Chemical and Petroleum Engineering, University of Kansas, Lawrence, KS, USA
Correspondence to: Michael S. Detamore, Department of Chemical and Petroleum Engineering, University of Kansas, 1530 West 15th Street, 4132 Learned Hall, Lawrence, KS 66045, USA. E-mail: firstname.lastname@example.org
Hydrogels used in biomaterials-based tissue engineering to regenerate articular cartilage offer great promise to address cartilage injuries (Azuma et al., 2007; DeKosky et al., 2010; Gong, 2010; Huang et al., 2008; Sant et al., 2010). Improving mechanical integrity, including both the deformation and failure of hydrogel-based constructs, is of great importance to cartilage regeneration. Although mechanical failure testing in outcome analyses is of crucial clinical importance regarding the success of engineered constructs, mechanical testing of hydrogel-based constructs to date has focused primarily on deformation rather than failure properties (Krupa et al., 2010; Lam et al., 2006; Langrana et al., 2010; Suljovrujic et al., 2011; Wang et al., 2011; Weichold et al., 2011).
Failure properties can be characterized by apparent fracture toughness, which reflects how much energy the material must absorb to fracture (Abdurrahmanoglu et al., 2009) and governs the response of materials to crack propagation (Stok and Oloyede, 2007). There exist various methods to determine the apparent fracture toughness in metals and plastic films, such as the single-edge notch test, trouser tear test, indentation test, etc. Because of the lack of much-needed studies on failure properties of hydrogel-based constructs, the overall objective of this study was to develop a method to evaluate comparable apparent fracture toughness for both articular cartilage and hydrogels in cartilage tissue engineering.
There are many standard fracture toughness tests for materials of various properties. One goal in standardized fracture toughness tests is to have a specimen size sufficiently large that the geometry of the specimen does not influence the results. In a tough anisotropic composite biological material such as cartilage, the anatomy of the cartilage inherently limits specimen size and shape. To compensate for this shortcoming, one technique that can be used is to reproduce the state of stress found in a defect in cartilage through use of an identically-shaped comparison specimen. In this way, the values of fracture toughness obtained for cartilage can be directly compared to those for other materials, such as hydrogels. Such a methodology is necessary to achieve the goals of this field, which is to create an effective hydrogel for cartilage tissue engineering. Ideally, the deformation properties should match those of articular cartilage, and the fracture properties should meet or exceed those of articular cartilage. While deformation properties (e.g. aggregate modulus, dynamic moduli) are well established for hydrogels and cartilage, and have heretofore been the primary focus of cartilage tissue-engineering studies with hydrogels, fracture properties are also important and stand alongside the modulus as an important design parameter for mechanical performance.
We recently reviewed fracture mechanics studies of both hydrogels and cartilage with the objective of identifying a common ground between these two disparate fields (Xiao et al., 2013). We concluded that among apparent fracture toughness tests, the modified single-edge notch test was preferred for application to both hydrogels and articular cartilage because it can avoid overstressed factors induced by grips, and it is derived from the single-edge notch test, which has solid fundamental models to analyse data. Chin-Purcell and Lewis (1996) first invented the modified single-edge notch test to measure the toughness of bovine patellar cartilage, based on the work of Mai and Atkins (1980) and Srawley and Gross (1967). Adams et al. (2003) supplemented their work by suggesting that the thickness of the specimen did not affect apparent fracture toughness. However, the models for analysing apparent fracture toughness data were geometry-dependent. Thus, after reviewing related ASTM standards, we adopted ASTM D 5045–99 (2007), a standard test for fracture toughness of plastics, to evaluate the apparent fracture toughness for both articular cartilage and hydrogels in cartilage tissue engineering.
We recently introduced interpenetrating networks (IPNs) with drastically improved mechanical performance relative to their constituent networks of agarose and poly(ethylene glycol) diacrylate (PEG-DA), capable of successful encapsulation of viable chondrocytes (DeKosky et al., 2010; Ingavle et al., 2012, 2014). However, mechanical properties were evaluated with compression, and it was unknown whether IPNs would also maintain superior failure properties following a testing method more relevant to cartilage fracture mechanics studies. The objective of this study was to develop a new method to not only compare the fracture toughnesses of the IPN and PEG-DA (agarose was not tested due to its low fracture strain) but, perhaps more importantly, to enable direct comparison of the hydrogels to cartilage, establishing a quantitative benchmark or ‘target’ for hydrogel toughness.
