- Email: [email protected]
Indentation stiffness does not discriminate between normal and degraded articular cartilage
Clinical Biomechanics 22 (2007) 843–848 www.elsevier.com/locate/clinbiomech
Indentation stiffness does not discriminate between normal and degraded articular cartilage Cameron P. Brown, Ross W. Crawford, Adekunle Oloyede
*
Institute of Health and Biomedical Innovation, Faculty of Built Environment and Engineering, Queensland University of Technology, 2 George St., Brisbane Q 4001, Australia Received 21 November 2006; accepted 25 April 2007
Abstract Background. Relative indentation characteristics are commonly used for distinguishing between normal healthy and degraded cartilage. The application of this parameter in surgical decision making and an appreciation of articular cartilage biomechanics has prompted us to hypothesise that it is difficult to define a reference stiffness to characterise normal articular cartilage. Methods. This hypothesis is tested for validity by carrying out biomechanical indentation of articular cartilage samples that are characterised as visually normal and degraded relative to proteoglycan depletion and collagen disruption. Compressive loading was applied at known strain rates to visually normal, artificially degraded and naturally osteoarthritic articular cartilage and observing the trends of their stress–strain and stiffness characteristics. Findings. While our results demonstrated a 25% depreciation in the stiffness of individual samples after proteoglycan depletion, they also showed that when compared to the stiffness of normal samples only 17% lie outside the range of the stress–strain behaviour of normal samples. Interpretation. We conclude that the extent of the variability in the properties of normal samples, and the degree of overlap (81%) of the biomechanical properties of normal and degraded matrices demonstrate that indentation data cannot form an accurate basis for distinguishing normal from abnormal articular cartilage samples with consequences for the application of this mechanical process in the clinical environment. 2007 Elsevier Ltd. All rights reserved. Keywords: Articular cartilage; Mechanical indentation; Stiffness; Osteoarthritis; Cartilage degradation
1. Introduction Due to the high degree of structure–function coupling in articular cartilage, the integrity of the constituents of the matrix, and their interactions with each other, will determine the mechanical behaviour of the tissue. Mechanical compression and indentation techniques, based around the stress–strain characteristic, have long been used to assess the functional performance of normal and degenerated articular cartilage in vitro (Harris et al., 1972; Hayes et al., 1972; Kempson et al., 1971), and have been reported *
Corresponding author. E-mail address: [email protected] (A. Oloyede).
0268-0033/$ - see front matter 2007 Elsevier Ltd. All rights reserved. doi:10.1016/j.clinbiomech.2007.04.013
to successfully track structural changes in artificial degeneration programs (Lyyra et al., 1999). Recent publications (Appleyard et al., 1999; Arokoski et al., 1994) have suggested that an alteration of the mechanical properties in an arthritic joint may be notable before any gross morphological change is apparent. This would potentially make the mechanical test a useful indicator of early stage degeneration and hence a viable tool in surgical decision making. It will, however, rely on an important assumption; namely that a clear demarcation exists between the biomechanical responses of loaded normal and degraded articular cartilage. It is well known that the mechanical properties of articular cartilage vary significantly both across a normal joint,
844
C.P. Brown et al. / Clinical Biomechanics 22 (2007) 843–848
and between joints depending on factors such as age and the predominating mechanical environment (Broom and Flachsmann, 2003; Kempson, 1982; Swann and Seedhom, 1993; Yao and Seedhom, 1993). Previous studies have further shown inconsistencies in the outcomes of degradation processes (Moody et al., 2006). It is not known, however, the extent to which this inherent variation in the properties of normal and degraded cartilage results in a property overlap between the two categories of tissue, with the possible consequence that differentiation may be difficult or unachievable. Following this observation, the use of the structure– load–deformation relationship for characterising articular cartilage degradation raises an important question; i.e., if a mechanical test can track structural changes in the tissue (Lyyra et al., 1999), can it necessarily be used to determine whether a tissue is normal or degraded? This question must be answered in two stages, firstly defining what constitutes a ‘‘normal’’ tissue and then using this to logically establish whether we can conclusively distinguish an abnormal one given the variation between subjects and the variations across the individual joint. In this study we hypothesise that if the variation in the indentation response of normal articular cartilage-on-bone is sufficiently large, it may not be possible to conclusively distinguish physiologically abnormal from normal tissue in the osteoarthritic joint from load–deformation characteristics alone. If the stress–strain properties of normal and degraded cartilage samples are plotted in the same graphical space, their overlap may be substantial enough to make the employment of such mechanical data for discriminating between physiologically normal and abnormal cartilage inaccurate and misleading. In order to test our hypothesis we investigated the loadinduced characteristics of normal articular cartilage-onbone samples in a controlled in vitro environment before and after exposure to a well established degradation program of trypsin treatment (Borthakur et al., 2000; Duvvuri et al., 2002; Korhonen et al., 2003; Moody et al., 2006) to determine whether the differences exhibited in their stress– strain data will sufficiently and adequately differentiate samples in the two conditions. Further, we will compare the results to those of cartilage-on-bone samples from osteoarthritic joints. 2. Methods Macroscopically normal and intact, and degenerate (nominally ICRS grade 1) (International Cartilage Repair Society, 2000) bovine patellae were harvested from prime oxen within 24 h of slaughter, wrapped in a 0.15 M saline soaked cloth and stored at 20 C. Prior to treatment, the joints were thawed in saline, sectioned into 20 · 20 mm samples. The thickness of the cartilage-on-bone specimens were measured with a digital linear variable displacement transducer (LVDT), embedded in Palapress (Heraeus Kulzer GmbH & Co. Hanau, Germany) dental
