Institute of Health and Biomedical Innovation, Queensland University of Technology 60 Musk Ave., Kelvin Grove, QLD, 4059, Australia
George W. Woodruff School of Mechanical Engineering, Georgia Institute of Technology, 315 Ferst Drive Atlanta, GA 30332, USA
Correspondence to: Prof. Dietmar W. Hutmacher, PhD, MBA, Institute of Health and Biomedical Innovation, Queensland University of Technology, 60 Musk Ave., Kelvin Grove, QLD 4059, Australia. Fax: +61-7-3138-6030. E-mail: firstname.lastname@example.org
Patients affected with traumatic osteochondral lesions, osteochondritis dissecans, and osteochondroses have osteochondral defects with limited spontaneous healing. When left untreated, these lesions cause pain and limited mobility, and eventually lead to degenerative conditions such as osteoarthritis. Current treatment strategies such as osteochondral autograft transplantation (OATS)/mosaicplasty and osteochondral allograft transplant have resulted in good cartilage/bone repair. However, these techniques can result in further complications such as donor site morbidity and graft rejections (Paul et al., 2009), and further degenerative changes in some patients (Hangody et al., 2010). By engineering osteochondral grafts using a patient's own cells and osteochondral scaffolds designed to facilitate cartilage and bone regeneration, osteochondral defects may be treated with less complications and better long-term clinical outcomes.
Over the last decade, there have been many new developments in various aspects of scaffold manufacturing. Computer-aided designs and manufacturing technologies are being used to fabricate personalized scaffolds for irregularly shaped defects (Hung et al., 2003; Swieszkowski et al., 2007). Materials used for scaffolds and matrices are becoming smarter and more versatile, and can be modified to incorporate bioactive peptides (Grafahrend et al., 2011). Even though scaffold manufacturing technologies are advancing at a rapid pace, no engineering strategies used to date can fully recapitulate the biochemical and physical characteristics of the native osteochondral tissue. While it is usually helpful to simplify the approach to in vivo repair from an engineering point of view, for a successful in vivo outcome, the biological complexities that take place within the joint must also be accounted for in the design. Furthermore, it is important to evaluate and analyze in vivo successes and failures to improve the design of future implants.
Osteochondral tissue has a heterogeneous, multilayered structure comprised of non-calcified cartilage (superficial, middle, deep zone), calcified cartilage, and subchondral bone. The developing trend in osteochondral tissue engineering is the utilization of multiphasic scaffolds to recapitulate the multiphasic nature of the native tissue (Keeney and Pandit, 2009). There are a number of different multiphasic approaches which involve different materials, cells, and bioactive molecules. In this review, we will discuss the emergence of multiphasic osteochondral tissue engineering concepts, then shift our focus toward in vivo successes and failures of multiphasic scaffolds to date and highlight future trends that are geared toward more clinically relevant osteochondral tissue engineering strategies.
PREVALENCE OF MULTIPHASIC APPROACH
Different types of biomaterials can be used to fabricate osteochondral scaffolds, and both hydrogels/matrices and polymeric scaffolds have been used in the past. Hydrogels are synthetic or natural polymeric matrices with an excellent ability to attract and retain water. Mechanical properties of hydrogels can be influenced by the polymer content and crosslink density. Hydrogels are used extensively in many different medical applications, including tissue engineering, and there is a good review on this topic (Drury and Mooney, 2003). Polymeric scaffolds can take on a variety of shapes and porosity, and they can be manufactured into woven/non-woven fabrics, meshes, sponges, and porous scaffolds with defined three-dimensional structures. Different types of polymers have been used for cartilage and bone tissue engineering scaffolds, and has been reviewed by Hutmacher et al (Hutmacher, 2000). The cartilage compartment is generally composed of hydrogels (collagen type I or II, hyaluronan, chitosan, alginate, gelatin, agarose) although biodegradable synthetic polymers such as polylactic acid (PLA), poly(lactic-co-glycolic acid) (PLGA), and polycaprolactone (PCL) have also been used for this application. The bone comportment is usually made from synthetic polymers or stiff materials (e.g., hydroxyapatite, tri-calcium phosphate, calcium polyphosphate, bioglass). There are good reviews available on commonly used biomaterials for scaffolds used in osteochondral tissue engineering (Martin et al., 2007; O'Shea and Miao, 2008; Keeney and Pandit, 2009), and they will not be covered in detail in this review.