It must not escape mention that cartilage failure in osteoarthritis is typically considered as a biological phenomenon following an initial impact injury, characterized by a cascade of signalling events over an extended period of time that result in the degeneration of the cartilage structure and thus the loss of mechanical integrity. However, in the fabrication of hydrogels in cartilage repair, the hydrogels must be able to withstand the mechanical environment if there is hope of patients returning to weight-bearing activity without extended delay, and in this context we therefore examined cartilage as a material. Based on the modified single-edge notch test, we established the groundwork for linking methodologies between fracture testing of cartilage and hydrogels in an effort to evaluate fracture properties for hydrogels in cartilage tissue engineering. Our hypothesis was that interpenetrating networks (IPNs) of agarose and poly(ethylene glycol) diacrylate (PEG-DA) would be significantly tougher than PEG-DA alone.
2 Materials and methods
2.1 Cartilage specimen preparation
Hog ankles from eight different individuals (n = 8), all males from either York or Barron strains, 5–7 months old and weighing 97–195 kg, were obtained from a local slaughterhouse. The ankles were carefully opened within 24 h of the hogs’ death. The joint compartments were opened (Figure 1), wrapped in Kimwipes, soaked in 0.01 m phosphate-buffered saline (PBS) containing 0.138 m sodium chloride and 0.0027 m potassium chloride, and stored at −20°C (Allen and Athanasiou, 2005) (storage time < 1 month). Cartilage specimens (thickness 0.901–1.185 mm; Figure 2) were sectioned from the central portion of each ankle of the eight different hogs using a precision diamond saw (Isomet 1000, Buehler, Lake Bluff, IL, USA).
2.2 Hydrogel specimen preparation
An effort was made to make the hydrogel geometries as similar as possible to the articular cartilage specimen, so as to enable apparent fracture toughness comparisons as closely as possible between cartilage and hydrogels. Gels were synthesized following the basic methods of our previous work (DeKosky et al., 2010; Ingavle et al., 2012, 2013). Briefly, 2-hydroxyethyl agarose (type VII) was obtained from Sigma-Aldrich (St. Louis, MO, USA). PEG-DA (MW 6000 Da) was obtained from SunBio (Anyang City, South Korea). The photoinitiator Irgacure 2959 (I-2959) was purchased from Ciba (Basel, Switzerland). Agarose powder (0.2 g) was added to 10 ml PBS and autoclaved for 30 min to yield a 2% w/v agarose solution. When the agarose had cooled to 39°C, it was pipetted into rectangular silicon rubber moulds (10 mm wide × 20 mm long × 1 mm high) between glass plates. After 10 min cooling at 4°C, the gels were removed and added to a reservoir of PBS to equilibrate for at least 24 h before synthesizing PEG-DA and IPNs. A solution of 0.1% w/v I-2959 photoinitiator in deionized (DI) water was dissolved in 20% w/v PEG-DA in PBS at room temperature. One rectangular agarose gel (10 mm width, 20 mm length, 1 mm height) was cut evenly into eight pieces and placed into a 2 ml Falcon tube. 1 ml monomer solution was added into the 2 ml Falcon tube to soak those eight pieces under constant agitation using a rocker for 6 h. The eight soaked pieces were then placed four at a time into a rectangular silicon mould between optical glass microscope slides (four pieces/mould) and the surrounding space was filled with monomer solution. The gels were exposed to ultraviolet light for 5 min on each side, using 312 nm light, 3.0 mW/cm½ (XL-1000; Spectronics Corp., Lincoln, NE, USA). Four pieces of IPNs/mould were cut using a razor blade. The PEG-DA gels were made by the same exposure procedure, except for filling the moulds with only monomer solution. In total, for each outcome measure, five separate batches of hydrogels were prepared.
