Osteochondral defects affect the articular cartilage and the subchondral bone and are frequently a result of traumatic events or degenerative processes (Davis et al., 2021). As cartilage presents limited natural regenerative potential, the pre-injury tissue configuration is rarely restored by spontaneous repair processes (Hunziker et al., 2015). Osteochondral defects are associated with relevant pain and limit the joint function and mobility of patients. Additionally, they might initiate degenerative processes in surrounding tissues, eventually leading to further degeneration of the complete joint (Hunziker et al., 2015). To stop the degenerative progression of osteochondral defects, a timely and effective treatment is of great importance. However, current clinical treatments are unable to restore healthy articulation or are associated with severe limitations (Davis et al., 2021), such as the need for multiple interventions or a triggering of degeneration in previously unaffected cartilage areas (Nukavarapu & Dorcemus, 2013; Hunziker et al., 2015). The development of scaffold-based tissue engineering (TE) strategies has the potential of supporting osteochondral defect healing, thereby complementing current clinical treatment options with a truly regenerative strategy.

The regeneration of osteochondral defects might be achieved by TE scaffolds imparting specific mechanical cues to guide the differentiation of mesenchymal stromal cells (MSCs) (Davis et al., 2021). In fact, numerous in vitro evaluations have shown that MSCs can sense and respond to mechanical stimulation (Delaine-Smith & Reilly, 2012; Schreivogel et al., 2019). Moreover, computational models employing mechano-biological rules to determine tissue formation have highlighted the importance of mechanics on the healing process of osteochondral defects. For example, the repair of osteochondral defects in minipigs was reproduced by describing tissue formation under specific ranges of minimum principal strain (Duda et al., 2005). Moreover, typical features of the in vivo repair pattern were obtained simulating tissue formation based on a mechanical stimulus, which was calculated from octahedral shear strain and fluid velocity (Kelly & Prendergast, 2005; Tortorici et al., 2021). To support the mechanics-dependent healing of osteochondral defects, scaffolds with different mechanical properties have been compared in vivo. For example, poly (lactide-co-glycolide) (PLGA) scaffolds with high and low stiffness were implanted in osteochondral defects in sheep, resulting in better subchondral bone formation with the stiffer scaffolds, although without improvements of cartilage formation (Schlichting et al., 2008). In another case, three PLGA scaffolds with different elastic moduli were tested in osteochondral defects in rabbit, observing that the two scaffolds with lower elastic moduli better supported the repair process compared to the stiffer one (Ikeda et al., 2009). In both cited examples, however, the different mechanical properties of the scaffolds were achieved by different scaffold porosities. Therefore, a clear distinction of mechanical stimuli from morphological cues (e.g., pore size) was not possible. To separately evaluate the influence of scaffold stiffness and morphology on osteochondral defect repair, an in silico model was recently developed (Tortorici et al., 2021). By investigating in silico the mechanical behavior of simple scaffold designs upon implantation into an osteochondral defect, it was possible to suggest scaffold properties that are supportive for the treatment of osteochondral defects. In fact, the computational results indicated that a scaffold with material elastic modulus in the low GPa range and an architecture reducing both compressive and radial displacements has the potential to improve osteochondral defect repair compared to an untreated defect (Tortorici et al., 2021). However, only non-degradable scaffolds were investigated in this model.