Scaffolds can influence the development and structure of the engineered tissue, and there is an increased awareness that osteochondral tissue engineering concepts need to take the in vivo complexities into account in order to increase the likelihood of a successful osteochondral tissue repair. Tissue structure is often linked to tissue function, and from a tissue engineering perspectives, the ability to recapitulate the structural aspects of the tissue can be important in order to restore function. Osteochondral scaffold designs can be generally categorized as monophasic, biphasic, or triphasic (Fig. 2). Monophasic approaches feature constructs fabricated from one material or one type of composite material with no spatial variations. In cellular or biological perspective, monophasic scaffolds do not have more than one cell type throughout the construct, and have no variation in the distribution of bioactive molecules. On the other hand, biphasic or triphasic approaches utilize two or three different materials, composites, or architecture to create a multilayered structure. One material can also be used to create biphasic or triphasic scaffolds if significant depth-dependent variations in physical properties exist between the different layers. Seeding different types of cells in the different scaffold layers or having more than one biological environment through incorporation of bioactive molecules can also result in biphasic/triphasic scaffolds. While monophasic approaches are still being used in some osteochondral regeneration studies, research focus is currently shifting toward multiphasic designs. The major advantage of multiphasic approaches is the ability to provide physical or biochemical cues which specifically targets cartilage or bone development. Cartilage and bone have different structural, mechanical, and biochemical microenvironments, and there are obvious limitations when osteochondral scaffold designs do not address such differences. Although scaffolds do not necessarily have to replicate all such parameters, tailoring osteochondral scaffolds with tissue-specific architecture may aid in generating functional osteochondral constructs within a shorter timeframe.
Implantation of tissue-engineered osteochondral constructs in vivo presents various challenges that are absent in an in vitro environment. The implanted construct is exposed to many different biological, chemical, and physical constraints which greatly influence neo-tissue survival, development, and functionality. Cells within the implanted construct will be exposed to various soluble factors secreted by the other cells (e.g., chondrocytes, osteoblasts, and synovial cells), oxygen and nutrient gradient within the joint tissues, and the dynamic mechanical environment within the joint can not only contribute to the development of the osteochondral grafts, but also their failures. Therefore, it is important that follow-up and evaluation studies are performed on the scaffold designs that were particularly promising in vivo, and in case of failures, discuss how they can be improved for future success. This point also leads us to the fact that while extensive in vitro studies on scaffold properties and cell-culture work can provide some key implications about the scaffold, research efforts must continue toward in vivo studies in order to gauge its potential impact on osteochondral tissue regeneration.