Hydrogel samples were then trimmed to strips using two razor blades evenly bonded together (distance between the two blades was 1 mm). Each hydrogel strip (1 × 1 × 20 mm) was then glued to a piece of marked closed-cell polyurethane foam (Thompson et al., 2003) (Pacific Research Laboratories, Vashon, WA, 10 pcf) 20 mm wide, 40 mm long and 1 mm thick (Figure 3) with a cyanoacrylate adhesive. This foam is commonly used to model the elastic properties of cancellous bone (Thompson et al., 2003). The crack was made through the closed-cell foam into the gels with the same series of cutting tools (Figure 4). A new razor blade was used for each sample to maintain a consistent sharp crack tip inside the hydrogel. When pressing the blade, the specimen was cut through to create the crack with a certain part of the hydrogel remaining inside the notch that was not cut. In such a way, crack lengths were reproducible.
Care was taken to ensure that hydrogel specimen dimensions matched cartilage specimen dimensions as closely as possible, which we emphasize is important for the current study and for future comparisons back to data obtained here for hydrogels and cartilage.
The thickness of the hydrogel (B′), the width of the hydrogel (w′), the width of the whole specimen (W′) and the crack length (a′) were measured with a micrometer under a stereomicroscope (~ × 10 magnification).
2.3 Solids content characterization
Swelling characteristics and solids content of the hydrogels were analysed to verify their suitability as a three-dimensional (3D) scaffold for tissue engineering, as solids content can help to provide a better understanding and context for mechanical properties. Cylindrical hydrogel constructs (3 mm diameter, ~1.9 mm height) were synthesized for solids content characterization, using the methods described above. Five independent batches were made for each type of hydrogel. The swollen gel mass (ms) was measured after soaking in excess deionized water for 24 h, and the dry gel mass (md) was measured after drying in a desiccation chamber over calcium sulphate for 72 h. Measurements were made in triplicate (three separate samples) for each batch, using a high-precision balance (Shimadzu Corp., Kyoto, Japan). The swelling degree (Q) and solids content were calculated by:
The conversion efficiency was evaluated in terms of the conversion of the PEG-DA in the PEG-DA and IPN gels. The conversion percentage of PEG-DA in pure PEG-DA gels was calculated by:
The denominator represents the theoretical amount of PEG-DA in the gel for 100% conversion efficiency, where φ0 is the w/v concentration of PEG-DA solution before photopolymerization (0.2) and V is the calculated volume of the preswollen sample, based on the biopsy punch diameter and mould thickness. The conversion of PEG-DA in IPNs was calculated by:
where the numerator is the estimated mass of PEG-DA in the dried IPN, κ is the partition coefficient (assumed to be 1) and the overbar denotes the average value for a given batch. Similarly, the percentage of PEG-DA in dried IPNs was also calculated:
2.4 Toughness test for articular cartilage
A crack was made through the bone to the cartilage with a series of custom-designed cutting tools (there were three cutting tools with three different notch distances, 0.25, 0.36 and 0.48 mm). The thickness of the specimen (B), the width of the articular cartilage (w), the width of the whole specimen (W) and the crack length (a) were then measured with a micrometer under a stereomicroscope (~ × 10 magnification). A new razor blade was used for each specimen to maintain a sharp crack tip. As with the hydrogel specimen preparation, part of the specimen was positioned inside the notch and remained uncut, to ensure that crack lengths were reproducible (Figure 4).
Each specimen was then marked (Figure 5), placed in a custom-built bath-grip assembly (Singh and Detamore, 2008) in an Instron Model 5848 (50 N load cell; Canton, MA), and lined up (Figure 5) in a hydrated environment with PBS at 37ºC (Figure 6). For each sample, the crack tip was consistently 30 mm away from the centre of the grips, consistent with gel specimens as noted below. Tensile loading was applied with a displacement rate of 1.5 mm/min until the specimen pulled apart. The ratio a:W was varied from 0.95 to 0.98, consistent with gel specimens as noted below.
2.5 Toughness test for hydrogels
Five independent batches were made for each hydrogel group (PEG-DA and IPN) to provide a sample size of n = 5. Eight samples were tested from each batch. Each specimen was marked, placed in the same custom-built bath-grip assembly in the Instron Model 5848, lined up and immersed in fresh PBS. For each sample, the crack tip was consistently 30 mm away from the centre of the grips, consistent with cartilage specimens as noted above (Figure 7). The same tensile loading was applied with the displacement rate of 1.5 mm/min until the specimen pulled apart. Load and displacement were measured. The ratio a′:W′ was varied from 0.95 to 0.98, consistent with cartilage specimens as noted above.
