Indentation testing of human cartilageSensitivity to articular surface degeneration.код для вставкиСкачать
ARTHRITIS & RHEUMATISM Vol. 48, No. 12, December 2003, pp 3382–3394 DOI 10.1002/art.11347 © 2003, American College of Rheumatology Indentation Testing of Human Cartilage Sensitivity to Articular Surface Degeneration Won C. Bae,1 Michele M. Temple,1 David Amiel,1 Richard D. Coutts,1 Gabriele G. Niederauer,2 and Robert L. Sah1 degeneration as indicated by macroscopic appearance, India ink staining, and histopathology score. Conclusion. The indentation stiffness of the normal to mildly degenerate samples tested in this study was sensitive to mild degeneration at the articular surface and was insensitive to changes associated with normal aging or to slight variations in cartilage thickness. This suggests that indentation testing may be a useful clinical tool for the evaluation of early-stage degenerative changes in articular cartilage. Objective. To determine, for clinical indentation testing of human articular cartilage, the effects of aging and degeneration on indentation stiffness and traditional indices of cartilage degeneration; the relationship between indentation stiffness and indices of degeneration; and the sensitivity and specificity of indentation stiffness to cartilage degeneration. Methods. Osteochondral cores from femoral condyles of cadaveric human donors were harvested. Samples were distributed into experimental groups based on donor age (young [20–39 years], middle [40–59 years], and old [>60 years]), and a macroscopic articular surface appearance that was either normal or mildly degenerate, without deep erosion. Samples were analyzed for indentation stiffness, cartilage thickness, India ink staining (quantitated as the reflected light score), and Mankin-Shapiro histopathology score. Results. Indentation stiffness, India ink staining, and the histopathology score each varied markedly between normal-sample and degenerate-sample groups but varied relatively little between normal samples obtained from different age groups. A decrease in indentation stiffness (softening) correlated with a decrease in the reflectance score and an increase in the overall histopathology score, especially the surface irregularity component of the histopathology score. Receiver operating characteristic analysis suggested that the indentation testing could accurately detect cartilage Short-duration indentation testing is being considered as a diagnostic tool for assessing the biomechanical properties of human cartilage (1–6). In the laboratory setting, indentation tests of long duration have been performed routinely on human or animal tissue as an end point (post mortem) method of evaluation (7–13). In the clinical setting, however, such traditional laboratory tests are not practical because of the long duration of the test and the large size of the test apparatus. Short-duration indentation testing, performed using an arthroscopic probe, may be useful and appropriate in the clinical setting. A short-duration indentation test of articular cartilage in adult humans may be affected by both degenerative changes and normal age-associated changes in the cartilage. Although several biomechanics studies have analyzed human articular cartilage using a rapid indentation test (2,4–6,12,14–17), and others have examined the relationships between short-duration indentation stiffness and traditional measures of cartilage degeneration (5,14,17,18), the effects of normal aging and mild cartilage degeneration remain to be established. The establishment of baseline indentation measurements for articular cartilage in the setting of normal aging would be useful to help interpret an indentation test result in the clinical setting. Determination of the 1 Won C. Bae, MS, Michele M. Temple, MS, David Amiel, PhD, Richard D. Coutts, MD, Robert L. Sah, MD, ScD: University of California–San Diego, La Jolla; 2Gabriele G. Niederauer, PhD: OsteoBiologics, Inc., San Antonio, Texas. Address correspondence and reprint requests to Robert L. Sah, MD, ScD, Department of Bioengineering, 9500 Gilman Drive, Mail Code 0412, University of California–San Diego, La Jolla, CA 92093-0412. E-mail: firstname.lastname@example.org. Submitted for publication October 8, 2002; accepted in revised form August 8, 2003. 3382 SENSITIVITY OF INDENTATION TESTING TO CARTILAGE DEGENERATION sensitivity to features of mild degeneration would help define the efficacy of indentation testing as a method of cartilage evaluation and would also help determine the basis for variations in indentation stiffness. The extent of articular cartilage degeneration has traditionally been analyzed by macroscopic viewing and histopathologic grading. Macroscopically, the grading system described by Collins (19) distinguishes early degeneration by defining grade 0 as cartilage with a smooth surface and grade 1 as cartilage with fibrillation limited to the superficial zone; grades 2–4 describe cartilage with increasing levels of erosive degeneration. The application of India ink highlights surface abnormalities and characterizes the increase in cartilage degeneration with increasing age (20). India ink contains carbon black particles and clusters with a diameter of ⬃40–100 nm (21,22). Such India ink particles are prevented from entering into normal articular cartilage, the average pore size of which is ⬃6 nm (23), but are able to stain cartilage that has articular