary Strategies"**

As widely demonstrated and known, adult articular cartilage exhibits a very poor self-healing capacity once injured. This is due to the complex multilayered morphological structure and an avascular, aneural, and hypocellular nature that characterize this tissue. A minimal damage or lesion may lead to cartilage tissue degeneration and osteoarthritis (OA) development, resulting in significant pain and disability.

Several approaches centered on cell-based therapies and cartilage engineering techniques have attempted to repair chondral or osteochondral defects that still remain a significant challenge in clinical practice. The failure in the use of these technics is often due to the improper mechanical properties of newly formed cartilage-like tissue, the early entrance of the lately differentiated chondrocytes in hypertrophic stage, characterized by a senescent-associated secretory-like phenotype, or the insufficient insertion into the host tissue, often characterized by the inflammatory, osteoarthritic milieu.

While several advances have been made in recent decades, the complexity and the multifactorial aspect of articular cartilage degeneration and a consequent, apparently unstoppable OA onset and progression suggest that a multidisciplinary approach will likely be optimal to address the challenge of preserving the articular cartilage in early stages and/or developing a functional cartilage replacement in advanced degenerative stages. This kind of approach is based on a combination of several disciplines, such as biomechanics and mechanobiology (bioreactors/moderate physical activity, etc.), innovative biomaterials functionalized with growth factors, exogenous enhancers and biomolecules exploiting pharmacological activities, cells from different sources (adult stem cells, chondrocytes, co-cultures), epigenetic modifications, functional foods, etc.

In the context of developing cartilage repair and regeneration strategies, the main goal of the present Special Issue was to invite original contributions, review articles, communications, and concept papers that address these challenges. The suggested focus and the goal of a multidisciplinary strategy is to realize a clinically relevant tool for cartilage repair or regeneration that is more likely to be successful, obtained by controlling both the formation of a new suitable tissue replacements and the damaged joint tissues environment on the local and systemic level.

The articles published within the present *Special Issue* deal with the innovative multi-disciplinary therapeutic approaches for musculoskeletal diseases, spacing from advanced 3D bioprinting technology to obtain a scaffold with different zonal cell densities, biphasic scaffold (ChondroMimetic) construction, passing through the comparison of different techniques for cartilage regeneration such as of mosaicplasty and matrix-assisted autologous chondrocyte transplantation (MACT) and histopathological features of osteochondral units, and ending with the considerations regarding development of bioreactors able to mimic the biomechanical load on chondrocytes in vitro, giving some interesting insights in this specific scientific field.

In the end, we would like to take this opportunity to express our most profound appreciation to the MDPI Book staff, the editorial team of Applied Sciences journal, especially Mr. Steph Ke, the assistant editor of this Special Issue, talented authors, and hardworking and professional reviewers.

## **Marta Anna Szychlinska and Giuseppe Musumeci**

*Editors*

## *Editorial* **Cartilage Repair and Regeneration: Focus on Multi-Disciplinary Strategies—Highlight on Magneto-Responsive Techniques**

**Marta Anna Szychlinska**

Department of Biomedicine, Neuroscience and Advanced Diagnostics (BiND), Human Anatomy Section, University of Palermo, 90127 Palermo, Italy; marta.sz@hotmail.it

**Abstract:** This editorial focuses on the interesting studies published within the present *Special Issue* and dealing with the innovative multi-disciplinary therapeutic approaches for musculoskeletal diseases. Moreover, it highlights the noteworthy magneto-responsive technique for a cartilage regeneration scope and reports some interesting studies and their outcomes in this specific field.

**Keywords:** tissue engineering; 3D bioprinting; ChondroMimetic; cartilage regeneration; osteochondral repair; mosaicplasty; matrix-assisted autologous chondrocyte transplantation; magneto-responsive techniques; biomechanical stimuli; multi-disciplinary approach

## **1. Introduction**

The articular cartilage represents an incredibly complex multi-layered tissue, characterized by avascular and aneural structure, which limits its regenerative properties. Once injured, cartilage leads to its progressive degeneration with severe consequences such as the onset of chronic degenerative disorders like osteoarthritis (OA). The latter determines articular pain and stiffness, until the total disability of the joint in advanced stages [1]. Until now, no therapeutic strategy exists for this complex disease and the necessity to find the optimal approach for the cartilage regeneration still represents a big challenge.

Recently, a lot is known concerning the onset and the triggering factors of OA, as well as the main events at the base of its progression. It is well known that this severe disorder represents a multifactorial, progressively degenerating pathologic event that, principally, affects the cartilage tissue, but that expands to all the tissues of the joint [2]. It appears evident that the most promising therapeutic strategy for this complex disorder, is represented by a multi-disciplinary and multi-targeted approach.

## **2. Highlights on the Studies Published in the Present Special Issue: Emerging Therapies for Osteochondral Regeneration**

The most promising approach for osteochondral repair is certainly represented by the tissue engineering, which aim is to create a cartilage and bone tissues able to replace the injured ones. This technique seems very encouraging, if it were not for the fact that the articular cartilage tissue has a multilayered complex structure, where every layer possesses its own spatial heterogeneity, different cell distribution and different mechanical properties [1]. With the advancement of the 3D bioprinting, the engineered grafts and the fabrication of the gradient scaffolds, enhanced their biomimicry and, consequently, their functionality and efficacy. In an interesting study by Dimaraki et al. [3] the authors bioprinted a scaffold with different zonal cell densities to mimic the organization of the complex three-layered articular cartilage structure. They observed a successful formation of a new cartilage-like tissue with a cell-density dependent zonal gradient. In another study by Berta et al. [4] a cell-free biphasic scaffold (ChondroMimetic) was evaluated for long-term outcomes in the treatment of osteochondral defects. The authors observed a cartilage-like repair tissue formation and clinical improvement at 7.9 years post-implantation. Zaffagnini et al. [5]

**Citation:** Szychlinska, M.A. Cartilage Repair and Regeneration: Focus on Multi-Disciplinary Strategies—Highlight on Magneto-Responsive Techniques. *Appl. Sci.* **2021**, *11*, 11092. https:// doi.org/10.3390/app112311092

Received: 3 November 2021 Accepted: 20 November 2021 Published: 23 November 2021

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

**Copyright:** © 2021 by the author. 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/).

compared the clinical outcomes of mosaicplasty and matrix-assisted autologous chondrocyte transplantation (MACT) at long-term follow-up (12 years post-surgery), concluding that both of these surgical procedures give satisfactory clinical results. Moreover, the authors suggest that MACT is the most suitable approach for the treatment of larger lesions. Indeed, many factors have to be considered when considering different strategies for the osteochondral defect treatment. The studies conducted in the field of histopathological aspects of musculoskeletal diseases, represent a pilot studies for the development of successful regenerative medicine approaches. In the study carried out by Desando et al. [6], which compared the histopathological features of osteochondral units, obtained from patients with both non-traumatic femoral head and with post-traumatic femoral head osteonecrosis. The authors reported substantial differences among them and suggested a multi-disciplinary and multi-targeted approach for osteonecrosis treatment based on its etiology.

Moreover, developing satisfactory strategies for cartilage regeneration requires deeper knowledge on biological systems. When considering the cartilage engineering strategy based on the use of mesenchymal stem cells (MSCs), the environment in which these cells are destined to promote chondrogenesis, has to be well-thought-out. The mechanical stimuli experienced by chondrocytes within the joint play a pivotal role in chondrogenesis and the development of bioreactors able to mimic the biomechanical load on cells in vitro, becomes of fundamental importance in developing new, multi-disciplinary strategies for cartilage regeneration approaches as suggested by Ravalli et al. [7].

#### **3. Magneto-Responsive Techniques for Cartilage Regeneration**

With the development of biotechnology, other promising techniques have been developed such as magnetic cell manipulation, achieved by the synergy between magnetic objects and magnetic field [8,9]. In general, there are four techniques based on the magnetic cell manipulation for the tissue engineering approaches. The first one includes the magnetic field-based guiding of the cells to the targeted site, which permits to the relatively small number of cells (i.e., magnetic mesenchymal stem cells) to accumulate at the level of the defect site and promote cartilage regeneration [10,11]. The second one is based on the enhancement of the seeding ability of cells within the scaffold, which permits the cells to migrate symmetrically and to promote the cell condensation, providing a suitable environment for cell proliferation and differentiation [12]. The third technique regards the formation of magnetic scaffolds, where the magnetic force is used to assemble the 3D structure to mimic the native tissue [13,14]. This technique based on magnetic patterning, works across a range of materials (e.g., hydrogels) and diamagnetic objects (e.g., living cells, drug delivering microspheres, etc.), characterized by differential magnetic susceptibility, with the potential to predictably position these objects in 3D materials, in response to brief magnetic field application. It confers several advantages, including remote control ability, sufficient cell density and cell adhesion enhancement, permitting one to achieve a very good grade of engineered tissue biomimicry [8,9]. Zlotnik et al. [15], demonstrated that a naturally diamagnetic objects, comprising living cells, can be predictably positioned throughout the 3D hydrogel. In this study, the magnetic susceptibility of the latter was enhanced by the addition of magnetic contrast agent (gadodiamide). After the cells achieved the required position, by the brief exposure to magnetic field, they were 'locked in' by a photo-crosslinking method. Afterwards, the magnetic contrast agent was washed out of the hydrogel to not interfere with the long-term cell viability. In the study, the authors applied this method to engineer cartilage constructs with a depth-dependent cellularity, mirroring that of the native tissue. The fourth strategy is based on guiding cell assembly into sheet-like structures to stack layer-by-layer, used for the formation of scaffold-free 3D cell culture. In this way, magnetic-labeled cells can be guided to a targeted location and form 3D arrangements in a convenient microenvironment to mimic tissue properties without the use of scaffolds [16,17].

In addition, it has been widely demonstrated that mechanical forces acting as an additional tool to mimic the in vivo environment are also applied to improve cartilage reconstruction as suggested above [7,18]. Magnetic nanoparticles represent the excellent candidates to apply remote magnetic-induced mechanical stimulation. Luciani et al. [19] used magnetic MSCs to enhance their seeding density and condensation into the scaffolds subjected to dynamic bioreactor. The results showed that MSC differentiation was markedly improved. Son et al. [20] exposed magnetic nanoparticle-labeled MSCs to static magnetic field and magnet-derived shear stress, demonstrating higher chondrogenic differentiation efficiency and no hypertrophic effects. Further, Hou et al., [21] demonstrated that the multifunctional hyaluronic acid-graft-amphiphilic gelatin microcapsules, loaded with the superparamagnetic iron oxide nanoparticles and chondrocytes, subjected to static magnetic field and magnet-derived shear stress, were able to stimulate chondrogenesis and fabricate cartilage tissue-mimetic pellets.

#### **4. Conclusions**

Several approaches have been evaluated for the cartilage regenerative outcomes including 3D bioprinting, cell-free biphasic scaffolds, mosaicplasty, MACT, and stem cell therapy [3–5,19,22]. Many of them giving encouraging results. However, the innovative multi-functional approaches in this field are still needed to overcome the existing limits. The most promising strategy for the cartilage regeneration seems to be represented by a multi-disciplinary approach based on tissue engineering combining innovative techniques such as formation of magneto-guided zonal cell gradient 3D structures to mimic the native tissue, application of biomechanical stimulation to reproduce the native environment of the joints, and the use of exogenous biomolecules (i.e., drug delivery scaffolds) able to stimulate cell differentiation and counteract the pathologic milieu of the affected joints.

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

**Acknowledgments:** This publication was only possible with the valuable contributions from the authors, reviewers, and the editorial team of *Applied Sciences*.

**Conflicts of Interest:** The author declares to be a guest editor of the present Special Issue and there are no other conflict of interest.