2.6 Calculation of apparent fracture toughness
The models from the American Society for Testing and Materials (ASTM) method D 5045–99 (2007) were adopted. Apparent fracture toughness was characterized as KQ in units of MPa/mm½. KQ was calculated as follows:
where 0.2 < x < 0.8, and f(x) is defined as follows:
where PQ = load determined in ASTM D 5045–99 (2007), B = specimen thickness, W = specimen width, a = crack length and x = a/W.
Note that, due to the inherent geometry of articular cartilage, neither the range of a:W nor the specimen thickness would meet the dimension requirements of the ASTM standard. The use of the undersized specimen results in an overestimation of the actual plane strain fracture toughness. This is recognized as a limitation of the method, but is required because of the physical limitations of the biological material. For this reason, identical geometries are used for comparison tests and the term ‘apparent fracture toughness’ is used to describe the fracture toughness properties.
2.7 Statistical analyses
To compare experimental groups, a single-factor analysis of variance (ANOVA) was performed, followed by a Tukey's Honestly Significant Difference post hoc test when significance was detected. Analysis was performed using the SPSS/PASW 17.0 statistical software package. All quantitative results were expressed as mean ± standard deviation (SD).
3.1 Solid content analysis
The solid content analysis of the hydrogels is shown in Table 1 and the swelling degrees are reported in Figure 8. The total solids content in the IPNs was determined to be 11.41 ± 0.23% (n = 5). As significant differences were not observed between mould and final swelled gel dimensions (i.e. the agarose gel does not swell or deswell in the PEG-DA solution), the PEG-DA content in IPN gels could be estimated by subtracting the average solids content of the pure agarose gels (complete conversion of agarose is expected; DeKosky et al., 2010) from the solids content of the IPNs. Agarose gels were composed of ~2.0% agarose, so the PEG-DA content of IPN gels was ~9.4%. The swelling degree of PEG-DA was 13% higher than that of IPNs (p < 0.05). The conversion efficiencies in the PEG-DA and IPN gels were 82.2% and 69.1%, respectively (Table 2) (n = 5).
Representative load–extension curves for the two hydrogel groups and cartilage are presented in Figure 9. The apparent fracture toughness for articular cartilage (n = 8) was 348 ± 43 MPa/mm½. The apparent fracture toughness of PEG-DA was 7.80 ± 0.93 MPa/mm½ and that of the IPN was 10.8 ± 1.4 MPa/mm½ (Figure 10). The apparent fracture toughness for articular cartilage was more than 30 times higher than that of the IPN, and the apparent fracture toughness of the IPN was 1.4 times greater than that of the PEG-DA (p < 0.05). Note that Figure 10 includes the apparent fracture toughness for each batch to demonstrate the range of values observed.
To the best of our knowledge, this was the first effort to apply the same method to both articular cartilage and hydrogels with analogous methods for determining apparent fracture toughness. The current study introduced the approach of adhering each hydrogel strip to a piece of thin closed-cell foam in an effort to provide a reasonable model of an osteochondral specimen, and then testing these hydrogels with the modified single-edge notch test. The shape of the hydrogel specimen was trimmed in exactly the same way as articular cartilage, and the distance from the grip centreline to the crack tip was consistent, thus keeping the state of stress in the crack tip consistent in the specimens, in an effort to facilitate direct comparisons of the apparent fracture toughness values.
The IPN was tougher than the PEG-DA, as hypothesized. Using the modified single-edge notch test method in the current study, the IPN was 1.4 times tougher (p < 0.05), as compared to being 3.3 times tougher using the unconfined compression in our previous work (not statistically significant) (DeKosky et al., 2010). The total solids content in IPNs was ~11% higher than that of the PEG-DA (p < 0.05), which may have contributed in part to the higher apparent fracture toughness of the IPNs (Gong et al., 2005, 2009). However, IPNs had a lower apparent PEG-DA conversion (69% compared to 82% in PEG-DA gels). Had the two hydrogel groups contained an equal amount of PEG-DA content, it is likely that the difference in their apparent fracture toughness would have been even greater.