The ability to degrade in the environment of the body (see Table 1 for definitions) is often listed amongst the properties of an ideal scaffold or material for TE (Bose et al., 2012; Middleton & Tipton, 2000; Turnbull et al., 2018). In fact, a biodegradable material has the advantage of not requiring a second surgical intervention for its removal (Casalini, 2017), while avoiding concerns on its long term influence at the site of implantation, e.g., the elicitation of a foreign body response (Griffith, 2002) or the stress shielding of the adjacent tissues (Muschler et al., 2004). Material degradation can be even employed for the timely delivery of specific chemical cues or drugs (Casalini, 2017). However, the process of degradation may result in modifications of the chemical, morphological, biological, and mechanical properties of a material (Nair & Laurencin, 2007), and thereby needs to be carefully tuned for the individual application. Synthetic biocompatible and biodegradable polymers (see Table 1 for definitions) typically degrade by hydrolysis (Middleton & Tipton, 2000), i.e., by a shortening of the polymeric chains caused by a reaction with water molecules (Casalini, 2017). Depending on degradation kinetic and small molecule diffusion, two types of hydrolytic degradation are possible: surface erosion or bulk degradation (Casalini, 2017). In surface erosion, the hydrolytic reaction is faster than water diffusion into the material, resulting in a thinning of the scaffold features without a reduction in molecular weight (Middleton & Tipton, 2000; Woodruff & Hutmacher, 2010). Surface erosion is typically observed in polyanhydrides and polyorthoesters (Nair & Laurencin, 2007). In bulk degradation, the hydrolytic reaction is slower than water diffusion into the material, resulting in a uniform reduction of molecular weight in the scaffold without weight loss nor volume modifications (Woodruff & Hutmacher, 2010; Casalini, 2017). Polyesters are known to degrade by bulk degradation with acidic degradation by-products (Casalini, 2017). If the production of the acidic degradation by-products is faster than their diffusion away from the material, these acidic by-products speed up the reduction in polymeric molecular weight in the bulk of the device (Middleton & Tipton, 2000), a phenomenon known as autocatalysis. The establishment of autocatalysis depends not only on polymer chemistry, but also on architectural features of the device, e.g., its thickness (Grizzi et al., 1995). Bulk degradation with autocatalysis has been observed in vivo in devices made from poly (lactide-co-glycolide) (PLGA) and poly (lactic acid) (PLA) (Therin et al., 1992), but not from poly (ε-caprolactone) (PCL) (Woodruff & Hutmacher, 2010). The experimentally observed degradation behaviors of some of the polymers that are commonly investigated for TE applications are summarized in Table 2.

As architecture is, next to the material type, the most important factor determining the mechanical properties of a scaffold, e.g., its compressive modulus (Marchiori et al., 2019), modifications in mechanical behavior can be expected as a consequence of a surface erosion process. Moreover, the molecular weight of polymers is closely related to their mechanical properties (Nunes et al., 1982). Consequently, a reduction in molecular weight resulting from a bulk degradation process (with or without autocatalysis) may modify the mechanical properties of a scaffold, as previously observed experimentally, e.g., in PLGA (Wu & Ding, 2004) and PCL (Lam et al., 2008) scaffolds. Given the link between scaffold-dependent mechanical cues and tissue formation discussed above, it is important to evaluate how modifications in scaffold mechanical properties due to degradation would influence the healing process. However, an experimental investigation of this phenomenon presents significant technical challenges, such as the need for long term in vitro and in vivo experiments (Pan, 2015). Computational models can offer first evaluations in a simplified environment, providing indications for future, more complex experimental tests. A number of computational studies have already explored tissue growth in presence of a degradable scaffold in the context of bone formation (Adachi et al., 2006; Chen et al., 2011; Gorriz et al., 2015). To the best of our knowledge, such an evaluation is so far lacking for osteochondral defects.

Here, we employed a previously developed computational model of scaffold-supported osteochondral defect repair (Tortorici et al., 2021) to study the consequences of scaffold degradation on the mechanics-dependent repair process. To do so, an “artificial” scaffold with strut-like architecture was modelled, whose material and architectural properties prior to degradation were those of the simplest design fostering an improved osteochondral defect repair compared to the untreated case, as previously established (Tortorici et al., 2021). The scaffold was defined “artificial” because its axisymmetric architecture and its material properties do not model a specific scaffold nor material currently investigated for tissue engineering. Three types of polymeric scaffold degradation, i.e. surface erosion, bulk degradation, and bulk degradation with autocatalysis, as well as different degradation rates were studied. Although ceramic (Marques et al., 2021) and metallic (Lv et al., 2021) degradable scaffolds have also been developed, their degradation modalities are different from the ones of polymeric materials and were not investigated here.