#### **References**


## *Article* **Bioprinting of a Zonal-Specific Cell Density Scaffold: A Biomimetic Approach for Cartilage Tissue Engineering**

**Angeliki Dimaraki 1,†, Pedro J. Díaz-Payno 1,2,†, Michelle Minneboo 1, Mahdiyeh Nouri-Goushki 1, Maryam Hosseini 1,3, Nicole Kops 2, Roberto Narcisi 2, Mohammad J. Mirzaali 1, Gerjo J. V. M. van Osch 1,2,4, Lidy E. Fratila-Apachitei <sup>1</sup> and Amir A. Zadpoor 1,\***


**Abstract:** The treatment of articular cartilage defects remains a significant clinical challenge. This is partially due to current tissue engineering strategies failing to recapitulate native organization. Articular cartilage is a graded tissue with three layers exhibiting different cell densities: the superficial zone having the highest density and the deep zone having the lowest density. However, the introduction of cell gradients for cartilage tissue engineering, which could promote a more biomimetic environment, has not been widely explored. Here, we aimed to bioprint a scaffold with different zonal cell densities to mimic the organization of articular cartilage. The scaffold was bioprinted using an alginate-based bioink containing human articular chondrocytes. The scaffold design included three cell densities, one per zone: 20 <sup>×</sup> <sup>10</sup><sup>6</sup> (superficial), 10 <sup>×</sup> <sup>10</sup><sup>6</sup> (middle), and 5 <sup>×</sup> <sup>10</sup><sup>6</sup> (deep) cells/mL. The scaffold was cultured in a chondrogenic medium for 25 days and analyzed by live/dead assay and histology. The live/dead analysis showed the ability to generate a zonal cell density with high viability. Histological analysis revealed a smooth transition between the zones in terms of cell distribution and a higher sulphated glycosaminoglycan deposition in the highest cell density zone. These findings pave the way toward bioprinting complex zonal cartilage scaffolds as single units, thereby advancing the translation of cartilage tissue engineering into clinical practice.

**Keywords:** bioprinting; biofabrication; tissue engineering; articular cartilage; human chondrocytes; cell density; cell gradient

## **1. Introduction**

Articular cartilage (AC) may be damaged due to aging, disease, or trauma. AC has a limited regenerative capacity that has been attributed to the lack of innervation and the avascular nature of the tissue [1]. AC defects may progress to the subchondral bone to create an osteochondral defect. If left untreated, such defects can potentially lead to the development of osteoarthritis [2,3], a cartilage-degenerating disease affecting over 300 million worldwide and representing a burden on the healthcare systems globally [4,5]. The treatment of AC defects often results in a fibrocartilage tissue with poor long-term outcomes [6], thus presenting a significant clinical challenge. This is partially because the current regenerative and tissue engineering strategies fail to recapitulate the native

**Citation:** Dimaraki, A.; Díaz-Payno, P.J.; Minneboo, M.; Nouri-Goushki, M.; Hosseini, M.; Kops, N.; Narcisi, R.; Mirzaali, M.J.; van Osch, G.J.V.M.; Fratila-Apachitei, L.E.; et al. Bioprinting of a Zonal-Specific Cell Density Scaffold: A Biomimetic Approach for Cartilage Tissue Engineering. *Appl. Sci.* **2021**, *11*, 7821. https://doi.org/10.3390/app11177821

Academic Editor: Rossella Bedini

Received: 20 July 2021 Accepted: 20 August 2021 Published: 25 August 2021

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

**Copyright:** © 2021 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/).

organization of the AC tissue. AC is a graded tissue that has three main zones: superficial, middle, and deep. These zones are associated with differences in matrix composition, matrix structural organization, and cell number, which result in gradients with complex physical, mechanical, and biological properties [7]. In adult healthy human articular cartilage, the superficial zone is characterized by thickly packed collagen type II fibers, a high tensile strength, and a high cell density, while the deep zone is characterized by a high sulfated glycosaminoglycan (sGAG) content, high compressive strength, and a low cell density as compared to the other layers [8].

Different biofabrication strategies, such as bioprinting, have been exploited to recapitulate few of the gradients connecting the three zones [9]. Graded scaffolds aim to mimic the highly organized native tissue. Several studies have explored the fabrication of multilayered constructs exhibiting differences in biological [10] or chemical composition [11–14] and physical or mechanical properties [15–17]. Different gradient strategies have also been combined to tailor physical (stiffness) and biochemical (growth factor) gradients that better control stem-cell behavior [18].

To the best of our knowledge, there has been only one study introducing a cell density gradient in a bioprinted construct [19], where swine articular cartilage-derived constructs with different density gradients of rabbit chondrocytes were compared to constructs with equivalent homogeneous cell distribution. In the study, gene expression and sGAG deposition were quantified for the overall constructs. However, limited information was presented regarding the extracellular matrix (ECM) that was deposited by the cells at different densities in each zone. Therefore, further research is warranted to characterize the cell-derived tissue deposition in each of the zones within a graded construct, especially using human material.

The aims of this paper were (i) to design and fabricate scaffolds with three-zone cell density using bioprinting with an alginate-based bioink containing human articular chondrocytes and a poly-ε-caprolactone (PCL) support structure, (ii) to characterize the cell viability and cell gradient stability overtime, and (iii) to evaluate the time-dependent deposition of ECM by the cells embedded in the scaffolds. To achieve the cell density gradient, the design of the scaffolds included three zones with three different cell densities. The different zones aimed to mimic both the dimensions and cell density of the superficial, middle, and deep zones as seen in the native structure of articular cartilage.

## **2. Materials and Methods**

#### *2.1. Scaffold Design*

Solidworks was used to design a standard tessellation language (.STL) file of scaffold solid form that incorporated both the PCL outer skeleton and the cell-embedded bioink part. The PCL skeleton design was composed of a square base of 8 × 8 mm<sup>2</sup> with a thickness of two layers (0.4 mm) and four vertical pillars at the corners with a height of 3 mm. The scaffold was designed as a 7.2 mm × 7.2 mm × 3 mm cube. The designed STL file was then processed in Slic3r [20] to generate the gcode file used by the bioprinter (BIO-X bioprinter, Cellink, Sweden). Briefly, the parameters used in Slic3r were as follows: infill pattern, rectilinear; infill density, 100% for PCL frame design and 10% for scaffold design; infill angle, 90◦; solid infill threshold area, 10; layer height, 0.2 mm; nozzle diameter, 0.2 mm. Two types of scaffolds were designed: a scaffold corresponding to a homogeneous cell density with one zone of 15 layers and a scaffold with three different zones corresponding to the three different cell densities. The dimensions of the zones were chosen to mimic the dimensions of the three different zones in the human articular cartilage. The top zone had a thickness of three layers (0.6 mm), the middle zone had a thickness of seven layers (1.4 mm), and the bottom zone had a thickness of five layers (1 mm), resulting in a total of 15 layers per scaffold.

## *2.2. Bioink Formulation*

An alginate-based biomaterial (IK1020000303, Cellink Bioink, Cellink, Sweden) was mixed with human chondrocytes according to the manufacturer's instructions to create the bioink. Briefly, human articular chondrocytes isolated from hyaline cartilage in the knee and expanded to the second passage were purchased from Lonza (NHAC-kn, Lonza Bioscience, Breda, The Netherlands). The cells were expanded until the third passage in chondrocyte growth medium (CC-3216, Lonza Bioscience, Breda, The Netherlands) according to the manufacturer's instructions, in an incubator at 37 ◦C, 5% CO2, and 90% humidity. The cells were trypsinized when confluency was reached, counted, and re-suspended in a growth medium. The hydrogel was taken up from the stock cartridge by a syringe and was mixed gently 10:1 with the cell suspension using a sterile female–female luer lock, and a second syringe was used to homogenize the cells in the bioink. The mixture was transferred to a new cartridge for bioprinting. According to the study by Hunziker et al., the total cell density within the human articular cartilage of a medial femoral condyle is <sup>10</sup> × 103 cell/mm<sup>3</sup> [21]. Therefore, the homogeneous scaffold was given a biomimetic total cell density of 10 × <sup>10</sup><sup>6</sup> cell/mL. Hunziker et al. also described the different cell densities in the different zones of articular cartilage in which it was established that the cell densities corresponding to the superficial, middle, and deeps zones are 24 ± <sup>3</sup> × 103 cells/mm3, <sup>10</sup> ± 0.5 × 103 cells/mm3, and 7 ± 0.5 × 103 cells/mm3, respectively [21]. The gradient scaffold was given a similar biomimetic cell density corresponding to each zone of the articular cartilage: superficial (top), middle (middle) and deep (bottom) zones. The top, middle, and bottom zones were designed to respectively have 20 × 106 cells/mL, <sup>10</sup> × <sup>10</sup><sup>6</sup> cells/mL, and 5 × <sup>10</sup><sup>6</sup> cells/mL, respectively (Figure 1).

**Figure 1.** A schematic representation of the fabrication methodology and the in vitro setup of the study. A PCL-reinforced alginate-based scaffold containing human chondrocytes was 3D bioprinted with different cell densities mimicking the cell gradient of the human articular cartilage zones. The scaffold was ionically crosslinked post bioprinting before supplementing with chondrogenic medium for 4 weeks. After in vitro culture, the samples were analyzed for cell viability and for matrix deposition through histological staining.

#### *2.3. Bioprinting*

The BIO-X bioprinter (Cellink, Sweden) was UV-sterilized and used inside a cell culture hood to ensure that the bioprinting process would run under sterile conditions. After sterilization, the gcode file was used to print the scaffold. Briefly, the printing surface, nozzle size, temperature, pressure, and speed of each print-head was selected through the bioprinter's interface. The selected parameters for each print-head are presented below (Table 1). Next, the three printheads were manually calibrated to the same spot in the printing surface to ensure a complete match when swapping printheads during the printing process for the different bioinks and biomaterials printed (Video S1). A 10 cm petri dish was used as the printing surface to provide better visibility between the nozzle tip and the printing surface during the calibration phase. The scaffolds were printed on 16 mm coverslips which were placed radially into the 10 cm petri dish. After printing, the scaffolds were treated with 1 mL of crosslinking solution (90 mM CaCl2, Cellink, Sweden) for 2 min.


**Table 1.** The parameters used for each printhead during bioprinting. Printheads 1 and 2 were used for the hydrogel layers, while printhead 3 was used to print the outer frame of PCL (RT: room temperature).

#### *2.4. In Vitro Culture*

The bioprinted constructs were cultured in chondrocyte differentiation basal medium (CloneticsTM CDMTM BulletKitTM, Lonza, Delft, The Netherlands) containing FBS, insulin, R3-IGF-1, gentamicin/amphotericin B, transforming growth factor-β1, and transferrin (undisclosed concentrations by Lonza). The differentiation medium was further enriched with 10 ng/mL fresh transforming growth factor-β3 (SRP3171, Sigma, Delft, The Netherlands) and 70 mM L-ascorbic acid 2-phosphate (A8960, Sigma, The Netherlands) for each medium change, according to the manufacturer's instructions (TS-CC-112-7 02/20, Lonza). The differentiation medium was changed three times a week for 25 days using 2 mL of medium per well (each well contained one scaffold), which was kept in a 24-well plate in an incubator at 37 ◦C, 90% humidity, and 5% CO2. The experiment included three scaffolds per condition (graded, homogeneous) per timepoint (day 0 and day 25).

### *2.5. Mechanical Characterization*

The 3D printed scaffolds (cell-free) were mechanically characterized using a uniaxial unconfined compression test using the LLOYD Instruments LR5k compression machine (AMETEK test & calibration instruments). The PCL frame, hydrogel, and the combined scaffolds (PCL and hydrogel) were tested separately using a 100 N load cell with a 0.1 N preload, a 1 mm deflection, and a strain rate of 0.002 s−<sup>1</sup> (i.e., a crosshead speed of 0.36 mm/min). From the load–deflection curve generated by the machine on the Nexygen software, 200 data points were exported per test for further analysis. Each of the data points included the recorded time (s), load (N), crosshead travel (mm), and deflection from the preload (mm), which were used for the generation of the stress–strain curves and the calculation of the compressive stiffness of the scaffolds. The stress was calculated by dividing the compression force by the cross-section area, and the strain was defined as the ratio of the crosshead travel to the initial length of the specimens. The stiffness calculations were performed using a moving regression algorithm generated with Gnu R [22] that was used to calculate the linear line with the steepest slope fit of the stress–strain curve. The slope of the linear curve was taken as the value for the compressive stiffness of the scaffolds (E = σ/ε, MPa).