Cartilage and bone have different healing timeframe and growth factors that directly influence tissue morphogenesis. By targeting them separately with biphasic distribution of growth factors, it is possible to accelerate and enhance the overall repair of the osteochondral defect. Integrating bioactive components into the scaffold designs have become one of the most interesting developments, and there is a recent review on this topic (Santo et al., 2013a, 2013b). Biomaterials can be synthesized or manufactured to provide not only the structural support, but also biological cues to influence cellular activity and tissue development. Incorporation of transforming growth factor beta (TGF-β1) and bone morphogenic protein-2 (BMP-2) in scaffolds using PLGA microspheres have shown sustained growth factor release over 3–6 weeks (Dormer et al., 2010; Reyes et al., 2012), and have resulted in bone development as early as week 6 post implantation in rabbit osteochondral defects (Dormer et al., 2012). A slightly different approach involves altering the biomaterials themselves to facilitate stronger affinity to bioactive peptides and growth factors to improve their retention (Grafahrend et al., 2011). Recently, chemically modified alginate hydrogels with increased affinity toward growth factors have been used to create biphasic osteochondral constructs with good in vivo outcomes (Re'em et al., 2012). This group reported that their bilayer constructs with affinity bound TGF-β1 and BMP-4 resulted in rapid cartilage and bone formation in as little as 4 weeks post-implantation, with good collagen type II distribution in the cartilage compartment. Another interesting and successful approach in this regard is to integrate biphasic scaffolds with growth factor plasmids to alter cellular activity. Chen et al. (2011) have used TGF-β1 and BMP-2 plasmids to influence cartilage and bone regeneration, respectively, in biphasic scaffolds made of gelatine, chitosan, and hydroxyapatite, and found that biphasic scaffolds with the plasmids had excellent osteochondral regeneration by week 4 in rabbit osteochondral defect model. With the current trend of combining biological functionality with multiphasic scaffolds, it is likely that osteochondral tissue engineering approaches will become much more comprehensive and multifaceted to augment the osteochondral repair in vivo.
Reports show that allowing osteochondral grafts to develop in vitro prior to in vivo implantation can have a positive influence in improving the quality of the repair following implantation in vivo. Miot et al. (2012) have reported that chondrocyte-seeded biphasic scaffolds that were pre-cultured for at least 2 weeks had significantly improved tissue morphology compared to those that were implanted without pre-culture following 8 months of in vivo implantation in a caprine model. Pre-culture in differentiation media may especially be beneficial when using multipotent progenitor cells or dedifferentiated chondrocytes or osteoblasts. In vitro culture in the presence of specific growth factors (e.g., TGF-β1/BMP-2) allows the progenitor cells to commit to either the chondrogenic or osteogenic pathway, inducing tissue-specific developments. This approach has been adopted by Sheehy, et al., and they have reported that pre-differentiation of agarose-encapsulated bone marrow stromal cells (BMSCs) in either chondrogenic or osteogenic medium (3 weeks) prior to biphasic assembly enhanced the accumulation of glycosaminoglycans (GAG) and calcium in vivo (Sheehy et al., 2013). The maturation of tissue engineered constructs in vitro allows them to gain matrix molecules such as GAG and collagen, which play important roles in withstanding the physiological mechanical load. Some modeling studies have shown that the maturity of tissue at the time of implantation has a large influence not only on the implanted construct itself, but on the surrounding cartilage, and affects collagen remodeling of the tissue (Nagel and Kelly, 2013). While culturing osteochondral grafts in vitro prior to in vivo implantation can be labor-intensive and time consuming, they can potentially influence the long-term repair, making them an important consideration in an in vivo study.
Need for Future Improvement
Analysis of mechanical properties
Clinically, an osteochondral injury from trauma usually occurs on the load-bearing part of the joint. As a result, in most animal models, osteochondral defects are created on the femoral condyles, which are subjected to various types of mechanical loading such as compression, shear, and hydrostatic pressure. It is commonly accepted that critical-sized osteochondral defects can induce significant degenerative changes to the surrounding cartilage and bone, possibly due to mechanical destabilization originating from from the defect region not being able to carry the load (Schinhan et al., 2012). In this regard, a newly developing osteochondral construct with inferior mechanical properties can also contribute to the mechanical imbalance near the defect region until their mechanical properties have matured (Khoshgoftar et al., 2013). Mayr, et al., have reported that the cartilage component of the osteochondral graft had only half the stiffness of the surrounding cartilage 6 months post implantation Mayr et al. (2013). The longer the osteochondral graft takes to fully mature in mechanical properties, the longer the surrounding cartilage is exposed to excessive load, which can contribute to degenerative processes. Unfortunately, few in vivo studies investigate the mechanical stiffness of the repaired tissue or the surrounding tissue. Considering the effort and cost of an in vivo experiment, it should be emphasized that thorough evaluation of the in vivo explants, above and beyond the histological means, could provide important data sets that can be used to improve osteochondral tissue engineering designs used in future studies.