acrylic and mounted in stainless steel sample holders. Care was taken to ensure that the cartilage remained hydrated during this process. Each sample was then placed in saline for at least 90 min. Before testing, the sample height was remeasured to ensure that its thickness had recovered to that prior to preparation. Eighteen normal samples were subjected to compressive loading on a Hounsfield testing facility (Tinius Olsen, Salfords, England) to 33% strain at loading rates of 0.1 s 1 to represent the usual rate of loading in the clinical environment (Franz et al., 2001), and compare the results to those obtained at 0.025 s 1 to provide an insight into the way that different rates of loading might determine the outcome of an indentation process. The loads were applied via a 4 mm plane-ended circular indenter in the centre of the sample area. The samples were unloaded and allowed to recover for 2 h in saline between tests and checked to ensure that full thickness was regained. The stress–strain behaviour of cartilage was obtained from the load–displacement curves and the stiffness of each sample was calculated as the first derivative of the stress–strain curve at nominated strains. By taking geometry into account we allow a more reliable parameter for cross comparison than the reaction force–displacement measurement alone, as the geometry of the specimen, particularly thickness, will affect their deformation characteristic (Hayes et al., 1972; To¨yra¨s et al., 2001; Zhang et al., 1997). Following mechanical testing, the samples were depleted of a large amount of their proteoglycan content to represent the loss of proteoglycan in the disease process (Rieppo et al., 2003). This type of modification was chosen over other degradation protocols for its reported comparatively large and consistent mechanical consequence on the cartilage matrix (Lyyra et al., 1999). To remove proteoglycans from the articular cartilage general matrix, the samples were immersed in PBS, pH 7.5 (P4417, Sigma–Aldrich, Sydney, Australia) solution containing 0.1 mg/mL trypsin for 1 h. After the period of exposure was completed, the samples were blotted dry and returned to the 0.15 M saline bath before retesting under the same compression conditions. We note at this juncture that the 1 h treatment will lead to different degraded states in the samples in accordance with the work of Moody et al., 2006. Osteoarthritic degradation, by nature, will similarly lead to different conditions in different samples. Of importance is the fact that a 1 h trypsin treatment will result in cartilage degradation and a changed mechanical response that is representative of this condition. Further, the results from normal and trypsin-treated samples were then compared to those of osteoarthritic cartilage-on-bone. The nine samples taken from degenerate patellae, characterised by superficial cracking and fibrillation (nominally ICRS grade 1 degradation (International Cartilage Repair Society, 2000)), were tested under the same conditions as the other sample categories. The microscopic and image processing techniques used are fully explained in a previous paper (Moody et al.,
C.P. Brown et al. / Clinical Biomechanics 22 (2007) 843–848
845
2006). In summary, samples were sectioned at 7 lm and stained with safranin-O using standard histological procedures. Optical absorbance profiles were taken using a microscope and a monochromatic light source, and processed with IMAGEJ software (1.33u, National Institutes of Health, Bethesda, USA) to estimate the proteoglycan concentration as a function of distance from the articular surface.
3. Results Typical absorbance profiles, showing the distribution of proteoglycans over the depth of the normal and degraded cartilage samples are presented in Fig. 1. The loss of proteoglycans after trypsin treatment tended to show similar features such as the distinct wavefront of depletion from the near zero proteoglycan content to the original concentration. The osteoarthritic specimens were characterised by a more uniform loss of proteoglycan over the depth of the cartilage matrix in addition to minor surface disruption. The mechanical results presented a large variation in the mechanical behaviour of the normal samples as shown in Figs. 2–4. At a strain rate of 0.1 s 1, trypsin treatment produced an average reduction in stiffness of approximately 25% at strains of 10% and 30%. Only 17% of the stress– strain data of trypsin-treated samples, however, fell outside of the range of the stress–strain behaviour of the normal samples, while only 22% of the data from normal samples fell outside of the envelope of the stress–strain behaviour of the trypsin-treated samples. This trend was consistent across each of the values of strain. Mechanical loading at the lower strain rate of 0.025 s 1 produced a similar trend with an average reduction in stiffness after degradation of 15–20% (Fig. 3). Again, only three specimens from the osteoarthritic and trypsin-treated groups fell below
Fig. 1. The concentration of proteoglycans by absorbance value for normal, trypsin-treated and osteoarthritic samples. Normal samples followed similar patterns, while trypsin treatment produced a wavefront of proteoglycan depletion. Samples from osteoarthritic joints were characterised by a more uniform depletion of proteoglycans over the depth of the tissue.
Fig. 2. The stress–strain curves in (a) showed a high variation in stiffness for normal (N) samples at 0.1 s 1. The labels ‘‘high’’ and ‘‘low’’ refer to the range of the samples in each group. Most osteoarthritic (OA) samples and samples degraded by trypsin treatment (T) fall within the normal variation. A loading rate of 0.025 s 1 (b) produced a similar range of behaviour to that of 0.1 s 1. The change in the stress–strain characteristic after trypsin treatment was similar to that of 0.1 s 1, but a greater overlap of normal and osteoarthritic samples was observed.
the range of stress–strain behaviour of normal samples. The stress–strain characteristic of the osteoarthritic samples showed a greater deviation from those of the normal at lower strains with 33% of the samples falling outside those for the normal at 10% strain. At 30% strain, however, the stress strain behaviour of the osteoarthritic samples overlap with the normal samples. Fig. 3a and b show the overlap of the stress–strain characteristics at low strain (a) and high strain (b). A similar mechanical effect to those of trypsin-treated samples was also obtained in the normal samples when the loading rate was changed, as shown in Fig. 2. This change produced an average reduction in stiffness of 20– 26%, increasing with increasing strain. The stiffness values for the samples produced similar results to the patterns of the stress–strain behaviour, as shown in Fig. 4. At 10% strain, 6% of the trypsin-treated
846
C.P. Brown et al. / Clinical Biomechanics 22 (2007) 843–848
Fig. 4. The stiffness of normal, trypsin-treated and osteoarthritic samples at strains of 0.1 (a) and 0.3 (b), tested at 0.1 s 1.