#### *2.6. Live/Dead Assay*

Cell viability was assessed at day 0 (post-printing) and day 25 after culture using live/dead staining (LIVE/DEAD® Viability/Cytotoxicity Kit, ThermoFisher, Delft, The Netherlands). Briefly, the samples were washed twice with 1 × PBS for 5 min before supplementing the scaffolds with 2 mM ethidium homodimer-1 (red, for dead cells) and 5 mM calcein-AM (green, for live cells) in 1 × PBS. The samples were allowed to incubate for 1 h at 37 ◦C before being washed twice in 1 × PBS and being imaged under a fluorescent microscope (ZOE fluorescent cell imager, Biorad, Delft, The Netherlands).

#### *2.7. Histology Staining*

All the specimens were fixed overnight in 4% paraformaldehyde at 4 ◦C with a tissue– fixative volume ratio of 1:20. Next, the scaffolds were washed twice with PBS, and were placed in a 1:1 solution of 100% EtOH:90 mM Cellink crosslinking agent (final 45 mM CaCl2) for storage. The samples were paraffin-embedded and were sectioned at a thickness

of 6 μm. The sections were stained with hematoxylin (HHS32, Sigma, The Netherlands) and eosin (HT110232, Sigma, The Netherlands) to examine their cell distribution, with 1% Alcian Blue (TMS-010-C, Sigma, The Netherlands) to analyze their sGAG content, and with Picrosirius Red (365548, Sigma, The Netherlands) for collagen deposition. The stained histological slices were imaged under a DM500 optical Leica microscope.

#### *2.8. Image Analysis*

The histological images were analyzed for the cell density in each of the three zones, as well as for the neo-tissue formation (sGAG and collagen deposition) within the hydrogel for all three zones. The analysis of the data was performed using Image J v2.0.0 software. The images were spatially calibrated using their integrated scale bar. For the analysis of H&E staining, the microscopic images were first split into the three-color channels RGB. Then, the green channel was selected as the optimum channel for the identification of the nuclei. Next, the threshold was adjusted before using the "analyze particles" plugin. The minimum size was set between five and 35 pixels depending on the image, while the maximum size was set at 300 pixels. The circularity range was set between 0.45 and 1.00 for all the images. This analysis of the H&E images was performed using the ROI (region of interest) manager to select three regions of interest corresponding to the three different zones and to analyze them separately using a different threshold and size spectrum depending on their specific data. For the samples stained with Alcian Blue and Picrosirius Red, the image analysis was performed with the use of the Trainable Weka Segmentation plugin [23]. This plugin utilizes a segmentation algorithm that combines a collection of machine learning algorithms with a set of selected image features to produce pixel-based segmentations. Three classes were used for the training of the algorithm and the classification of the results: (a) a class for the negative background staining of the alginate, (b) a class for the stained cell nuclei, and (c) a class for the positively stained areas (areas with darker blue or red color for AB and PR, respectively). The training of the algorithm was performed by manually selecting multiple areas belonging to each of the classes using the Image J's selection tool and categorizing them according to their corresponding class. After that, the features of the input image were extracted and converted to a set of vectors of float values (format for Weka classifiers). Finally, the plugin created and displayed the resulting image. This image was equivalent to the current overlay (8 bit color with each color corresponding to a specific class). The training features used for the training of the algorithm were Gaussian blur, Sobel filter, Hessian matrix, and the difference of the Gaussians. After training, the same classifier was applied to all the images of the same staining. The plugin performed the image segmentation on the basis of the current classifier and the selected features and created a stack of images, each one highlighting one of the selected features. By selecting the desired 8 bit color image from the stack and setting an appropriate threshold, we performed area measurements on the objects within the images to evaluate the sGAG content of the areas positively stained for collagen (for all the zones and timepoints). The ratio between positive area and total area for the samples analyzed is presented as a percentage.

#### *2.9. Statistical Analysis*

All quantitative results are presented as the mean ± standard deviation. Statistical analysis was performed using GraphPad Prism 8 (GraphPad Software, San Diego, CA, USA, Version 8.0.2). For the analysis of the cell density, sGAG content, and collagen content, the experimental groups were analyzed for significant differences using a two-way analysis of variance (ANOVA) and the results were corrected for multiple comparisons using Bonferroni's post hoc test. For the comparison of the stiffness, as well as for the differences in the cell viability and cell density between the two different scaffold designs, an unpaired *t*-test or one-way ANOVA was performed. Probability *p*-values <0.05 were considered statistically significant.

## **3. Results**

#### *3.1. PCL-Reinforced Alginate Scaffolds with Different Cell Density Zones Can Be Successfully Fabricated as Single Units Using Bioprinting*

Firstly, we designed a structure that could combine both an outer frame of stiff PCL and the soft alginate-based bioink with an overall size of 8 mm × 8 mm × 3 mm (Figure 2a–c). For the PCL frame, two different designs were tested: a closed design (Figure S1a) and an open design (Figure 2a and Figure S1b). The open design resulted in a higher viability of the cells in the bioink (Figure S1c–e) and was, therefore, chosen for further experiments. For the bioink, a 10% infill density was selected in order to create channels in the *z*-direction (Figure 2e) that resulted in visible pores in the scaffold of ~0.230 mm<sup>2</sup> (Figure 2c) to allow for sufficient nutrient diffusion to all the layers of the cell-laden hydrogel. The different parts of the zonal scaffold showing the PCL frame in yellow and the three different cell density zones in red were sliced into a printing pattern suitable for 3D printing (Figure 2d). The design used for the fabrication of the biomimetic cartilage scaffolds was bioprinted monolithically as a single unit (Figure 2e). The mean compressive stiffness of the scaffolds (PCL + hydrogel) was 8.35 ± 0.43 MPa. This was mostly attributed to the PCL framework, since the mean compressive stiffness of the PCL framework alone was 8.02 ± 0.69 MPa while the hydrogel alone was 0.23 MPa ± 0.01 (Figure 2f).

The next step was to verify that we could 3D print the different zones (top, middle, and bottom) with different cell densities of human chondrocytes (i.e., 20 × <sup>10</sup>6, 10 × <sup>10</sup>6, and 5 × 106 cells/mL, respectively), recapitulating some aspects of the cytocomplexity of the human hyaline articular cartilage. Live/dead staining at day 0 post bioprinting demonstrated that it was possible to control such cell distribution, as evidenced by a higher cell density in the top zone and the lowest cell density in the bottom zone (Figure 2g). Overall, a high viability (>90%) of the bioprinted cells was observed throughout the different zones of the scaffolds (Figure 2h).

#### *3.2. Cell Density Can Be Maintained in the Different Zones Overtime In Vitro*

To investigate the maintenance of the zonal distribution of the cells over time, we cultured human chondrocytes in the hydrogel for 25 days. We compared the scaffolds with different cell densities, herein called the zonal scaffolds, with the scaffolds in which the cell density (10E6 cells/mL) was constant throughout the entire scaffold. At day 0, right after the bioprinting process took place, a high cell viability was detected for the embedded cells (Figure 3a). In addition, the quantification of the H&E staining at day 0 showed significant differences between the bottom/middle, middle/top, and bottom/top zones of the graded scaffolds (Figure 3b, left), whereas no difference in the cell density was observed throughout the homogeneous scaffolds (Figure 3b, right). After 25 days of culture, cell viability remained high for all the specimens (Figure 3c). In addition, a decrease in the cell density of the middle zone was observed, rendering the cell density between the middle and the bottom zones as not significantly different, while the cell density in the top zone remained significantly higher than the other zones (Figure 3d, left). The cell density of the homogeneous scaffolds remained constant throughout the scaffold thickness for all the timepoints (Figure 3d, right).

**Figure 2.** PCL-reinforced alginate scaffolds with different cell density zones can be bioprinted as a single unit. (**a**) The solid structure created in Solidworks corresponding to the outer PCL frame; (**b**) the solid structure corresponding to the hydrogel part; (**c**) a macroscopic image of the top view of a 3D printed scaffold after ionic crosslinking; (**d**) the different solid structures combined by Slic3r software, showing the PCL in yellow and the hydrogel in different shades of red for the different cell density zones; (**e**) Slic3r design of the printing pattern showing a 100% infill for the PCL and a 10% infill for the hydrogel; (**f**) a bar chart showing the mechanical properties (stiffness) of the outer PCL frame, the combined PCL frame + hydrogel, and the hydrogel alone printed scaffolds (*n* = 3); (**g**) a fluorescent microscopic image of cell viability with live/dead staining (green: live; red: dead) showing the cell distribution in the scaffold bioprinted with three different cell densities for each of the zones: bottom (5E6 cell/mL), middle (10E6 cells/mL), and top (20E cells/mL) at day 0. Lines are drawn on the image to illustrate the approximate division of the zones. Scale bar = 200 μm; (**h**) the corresponding quantification of the overall live and dead cells (*n* = 4).

**Figure 3.** Gradient cell density can be maintained in the zonal scaffold over time in in vitro culture; (**a**) fluorescent microscopic images indicate cell viability based on live/dead staining (green: alive; red: dead) and bright-field microscopy images of the H&E staining (matrix: violet; nuclei: dark purple) showing the cell distribution for the three different zones, bottom, middle, and top, of the zonal and homogeneous bioprinted scaffolds at day 0; (**b**) the quantification of the number of the cells from the H&E staining at day 0; (**c**) the live/dead and H&E images of the zonal and homogeneous scaffold at day 25; (**d**) the quantification of the number of cells from the H&E staining at day 28 (*n* = 3; *p*-values < 0.05 are shown). Scale bar: 100 μm.

## *3.3. Bioprinted Zonal Scaffold with Different Cell Densities Leads to a Gradient Cell-Derived ECM Deposition*

The evaluation of the sGAG and collagen deposition in the bioprinted scaffolds was performed on the basis of the analysis of Alcian Blue and Picrosirius Red histological staining of the scaffold cross-sections. At day 0, the histological analysis showed the background staining for the basal bioink material (alginate and methylcellulose) with a weak staining for Alcian Blue (Figure 4a) and a strong pink (negative for collagen) staining for Picrosirius Red (Figure 4c). A semiquantitative analysis of the stained samples was performed using an optimized threshold to subtract the background staining from positively stained areas (Figure 4b,d).

**Figure 4.** Bioprinted zonal scaffolds with cell density gradients result in differential cartilage-like tissue deposition; (**a**) histological images of zonal or homogeneous scaffold bioprinted with human chondrocytes and evaluated for sGAG with Alcian Blue staining (AB) at day 0; (**b**) the corresponding semiquantitative zonal analysis of the AB staining; (**c**) histological images evaluated for collagen deposition with Picrosirius Red staining (PSR) at day 0; (**d**) the corresponding semiquantitative zonal analysis of the PSR staining; (**e**) the histological images evaluated for sGAG with AB after 25 days in culture; (**f**) the corresponding semiquantitative zonal analysis of the AB staining; (**g**) histological images evaluated for collagen with PSR; (**h**) the corresponding semiquantitative zonal analysis of the PSR staining (*n* = 3; *p*-values <0.05 are shown). Scale bar: 100 μm.