An iterative computational model simulated the mechanics-dependent tissue formation within an osteochondral defect implanted with a scaffold in a knee femoral condyle. The model represented a focal osteochondral defect resulting from surgical intervention to treat damaged chondral or bone tissue, e.g., in consequence of trauma, early osteoarthritis or disease-related bone lesions. Therefore, the tissues outside the defect were simulated as healthy. A detailed description of the model has been provided elsewhere (Tortorici et al., 2021) and is here briefly summarized. Amongst the scaffolds evaluated in the cited model, the scaffold with the simplest architecture that resulted in improved repair compared to the untreated osteochondral defect was selected for the present investigation. As an additional feature compared to the previously published model, scaffold degradation was implemented here, as described in the following sections.

The influence of polymeric scaffold degradation on the mechanics-dependent repair of osteochondral defects was investigated in silico by simulating three modalities of hydrolytic degradation: surface erosion, bulk degradation, and bulk degradation with autocatalysis. For each degradation modality, three speeds were investigated: fast, matched, and slow, causing the complete degradation of the scaffold in 25, 50, and 100 days, respectively. The repair outcome was evaluated at day 125. Moreover, bulk degradation with an exponential loss of mechanical competence was implemented, simulating the experimentally-observed degradation rates of porous scaffolds produced from PCL or PDLLA (which were reported to have comparable exponential degradation rates) and from PLGA. The following sections report the results obtained for each degradation modality and each degradation speed.

Scaffold-based tissue engineering strategies aim at supporting the healing of osteochondral defects by overcoming the limitations of current clinical treatments (Davis et al., 2021). However, an identification of the ideal mechanical and architectural features of a scaffold for osteochondral defect healing is still lacking. Promising mechanical and architectural properties of a scaffold for osteochondral defect healing were previously identified in silico by employing an axisymmetric model of a knee femoral condyle, in which tissue formation was determined by a mechanical stimulus computed from octahedral shear strain and fluid velocity (Tortorici et al., 2021). However, this model did not provide any insights concerning the influence of scaffold degradation, which has the potential to alter both the mechanical and the architectural characteristics of a device. In this study, the cited model was further developed to simulate the degradation of a synthetic polymeric scaffold with strut-like architecture by three modalities (surface erosion, bulk degradation, and bulk degradation with autocatalysis) and with three different degradation speeds (faster, equal or slower than the tissue repair process). The in silico evaluations suggested that, in the absence of autocatalysis, the speed of scaffold degradation, rather than its modality, is the factor having a predominant influence on the mechanics-dependent tissue formation. Specifically, times of full scaffold degradation longer than the expected time of completion of the tissue repair process were found to foster the best repair outcomes in the investigated model.

Initially, the three different modalities of polymeric scaffold degradation were compared to assess whether one of them presented particular advantages or disadvantages for the repair of osteochondral defects. Specifically, simplified algorithms with linear reductions in either volume fraction or elastic modulus for surface erosion and bulk degradation (with or without autocatalysis), respectively, were investigated. For bulk degradation and bulk degradation with autocatalysis, the scaffold begun the erosion process (leading to the complete disappearance of the device) when its material elastic modulus reached 10% of the initial value. Thus, the simplified algorithms reproduced the experimental observations that surface erosion causes thinning of the scaffold features without alterations of molecular weight (here considered directly proportional to the material elastic modulus, see Equation 4) and that bulk degradation results in a homogeneous reduction in molecular weight, and thereby of elastic modulus according to Equation 4, without alterations in the volume of the device up to an advanced state of degradation (Middleton & Tipton, 2000; Casalini, 2017). When autocatalysis was modelled, the elements in the bulk of the scaffold struts had a faster reduction in molecular weight, i.e. in elastic modulus, than those on the surface, also in this case reproducing experimental observations (Therin et al., 1992). Therefore, surface erosion and bulk degradation resulted in a transfer of the load-bearing function from the scaffold to the newly formed tissues in the model by two different mechanisms. In surface erosion, the biological tissues could gradually grow in areas previously occupied by the scaffold, whose load-bearing ability was reduced by the physical disappearance of the scaffold material. In bulk degradation, the amount of scaffold material remained unaltered for the majority of the degradation process and the transfer of the load-bearing function to the biological tissues happened by a softening of the scaffold material itself.