Fig. 3. (a) shows the overlap at low strain and (b) at high strain at 0.1 s 1. The percentage figures represent the number of samples that fall outside of the envelope at each strain.
samples and 22% of the osteoarthritic samples exhibited responses that fell below those obtained for normal samples. At 30% strain, 22% of the trypsin-treated samples and no osteoarthritic samples responded with stiffness values that were lower than and outside the envelope of those for the normal samples. 4. Discussion Mechanical test results from the indentation of normal cartilage-on-bone samples showed a large variation in stiffness. This has also been observed in previously published in vivo (Vasara et al., 2005) and in vitro (Broom and Flachsmann, 2003; Kempson, 1982) investigations. A change in the loading rate produced a further variation in mechanical behaviour in accordance with the previously published work of Oloyede et al. (1992). Given the variation in
normal samples, these results lead to the question of whether the measurements from a mechanical indentation test are meaningful for distinguishing between normal and degraded articular cartilage. For the purpose of comparing the effects of the rate of loading on our results, we have driven the applied loads at the rates of 0.025 s 1 and 0.1 s 1, where the latter velocity of loading approximates that applied in clinical indentation assessments (Vasara et al., 2005). The different loading rates produced similar patterns of results, with an average change in stiffness after the 1 h trypsin treatment of 25%. This large change, however, only produced a stress–strain characteristic outside the envelope of the same parameter for the normal samples in approximately 17% of the degraded samples at the loading rates of 0.1 s 1 and 0.025 s 1 used in this study. Such ambiguity would potentially complicate, and introduce errors into tissue differentiation or classification based on mechanical data. From Figs. 2–4 we can see that 78% of the normal samples fell within the envelope of the stress–strain characteristic of the trypsin-treated samples and may therefore be classed as degraded by a mechanical test. Likewise, 83% of the trypsin-treated samples fell within the envelope of normal samples suggesting that a large number of degraded tissues could be classed as normal. Similar results were obtained for osteoarthritic samples at 10% strain. Applying a 90% confidence interval reduced these figures to 66% and 73%, respectively. The results presented in Fig. 2 suggest that if the loading rate cannot be controlled, the resultant stiffness
C.P. Brown et al. / Clinical Biomechanics 22 (2007) 843–848
may change considerably, resulting in an increase in the overlap that will add to the already large error. Although the normal and degraded groups showed a large variation in their stiffness, the stress–strain behaviour maintained similar basic attributes. The osteoarthritic samples were generally characterised by slightly lower resultant stresses in the toe region and higher stresses at the later stages of deformation when compared to the proteoglycan-depleted samples. This effect is most likely due to surface disruption in the osteoarthritic samples, which we argue, would act to decrease the stiffness at very low strains. Overall, a large proportion of the stress–strain curves from the normal and degraded groups showed similar characteristics across the three groups of tissue tested, indicating that a normal sample may behave similarly to an osteoarthritic sample or a cartilage sample devoid of a large portion of its proteoglycans. Earlier studies investigating the application of load– deformation or stress–strain-based data in normal and degraded cartilage (Lyyra et al., 1999) suggest the possibility of obtaining a relative change in degraded tissue properties from a known normal. Since the mechanical characteristics of normal cartilage samples vary and can overlap significantly with degraded tissue, it can be concluded that it is almost impossible to define a normal/ healthy tissue reference state against which degradation can be determined, even within the same joint sample. For example, it is possible to choose at random, a single specimen from the 36 normal and proteoglycan-depleted samples. Even with the known extremes of the data set and a single, controlled loading rate, the probability of distinguishing such a sample as either normal or abnormal is only 19.4%. The probability of distinguishing the same specimen with 90% confidence is 30.5%. For the osteoarthritic samples, the probablity of distinguishing a random sample as either normal or abnormal using our present mechanical data is 38%. Such an analysis can also be applied to the published in vivo results from human cartilage (Vasara et al., 2005) in which a commercially available arthroscopic indenter was used to compare the ‘‘stiffness’’ of normal and ICRS grade 1 tissue. In this case, the probability of distinguishing either a normal or abnormal tissue, taking into account the variations between the different joint surfaces in the knee, is 22%. Despite a statistically significant trend being observed, the vast majority of indentation results from the osteoarthritic tissue fell within the range of the results for normal samples. It should be noted that the in vivo test was performed on ICRS grade 1 sites which by definition (International Cartilage Repair Society, 2000) are characterised by softening or visible surface cracking. It seems therefore that the surgeon, unaided, can distinguish degraded tissue more reliably than an indentation device, leading to the argument that such a method might not be capable, on its own, of enhancing the arthroscopic evaluation of cartilage defects. It should be noted that the indenter diameter used in the experimental investigation is larger than that used in the
847