At day 25 of culture, both types of staining demonstrated that the scaffolds supported the deposition of cartilage-like tissue components. In the zonal scaffold, Alcian Blue staining revealed an increased deposition of sGAG in the top zone of the scaffold at day 25, as indicated by a strong Alcian Blue signal (Figure 4e, zonal, AB, top). The semiquantification of the staining revealed the zonal manner in which sGAG was deposited within the scaffolds over time, displaying a significantly increase from the bottom to the top zone (Figure 4f, left), which was in accordance with the cell density distribution within the scaffolds. The homogeneous scaffolds deposited a similar amount of sGAG in all the different zones at day 25 (Figure 4e, homogeneous, AB and Figure 3f, right). For the collagen deposition, it was observed that chondrocytes appeared to mainly secrete small amounts of pericellular collagen according to the intense orange-red staining observed (which is positive for collagen staining) after 25 days of culture (Figure 4g). Similar to the result of Alcian Blue staining, some samples exhibited a trend toward stronger staining in the top zone of the specimens. By comparison, the homogeneous scaffolds revealed no differences in their staining throughout the section. The statistical analysis of the semiquantitative values corresponding to Picrosirius Red revealed a trend of increasing collagen deposition in the top zone of some of the zonal scaffolds. However, no significant differences were found between the groups after 25 days of culture (Figure 4h).

#### **4. Discussion**

In this study, we aimed to generate a cartilage scaffold of a clinically relevant size that mimics the gradient distribution of cells that is observed across the three distinct zones of the native tissue in hyaline articular cartilage of the human condyle and to analyze the cell viability and cell-derived matrix deposition. Over time in culture, the generated zonal scaffolds led to a gradient in cartilage-like matrix deposition. For printing each zonal scaffold, four different inks (one ink and three bioinks) were used to print different parts of the scaffold. Every printing part used its own printhead except for the middle and top zones where the bioink-containing cartridge had to be interchanged for every printed scaffold. This allowed us to simultaneously 3D print a PCL frame and the different chondrocyte-embedded zones with clinically relevant thicknesses and zone-specific cell densities. Few papers have investigated the 3D printing of more than one bioink for a single construct [10,19,24–26], probably due to the lack of several printheads or impracticality due to the constant changing of the printing cartridges that results in a time-consuming process. Recent studies suggest the use of coaxial needles [27] or continuous chaotic printing [28] as alternatives to generate multicellular heterogeneous systems. Several studies have combined PCL with bioinks to enhance the compressive stiffness of hydrogels [29–31]. In our study, the PCL frame helped the maintenance of the structural integrity of the scaffolds during bioprinting and subsequent in vitro culture. In addition, a PCL frame can be useful for future in vivo implantation [32,33]. Previous studies have shown that different cell densities can lead to different mechanical properties which are correlated with the amount of sGAG produced. In this study, we did not perform mechanical characterization of the hydrogel overtime. Therefore, future studies should also focus on the effect of different cell densities on the mechanical properties of the scaffold overtime. Regarding the PCL reinforcement, in most studies, this reinforcement is done by printing a mesh of PCL fibers in between hydrogel fibers, where the fiber orientation alternates orthogonally after every layer. In this study, however, it was determined that passing the hot (210 ◦C) thermoplastic nozzle right next to the cell-embedded hydrogel fibers negatively affects the chondrocyte viability, which has been also noted in other studies [34]. Therefore, another design approach was chosen in which the PCL would act as an outer skeleton for the hydrogel, concentrating the cell-embedded volume of the bioink at the center and maintaining its shape throughout its thickness. This approach was able to significantly increase the bulk mechanical properties compared to the bioink alone, as others have demonstrated [29–31]. Other studies also explored the pre-printing of PCL structures and then either cast or *z*-printed the bioink in the channels of the PCL mesh [35]. Despite

this option giving more freedom, as it decouples the fabrication process of the outer frame and the bioprinting of the bioink, *z*-print is not yet an option available in all 3D printing systems. After fabrication of such complex structures, live/dead and histological analysis demonstrated that human chondrocytes still retained high viability and zonespecific cell densities. In addition, this cell density gradient was shown to be maintained during in vitro culture for 25 days, resulting in a gradient in tissue deposition as compared to homogeneous scaffolds. Previous studies aiming to fabricate similar constructs with different cell density layers through bioprinting have not shown such clear cell distribution or its maintenance over time [19]. For example, it has been reported that cell gradient in scaffolds of 3 mm thickness fail due to the fusion of the zones during bioprinting. Moreover, the analysis of that study was focused on the overall scaffold rather than on the individual zones [19]. It is noteworthy that the difference in cell number between the middle and the bottom zones in the gradient scaffold is less apparent after culture in differentiation medium. Although cells do not tend to migrate in alginate, this particular bioink formulation containing methylcellulose may allow for some cellular migration [36] from the middle to the bottom layer, perhaps in the search for available nutrients and oxygen. This potential migration may have been caused by the cell gradient itself, as no changes in cell number were observed in the homogeneous scaffold. Further research is needed to verify this hypothesis. Despite the presence of the tissue deposition gradients, low amounts of glycosaminoglycan and collagen were seen in all the groups. This may be due to the alginate-based bioinks that offer good printability [25,30,35,37–41] but do not offer enzymatic-derived tissue remodeling [42], which have been suggested to potentially act as a physical barrier for tissue secretion in vitro, as also observed by others [43–45]. On the other hand, alginate-based biomaterials have been shown to be capable of generating cartilage when implanted in vivo [46,47], suggesting that other factors may also be playing a role in this regard. For instance, chondrocytes from only one donor were used in our study, and it is known that there can be some variability between donors or even between chondrocytes from different zones [45]. Regarding tissue deposition, it is interesting to point out that the superficial zone of native mature articular cartilage is characterized by a lower sGAG presence than the other zones. Our scaffold demonstrated that having a higher cell density in the superficial zone will result in higher sGAG. However, the tissue formed in our scaffold is still in development and yet far from what is seen in articular cartilage tissue. It will be necessary to evaluate how applications of these different zones would influence eventual tissue that is formed using in vivo models. Our study is in agreement with the previous studies showing that constructs with a higher cell density result in higher cartilage-like matrix deposition [48,49] and enhanced gene expression of cartilage markers [19]. This study focused on the recapitulation of the cell gradient observed in native hyaline articular cartilage. However, there are other gradients that could be interesting to investigate or even to combine. For instance, AC has been shown to have a gradient in stiffness [50–53]. Previous research has aimed to combine a stiffness gradient achieved by photo-crosslinking with a growth factor gradient achieved by mixing both TGF-β1 and BMP-2 in the hydrogel (before crosslinking). These gradients proved capable of regulating the number of embedded hMSCs and their differentiation toward bone- or cartilage-like matrix-producing cells [18]. In addition, from a developmental biology perspective, cartilage starts as a soft template. Therefore, it may be interesting to explore the generation of the stiffness gradients that occur in time. This points to the need for the development of new 4D bioprinting approaches that investigate the use of stimulus-responsive biomaterials to answer such complex questions [54].

#### **5. Conclusions**

In this study, zonal cartilage scaffolds with clinically relevant sizes and cell densities were successfully bioprinted as single units. The different zonal cell densities were partially maintained in vitro for 25 days. In addition, the scaffolds led to a gradient in the extracellular matrix components produced by the embedded human chondrocytes. These findings

shed some light on the effects of zonal cell gradients on the formation of new tissue matrix and help in addressing the remaining obstacles in order to accelerate the translation of cartilage tissue engineering into clinical practice.

**Supplementary Materials:** The following are available online at https://www.mdpi.com/article/ 10.3390/app11177821/s1: Figure S1. PCL design can affect cell viability; Video S1. A clip of the bioprinting process.

**Author Contributions:** Conceptualization, A.D., R.N., L.E.F.-A. and A.A.Z.; methodology, P.J.D.-P., A.D., M.M., M.H. and N.K.; software, P.J.D.-P. and A.D.; formal analysis, P.J.D.-P. and A.D.; investigation, P.J.D.-P. and A.D.; writing—original draft preparation, P.J.D.-P. and A.D.; writing—review and editing, P.J.D.-P., G.J.V.M.v.O., L.E.F.-A. and A.A.Z.; supervision, P.J.D.-P., M.N.-G., M.J.M. and L.E.F.-A.; project administration, L.E.F.-A. and A.A.Z.; funding acquisition, G.J.V.M.v.O. and A.A.Z. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by the Dutch Medical Delta project: RegMed4D.

**Data Availability Statement:** The data that support the findings of this study are available from the corresponding author, [A.A.Z.], upon reasonable request.

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

#### **References**


## *Article*

## **Follow-Up Study Evaluating the Long Term Outcome of ChondroMimetic in the Treatment of Osteochondral Defects in the Knee**

**Agnes Berta 1,**†**, Matthew S. Shive 2,\*,**†**, Andrew K. Lynn 3, Alan Getgood 4, Saara Totterman 5, Grahame Busby 6, Jerome Hollenstein 6, Gábor Vásárhelyi 1, Imre Kéki <sup>7</sup> and László Hangody 1,8**


Received: 7 June 2020; Accepted: 11 August 2020; Published: 14 August 2020

**Abstract:** Scaffolds are thought to be a key element needed for successful cartilage repair treatments, and this prospective extension study aimed to evaluate long-term structural and clinical outcomes following osteochondral defect treatment with a cell-free biphasic scaffold. Structural outcomes were assessed using quantitative 3-D magnetic resonance imaging (MRI) and morphological segmentation to determine the percentage of defect filling and repair cartilage T2 relaxation times, and clinical outcomes were determined with the modified Cincinnati Rating System, and the Knee Injury and Osteoarthritis Outcome Score (KOOS). Seventeen subjects with osteochondral defects in the knee were treated with ChondroMimetic scaffolds, from which 15 returned for long-term evaluation at a mean follow-up of 7.9 ± 0.3 years. The defects treated were trochlear donor sites for mosaicplasty in 13 subjects, and medial femoral condyle defects in 2 subjects. MRI analysis of scaffold-treated defects found a mean total defect filling of 95.2 ± 3.6%, and a tissue mean T2 relaxation time of 52.5 ± 4.8 ms, which was identical to the T2 of ipsilateral control cartilage (52.3 ± 9.2 ms). The overall modified Cincinnati Rating System score was statistically significant from baseline (*p* = 0.0065), and KOOS subscales were equivalent to other cartilage repair techniques. ChondroMimetic treatment resulted in a consistently high degree of osteochondral defect filling with durable, cartilage-like repair tissue at 7.9 years, potentially associated with clinical improvement.

**Keywords:** biphasic scaffold; osteochondral defect; cartilage repair; quantitative MRI; calcium phosphate

## **1. Introduction**

Articular cartilage damage continues to present a therapeutic challenge, and despite multiple and differing approaches ranging from bone marrow stimulation [1], autologous chondrocyte implantation (ACI) [2], and various grafting procedures incorporating allografts and autografts [3,4], none are recognized as sufficient. Recent reviews demonstrate that clinical research has failed to substantiate the

superiority of one technique or product, in large part because of heterogeneous patient demographics, the differing etiologies of defects, and the varying study designs and outcome measures [5–7]. The lack of longer term (5–10 years) clinical evidence in cartilage repair further limits evidence-based treatment algorithms. Nonetheless, the fundamental goal of any cartilage repair treatment should be to avoid the progression to secondary osteoarthritis [8,9], by achieving structural repair that is comparable to native hyaline cartilage, and thereby assuring long-term durability, joint function, and pain relief.

ChondroMimetic (Collagen Solutions, Plc) was developed to be a cell-free, osteochondral scaffold for use in cartilage repair. The biphasic implant comprises a 2 mm, unmineralized, type I bovine collagen and chondroitin-6-sulphate glycosaminoglycan (collagen/GAG) chondral layer, coupled to a 6mm, mineralized, subchondral bone layer, containing collagen/GAG and calcium phosphate brushite. These two distinct layers are separated by a continuous or 'soft' interface that closely mimics the compositional transition between mineralized bone and unmineralized cartilage at the tide mark [10]. As an open-cell foam, the ChondroMimetic scaffold which allows compression during surgical handling for press-fit implantation, followed by auto-expansion to fit the defect shape, bridging any gaps or irregularities [11]. In a defect with stable margins, ChondroMimetic offers structural and mechanical protection for the marrow-derived blood clot, similarly to mechanisms claimed for other techniques (e.g., autologous matrix-induced chondrogenesis (AMIC) [12], BST-CarGel [13]). Large animal preclinical studies with ChondroMimetic demonstrated a consistency in osteochondral repair characterized by a high degree of cartilage defect filling with a hyaline-like repair tissue in both medial femoral condyle and trochlear defect sites [14–16].