Previous analyses performed with the same computational model studied the mechanics-dependent repair process of an untreated osteochondral defect and of an osteochondral defect implanted with a non-degradable scaffold, which had the same mechanical and morphological properties of the scaffold investigated here (Tortorici et al., 2021). In the mentioned study, the repair of untreated osteochondral defects was predicted to develop fibrous tissue at the articular surface, with bone resorption mixed with bone formation in the middle and proximal regions, and lateral bone apposition. The non-degradable scaffold fostered an improved repair, as cartilage formed in most of the defect (and specifically at the articular surface) and a thin subchondral bone layer was established throughout the proximal base, although fibrous tissue was predicted to form at the interface with the surrounding healthy tissues. The improved repair outcome fostered by the scaffold was ascribed to its mechanical support, which reduced the strain in the defect, and to its architecture, which enabled load transmission to the defect base.

When a fast bulk degradation of the scaffold was implemented, the repair outcome was analogous to the one predicted for the untreated osteochondral defect (Figure 4B left), indicating that the mechanical support provided by the scaffold lasted a time too short to significantly steer tissue formation. On the other hand, both matched and slow bulk degradation resulted in a repair outcome similar to the non-degradable scaffold, although with higher amounts of fibrous tissue (Figure 4B middle and right, respectively). The increased formation of fibrous tissue could be ascribed to higher strains generated by the progressive decrease in stiffness of the scaffold (Figure 4C). Interestingly, the mechanical failure of the scaffold was never predicted when bulk degradation was implemented, independently of the degradation speed. This result was ascribed to the gradual and homogenous change in mechanical properties of the scaffold, which avoided the excessive straining of individual scaffold elements.

When the scaffold degraded by surface erosion, the fast degradation speed was predicted to induce mechanical failure of the scaffold (Table 6). As the failure occurred 1 day prior to complete scaffold degradation, it could be ascribed to the low thickness of the struts and their extremely increased porosity, which derived from the reduction in the volume fraction occupied by the scaffold material (Figure 3C). Mechanical failures were observed neither with the matched nor with the slow surface erosion. The repair outcome with the matched surface erosion was intermediate between the ones previously obtained in the untreated defect and with the non-degradable scaffold (Tortorici et al., 2021). In fact, a continuous subchondral bone layer formed and cartilage was found in the middle of the defect (Figure 3B left); however, the articular surface was occupied by fibrous tissue. Thus, the scaffold undergoing matched surface erosion did not provide sufficient mechanical support to the defect to avoid major fibrous tissue formation. Interestingly, the repair outcome with slow surface erosion was analogous to the one with slow bulk degradation (compare Figure 3B right to Figure 4B right). Although the overall stiffness of surface-eroding and bulk-degrading scaffolds was comparable between all imposed degradation speeds throughout the simulation (compare Figure 3C and Figure 4C), a similar repair outcome was achieved only with the implementation of the slow degradation speed. This result seems to indicate that the modality of degradation (either surface erosion or bulk degradation without autocatalysis) for scaffolds with strut-like architecture does not have an influence on the mechanics-dependent repair process, as long as the scaffold provides mechanical support for a sufficiently long time. Notably, such a sufficiently long time was identified in this computational model not as the time of expected defect healing, i.e. 50 days, but as long as 100 days. The similarity of overall scaffold stiffness during bulk degradation and surface erosion observed in this model was ascribed to the low thickness of the strut-like architecture, where the strut width was composed of 50% surface and 50% bulk elements. Thus, the reduction in scaffold material during surface erosion had a comparable influence on the overall mechanical properties to a homogenous reduction in material elastic modulus. This result is not expected to hold true for bulkier devices, for which surface erosion might have a reduced influence on the overall mechanical properties. It is important to notice that when both surface erosion and bulk degradation (without autocatalysis) with slow degradation speed were implemented, a layer of fibrous tissue developed at the interface between soft and hard tissues, probably deriving from increased strain values in this area resulting from the progressive loss of mechanical support from the scaffold. To avoid this effect, the architecture of the scaffold could be modified to one that provides a higher resistance against deformation in the early phases of degradation: for example, a grid-like architecture that reduces displacements both in the direction of the applied load and perpendicular to it has been previously indicated as favorable (Tortorici et al., 2021).