The relative magnitude is less important than the advantage gained by the significant improvement in reproducibility afforded by the modified single-edge notch test. In fact, the large standard deviations with compression to failure in our previous study (80–95% of the mean!) were cited as a limitation of that testing modality (DeKosky et al., 2010), as small changes in failure strain lead to large variations in toughness measurements, due to the non-linear stress–strain behaviour at very high strains under compression. This limitation of large standard deviations was the impetus for exploring a more suitable method to evaluate the apparent fracture toughness, which ultimately led to our recent review to find the common ground between cartilage and hydrogel fracture mechanics (Xiao et al., 2013), which in turn naturally evolved to the current study. With the modified single-edge notch test in the current study, we were able to achieve standard deviations in fracture toughness that were only 12–13% of the mean, a significant improvement over our previous compression-to-failure method (DeKosky et al., 2010), allowing for a statistically significant demonstration of the superior toughness of IPNs compared to PEG-DA. By evaluating different batches, the reproducibility of the apparent fracture toughness from batch to batch was confirmed for both the PEG-DA and the IPN.
The major task of this study was to evaluate the apparent fracture toughness for both hydrogels and articular cartilage. In the current study, the apparent fracture toughness for porcine articular cartilage was 348 ± 43 MPa/mm½ using the modified single-edge notch test, which was more than 200 times higher than that for bovine articular cartilage measured by the single-edge notch test (Stok and Oloyede, 2007). The single-edge notch test was a different test in which only the cartilage was gripped and loaded under tension. The apparent fracture toughness of adult canine patellar cartilage in a modified single-edge notch test from ChinPurcell and Lewis (1996) was characterized by a J integral with the value of 0.14 ± 0.08 kN/m, which cannot be directly compared with the current study results, as here we developed a new model based on an ASTM standard that could be used for both cartilage and hydrogels.
The current study showed that the apparent fracture toughness of articular cartilage was 31 times higher than that of the IPN, which was in turn significantly tougher than the PEG-DA alone. With this perspective, the failure properties of synthesized hydrogels in tissue-engineering regeneration clearly need to be enhanced. Thus, factors that can significantly improve the apparent fracture toughness for the hydrogels need to be identified. It is recognized that the relative importance of scaffold fracture toughness compared to other mechanical properties, such as compressive strength and stiffness, is not yet clear and warrants further investigation. It is further recognized that compressive moduli, typically relaxed/aggregate moduli, are also a crucial design parameter for scaffolds in cartilage tissue engineering, to stand alongside, but certainly not be replaced by, fracture properties as mechanical performance design criteria.
There were some limitations with regard to testing both articular cartilage and hydrogels. For testing articular cartilage, the geometry was a limitation in terms of being unable to fit the ASTM standard geometry. However, because of the physical size and structure of cartilage, this limitation is unavoidable, and the method was developed to ensure that the geometry of the specimens was as close to the ASTM standard geometry as possible while maintaining similar states of stress at the crack tip. In the future, longer-term studies will be valuable to evaluate whether extracellular matrix production by cell-seeded gels will significantly improve apparent fracture toughness in this testing regime. Moreover, it is clearly difficult to create apparent fracture toughness tests that will be identical for comparing hydrogels and articular cartilage, as cartilage has a complex gradient boundary and changing properties at the interface with the subchondral bone. Nevertheless, the approach developed here took into consideration a judiciously selected compromise among testing methods available for each material, so that boundary conditions of the cartilage in the in vivo state were maintained. The hydrogel testing method was carefully designed to match as closely as possible the modified single-edge notch test applied to cartilage.
Overall, a new method was developed to evaluate the failure properties of both articular cartilage and hydrogels in the context of cartilage tissue engineering. The IPNs possessed a greater toughness relative to the PEG-DA. However, their apparent fracture toughness was over an order of magnitude less than articular cartilage, so clearly there is much work to do in improving hydrogel fracture properties, unless cells are able to significantly close this gap with the production of functional extracellular matrix in clinically relevant time periods. Therefore, by developing standard testing methods that could accommodate both hydrogels and cartilage, we have a clearer target for the extent to which the failure properties of hydrogels must be improved, which will ultimately lead to better tissue replacements for damaged articular cartilage.
Conflict of interest
The authors have declared that there is no conflict of interest.
The authors would like to acknowledge funding from the NIH/NIBIB (Grant No. R21 EB008783).