Since long term data (>5 years post treatment) is fast becoming a required component for clinical decision-making in cartilage repair, the aim of this prospective extension study was to determine the long-term effectiveness of ChondroMimetic treatment of osteochondral defects in subjects enrolled in a previous short-term study. Using 3D quantitative magnetic resonance imaging (MRI) techniques, a consistently high degree of osteochondral defect filling with durable, cartilage-like repair tissue was found at 7.9 years and potential clinical benefit.

## **2. Methods**

#### *2.1. Study Design and Participants*

This study was an extension of a prospective, interventional study which enrolled 17 subjects between February 2009 and March 2010 at a single clinical site (formerly unpublished). The 17 eligible subjects (8 males, 9 females) were between 18 and 50 years of age with at least one osteochondral defect measuring <12 mm in the knee. Subjects with a body mass index >40 kg/m<sup>2</sup> or a previous cartilage repair treatment in the index knee were excluded. All subjects who participated in the original interventional study were eligible to be enrolled in the extension study, which was designed to provide long-term follow up at a single time point which included a clinic visit and a single MRI scan. All subjects were asked to provide written informed consent prior to extension study activities, which were approved by national and local ethics review boards. Both studies were performed in accordance with guidelines for Good Clinical Practice. This extension study was registered with ClinicalTrials.gov (NCT03385642).

#### *2.2. Interventional Details*

During the original interventional study, following osteochondral defect preparation, the preloaded implant was hydrated with saline through the hydration portal in the delivery device, which was then positioned perpendicular to the defect with the tip inserted to the to the full defect depth. Simultaneous advancement of the implant and withdrawal of the device tip left the implant to expand into the defect volume. ChondroMimetic scaffolds were used to treat mosaicplasty donor sites in the trochlea in 15/17 subjects concurrent with mosaicplasty treatment of 12 femoral condyle defects and 3 ankle defects. ChondroMimetic scaffolds were also used to treat primary femoral condyle defects

in 2/17 subjects. Ultimately, scaffolds chosen for use ranged from 6.5–8.5 mm in diameter. None of the subjects included in the study had tibial articular defects.

#### *2.3. Outcome Measures*

Neither the subjects nor the investigator were blinded to treatment in this study, but the administration of identical validated questionnaires, standardized acquisition of MRI scans, and subsequent semi-automated, quantitative MRI analysis of repair tissue structure by an independent 3rd party minimized method and analysis bias.

#### 2.3.1. Structural Repair

Repair tissue structure, defined as the quantity (i.e., degree of defect filling (%)) and quality (T2 relaxation time) of new cartilage tissue, was assessed by three-dimensional (3D) quantitative MRI conducted at an imaging core lab (Qmetrics Technologies, Rochester, NY, USA) using, validated segmentation and 3D reconstruction techniques [17,18] which have been applied in Phase 3 clinical trials of cartilage repair and multiple osteoarthritis studies [19–21]. Standardized MRI scans were acquired for each subject upon enrolment into the extension study with a Siemens 3T MAGNETOM® Verio, A Tim+Dot System and dedicated Tx-Rx CP extremity coil at a single, prequalified and trained MRI clinic. Technician training and scanner magnet qualification were conducted prior to trial activities. A uniformity and linearity phantom scan was performed for quality control. Imaging phantoms within the field of view served as internal controls.

The imaging protocol was designed to provide very high spatial resolution with high in-plane resolution and thin slices for all sequences. The study images were acquired using a fat-suppressed 3D spoiled gradient-echo (FS SPGR) sequence in sagittal and axial planes (slice thickness: 2 mm), and a 2D dual-echo fat-suppressed fast spin-echo (FS dual echo) sequence in the sagittal plane (slice thickness 1.2 mm). For morphological analyses of cartilage, cartilage defects, bone and bone defects, sagittal and axial FS SPGR series were used. Sagittal FS dual-echo series were used for T2 analyses. All scans were sent to an imaging core lab for centralized scan quality review, storage, and quantitative analysis. The sagittal 3D FS SPGR series and the sagittal FS dual echo series were segmented for bone and cartilage using a semi-automated atlas-based segmentation method [17]. The segmentations were reviewed and edited, if needed, by a musculoskeletal radiologist with expertise in cartilage repair. Original cartilage and bone defect boundaries for each subject were manually established on FS SPGR and FS dual echo segmentations using edited segmentation of MRI scans (sagittal dual echo) obtained 10 days post-operatively in the original interventional ChondroMimetic study.

The segmented 3D volume of new repair tissue at follow-up was then used to quantitatively determine the degree of defect filling (%) and the mean T2 relaxation time of cartilage repair tissue. Filling was calculated as a ratio of new repair tissue (bone and cartilage) to the original osteochondral defect volume. In cases where multiple donor sites in the same knee were treated, confluence of defects was observed in the trabecular bone resulting in a single defect volume and a single fill ratio. A radiologist-selected region of interest (ROI) of cartilage on an untouched area of the trochlea in the same knee was segmented and analyzed for each subject as a native T2 control.

## 2.3.2. Clinical Benefit

Clinical outcomes at the extension study timepoint were evaluated using two subject self-reported questionnaires. The modified Cincinnati Rating System has 8 components in the score, assessing pain, swelling, giving way, overall activity levels and walking, stairs, running activity, and jumping/twisting. With this instrument, a higher score is a better result and a lower score implies greater disability. Grading of the scores followed that proposed by Bentley et al. [22], where a score of <30 as "poor", 30–54 as "fair", 55–79 as "good" and >80 as "excellent".

The Knee injury and Osteoarthritis Outcome Score (KOOS) [23] in the Likert format has 5 subscales: Pain, other Symptoms, Function in daily living, Function in Sport and Recreation and knee-related

Quality of Life. Scores are attributed by dimension, ranging from 0 to 100 (high score better outcome). The KOOS score was not administered in the original interventional study, and therefore baseline scores were not available.

## *2.4. Statistical Considerations*

All statistical analyses were conducted using SAS 9.4 or above (SAS Institute Inc., Cary, NC, USA). All graphs were produced in Microsoft Excel 2016 (Microsoft Inc., Redmond, WA, USA). For statistical analysis purposes, baseline was defined for clinical outcome data as data collected pre-operatively, and for MRI analyses, 10 day post-operatively. Quantitative structural repair outcomes underwent univariate and descriptive analyses, investigating the scores at the single study time point for the degree of defect filling (total volume), and for repair tissue quality (T2). Clinical outcome data underwent univariate and descriptive analyses, for both modified Cincinnati Rating System and KOOS subscales. Statistical comparisons used Student's *t*-tests.

## **3. Results**

## *3.1. Enrolment and Subject Characteristics*

Enrolment into the extension study targeted 8-year follow-up for the 17 subjects in the original clinical study. Ultimately, 15 of the original 17 subjects completed the single extension study visit, and their demographic characteristics are shown in Table 1. The study population was balanced by gender, and subjects received from 1 to 5 ChondroMimetic scaffolds, per knee, to treat trochlear donor sites for mosaicplasty in 13 subjects, and medial femoral condyle defects in 2 subjects. A total of 14 additional procedures since enrolment into the original interventional study were recorded by 12 individual subjects. The majority of these interventions involved either arthroscopy of knees or ankles or intra-articular injections. No interventions were specifically attributed to the ChondroMimetic treatment by the Investigator.


**Table 1.** Follow-up Characteristics of Trial Subjects ‡.

‡ Plus-minus values are means ± standard deviation unless otherwise indicated. ¶ Determined using quantitative 3D magnetic resonance imaging (MRI) of 10 days postoperative scans. ˆ Hyaluronic acid injection (4); microfracture of new primary lesion (3); second-look arthroscopy (5). \* From physical examination and standardized interview during the 8-year study visit.

There were no withdrawals from the extension study. The only missing data in the study was MRI analysis from 1 subject who did not yield usable scans due to movement.

#### *3.2. Repair Outcomes*

#### 3.2.1. Structural Repair

The quantity and quality of ChondroMimetic repair tissues was determined for 14 subjects (the MRI from the fifteenth subject was unusable due to movement; see Table 2). ChondroMimetic-treated osteochondral defects at 7.9 years post-treatment demonstrated a substantial degree of defect filling of 95.2 ± 3.6%. Figure 1 shows the Total Defect Fill (%) of individual subjects and illustrates a low variability and consistently high degree of defect filling following ChondroMimetic treatment. In addition, subjects treated with ChondroMimetic had a mean T2 relaxation time of 52.5 ± 4.8 ms for the cartilage repair tissue ROI, which was nearly equal to the T2 relaxation time of the ipsilateral native control cartilage regions (52.3 ± 9.2 ms (*p* = 0.94)), and suggestive of a very similar tissue biochemical make-up and structure. The T2 standard deviations calculated for each repair tissue volume were comparable to native cartilage (17.3 ± 3.3 vs. 13.1 ± 5.3 ms), with a variance that was statistically equal to that of native cartilage (*p* = 0.11 by F-test).

**Table 2.** Structural repair outcomes by 3 dimensional MRI at 7.9 years follow-up (n = 14).


**Figure 1.** Total Defect Fill (%) by quantitative MRI for individual subjects at a mean followup of 7.9 years after ChondroMimetic treatment. Total Defect Fill (%) was determined using 3D quantitative magnetic resonance imaging by calculating the ratio of the new repair tissue (bone and cartilage) volume at followup, to original defect (bone and cartilage) volume acquired from 10 days post-operative scans (baseline). (n = 14).

Using each scan series, full 3D reconstructions were generated of treated knees at the 10-day post-operative baseline, and at follow-up, to assist in visualizing the morphological impact of

ChondroMimetic repair. Figure 2 illustrates a representative reconstruction conducted for a 29-year-old female subject who received 5 ChondroMimetic scaffolds in the trochlea, which are clearly visible filling the newly treated osteochondral defects in Figure 2A by 2D MRI, and are similarly visible in the 3D rendering (Figure 2B). At the follow-up after 8 years and 3 months, both the subchondral bone and cartilage aspects of the original defects appear as near-normal in both Figure 2C,D, compared with ipsilateral control regions in Figure 2E.

**Figure 2.** Longitudinal MRI reconstruction of a ChondroMimetic-treated subject knee at 10 days (**A**,**B**) and 8 years, 3 months (**C**,**D**). (**A**) By 2D MRI, 5 ChondroMimetic implants can be observed with hyper-intense peri-implant boundaries, and lack of cartilage-like signal (white arrow). (**B**) 3D rendering of knee yields ChondroMimetic implants (green) surrounded by cartilage (brown) by automated software based on MRI signal. (**C**) At the follow-up after 8 years and 3 months, treated bone defects are filled with bony repair tissue and the treated cartilage defect is filled with cartilaginous repair tissue. (**D**) Automated rendering demonstrates cartilaginous tissue in repaired defects congruous with surrounding cartilage. (**E**) The trochlear control region of interest (filled red circle) of native cartilage used for T2 relaxation time comparison. The open purple circle shows an unused, alternate control region of interest.

## 3.2.2. Clinical Benefit

The modified Cincinnati Rating System overall score at the study timepoint was 80.9 ± 15.5, which was statistically significant from baseline (*p* = 0.0065) and represents a grading of 'excellent'.

The KOOS subscale data are presented in Figure 3. The mean pain score was 90.0 ± 11.4. The mean symptom score was 80.0 ± 15.8. The mean function in daily living (ADL) score was 93.4 ± 9.1. The mean function in sports and recreational activities (Sports/Rec) score was 68.7 ± 24.7. The knee related quality of life (QOL) mean score was 66.7 ± 22.2.

**Figure 3.** Knee Injury and Osteoarthritis Outcome Scores (KOOS) for ChondroMimetic subjects at a mean follow-up of 7.9 years. Data represents mean + standard deviation (n = 15).