The results of the computational model clearly indicated that the instauration of autocatalysis is critical, as the mechanical failure of the scaffold with strut-like architecture was consistently predicted prior to full repair of the osteochondral defect, independently of the imposed degradation speed (Table 6). Such a markedly negative effect depended also on the architecture of the scaffold, which was composed only of thin (0.5 mm thickness) vertical struts (Figure 1A). As the bulk of the scaffold struts was completely degraded, the two surface layers of each strut effectively became independent pillars with the same height of the original strut (5 mm), but a much reduced thickness of 0.125 mm, corresponding to the width of one element in the mesh of the finite element model. This alteration of the height-to-thickness ratio, combined with the progressive decrease in material elastic modulus, made the scaffold struts more susceptible to buckling phenomena or to breakages due to excessive deformation. In particular, buckling strongly depends on the aspect ratio of the loaded body and may cause failures even if the yield stress of the material has not been exceeded (Beer et al., 2015). Therefore, scaffold architectures different than the one investigated here might better sustain loads even in presence of autocatalysis, for example if featuring connecting elements between vertical struts and/or having oblique rather than vertical struts. Nonetheless, the transformation of the scaffold into a structure consisting only of a thin outer layer, and the deriving reduction in mechanical competence, could be expected when employing materials that undergo bulk degradation with autocatalysis. The precise thickness of the remaining outer layer depends on several factors, e.g. the diffusion speed of the involved molecules or oligomers and the cleavage rate of the chemical bonds (Cameron & Kamvari-Moghaddam, 2008).

Bulk degradation without autocatalysis was also modelled as an exponential decrease in molecular weight (and thereby in elastic modulus, see Equation 4) based on the available literature data. This analysis enabled an evaluation of the influence of degradation that came closer to an experimental case. Specifically, the degradation behavior of porous scaffolds made from PCL or PDLLA and from PLGA was implemented. When the PCL- or PDLLA-like degradation was studied (slow exponential degradation rate), the complete disappearance of the scaffold was calculated to happen after more than 2 years. Therefore, the scaffold material was still present at the last investigated time point (day 125). However, tissue formation had already reached equilibrium at day 75, meaning that variations in tissue distributions from the results here reported might be expected only after the onset of erosion. As the scaffold material maintained a high elastic modulus (>600 MPa, Figure 5C) throughout the investigated time, the mechanics-dependent repair of the osteochondral defect (Figure 5B right) was analogous to the one obtained with the non-degradable scaffold (Tortorici et al., 2021), which was regarded as a positive outcome. It remains to be ascertained whether the onset of erosion would drastically alter this result. However, in the in vivo case, the production of extra-cellular matrix would likely stabilize the tissue configuration that was built during the long time of scaffold degradation. Thus, the repair tissues would gradually become independent of the mechanical support of the scaffold, reducing the risk of large variations in tissue distribution in consequence of the full degradation of the scaffold material.

A similar repair of the osteochondral defect was predicted with the material degrading as a porous PLGA scaffold (fast exponential degradation rate), although more fibrous tissue formation and regions of bone resorption were predicted (Figure 5B left). In this case, the degradation was faster and the scaffold material had completely disappeared at the last investigated time point (day 125, Figure 5C). The results obtained with the experimentally-derived bulk degradation rates confirmed the observation performed with the linear degradation rates, i.e. that long times of scaffold degradation are particularly advantageous to the mechanics-dependent tissue formation.