commercially available arthroscopic indenter. It is further possible that different loading paths/regimes used to achieve the same level of indentation may produce a modification in the patterns of variation. It was observed in this analysis, however, that the variations obtained in our results were very similar to those observed from studies using a commercially available arthroscopic indenter (Vasara et al., 2005) and therefore argue that they form a reasonable representation of the patterns of overlap from clinical indentation. Furthermore, similar patterns of overlap were obtained across different levels of indentation and at different loading velocities. The large variation in the mechanical behaviour of normal tissues and the lack of discrepancy between normal and degraded behaviours have strong implications for the use of a mechanical test for characterising functional abnormality in cartilage, with significant consequences for surgical decisions. In conclusion, our investigation demonstrates that while the mechanical test remains a valuable method for tracking progressive changes in structure via sequential testing, it cannot reliably distinguish either normality or abnormality in a given joint. Acknowledgements The authors wish to thank the Australian Research Council and Smith & Nephew for supporting this research. References Appleyard, R.C., Ghosh, P., Swain, M.V., 1999. Biomechanical, histological and immunohistological studies of patellar cartilage in an ovine model of osteoarthritis induced by lateral meniscectomy. Osteoarthritis and Cartilage 7, 281–294. Arokoski, J., Jurvelin, J., Kiviranta, I., Tammi, M., Helminen, H.J., 1994. Softening of the lateral condyle articular cartilage in the canine knee joint after long distance (up to 40 km/day) running training lasting one year. International Journal of Sports Medicine 15, 254–260. Borthakur, A., Shapiro, E.M., Beers, J., Kudchodkar, S., Kneeland, J.B., Reddy, R., 2000. Sensitivity of MRI to proteoglycan depletion in cartilage: comparison of sodium and proton MRI. Osteoarthritis and Cartilage 8, 288–293. Broom, N.D., Flachsmann, R., 2003. Physical indicators of cartilage health: the relevance of compliance, thickness, swelling and fibrillar texture. Journal of Anatomy 202, 481–494. Duvvuri, U., Kudchodkar, S., Reddy, R., Leigh, J.S., 2002. T(1rho) relaxation can assess longitudinal proteoglycan loss from articular cartilage in vitro. Osteoarthritis and Cartilage 10, 838–844. Franz, T., Hasler, E.M., Hagg, R., Weiler, C., Jakob, R.P., Mainil-Varlet, P., 2001. In situ compressive stiffness, biochemical composition, and structural integrity of articular cartilage in the human knee joint. Osteoarthritis and Cartilage 9, 582–592. Harris, E.D.J., Parker, H.G., Radin, E.L., Krane, S.M., 1972. Effects of proteolytic enzymes on structural and mechanical properties of cartilage. Arthritis and Rheumatism 15, 497–503. Hayes, W.C., Keer, L.M., Herrmann, G., Mockros, L.F., 1972. A mathematical analysis for indentation tests of articular cartilage. Journal of Biomechanics 5, 541–551. International Cartilage Repair Society, 2000. ICRS Cartilage Injury Evaluation Package. In: ICRS 2000 Standards Workshop. International Cartilage Repair Society: Switzerland.
848
C.P. Brown et al. / Clinical Biomechanics 22 (2007) 843–848
Kempson, G.E., 1982. Relationship between the tensile properties of articular cartilage from the human knee and age. Annals of the Rheumatic Diseases 41, 508–511. Kempson, G.E., Spivey, C.J., Swanson, S.A.V., Freeman, M.A.R., 1971. Patterns of cartilage stiffness on normal and degenerative human femoral heads. Journal of Biomechanics 4, 597–609. Korhonen, R.K., Laasanen, M.S., Toyras, J., Lappalainen, R., Helminen, H.J., Jurvelin, J.S., 2003. Fibril reinforced poroelastic model predicts specifically mechanical behaviour of normal, proteoglycan-depleted and collagen degraded articular cartilage. Journal of Biomechanics 36, 1373–1397. Lyyra, T., Arokoski, J.P.A., Oksala, N., Vihko, A., Hyttinen, M., Jurvelin, J.S., Kiviranta, I., 1999. Experimental validation of arthroscopic cartilage stiffness measurement using enzymatically degraded cartilage samples. Physics in Medicine and Biology 44, 525–535. Moody, H.R., Brown, C.P., Bowden, J.C., McElwain, D.L.S., Crawford, R.W., Oloyede, A., 2006. In vitro degradation of articular cartilage – does trypsin treatment produce consistent results? Journal of Anatomy 209, 259–267. Oloyede, A., Flachsmann, R., Broom, N.D., 1992. The dramatic influence of loading velocity on the compressive response of articular cartilage. Journal of Connective Tissue Research 27, 1–15.
Rieppo, J., Toyras, J., Nieminen, M.T., Kovanen, V., Hyttinen, M.M., Korhonen, R.K., Jurvelin, J.S., Helminen, H.J., 2003. Structure– function relationships in enzymatically modified articular cartilage. Cells Tissues Organs 175, 121–132. Swann, A.C., Seedhom, B.B., 1993. The stiffness of normal articular cartilage and the predominant acting stress levels. British Journal of Rheumatology 32, 16–25. To¨yra¨s, J., Lyyra-Laitinen, T., Niinima¨ki, M., Lindgren, R., Nieminen, M.T., Kiviranta, I., Jurvelin, J.S., 2001. Estimation of the Young’s modulus of articular cartilage using an arthroscopic indentation instrument and ultrasonic measurement of tissue thickness. Journal of Biomechanics 34, 251–256. Vasara, A.I., Jurvelin, J., Peterson, L., Kiviranta, I., 2005. Arthroscopic cartilage indentation and cartilage lesions of anterior cruciate ligament-deficient knees. International Journal of Sports Medicine 33, 408–414. Yao, J.Q., Seedhom, B.B., 1993. Mechanical conditioning of articular cartilage to prevalent stress. British Journal of Rheumatology 33, 16–25. Zhang, M., Zheng, Y.P., Mak, A.F., 1997. Estimating the effective Young’s modulus of soft tissues from indentation tests – nonlinear finite element analysis of effects of friction and large deformation. Medical Engineering and Physics 19, 512–517.