Integration between the host tissue and the osteochondral graft is critical in the long-term survival of the implant. However, incomplete integration is frequently observed, especially in the cartilage compartment. There are a number of potential factors which can impede integration. PRG4, a lubricating molecule abundantly present in synovial fluid, has been shown to inhibit cartilage integration (Englert et al., 2005). In addition, studies show that cell death around the edge of the tissue has been found to hamper tissue integration (Gilbert et al., 2009). In order to facilitate cell-mediated bridging between the host cartilage and the engineered graft, chondrocyte-seeded collagen gel can be used to seal the gap between the host cartilage and the engineered graft (Pabbruwe et al., 2009).
It can also be postulated that the typical lack of collagen fiber orientation in the repaired cartilage also has a role in preventing strong integration at the cartilage level. The superficial zone of the cartilage in normal cartilage is aligned horizontally, parallel to the direction of joint articulation. However, within the repaired cartilage, this arrangement is often missing, hence the edge adjoining the native and engineered cartilage tissue is prone to breakage. Vertical orientation of collagens near the subchondral bone has been attributed to anchoring the cartilage tissue against large strains (Shirazi et al., 2008). Lack of collagen orientation in a dynamic loading environment of the joint is likely to have a role in in vivo failure of implanted constructs and diminished integration. Additionally, it is well-documented that cartilage stiffness is depth-dependent, and that superficial cartilage layer deforms much more than the deeper layers (Schinagl et al., 1997). In this regard, when the cartilaginous component of the osteochondral scaffolds lack the depth-varying deformation patterns, mismatched compressive strain levels between the cartilage and the implant is likely to result in increased shear stress at the interface region, causing a tear. Tissue engineering cartilage grafts with depth-varying compressive properties have also been proposed in the past (Klein et al., 2009), and can be incorporated in future osteochondral construct designs. Scaffold architecture and the presence of migratory cells within or immediately surrounding the osteochondral tissue graft will likely to play a key role in integration and tissue repair, and have important implications for future osteochondral tissue engineering approaches.
One of the problems often observed in vivo is that the implanted scaffold sinks into the bone (Fig. 3). This has been reported in studies using stiff scaffold materials (e.g., tantalum metal (Mrosek et al., 2010), PCL (Ho et al., 2010)) as well those using soft materials (e.g., alginate (Heiligenstein et al., 2011)) for their in vivo implants. Reasons behind scaffold-sinking most likely come from a few different factors. Some of the likely contributing factors include increased macrophage activity (Pei et al., 2009) as well as the resorption of bone on the bottom of the osteochondral defect through increased osteoclast activity related to mechanical stress (Duda et al., 2005). The types of cells seeded in the implanted osteochondral construct may also affect tissue development by attracting macrophages. One study which used xenogenic synovium-derived cells to seed the osteochondral implant reported positive staining of macrophages around the osteochondral graft which are likely to have caused scaffold sinking and ultimately implant failure (Pei et al., 2010). In order to improve the integration and functionality of the implanted osteochondral constructs, further investigation is needed to determine the exact causes of scaffold sinking.