A verification of the computational results by comparison with the in vivo tissue formation in osteochondral defects is challenging. Although the in vivo degradation behavior of scaffolds or biomaterials has been often studied in subcutaneous (Lam et al., 2008), subdermal (Pitt C et al., 1981; Pitt G et al., 1981), intramuscular (Lam et al., 2008; Therin et al., 1992) or calvarial (Lam et al., 2008) defects, extensive evaluations of scaffold degradation in osteochondral defects, and thereby of degradation-dependent tissue formation, are, to the best of our knowledge, still lacking. When testing biodegradable or bioresorbable scaffolds for osteochondral defect healing, the presence or absence of scaffold remnants at a specific time point is a frequently reported observation on the in vivo degradation behavior of the specimens (Athanasiou et al., 1997; Barron et al., 2015; Holland et al., 2005; Hsieh et al., 2018; Ikeda et al., 2009; Reyes et al., 2012; Schlichting et al., 2008). Although important to assess the time of complete disappearance of the investigated device, this observation is not sufficient to establish a comparison with the results of the here presented model in terms of mechanics-dependent tissue formation. Some studies offer an evaluation of the in vivo degradation of scaffolds in osteochondral defects based on previous in vitro tests. For example, a two-phase PLGA scaffold implanted in goat failed to support tissue growth in the center of the osteochondral defects, an outcome ascribed to the faster degradation of the scaffold core due to the establishment of autocatalysis, as observed in vitro (Athanasiou et al., 1997). However, some authors reported differences in the in vivo degradation of the scaffolds compared to the results expected from previous in vitro evaluations (Holland et al., 2005), or that expected differences in degradation amongst specimens could not be confirmed in vivo by visual inspection only (Wang et al., 2014). In another study, two types of PLGA scaffolds were tested in sheep, with the stiffer scaffold with slower degradation resulting in improved subchondral bone formation (Schlichting et al., 2008). Interestingly, the authors suggested that mechanical failures due to the faster degradation of the softer scaffold may have negatively influenced tissue formation, thereby supporting our in silico predictions. However, the differences in the initial stiffness and porosity of the investigated scaffolds prevent a clear comparison with our model in terms of mechanics-dependent tissue growth. Therefore, the results here obtained in silico can be considered as an indication, while further experimental assessments are needed to confirm these findings. For example, direct comparisons between in vivo and simulated tissue repair have been reported for untreated osteochondral defects (Duda et al., 2005) and scaffold-supported large bone defects (Perier-Metz et al., 2020) and could be in future performed also for osteochondral defects treated with biodegradable architectured scaffolds.