Indentation stiffness does not discriminate between normal and degraded articular cartilage Cameron P. Brown, Ross W. Crawford, Adekunle Oloyede
*
Institute of Health and Biomedical Innovation, Faculty of Built Environment and Engineering, Queensland University of Technology, 2 George St., Brisbane Q 4001, Australia Received 21 November 2006; accepted 25 April 2007
Abstract Background. Relative indentation characteristics are commonly used for distinguishing between normal healthy and degraded cartilage. The application of this parameter in surgical decision making and an appreciation of articular cartilage biomechanics has prompted us to hypothesise that it is difficult to define a reference stiffness to characterise normal articular cartilage. Methods. This hypothesis is tested for validity by carrying out biomechanical indentation of articular cartilage samples that are characterised as visually normal and degraded relative to proteoglycan depletion and collagen disruption. Compressive loading was applied at known strain rates to visually normal, artificially degraded and naturally osteoarthritic articular cartilage and observing the trends of their stress–strain and stiffness characteristics. Findings. While our results demonstrated a 25% depreciation in the stiffness of individual samples after proteoglycan depletion, they also showed that when compared to the stiffness of normal samples only 17% lie outside the range of the stress–strain behaviour of normal samples. Interpretation. We conclude that the extent of the variability in the properties of normal samples, and the degree of overlap (81%) of the biomechanical properties of normal and degraded matrices demonstrate that indentation data cannot form an accurate basis for distinguishing normal from abnormal articular cartilage samples with consequences for the application of this mechanical process in the clinical environment. 2007 Elsevier Ltd. All rights reserved. Keywords: Articular cartilage; Mechanical indentation; Stiffness; Osteoarthritis; Cartilage degradation
1. Introduction Due to the high degree of structure–function coupling in articular cartilage, the integrity of the constituents of the matrix, and their interactions with each other, will determine the mechanical behaviour of the tissue. Mechanical compression and indentation techniques, based around the stress–strain characteristic, have long been used to assess the functional performance of normal and degenerated articular cartilage in vitro (Harris et al., 1972; Hayes et al., 1972; Kempson et al., 1971), and have been reported *
Corresponding author. E-mail address: [email protected] (A. Oloyede).
0268-0033/$ - see front matter 2007 Elsevier Ltd. All rights reserved. doi:10.1016/j.clinbiomech.2007.04.013
to successfully track structural changes in artificial degeneration programs (Lyyra et al., 1999). Recent publications (Appleyard et al., 1999; Arokoski et al., 1994) have suggested that an alteration of the mechanical properties in an arthritic joint may be notable before any gross morphological change is apparent. This would potentially make the mechanical test a useful indicator of early stage degeneration and hence a viable tool in surgical decision making. It will, however, rely on an important assumption; namely that a clear demarcation exists between the biomechanical responses of loaded normal and degraded articular cartilage. It is well known that the mechanical properties of articular cartilage vary significantly both across a normal joint,
844
C.P. Brown et al. / Clinical Biomechanics 22 (2007) 843–848
and between joints depending on factors such as age and the predominating mechanical environment (Broom and Flachsmann, 2003; Kempson, 1982; Swann and Seedhom, 1993; Yao and Seedhom, 1993). Previous studies have further shown inconsistencies in the outcomes of degradation processes (Moody et al., 2006). It is not known, however, the extent to which this inherent variation in the properties of normal and degraded cartilage results in a property overlap between the two categories of tissue, with the possible consequence that differentiation may be difficult or unachievable. Following this observation, the use of the structure– load–deformation relationship for characterising articular cartilage degradation raises an important question; i.e., if a mechanical test can track structural changes in the tissue (Lyyra et al., 1999), can it necessarily be used to determine whether a tissue is normal or degraded? This question must be answered in two stages, firstly defining what constitutes a ‘‘normal’’ tissue and then using this to logically establish whether we can conclusively distinguish an abnormal one given the variation between subjects and the variations across the individual joint. In this study we hypothesise that if the variation in the indentation response of normal articular cartilage-on-bone is sufficiently large, it may not be possible to conclusively distinguish physiologically abnormal from normal tissue in the osteoarthritic joint from load–deformation characteristics alone. If the stress–strain properties of normal and degraded cartilage samples are plotted in the same graphical space, their overlap may be substantial enough to make the employment of such mechanical data for discriminating between physiologically normal and abnormal cartilage inaccurate and misleading. In order to test our hypothesis we investigated the loadinduced characteristics of normal articular cartilage-onbone samples in a controlled in vitro environment before and after exposure to a well established degradation program of trypsin treatment (Borthakur et al., 2000; Duvvuri et al., 2002; Korhonen et al., 2003; Moody et al., 2006) to determine whether the differences exhibited in their stress– strain data will sufficiently and adequately differentiate samples in the two conditions. Further, we will compare the results to those of cartilage-on-bone samples from osteoarthritic joints. 2. Methods Macroscopically normal and intact, and degenerate (nominally ICRS grade 1) (International Cartilage Repair Society, 2000) bovine patellae were harvested from prime oxen within 24 h of slaughter, wrapped in a 0.15 M saline soaked cloth and stored at 20 C. Prior to treatment, the joints were thawed in saline, sectioned into 20 · 20 mm samples. The thickness of the cartilage-on-bone specimens were measured with a digital linear variable displacement transducer (LVDT), embedded in Palapress (Heraeus Kulzer GmbH & Co. Hanau, Germany) dental