#### **4. Discussion**

The main finding of the current study is that voluminous, high quality repair cartilage results from ChondroMimetic treatment of osteochondral defects after 7.9 years, supporting the hypothesis that ChondroMimetic brings consistent and durable long-term structural repair with an appropriate safety profile. Improvements in clinical outcomes were highly significant over the baseline (*p* < 0.0065) for the 7.9-year follow-up, which suggests a benefit of ChondroMimetic treatment when used as backfill for mosaicplasty donor sites. The use of validated, three-dimensional quantitative MRI [17,18], which assessed the repair tissue quantity and quality, is a particularly valuable aspect of this study, with a high level of standardization and precision and conducted under Good Clinical Practice guidelines.

The high degree of defect filling by new bone and repair cartilage seen in ChondroMimetic defects at 7.9 years (95.2%) is a critical outcome, as any joint resurfacing should aim to reestablish mechanical homeostasis and anatomical shape with an integrated surface, particularly since chondrocyte-mediated biosynthesis, remodeling, and either tissue repair or degradation over time is dependent on the mechanical loading conditions of cartilage [24]. Normal surface morphology and joint articulation would be expected to improve biomechanical conditions and inhibit the degeneration of defect and peri-defect tissues and slow the progression of secondary osteoarthritis [8,9]. This particular outcome was not fully unexpected, since critical size osteochondral defects in goat knees treated with ChondroMimetic scaffolds yielded similarly striking filling [14]. Furthermore, this level of defect filling is greater than that resulting from another off-the-shelf biological scaffold, BST-CarGel, at 5 year spost-treatment, which produced %Fill of 93.79%, as determined by very similar quantitative MRI methods [13].

Quantitative MRI also identified durable, high quality cartilage repair tissue by T2 relaxation time for ChondroMimetic-treated defects. T2 relaxation time is well known to be sensitive to, and highly dependent on, the extracellular cartilage matrix and particularly the collagen network structure, orientation, as well as macromolecular concentration, and tissue hydration [25–29]. When acquired under standardized conditions coupled with three-dimensional segmentation of regions of interest, T2 relaxation time has been an effective, quantitative method for comparing repair tissue to native articular cartilage, and/or to repair tissue resulting from another treatment [13,30,31]. Here, the determination of the quality of repair is evidenced by the closeness of measured T2 values to that found from the ipsilateral normal articular cartilage (52.3 vs. 52.5 ms, *p* = 0.94), indicating a near-normal level of tissue quality after 7.9 years. The overall level of repair tissue organization achieved is further evidenced by the T2 standard deviations calculated for each repair tissue volume, which were comparable to native cartilage with a variance that was statistically equal to that of native cartilage. In contrast, a tissue with regions of differing organization, or of mixed hyaline-fibrocartilage tissue, would be expected to have a more widespread variance than normal cartilage. When comparing to T2 outcomes resulting from MRI studies with other scaffolds, ChondroMimetic repair tissue demonstrated

a similar closeness to native control cartilage as a study with Hyalograft C [32] in the knee, but was improved compared to the MaioRegen scaffold in the ankle [30], and BST-CarGel in the knee [13], neither of which resulted in T2 relaxation times similarly close to control cartilage. This finding also confirms consistency of repair in improved collagen-based organization and content, which are necessary components for long-term durability of cartilage since collagen breakdown is considered to be a critical step in the progression of osteoarthritis [33,34].

It may be considered a limitation of the study that ChondroMimetic was used as back-fill treatment for the majority of osteochondral defects (13/15) in the study, particularly since it is generally believed that mosaicplasty donor sites are: (1) not exposed to weight bearing forces; and (2) ultimately filled with a fibrocartilage through self-repair processes and are asymptomatic [35]. To the contrary, donor sites experience high loads with daily activities, depending on their size and location, from 1.3 times up to 7.8 times body weight [36]. Furthermore, Andrade et al. [37] reviewed 21 original mosaicplasty studies comprising 1726 patients and found that donor site morbidity reportedly occurs in 5.9–16.9% of cases and manifests through crepitation and knee stiffness, instability during activities of daily living, and persistent pain. The fact that in this study no adverse events or symptoms were reported to be related to donor sites or ChondroMimetic implants over the 7.9 year follow-up, and modified Cincinnati and KOOS scores demonstrated a high level of satisfaction, demonstrates that ChondroMimetic treatment was safe and may suggest some clinical benefit, further supported by the MRI-assessed high quality cartilage repair. Furthermore, the regeneration of bone and cartilage to near-native levels of fill and biochemical structure suggest that Chondromimetic may fit into an algorithm for treatment of full-thickness, primary cartilage lesions. Indeed, two full weight bearing primary defects on the medial femoral condyles were treated in this study with ChondroMimetic and demonstrated similarly high degrees of total defect fill (mean: 95.85%) and high quality repair tissue with T2 relaxation times comparable to native cartilage (means: 46.7 ms vs. 47.7 ms, respectively), in addition to satisfactory clinical benefit.

A true study limitation was the lack of an active control group, such as microfracture, which has been suggested by current regulatory guidance [38]. But use of an ipsilateral native cartilage control ROI for quantitative structural outcomes provided an appropriate internal control for making a determination of quality and durability of cartilage repair. Also, the small sample size of the study could limit the interpretation of the results. However, the low variance in both clinical and structural quantitative MRI outcomes nonetheless supports a consistency in repair that would likely be maintained with a larger sample size. Long-term follow-up studies can also present challenges that limit interpretation of study outcomes. For example, here, 14 procedures were conducted in the index knee of 12 subjects over the 8-year follow-up period. However, it can be concluded that it is highly unlikely that these procedures had any effect on the assessments conducted for this study. In 2 cases, the procedures were not knee related (e.g., lung biopsy). The other 12 included: (1) hyaluronic acid injections in 4 subjects performed at least 12 months prior to the extension study visit, and therefore would be expected to have lost any therapeutic effect; (2) microfracture of new primary lesions in 3 subjects, although this would represent a worst case scenario since the procedure could worsen the pain at the time of the extension study visit, or simply decrease the new pain associated with the new lesion back to baseline levels; and (3) second-look arthroscopy in 5 subjects, conducted 1–2 years post-operative by the investigator and unrelated to the study (e.g., to remove bone material from an HTO; to remove malleolar screws from ankles).

The ChondroMimetic scaffold was designed to provide a cell-free cartilage repair alternative, with biological components and architecture, conducive to the re-establishment of the subchondral bone whilst supporting regeneration of articular cartilage. With time and experience, it has been generally recognized that scaffolds are essential in supporting the processes of cartilage repair and indeed, almost all current approaches have incorporated some type of scaffold. So-called "augmented bone marrow stimulation" involves concurrent implantation of an exogenous scaffold (e.g., AMIC [12], BST-CarGel [13], Biocartilage [39]), 3rd generation ACI techniques incorporate membranes to support

cell delivery and residency (Hyalograft C [40], MACI [41]), and grafting techniques, by their nature, involve implantation of natural osteochondral scaffolds [42]. The need to have cells, at all, is being questioned when scaffolds alone, with their low cost and simplicity, bring equivalent results in a more cost-effective manner [43].

## **5. Conclusions**

Treatment of osteochondral defects in the knee with the ChondroMimetic scaffold was associated with a consistently high degree of filling with a cartilage-like repair tissue over a long-term (7.9 year) follow-up period, improved clinical scores, and noted lack of donor site morbidity, underscoring potential benefits of a cell-free, single step scaffold from both a clinical and a health economic standpoint.

**Author Contributions:** Conceptualization, A.B., M.S.S., G.B. and J.H.; Formal analysis, M.S.S. and S.T.; Funding acquisition, G.B. and J.H.; Investigation, A.B., G.V., I.K. and L.H.; Methodology, A.B. and M.S.S.; Project administration, A.B. and G.B.; Supervision, M.S.S. and L.H.; Visualization, M.S.S.; Writing—original draft, M.S.S.; Writing—review & editing, A.B., M.S.S., A.K.L., A.G., S.T., G.B., J.H. and L.H. All authors read and approved the final manuscript.

**Funding:** Collagen Solutions (study sponsor) was the sole funding source and was responsible for study design, data interpretation, and manuscript publication.

**Conflicts of Interest:** G.B. and J.H. were employees of the study sponsor at the time of the study. M.S.S., A.K.L. and A.G. receive consultancy fees from the study sponsor. A.K.L. holds equity in the study sponsor. A.G. receives consultancy fees and research support from Smith & Nephew and Ossur. A.B., S.T., G.V., I.K. and L.H. have no potential conflict of interest to declare.

**Ethical Approval:** Ethics approval was obtained from the National Scientific and Ethical Committee of the Medical Research Council of Hungary (ETT TUKEB) (Reference: 58813-/2016/EKU) and the local Research Ethics Committee of Uzsoki Hospital (Uzsoki IKEB) (Reference: 223-IK/2016).

## **List of Abbreviations**


## **References**


© 2020 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 (http://creativecommons.org/licenses/by/4.0/).

## **Mosaicplasty versus Matrix-Assisted Autologous Chondrocyte Transplantation for Knee Cartilage Defects: A Long-Term Clinical and Imaging Evaluation**

**Stefano Za**ff**agnini 1, Angelo Bo**ff**a 1,\*, Luca Andriolo 1, Davide Reale 1, Maurizio Busacca 2, Alessandro Di Martino <sup>1</sup> and Giuseppe Filardo 3,4**


Received: 26 May 2020; Accepted: 1 July 2020; Published: 3 July 2020

**Abstract:** Different surgical procedures have been proposed over the past few years to treat cartilage lesions. The aim of this study was to compare mosaicplasty and matrix-assisted autologous chondrocyte transplantation (MACT) at long-term follow-up. Forty-three patients were included: 20 mosaicplasty and 23 MACT. Patients were evaluated before and 12 years after surgery with the International Knee Documentation Committee (IKDC) subjective and objective scores for symptoms and function, and with the Tegner score for activity level. Magnetic Resonance Imaging (MRI) was used to evaluate repair tissue with the MOCART 2.0 score. Mosaicplasty and MACT showed good clinical and MRI results (IKDC subjective score 75.3 ± 21.8 and 81.8 ± 13.0, both p < 0.0005). Mosaicplasty presented a 10% reoperation rate and a 25% overall failure rate, while no failures were documented in MACT (p = 0.016). While size did not influence the results in the MACT group, mosaicplasty presented lower IKDC objective and Tegner scores in lesions bigger than 2 cm<sup>2</sup> (p = 0.031 and p = 0.014, respectively). Mosaicplasty and MACT presented both satisfactory clinical and MRI results at long-term follow-up. However, for larger lesions, MACT presented better subjective and objective outcomes, as well as less failures, which should be considered when choosing the most suitable treatment for patients affected by knee cartilage lesions.

**Keywords:** mosaicplasty; MACT; ACI; scaffold; osteochondral autologous transplantation; OAT; cartilage; knee

## **1. Introduction**

Articular cartilage lesions of the knee are a common issue, often affecting a young population and causing pain and functional impairment, with consequent high social impact [1,2]. Due to the limited cartilage healing potential and the difficulty to restore complex biomechanical features, chondral defects remain a challenging problem [3]. Thus, to avoid their deleterious consequences, such as the risk of further cartilage loss and osteoarthritis (OA) development, numerous surgical procedures have been proposed over the years to restore the articular surface with hyaline-like, durable repair tissue, from reconstructive techniques to the more ambitious regenerative strategies [4–7]. For these surgical

approaches, the first procedures described were osteochondral autologous transplantation (OAT) and autologous chondrocyte implantation (ACI), respectively [8,9]. These techniques demonstrated promising results up to long-term follow-up, but they also showed several drawbacks, such as significant donor site morbidity for OAT [10], and the risk of graft hypertrophy for ACI [11]. Moreover, both procedures were doomed by the complexity and morbidity of the surgical procedure, requiring an arthrotomy, and thus producing a higher risk of joint stiffness and arthrofibrosis [12,13].