Tissue-engineered osteochondral grafts do not always result in hyaline cartilage, and fibrocartilage developments have been reported in the past. Fibrocartilage refers to cartilage tissues that contain high percentage of collagen type I, which is not normally found in healthy hyaline cartilage. This type of cartilage has inferior mechanical properties, and has limited longevity in the demanding mechanical environment within the joint. It is one of the causes of treatment failures in patients that receive autologous cell implantations or subchondral drilling/microfracture, as a large percentage of such treatments result in the formation of fibrocartilage (Bentley et al., 2003; Williams and Harnly, 2007). While most investigators stain their tissue explants with Safranin-O (GAG) and hemotoxylene/eosin (nucleus, cytoplasm), it is important to also routinely stain or analyze for collagen type I and II (Fig. 4). Many of the in vivo osteochondral tissue engineering studies induce subchondral bone bleeding as they create full-thickness osteochondral defects, and with the implantation of the new scaffold, whether with or without the previous cell-seeding, the scaffolds may become infused with the blood from the subchondral bone. While there are advantages to this, mainly the recruitment of the progenitor cells, this could also lead to fibrocartilage formation and therefore it is important to perform systematic analysis to detect collagen type I in the repaired cartilage, and compare them to the untreated empty defect, which is most likely to be similar to the clinically used treatments such as subchondral drilling. It has been seen in the past that even when the repaired cartilage has good macroscopic morphology and GAG levels, it is possible to have high levels of collagen type I (Pei et al., 2010). Unfortunately, the majority of the recent in vivo studies do not report on collagen type I content in their repaired cartilage. Unless the newly implanted scaffolds can reduce fibrocartilage formation while improving overall tissue quality, it will be difficult to justify the efficacy of such treatment.
Furthermore, it is also important to investigate the structural organization of the collagen fibrils within the cartilage. While the cartilage may have decent GAG content, without the zone-specific collagen organization, the repaired cartilage grafts will not have similar mechanical properties as the native cartilage (Laasanen et al., 2003). Zone-dependent organization of collagen fibrils within the healthy articular cartilage is well documented, and such organization has a large influence on the mechanical properties of the cartilage and how it behaves under joint loading (Halonen et al., 2013). Polarized light microscopy can be used to visualize the birefringence from depth-dependent collagen fiber orientations within the cartilage tissue (Kiraly et al., 1997). More recently, small angle X-ray scattering (Moger et al., 2007), Fourier transform infrared spectroscopy (Boskey and Pleshko Camacho, 2007) and second harmonic generation multiphoton microscopy (Brown et al., 2003) have also been used to this end. It has proven difficult, however, to control how the collagen fibers are arranged and remodeled within the repaired cartilage, but its importance in the functionality of the repaired cartilage tissue is clear, and collagen fiber alignment needs to be examined when in vivo explants are evaluated. Apart from the role that collagen fibrils play in determining the mechanical properties of the cartilage, collagen fibers also help control proteoglycan (PG) swelling and retention within its framework (Poole et al., 2001). Lack of collagen network and its unique architecture may result in PG loss and tissue swelling (Khoshgoftar et al., 2013). One study in particular has reported that tissue engineered cartilage in minipigs begin to show depth-dependent collagen orientation at 26 weeks, and continues to develop until they reach native-cartilage like collagen orientation after 52 weeks (Paetzold et al., 2012). Since typical in vivo studies last for up to 6 months (26 weeks), it is likely that some collagen alignment is present within the tissue, and will offer additional information on the functionality of the repaired osteochondral tissue.
Natural healing of immature animals: Deciphering scaffold benefits
Immature animals have naturally excellent regenerative abilities, and there are a number of differences in the way osteochondral defects in immature animals heal compared to those of an adult animal. A study has shown that when an osteochondral defect was left untreated in an immature rabbit, it showed an excellent repair, comparable to normal cartilage, while the same untreated defect in mature rabbits did not heal (Mrosek et al., 2010). In this respect, it is important to state whether the animal used in the in vivo study is skeletally fully mature, as this may have even greater influence on the outcomes compared to the treatment groups that was being studied. In addition, it is also important to determine whether the osteochondral defect created is of critical size, that is, a defect that will not heal on its own without further treatment. This will serve as an important factor in deciphering the benefits of the newly implanted osteochondral construct, and it will make it more feasible to compare between other studies that have used similarly mature animal.