As limitation of the here presented model, the “artificial” scaffold configuration must be mentioned. The scaffold was defined “artificial” because it did not model a specific polymer nor scaffold architecture currently investigated in tissue engineering, but its simplified properties were shown to promote regeneration in a previous computational study (Tortorici et al., 2021). The “artificial” scaffold in each investigated case had the same mechanical and architectural properties in its non-degraded state, but the possibility to degrade by each one of the three investigated modalities in turn. This situation was also “artificial” because, in reality, the degradation modality of a polymeric device is determined by both its chemical and architectural features (von Burkersroda et al., 2002). Thus, polymeric scaffolds degrading by different modalities can hardly be expected to have equal properties prior to the onset of degradation. For example, changes in architecture might be required to obtain the same overall stiffness in scaffolds produced from different materials. Further limitations of this study include the simplified degradation behaviors that were implemented, as well as the evaluation of only mechanics-related phenomena. A first simplification in the implemented degradation behaviors was the clear-cut distinction between surface erosion and bulk degradation, while in reality polymeric degradation might involve a combination of the two modalities (Göpferich, 1996). A second simplification was the direct proportionality between decrease in polymeric molecular weight and decrease in material elastic modulus assumed here. Although the loss of mechanical competence is a degradation-related phenomenon, the kinetics of molecular weight reduction and of elastic modulus decrease might differ (Göpferich, 1996). For example, in highly porous PLGA scaffolds, the first stages of in vitro degradation were characterized by a constant or slightly increased elastic modulus, while the molecular weight was already subjected to reductions (Wu & Ding, 2004). Furthermore, the linear algorithms and the imposed degradation speeds employed for the comparison of the three degradation modalities were a third simplification, which has been addressed in the second part of this work. In fact, bulk degradation has been reported to follow an exponential behavior for several polymers, such as PCL, PDLLA, and PLGA (Pitt C et al., 1981; Pitt G et al., 1981; Wu & Ding, 2004). Moreover, the degradation coefficients measured experimentally, e.g. for PCL and PDLLA, would cause a much slower scaffold degradation than the one modelled with the fast, matched, and slow degradation speeds, as shown when the experimentally-derived bulk degradation was implemented (Figure 5C). Concerning the evaluation of the influence of degradation phenomena on osteochondral defect repair, this work focused on mechanics-dependent tissue formation. However, the biological influence of scaffold degradation might derive also from other factors, such as local variations in pH and the release of degradation by-products (Suganuma & Alexander, 1993; Taylor et al., 1994; Sung et al., 2004). Therefore, the here presented model does not provide a complete evaluation of the complex events involved in polymeric scaffold degradation and the consequent tissue formation. Nonetheless, the model offers valuable insights on the mechanical influences that can be expected from the degradation of a synthetic polymeric scaffold with strut-like architecture in osteochondral defects. These insights could be used as indications in the selection of biodegradable or bioresorbable materials for scaffolds to support the healing of osteochondral defects. The results of this study are expected to apply also to defects of greater width implanted with a scaffold composed of the same repetitive unit. In fact, a previous in silico model showed similar mechanics-dependent repair patterns in untreated osteochondral defects with radii of 5 mm, 7 mm, and 9 mm, although with more fibrous tissue and less bone formation with increasing defect width (Kelly & Prendergast, 2005). On the other hand, the repair outcome supported by degradable scaffolds with non-repetitive architectural units and/or different morphologies should be specifically evaluated. Particularly, scaffolds obtained via production techniques that do not enable a precise control on the resulting material distribution might generate diverse local mechanical environments in the defect, e.g., due to walls of varying thickness or pores of different sizes. However, the simulation possibilities of the model are currently limited to scaffolds with axisymmetric geometries, thereby precluding an accurate assessment of devices with non-uniform properties in three dimensions. This issue could be addressed in future by employing a three-dimensional non-axisymmetric model.

The here-presented model could be further developed to investigate the role of scaffold-supported tissue formation also for specific subpopulations, e.g., differing in age or sex. In fact, evidence suggests that age and sex might result in differences in tissue growth, e.g., when comparing chondrogenesis and osteogenesis of MSCs seeded in decellularized extra-cellular matrix taken from new-born, juvenile, and adult rabbits (Wang et al., 2020), and in the progression of diseases, e.g., osteoarthritis (Urquhart et al., 2008). The mechanobiological rules employed to simulate tissue formation may be adapted to reproduce the experimental outcomes obtained for various groups, as previously done in the context of bone regeneration when comparing adult and elderly mice (Borgiani et al., 2019). To specifically investigate osteochondral tissue affected by osteoarthritis, additional modifications of the model, such as its geometry and the assigned material properties, might be required. In fact, osteoarthritis has been observed to influence the properties of all the joint tissues and to cause structural changes, e.g., the narrowing of the joint space (Davis et al., 2021).

Taken together, our results suggest that, amongst the three possible degradation modalities of synthetic polymeric scaffolds with strut-like architecture, bulk degradation with autocatalysis and bulk degradation without autocatalysis cause the highest and the lowest mechanical instability of the scaffold, respectively (Figure 6). As the instauration of autocatalysis consistently resulted in the prediction of mechanical failures, we propose that the choice of material and architecture of a scaffold for osteochondral defect repair should aim at avoiding this phenomenon. Moreover, the results of the simulation indicate that scaffolds with strut-like architectures degrading by surface erosion or by bulk degradation without autocatalysis can provide an equally adequate support to the mechanics-dependent repair of osteochondral defects, but only if the scaffold material degrades in a much longer time than the time of expected completion of the repair process.