Therefore, these approaches have been further developed to overcome the aforementioned limitations. On the one hand, the introduction of the mosaicplasty technique involved the use of multiple autologous osteochondral plugs of smaller diameter compared to OAT, thus causing lower morbidity, transferred from lower weight-bearing areas to the cartilage defect [14]. On the other hand, matrix-assisted autologous chondrocyte transplantation (MACT) was introduced to overcome ACI drawbacks thanks to the use of different types of scaffolds for chondrocyte culture and the direct transplantation of a bioengineered tissue, a more manageable and stronger implant for an easier and more stable positioning of the cell-scaffold construct [15]. Moreover, both options can be performed through an arthroscopic approach, allowing to reduce patient's morbidity, surgical time, recovery, and complications related to open surgery [14,16]. Both techniques showed promising short-term results, later confirmed by studies at long-term follow-up [17,18]. Mosaicplasty was suggested to provide better clinical results compared to MACT at short-term [19], but data on long-term comparison between these two techniques are still lacking.

The aim of this study was to compare the clinical outcomes and the quality of repair tissue provided by mosaicplasty and MACT for the treatment of knee cartilage defects at long-term follow-up.

## **2. Materials and Methods**

#### *2.1. Patient Selection*

All patients undergoing a cartilage repair procedure were prospectively followed, and clinical outcomes were collected in an institutional database, approved by the Hospital Ethics Committee and Internal Review Board of the Rizzoli Orthopedic Institute, Bologna, Italy (prot. gen. n. 39667). Informed consent of all patients was obtained. Surgical indications for these procedures were as follows: focal cartilage knee defects graded III–IV according to International Cartilage Regeneration & Joint Preservation Society (ICRS) classification in patients complaining of clinical symptoms (pain, swelling, locking, and giving way) after failure of conservative treatments. Contra-indications for surgery were untreated limb misalignment or knee instability, diffuse arthritis or bipolar ("kissing") lesions, and other general medical conditions such as infectious, tumor, metabolic, rheumatic, and inflammatory diseases. A definitive diagnosis of chondral lesion and sizing was performed during the arthroscopic procedure. Patients who presented with an anterior cruciate (ACL) lesion at the time of surgery underwent the combined ACL reconstruction in the same surgical session with cartilage harvesting.

A statistician, blinded to the treatment outcome, extracted two homogeneous groups of patients from the general database treated with mosaicplasty or MACT. These groups were comparable both in terms of patient demographics and cartilage lesion features. In detail, male or female patients were included, aged between 18 and 40 years, with body mass index (BMI) between 18.5 and 30, treated for lesions involving only femoral condyles (excluding trochlea and tibial plateau), with defect size between 1.0 and 3.0 cm2, traumatic or degenerative etiology (excluding osteochondritis dissecans lesions), and evaluated up to a minimum 10-year follow-up. According to these criteria, 43 patients were included in the study: 20 mosaicplasty and 23 MACT. The specific characteristics of the two groups are reported in Table 1. Statistical analysis confirmed that the two treatment groups were homogeneous with regard to gender, age, BMI, activity level, defect size and location, etiology, combined and previous surgery.


**Table 1.** Comparison of patient and lesion characteristics in the two groups.

Statistical analysis showed no significant differences between the two groups. ACL, anterior cruciate ligament; BMI, body mass index; MACT, matrix-assisted autologous chondrocyte transplantation; NS, not significant; SD, standard deviation; y, years.

#### *2.2. Surgical Techniques and Rehabilitation Protocols*

Arthroscopic mosaicplasty was performed in one step using the Autogenous Osteochondral Grafting System (Smith & Nephew, London, UK) surgical instrumentation, as previously described [20]. In brief, damaged cartilaginous and fibrous tissue was excised, and the lesion was then measured for size and location. Recipient 15 mm deep holes were made perpendicular to the cartilage surface. After preparing the defect area, osteochondral grafts were harvested. The donor site was preferentially the superolateral ridge of the femoral condyle, whereas the superomedial ridge was used only when four grafts were necessary. Tubular chisels were used to harvest the graft and then, with a smooth cannula, the grafts were delivered into the defect. One plug was used in 4 patients, 2 plugs in 9 patients, 3 plugs in 5 patients, and 4 plugs in 2 patients. The stability of the plugs was tested by cyclic bending of the knee while grafts were visualized.

Arthroscopic MACT was performed in two steps as previously described [21]. In brief, the first procedure consisted of an arthroscopic biopsy of healthy cartilage, sent for chondrocyte cell culture and subsequent seeding onto a hyaluronic acid-based scaffold (Hyaff 11, Fidia Advanced Biopolymers Laboratories, Padova, Italy) to obtain the bioengineered tissue Hyalograft C (Fidia Advanced Biopolymers Laboratories, Padova, Italy). After 6 weeks, the second step was performed, and the bioengineered tissue was arthroscopically implanted through cannulated devices. Because of the physical adhesive characteristics of the graft, no fibrin glue or sutures were used to fix the implant. Under arthroscopic control, the stability of implanted patches was evaluated during cyclic bending of the knee.

Postoperative rehabilitation protocols were comparable for both treatments. Postoperative management focused on early mobilization to facilitate faster resolution of swelling, promote healing and joint nutrition, and prevent adhesions. Passive knee range of motion (ROM) exercises on a continuous passive motion (CPM) machine (0◦–30◦) and static quadriceps exercises were started few days after surgery. Usually, 90◦ of knee flexion was reached within 4 weeks and full ROM within 8 weeks after surgery. Toe-touch ambulation with crutches and braces in full extension was allowed the third or fourth week and was usually completed within 6 to 8 weeks after surgery. Progressive

recovery of functional daily activities and muscle strength were performed, with particular care during exercises in open kinetic chain, which were allowed only after 12 weeks. In the mosaicplasty group, full athletic activity was permitted after 4 months and contact and traumatic sports were allowed after 6 months. In the MACT group, high-impact activities were discouraged until 10 to 12 months after implantation.

## *2.3. Clinical and Radiological Evaluation*

Patients were evaluated before surgery and prospectively for a mean of 12 years of follow-up (140.2 ± 15.1 months, range 120–168 months) with the International Knee Documentation Committee (IKDC) subjective and objective scores for the evaluation of knee symptoms and function, and with the Tegner score for activity level. Effusion and passive motion deficit were used to determine the final functional grade of the knee (A—normal, B—nearly normal, C—abnormal, or D—severely abnormal). Failures were also recorded: the procedures were considered to have failed if the patient needed a reoperation because of symptoms related to the primary defect. For these patients, the scores of the latest follow-up available before reoperation were carried forward at the final follow-up. Besides surgical failures, patients without a clinically significant improvement (10 IKDC subjective points compared with the baseline evaluation, as previously described [22]) were considered to be a clinical failure.

Magnetic Resonance Imaging (MRI) was used to evaluate the repair tissue in 30 knees (15 of the mosaicplasty group and 15 of the MACT group) at long-term follow-up. MRI evaluation was performed using a 1.5-T superconducting magnet (General Electric Co, Fairfield, Connecticut) with a dedicated phased array (8 channels) HD knee coil. The following sequences were used for graft evaluation: sagittal fast spin echo proton density weighted with fat saturation (TR 3500, TE 22, FOV 16, slice thickness 3 mm, Matr 320 × 256); sagittal T2 MAP (TR 1000, echoes 8 FOV 16, Th. 3 mm Matr 320 × 224); sagittal 3D "Cube" proton density weighted (TR 2500, TE 30, FOV 20, slice thickness 0.8 mm Matr 288 × 288); coronal fast spin echo proton density weighted with fat saturation (TR 2600, TE 40, FOV 16, slice thickness 4 mm, Matr 320 × 224); axial fast spin echo proton density weighted with fat saturation (TR 2600, TE 40, FOV 16, slice thickness 4 mm, Matr 320 × 224). The magnetic resonance observation of cartilage repair tissue (MOCART) 2.0 scoring system was applied for graft evaluation, which considers seven variables to describe morphology and signal intensity of the repair tissue [23]. All imaging evaluations were performed by an orthopedic surgeon and a musculoskeletal radiologist experienced in cartilage regeneration procedures in consensus.

## *2.4. Statistical Methods*

All continuous data were expressed in terms of mean ± SD, categorical variables were expressed as proportions or percentages. The Shapiro-Wilk test was performed to test normality of continuous variables. The Levene test was used to assess the homoscedasticity of the data. Repeated measures ANOVA was performed to compare the scores at different follow-up times. The one way ANOVA test was performed to assess the between group differences of continuous and normally distributed and homoscedastic data; the Mann-Whitney test was used otherwise. The Spearman rank correlation was used to assess correlations between numerical scores and continuous data, the Kendall's tau correlation was used to assess correlations between ordinal scores and continuous data. Pearson's chi square exact test was performed to investigate relationships between grouping variables. For all tests p < 0.05 was considered significant. All statistical analysis was performed using SPSS v.19.0 (IBM Corp., Armonk, NY, USA).

## **3. Results**

Both groups showed a marked improvement of all clinical scores from baseline to the last follow-up. The IKDC subjective score improved significantly in the mosaicplasty group from the basal level of 38.1 ± 11.8 to 75.3 ± 21.8 (p < 0.0005) at the final follow-up. Similarly, in the MACT group there was

a significant improvement from the basal level of 37.6 ± 14.9 to the final follow-up of 81.8 ± 13.0 (p < 0.0005). No significant differences in the IKDC subjective scores were found between mosaicplasty and MACT groups (Figure 1).

**Figure 1.** Comparison of the International Knee Documentation Committee (IKDC) subjective score achieved at a final follow-up of 12 years by both groups of patients. The values are expressed in median and 25th and 75th percentiles. No significant differences were found between the two groups. FU, follow-up.

The IKDC objective score in the mosaicplasty group improved from baseline, where no knees were evaluated as normal or nearly normal (13 C, 7 D), to 16 normal or nearly normal knees at 12 years of follow-up (7 A, 9 B, 3 C, 1 D). In the MACT group, this score improved from 7 normal or nearly normal knees at baseline (1 A, 6 B, 9 C, 7 D) to 23 at the final follow-up (16 A, 7 B). The IKDC objective score improvement was significantly higher in the MACT group compared with the mosaicplasty group at 12 years of follow-up (p = 0.029). The Tegner score in the mosaicplasty group showed a significant improvement from 2.9 ± 1.4 at the pre-operative evaluation to 5.0 ± 2.4 at the final follow-up (p = 0.043); however, the mean activity level remained lower vs. pre-injury (7.1 ± 2.0, p = 0.081). Similarly, the Tegner score of the MACT group showed a statistically significant improvement (p = 0.001) from the pre-operative level (1.6 ± 1.5) to the final follow-up (5.3 ± 2.1), although also in this case without reaching the pre-injury level (7.0 ± 1.6, p = 0.081). No significant differences were found in the Tegner scores between mosaicplasty and MACT at the final follow-up, with only eight patients for each group achieving the pre-injury activity level 12 years after surgery (Figure 2).

Two patients of the mosaicplasty group failed (one with four plugs and one with three plugs) and were treated with MACT 2 years after the treatment, fixing a 10% reoperation rate. At the final mean follow-up of 12 years, three more patients in the mosaicplasty group (one patient with one plug, one with two plugs, and one with four plugs) were considered clinical failures, for an overall failure rate of 25%, significantly higher compared to the MACT group, where no surgical or clinical failures occurred (p = 0.016).

The MRI findings of 30 knees (15 mosaicplasty and 15 MACT), analyzed with the MOCART 2.0 scoring system, showed a high rate of cartilage defect filling in both groups, and a complete integration into the adjacent cartilage in half of the patients in both groups. However, the surface of the repair tissue was damaged in the majority of cases. The structure of the repair tissue was homogeneous in most cases, and the signal intensity was normal (isointense) in half of the patients in both groups. At subchondral bone level, a bony defect or bony overgrowth was reported in the majority of patients, while an edema-like marrow signal was reported in almost half of the patients in both groups (Figures 3

and 4). No significant differences were reported between the two groups in the overall MOCART 2.0 scores and in all the seven subscales (Table 2).