Scaling-Up Animal Models
Many different animal species have been used in the past for in vivo implantation of osteochondral constructs. Rabbits, pigs, sheeps, horses, dogs, minipigs, and even rats have been used for osteochondral defect models in the past (Table 1). The defect size is often limited by the size of the animal's joint, and while cost-effective, there are obvious limitations in extrapolating results from smaller animals to plan human clinical trials. As an example, average size of osteochondral defects made in rabbits and dogs were about 0.14 cm2 and 0.13 cm2, respectively (Table 2), while osteochondral defects in human patients were more than 10 times larger, ranging from 1.5–6 cm2 (Kon et al., 2011). While in vivo experiments using small animal models bring innovative osteochondral tissue engineering concepts one step closer to human clinical trials, it is important to follow up successful small animal experiments by scaling up and using animal models that have comparable joint dimensions and joint load. Animals such as sheeps, pigs, and horses have surgically created defect sizes which range from 0.29–0.79 cm2 and have average defect depths similar to humans at around 0.68–1 cm. Body weights of these animals are also either comparable or much heavier than humans, making them a better suited models to predict the outcomes in clinical trials.
Table 1. In Vivo Osteochondral Repair Evaluation in Studies Conducted Between 2009 and 2012
Table 2. Average Osteochondral Defect Area and Depth in Animal Models and Human Clinical Trials
In vivo osteochondral defect studies
Number of studies
Average defect area (cm2)
Average defect depth (cm)
0.14 ± 0.02
0.39 ± 0.11
0.29 ± 0.02
1.00 ± 0.91
0.46 ± 0.00
0.68 ± 0.25
0.01 ± 0.00
0.20 ± 0.14
0.13 ± 0.00
0.60 ± 0.00
0.16 ± 0.04
0.50 ± 0.00
0.28 ± 0.00
0.40 ± 0.42
0.79 ± 0.00
1.00 ± 0.00
Human (knee; N = 34; Average age = 35.3 ± 10.2 years)
The emergence of biphasic scaffold designs has enabled researchers to develop strategies to regenerate both cartilage and bone simultaneously, and there are many promising results to date. In the future, it is projected that increasing number of triphasic designs will emerge. Kon, et al., have already carried this concept to human clinical trials Kon et al. (2011), but it is still largely a novel approach. Triphasic constructs may target cartilage, calcified cartilage, and bone simultaneously, allowing the transitional layer to tie in cartilage and bone compartment smoothly. Additionally, triphasic concepts may also be interpreted as further developing the idea of stratified cartilage and incorporating them into osteochondral models. There are a number of groups that have developed multilayered cartilage constructs either with different chondrocyte subpopulations (Kim et al., 2003) or by using hydrogels with varying degree of stiffness to mimic the stratified nature of the cartilage (Nguyen et al., 2011). And it will be interesting to see whether osteochondral constructs with stratified cartilage design produce tissue grafts with similar properties as the natural cartilage.
Osteochondral tissue engineering is continuing to evolve, as are osteochondral scaffolds, at a fast rate. Active research is being carried out in order to fabricate smarter scaffolds. Osteochondral scaffolds are being designed to facilitate tissue-specific growth-factor delivery, mimic connective tissue ECM, be chondro- or osteo-inductive, and recapitulate the stratified nature of the osteochondral tissue through multiphasic designs. While there are considerable efforts in finding new ways to improve osteochondral scaffolds, evaluating their potential in a long-term in vivo study is equally important. In order to expedite our progress in developing successful tissue engineering strategies, our standards for in vivo experiments should be set higher. The scope of implant evaluation should not be limited to evaluating general morphologies and macroscopic/histological scores, but also include the assessment of collagen types, matrix organization, and mechanical properties, since their functionality is as important as their gross morphology. It is also highly encouraged that studies include appropriate controls (empty defects, cell-free group) or provide comparisons to current clinical treatment strategies (subchondral drilling, OATS, ACI) to highlight the benefits and novelty. While it is difficult to directly compare different tissue engineering strategies, standardized and systematic evaluation of in vivo experiments will surely benefit current efforts in osteochondral regeneration and promote efficiency in future bench-to-bedside concepts.