acrylic and mounted in stainless steel sample holders. Care was taken to ensure that the cartilage remained hydrated during this process. Each sample was then placed in saline for at least 90 min. Before testing, the sample height was remeasured to ensure that its thickness had recovered to that prior to preparation. Eighteen normal samples were subjected to compressive loading on a Hounsfield testing facility (Tinius Olsen, Salfords, England) to 33% strain at loading rates of 0.1 s 1 to represent the usual rate of loading in the clinical environment (Franz et al., 2001), and compare the results to those obtained at 0.025 s 1 to provide an insight into the way that different rates of loading might determine the outcome of an indentation process. The loads were applied via a 4 mm plane-ended circular indenter in the centre of the sample area. The samples were unloaded and allowed to recover for 2 h in saline between tests and checked to ensure that full thickness was regained. The stress–strain behaviour of cartilage was obtained from the load–displacement curves and the stiffness of each sample was calculated as the first derivative of the stress–strain curve at nominated strains. By taking geometry into account we allow a more reliable parameter for cross comparison than the reaction force–displacement measurement alone, as the geometry of the specimen, particularly thickness, will affect their deformation characteristic (Hayes et al., 1972; To¨yra¨s et al., 2001; Zhang et al., 1997). Following mechanical testing, the samples were depleted of a large amount of their proteoglycan content to represent the loss of proteoglycan in the disease process (Rieppo et al., 2003). This type of modification was chosen over other degradation protocols for its reported comparatively large and consistent mechanical consequence on the cartilage matrix (Lyyra et al., 1999). To remove proteoglycans from the articular cartilage general matrix, the samples were immersed in PBS, pH 7.5 (P4417, Sigma–Aldrich, Sydney, Australia) solution containing 0.1 mg/mL trypsin for 1 h. After the period of exposure was completed, the samples were blotted dry and returned to the 0.15 M saline bath before retesting under the same compression conditions. We note at this juncture that the 1 h treatment will lead to different degraded states in the samples in accordance with the work of Moody et al., 2006. Osteoarthritic degradation, by nature, will similarly lead to different conditions in different samples. Of importance is the fact that a 1 h trypsin treatment will result in cartilage degradation and a changed mechanical response that is representative of this condition. Further, the results from normal and trypsin-treated samples were then compared to those of osteoarthritic cartilage-on-bone. The nine samples taken from degenerate patellae, characterised by superficial cracking and fibrillation (nominally ICRS grade 1 degradation (International Cartilage Repair Society, 2000)), were tested under the same conditions as the other sample categories. The microscopic and image processing techniques used are fully explained in a previous paper (Moody et al.,
C.P. Brown et al. / Clinical Biomechanics 22 (2007) 843–848
845
2006). In summary, samples were sectioned at 7 lm and stained with safranin-O using standard histological procedures. Optical absorbance profiles were taken using a microscope and a monochromatic light source, and processed with IMAGEJ software (1.33u, National Institutes of Health, Bethesda, USA) to estimate the proteoglycan concentration as a function of distance from the articular surface.
3. Results Typical absorbance profiles, showing the distribution of proteoglycans over the depth of the normal and degraded cartilage samples are presented in Fig. 1. The loss of proteoglycans after trypsin treatment tended to show similar features such as the distinct wavefront of depletion from the near zero proteoglycan content to the original concentration. The osteoarthritic specimens were characterised by a more uniform loss of proteoglycan over the depth of the cartilage matrix in addition to minor surface disruption. The mechanical results presented a large variation in the mechanical behaviour of the normal samples as shown in Figs. 2–4. At a strain rate of 0.1 s 1, trypsin treatment produced an average reduction in stiffness of approximately 25% at strains of 10% and 30%. Only 17% of the stress– strain data of trypsin-treated samples, however, fell outside of the range of the stress–strain behaviour of the normal samples, while only 22% of the data from normal samples fell outside of the envelope of the stress–strain behaviour of the trypsin-treated samples. This trend was consistent across each of the values of strain. Mechanical loading at the lower strain rate of 0.025 s 1 produced a similar trend with an average reduction in stiffness after degradation of 15–20% (Fig. 3). Again, only three specimens from the osteoarthritic and trypsin-treated groups fell below
Fig. 1. The concentration of proteoglycans by absorbance value for normal, trypsin-treated and osteoarthritic samples. Normal samples followed similar patterns, while trypsin treatment produced a wavefront of proteoglycan depletion. Samples from osteoarthritic joints were characterised by a more uniform depletion of proteoglycans over the depth of the tissue.
Fig. 2. The stress–strain curves in (a) showed a high variation in stiffness for normal (N) samples at 0.1 s 1. The labels ‘‘high’’ and ‘‘low’’ refer to the range of the samples in each group. Most osteoarthritic (OA) samples and samples degraded by trypsin treatment (T) fall within the normal variation. A loading rate of 0.025 s 1 (b) produced a similar range of behaviour to that of 0.1 s 1. The change in the stress–strain characteristic after trypsin treatment was similar to that of 0.1 s 1, but a greater overlap of normal and osteoarthritic samples was observed.
the range of stress–strain behaviour of normal samples. The stress–strain characteristic of the osteoarthritic samples showed a greater deviation from those of the normal at lower strains with 33% of the samples falling outside those for the normal at 10% strain. At 30% strain, however, the stress strain behaviour of the osteoarthritic samples overlap with the normal samples. Fig. 3a and b show the overlap of the stress–strain characteristics at low strain (a) and high strain (b). A similar mechanical effect to those of trypsin-treated samples was also obtained in the normal samples when the loading rate was changed, as shown in Fig. 2. This change produced an average reduction in stiffness of 20– 26%, increasing with increasing strain. The stiffness values for the samples produced similar results to the patterns of the stress–strain behaviour, as shown in Fig. 4. At 10% strain, 6% of the trypsin-treated
846
C.P. Brown et al. / Clinical Biomechanics 22 (2007) 843–848
Fig. 4. The stiffness of normal, trypsin-treated and osteoarthritic samples at strains of 0.1 (a) and 0.3 (b), tested at 0.1 s 1.