Further analysis was performed to determine the parameters that influenced the clinical outcome at the final follow-up in the two treatment groups. Defect size was found to significantly influence the final IKDC scores between the two groups, with MACT producing significantly higher objective results (p = 0.031) and a tendency for higher subjective results compared to mosaicplasty (81.8 ± 18.3 vs 58.8 <sup>±</sup> 25.0, p = 0.092) for lesions > 2 cm2. Moreover, while size did not influence results in the MACT group, patients with lesions bigger than 2 cm<sup>2</sup> had a lower improvement in their activity level as measured by Tegner (p = 0.014) (Figure 5). Conversely, age, sex, BMI, site, etiology, number of plugs (for the mosaicplasty group), and previous or associated surgery did not significantly influence the final clinical outcome. No correlations were also found between MOCART 2.0 total scores or subscales and basal parameters, including age, BMI, and defect size. Moreover, no correlations were found between clinical scores (IKDC subjective and objective, and Tegner score) and MOCART 2.0 total scores or subscales at final follow-up.

**Figure 2.** Comparison of the Tegner scores achieved at a final follow-up of 12 years. In both groups, scores improved significantly from the pre-operative level to the final follow-up, even though the mean activity level remained significantly lower vs. pre-injury. No differences were found in sports activity levels between the two groups. FU, follow-up; Pre-inj, pre-injury; Pre-op, pre-operative.

**Figure 3.** MRI evaluation at 11 years' follow-up of a medial femoral condyle lesion treated with mosaicplasty in a 37-year-old man. Sagittal 3D "Cube" PD (**a**) and coronal PD FAT SAT (**b**) demonstrate a good filling of the defect (arrows) and a complete integration into adjacent cartilage. However, there is an inhomogeneous structure of the repair tissue with adjacent bony overgrowth.

**Figure 4.** MRI evaluation at 9 years' follow-up of a medial femoral condyle lesion treated with matrix-assisted autologous chondrocyte transplantation in a 39-year-old woman. Sagittal 3D "Cube" PD (**a**) and coronal PD FAT SAT (**b**) demonstrate a good filling of the defect (arrows) and integration into adjacent cartilage, a homogeneous structure and a normal signal intensity of the repair tissue. Moreover, no significant bony defect, bony overgrowth, or major subchondral changes were detected.


**Table 2.** Magnetic resonance Observation of Cartilage Repair Tissue (MOCART) 2.0 evaluation.

MACT, matrix-assisted autologous chondrocyte transplantation; MOCART, magnetic resonance observation of cartilage repair tissue; NS, not significant; SD, standard deviation.

**Figure 5.** Evaluation of the defect size effect on the International Knee Documentation Committee (IKDC) subjective score and the percentage Tegner Score improvement at a final follow-up of 12 years in both groups of patients. The asterisk (\*) indicates for the MACT group a tendency for higher subjective results compared to the mosaicplasty group for the defect area > 2 cm2 (p = 0.092). The double asterisk (\*\*) indicates a lower improvement in terms of Tegner Score for the defect area > 2 cm2 in the mosaicplasty group (p = 0.014).

#### **4. Discussion**

The main finding of this study is that both mosaicplasty and MACT techniques provided satisfactory clinical results at long-term follow-up in two homogeneous groups of patients affected by knee cartilage defects. Additionally, the MRI evaluation suggested a good and durable quality of repair tissue, although without correlations with the clinical outcomes. The comparative analysis of the two procedures suggested that while similar results were obtained for small lesions, MACT should be preferred versus mosaicplasty when addressing larger knee cartilage defects.

The overall clinical results reported in this study are in line with other recently published long-term studies, showing that both mosaicplasty and MACT may offer satisfactory and durable results when applied for the treatment of relatively small chondral lesions of knee femoral condyles [17,24–26]. In the largest available study of mosaicplasty that included 831 patients evaluated at 10 years of follow-up, Hangody et al. reported 92% good-to-excellent results with a low complication rate [26]. Similarly, Andriolo et al. performed a long-term evaluation at 15 years of follow-up in 113 patients treated with MACT, reporting a significant and stable clinical scores improvement. Even the failures rates reported in this study reflect the literature findings, with higher failure rates documented in patients treated with mosaicplasty technique compared to MACT. Pareek et al. [13] found that, at an average follow-up of 10 years, patients who underwent mosaicplasty had an overall failure rate of 28% and a reoperation rate of 19%, thus reflecting the results obtained in the current study (overall failure rate of 25% and a reoperation rate of 10%). The survival of MACT of the present study appears to be actually higher than previous findings. A systematic review reported a failure rate of 10.4% at mid-long term follow-up in patients treated with MACT [27]. However, these differences may be due to the different populations of patients considered; other studies demonstrated a higher failure rate in some patients categories, such as knee osteoarthritis and patellar defects [25,28], and these types of patients were excluded in the current study, where no failures were documented in the MACT group.

Long-term results of both mosaicplasty and MACT have also been evaluated in comparison with other cartilage repair techniques. Compared to microfractures, mosaicplasty resulted in significantly better outcomes in several studies at long-term follow-up [29–31]. In particular, even though both mosaicplasty and microfractures presented a worsening of results at 10 years, Gudas et al. [32] reported significantly better results in patients treated with mosaicplasty, which presented a higher

activity level, a lower failure rate, and a lower radiographic evidence of osteoarthritic degenerative progression. These results may be explained by the better quality of repair tissue provided by the use of autologous graft compared to fibrocartilage obtained with bone marrow stimulation techniques [32]. Similarly, MACT also presented better patient-reported outcome scores and lower failures compared to microfractures [33,34]. Additionally in this case, the excellent durability of MACT results could be explainable by the hyaline-like tissue restored [35], compared to the fibrocartilaginous repair tissue provided by microfractures as demonstrated by MRI studies [36,37].

The role of MRI in assessing cartilage repair tissue has been widely reported [38], and the imaging results of this study confirmed the literature findings in both treatment groups, with high MOCART 2.0 scores. However, several suboptimal imaging findings were also documented, both in terms of surface integrity and subchondral bone alterations, although these findings did not correlate with clinical results. To this regard, previous studies were controversial in regard the clinical significance of MRI findings. Tetta et al. [39] showed that mosaicplasty implants with a better imaging appearance also presented better objective and subjective scores, and Kon et al. [40] confirmed this correlation in patients treated with MACT. Conversely, Adrian et al. [41], while demonstrating a good-quality repair tissue at 10 years after MACT, found no correlation between clinical outcome and MRI results. These controversial findings have to be interpreted in light of a meta-analysis investigating the correlation between MRI and clinical outcomes after cartilage repair [42]. In fact, only 28% of the included studies found such correlation, leaving unanswered the question of whether MRI is reliable in predicting clinical outcomes after cartilage repair. Therefore, caution should be recommended when interpreting MRI findings and greater importance should be placed on the patient clinical evaluation. This is especially true in a comparative study such as the current one, aiming at determining the most suitable treatment in patients affected by cartilage lesions.

The present study compared both clinical and imaging outcomes of mosaicplasty and MACT at long-term follow-up. The results of these techniques in the literature, evaluated singularly, were confirmed, with good and long-lasting results and a low surgical failure rate. Mosaicplasty and MACT also showed similar overall clinical outcomes and comparable tissue repair quality at MRI evaluation. However, this comparative analysis underlined some interesting aspects. MACT presented better clinical results in relatively larger cartilage lesions, better objective findings, and a lower failure rate compared to mosaicplasty. These findings are inconsistent with the only available comparative analysis of these two techniques at short-term follow-up [19]. In detail, Clavé et al. reported a significantly greater symptom improvement 2 years after treatment for mosaicplasty compared to MACT in defects larger than 3.5 cm2, while no significant difference was found for smaller lesions [19]. On the other hand, in contrast to these results, Bentley et al. [43] reported a higher rate of failure in patients with large cartilage lesions treated with mosaicplasty compared to first-generation ACI at long-term follow-up, thus underlining that mosaicplasty is not a suitable option for the treatment of large lesions. This result was confirmed by other authors that investigated the significance of the lesion size for mosaicplasty technique, showing better clinical improvement in patients affected by small lesions [44,45]. An important aspect related to the lesion size is the number of plugs used to cover the defect being correlated with the clinical outcomes and the failure rate [45]. A limited number of plugs could correlate with better results probably because of the cylindrical shape of the grafts that does not allow optimal coverage of the lesion site by multiple grafts, leaving residual small uncovered areas that might impair or diminish the healing process [46,47].

Another important aspect that needs to be taken into account when evaluating cartilage procedures is the clinical profiling, aimed at determining the patient and lesion characteristics which may play a prognostic role on the final outcome [48]. Many studies evaluated several factors that can influence the failure rate of both these techniques. Among them, age, sex, etiology and size of the defect seem to modify the failure rate, with female patients, older patients and those affected by degenerative or larger lesions, complex cases, and patients undergoing salvage treatment being more at risk of failures [13,24,25,49,50]. Such profiling of patient characteristics which may influence the outcome were not possible in this series due to the low number of patients, although interesting findings could be underlined for the most suitable treatment indication based on the lesion size.

The current study has limitations that should be considered in the interpretation of the results. First of all, the retrospective design, with consequent lack of randomization, may entail an inclusion bias, with different patients treated with different procedures. Nevertheless, data were collected prospectively, and the strict patient selection by a statistician blinded to the study outcome allowed us to obtain two homogeneous groups to compare, thus allowing us to draw conclusions about the selected population. The absence of an intermediate evaluation is another limitation. Unfortunately, some patients of the two treatment groups were not evaluated at the intermediate follow-ups, hindering the possibility to better understand possible oscillations of the outcomes over time between the two techniques. Nevertheless, short and mid-term follow-ups of both procedures have already been reported in the literature. Secondarily, the relatively small number of patients evaluated hindered the possibility to confirm some correlations between outcome and influencing factors. Moreover, a slight (although not significant) heterogeneity was present between the two groups in terms of etiology, with some more degenerative cases in the mosaicplasty group. Nevertheless, the current survey represents the largest population evaluated in a comparative long-term analysis, and allowed meaningful clinical and MRI results, showing overall good results over time for both techniques, but lower results for mosaicplasty in lesions larger than 2 cm2. Other limitations are the absence of arthroscopic and histologic examinations, and of a radiologic follow-up, often considered the ideal technique for evaluating cartilage quality and OA evolution. Moreover, the imaging evaluation was not available for all patients, because some patients declined to undergo MRI exam. Nevertheless, the high rate of MRI evaluations at long-term follow-up and the use of the recent MOCART 2.0 tool offered important information regarding the state of the repair tissue and of the whole joint, leaving the analysis of the degenerative progression to future studies with longer follow-up times.

Alternative solutions have been explored to restore the articular surface while overcoming the drawbacks of mosaicplasty and MACT, aiming at lowering costs, simplifying the procedure, and improving long-term results [51,52]. The most recent developments involve the use of cell sources for one-step solutions or the application of biomaterials, such as a cell-free approach [53–56]. Nevertheless, while research is moving forward, it is fundamental to keep documenting the results of the most established procedures like mosaicplasty and MACT, to better understand how much and how long can patients benefit from these cartilage procedures, to detect the ideal candidates through a patient profiling, and to set a reference point to measure the potential of the other emerging solutions.

**Author Contributions:** Conceptualization, S.Z. and G.F.; methodology, M.B. and A.D.M.; data curation, D.R.; writing—original draft preparation, A.B. and L.A.; writing—review and editing, G.F.; supervision, A.D.M., M.B., G.F., and S.Z. All authors have read and agreed to the published version of the manuscript.

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

**Conflicts of Interest:** S.Z. has received institutional support from Fidia Farmaceutici, Cartiheal, IGEA Clinical Biophysics, Biomet, and Kensey Nash; grant support from I+; and royalties from Springer. 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. The other authors declare no conflict of interest.

### **References**


© 2020 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 (http://creativecommons.org/licenses/by/4.0/).

## *Article*