Fig. 3. (a) shows the overlap at low strain and (b) at high strain at 0.1 s 1. The percentage figures represent the number of samples that fall outside of the envelope at each strain.
samples and 22% of the osteoarthritic samples exhibited responses that fell below those obtained for normal samples. At 30% strain, 22% of the trypsin-treated samples and no osteoarthritic samples responded with stiffness values that were lower than and outside the envelope of those for the normal samples. 4. Discussion Mechanical test results from the indentation of normal cartilage-on-bone samples showed a large variation in stiffness. This has also been observed in previously published in vivo (Vasara et al., 2005) and in vitro (Broom and Flachsmann, 2003; Kempson, 1982) investigations. A change in the loading rate produced a further variation in mechanical behaviour in accordance with the previously published work of Oloyede et al. (1992). Given the variation in
normal samples, these results lead to the question of whether the measurements from a mechanical indentation test are meaningful for distinguishing between normal and degraded articular cartilage. For the purpose of comparing the effects of the rate of loading on our results, we have driven the applied loads at the rates of 0.025 s 1 and 0.1 s 1, where the latter velocity of loading approximates that applied in clinical indentation assessments (Vasara et al., 2005). The different loading rates produced similar patterns of results, with an average change in stiffness after the 1 h trypsin treatment of 25%. This large change, however, only produced a stress–strain characteristic outside the envelope of the same parameter for the normal samples in approximately 17% of the degraded samples at the loading rates of 0.1 s 1 and 0.025 s 1 used in this study. Such ambiguity would potentially complicate, and introduce errors into tissue differentiation or classification based on mechanical data. From Figs. 2–4 we can see that 78% of the normal samples fell within the envelope of the stress–strain characteristic of the trypsin-treated samples and may therefore be classed as degraded by a mechanical test. Likewise, 83% of the trypsin-treated samples fell within the envelope of normal samples suggesting that a large number of degraded tissues could be classed as normal. Similar results were obtained for osteoarthritic samples at 10% strain. Applying a 90% confidence interval reduced these figures to 66% and 73%, respectively. The results presented in Fig. 2 suggest that if the loading rate cannot be controlled, the resultant stiffness
C.P. Brown et al. / Clinical Biomechanics 22 (2007) 843–848
may change considerably, resulting in an increase in the overlap that will add to the already large error. Although the normal and degraded groups showed a large variation in their stiffness, the stress–strain behaviour maintained similar basic attributes. The osteoarthritic samples were generally characterised by slightly lower resultant stresses in the toe region and higher stresses at the later stages of deformation when compared to the proteoglycan-depleted samples. This effect is most likely due to surface disruption in the osteoarthritic samples, which we argue, would act to decrease the stiffness at very low strains. Overall, a large proportion of the stress–strain curves from the normal and degraded groups showed similar characteristics across the three groups of tissue tested, indicating that a normal sample may behave similarly to an osteoarthritic sample or a cartilage sample devoid of a large portion of its proteoglycans. Earlier studies investigating the application of load– deformation or stress–strain-based data in normal and degraded cartilage (Lyyra et al., 1999) suggest the possibility of obtaining a relative change in degraded tissue properties from a known normal. Since the mechanical characteristics of normal cartilage samples vary and can overlap significantly with degraded tissue, it can be concluded that it is almost impossible to define a normal/ healthy tissue reference state against which degradation can be determined, even within the same joint sample. For example, it is possible to choose at random, a single specimen from the 36 normal and proteoglycan-depleted samples. Even with the known extremes of the data set and a single, controlled loading rate, the probability of distinguishing such a sample as either normal or abnormal is only 19.4%. The probability of distinguishing the same specimen with 90% confidence is 30.5%. For the osteoarthritic samples, the probablity of distinguishing a random sample as either normal or abnormal using our present mechanical data is 38%. Such an analysis can also be applied to the published in vivo results from human cartilage (Vasara et al., 2005) in which a commercially available arthroscopic indenter was used to compare the ‘‘stiffness’’ of normal and ICRS grade 1 tissue. In this case, the probability of distinguishing either a normal or abnormal tissue, taking into account the variations between the different joint surfaces in the knee, is 22%. Despite a statistically significant trend being observed, the vast majority of indentation results from the osteoarthritic tissue fell within the range of the results for normal samples. It should be noted that the in vivo test was performed on ICRS grade 1 sites which by definition (International Cartilage Repair Society, 2000) are characterised by softening or visible surface cracking. It seems therefore that the surgeon, unaided, can distinguish degraded tissue more reliably than an indentation device, leading to the argument that such a method might not be capable, on its own, of enhancing the arthroscopic evaluation of cartilage defects. It should be noted that the indenter diameter used in the experimental investigation is larger than that used in the
847
commercially available arthroscopic indenter. It is further possible that different loading paths/regimes used to achieve the same level of indentation may produce a modification in the patterns of variation. It was observed in this analysis, however, that the variations obtained in our results were very similar to those observed from studies using a commercially available arthroscopic indenter (Vasara et al., 2005) and therefore argue that they form a reasonable representation of the patterns of overlap from clinical indentation. Furthermore, similar patterns of overlap were obtained across different levels of indentation and at different loading velocities. The large variation in the mechanical behaviour of normal tissues and the lack of discrepancy between normal and degraded behaviours have strong implications for the use of a mechanical test for characterising functional abnormality in cartilage, with significant consequences for surgical decisions. In conclusion, our investigation demonstrates that while the mechanical test remains a valuable method for tracking progressive changes in structure via sequential testing, it cannot reliably distinguish either normality or abnormality in a given joint. Acknowledgements The authors wish to thank the Australian Research Council and Smith & Nephew for supporting this research. References Appleyard, R.C., Ghosh, P., Swain, M.V., 1999. Biomechanical, histological and immunohistological studies of patellar cartilage in an ovine model of osteoarthritis induced by lateral meniscectomy. Osteoarthritis and Cartilage 7, 281–294. Arokoski, J., Jurvelin, J., Kiviranta, I., Tammi, M., Helminen, H.J., 1994. Softening of the lateral condyle articular cartilage in the canine knee joint after long distance (up to 40 km/day) running training lasting one year. International Journal of Sports Medicine 15, 254–260. Borthakur, A., Shapiro, E.M., Beers, J., Kudchodkar, S., Kneeland, J.B., Reddy, R., 2000. Sensitivity of MRI to proteoglycan depletion in cartilage: comparison of sodium and proton MRI. Osteoarthritis and Cartilage 8, 288–293. Broom, N.D., Flachsmann, R., 2003. Physical indicators of cartilage health: the relevance of compliance, thickness, swelling and fibrillar texture. Journal of Anatomy 202, 481–494. Duvvuri, U., Kudchodkar, S., Reddy, R., Leigh, J.S., 2002. T(1rho) relaxation can assess longitudinal proteoglycan loss from articular cartilage in vitro. Osteoarthritis and Cartilage 10, 838–844. Franz, T., Hasler, E.M., Hagg, R., Weiler, C., Jakob, R.P., Mainil-Varlet, P., 2001. In situ compressive stiffness, biochemical composition, and structural integrity of articular cartilage in the human knee joint. Osteoarthritis and Cartilage 9, 582–592. Harris, E.D.J., Parker, H.G., Radin, E.L., Krane, S.M., 1972. Effects of proteolytic enzymes on structural and mechanical properties of cartilage. Arthritis and Rheumatism 15, 497–503. Hayes, W.C., Keer, L.M., Herrmann, G., Mockros, L.F., 1972. A mathematical analysis for indentation tests of articular cartilage. Journal of Biomechanics 5, 541–551. International Cartilage Repair Society, 2000. ICRS Cartilage Injury Evaluation Package. In: ICRS 2000 Standards Workshop. International Cartilage Repair Society: Switzerland.
848
C.P. Brown et al. / Clinical Biomechanics 22 (2007) 843–848
Kempson, G.E., 1982. Relationship between the tensile properties of articular cartilage from the human knee and age. Annals of the Rheumatic Diseases 41, 508–511. Kempson, G.E., Spivey, C.J., Swanson, S.A.V., Freeman, M.A.R., 1971. Patterns of cartilage stiffness on normal and degenerative human femoral heads. Journal of Biomechanics 4, 597–609. Korhonen, R.K., Laasanen, M.S., Toyras, J., Lappalainen, R., Helminen, H.J., Jurvelin, J.S., 2003. Fibril reinforced poroelastic model predicts specifically mechanical behaviour of normal, proteoglycan-depleted and collagen degraded articular cartilage. Journal of Biomechanics 36, 1373–1397. Lyyra, T., Arokoski, J.P.A., Oksala, N., Vihko, A., Hyttinen, M., Jurvelin, J.S., Kiviranta, I., 1999. Experimental validation of arthroscopic cartilage stiffness measurement using enzymatically degraded cartilage samples. Physics in Medicine and Biology 44, 525–535. Moody, H.R., Brown, C.P., Bowden, J.C., McElwain, D.L.S., Crawford, R.W., Oloyede, A., 2006. In vitro degradation of articular cartilage – does trypsin treatment produce consistent results? Journal of Anatomy 209, 259–267. Oloyede, A., Flachsmann, R., Broom, N.D., 1992. The dramatic influence of loading velocity on the compressive response of articular cartilage. Journal of Connective Tissue Research 27, 1–15.
Rieppo, J., Toyras, J., Nieminen, M.T., Kovanen, V., Hyttinen, M.M., Korhonen, R.K., Jurvelin, J.S., Helminen, H.J., 2003. Structure– function relationships in enzymatically modified articular cartilage. Cells Tissues Organs 175, 121–132. Swann, A.C., Seedhom, B.B., 1993. The stiffness of normal articular cartilage and the predominant acting stress levels. British Journal of Rheumatology 32, 16–25. To¨yra¨s, J., Lyyra-Laitinen, T., Niinima¨ki, M., Lindgren, R., Nieminen, M.T., Kiviranta, I., Jurvelin, J.S., 2001. Estimation of the Young’s modulus of articular cartilage using an arthroscopic indentation instrument and ultrasonic measurement of tissue thickness. Journal of Biomechanics 34, 251–256. Vasara, A.I., Jurvelin, J., Peterson, L., Kiviranta, I., 2005. Arthroscopic cartilage indentation and cartilage lesions of anterior cruciate ligament-deficient knees. International Journal of Sports Medicine 33, 408–414. Yao, J.Q., Seedhom, B.B., 1993. Mechanical conditioning of articular cartilage to prevalent stress. British Journal of Rheumatology 33, 16–25. Zhang, M., Zheng, Y.P., Mak, A.F., 1997. Estimating the effective Young’s modulus of soft tissues from indentation tests – nonlinear finite element analysis of effects of friction and large deformation. Medical Engineering and Physics 19, 512–517.