surface fibrillation (24). India ink particles are entrapped in surface irregularities and also adhere to fibrillated cartilage (21). Because India ink particles absorb and scatter incident light (25), the reflection of light from cartilage depends on the degree of India ink staining. Application of digital video imaging and analysis to the surface of ink-stained cartilage samples allow for quantitation and localization of degeneration over broad surface areas of joints (26). Histopathologic analysis has provided more detailed structural information about cartilage degeneration, from superficial fraying and splitting of the articular surface in mild cases, to deep fissures and extensive erosion extending beyond the superficial layer of the cartilage in advanced cases (27). Histopathologic features of articular cartilage structure, cartilage cellularity, and tidemark integrity, as well as histochemical staining of glycosaminoglycan with Safranin O, are often tallied using the Mankin scoring system (28), or the modified Mankin scoring system described by Shapiro and Glimcher (29), to grade the severity of osteoarthritic cartilage degeneration. The relationships between the indentation stiffness and the material and structural properties of cartilage have been analyzed theoretically with simplifying model assumptions and have been examined experimentally in a number of studies. Theoretically, the indentation stiffness of articular cartilage has been related to the biomechanical material properties (i.e., Young’s modulus and Poisson’s ratio) of cartilage tissue as well as the thickness of the cartilage layer (30,31). As cartilage 3383 thickness increases to a sufficient degree, the indentation stiffness is predicted to become essentially independent of thickness. However, the indentation stiffness of human articular cartilage is likely to depend also on complex features, such as depth variation in tissue biomechanical material properties (32), especially the tensile modulus of the superficial region (33), which decreases with advancing age (34). Such variations, as well as structural features (e.g., fissures) associated with degeneration (27), are not easily incorporated into theoretical models. Experimentally, short-duration indentation stiffness has been related to structural measures of cartilage degeneration. The indentation stiffness of human femoral head samples was found to be markedly decreased in samples with a gross structural grade of osteoarthritis but variably decreased in samples with a gross grade of mild degeneration (17,18). The relationship between indentation stiffness and cartilage degeneration has been clarified somewhat by more detailed histopathologic analysis. Indentation stiffness was decreased in human patellar samples with surface fibrillation (14) and in relatively old (mean age 65 years) human femoral condyle samples with increasing Mankin scores (28). More detailed studies would be useful to assess further whether indentation stiffness is affected by age or surface-specific structural measures of cartilage degeneration, especially in early stages. The receiver operating characteristic (ROC) curve can be used to define the performance of a diagnostic test (35,36). A ROC plot illustrates the relationship between the proportion of true-positive and false-positive cases for a population of test samples and either specific test results or result thresholds that attempt to distinguish disease from nondisease cases. In addition, the shape of the ROC plot and the area under the ROC plot are features that indicate the diagnostic accuracy of a test (36,37). Such ROC analysis has not been applied previously to indentation stiffness to test for various indicators of cartilage degeneration. The objectives of this study on human articular cartilage were to determine 1) how indentation cartilage stiffness, measured using a handheld arthroscopic probe (3), varies within a sample as well as between samples, as a function of donor age and tissue degeneration, 2) how these indentation stiffness values correlate with structural indices of cartilage degeneration, including India ink staining, histopathology, and cartilage thickness, and 3) the sensitivity and specificity of indentation stiffness measurements to the indices of degeneration by ROC analysis. 3384 BAE ET AL MATERIALS AND METHODS Sample preparation. Using a surgical coring device (Osteochondral Autograft Transfer System; Arthrex, Naples, FL), 10-mm–diameter osteochondral cores were harvested from the anterior region of the lateral and medial femoral condyles of fresh (not previously frozen) cadaveric human knee joints, obtained from tissue banks. Each core was notched at the posterior edge to maintain orientation throughout testing, immersed in a solution of Dulbecco’s phosphate buffered saline (PBS; 2.667 mM KCl, 1.471 mM KH2PO4, 138 mM NaCl, 8.1 mM Na2HPO4–7H2O [pH 7.2]), and proteinase inhibitor (PI; 1 mM phenylmethanesulfonyl fluoride, 2 mM disodium ethylenediamine tetraacetate, 5 mM benzamidine– HCl, and 10 mM N-ethylmaleimide) (38) at 4°C for 1 hour, and then stored at ⫺70°C until the time of testing. Previous studies demonstrated that indentation stiffness measurements are not affected by a single freeze–thaw cycle (18,39–41). Indentation testing. Each sample was thawed by immersion in PBS plus PI at ⬃22°C and clamped at the bony region into a custom jig to stabilize the sample. Then, indentation testing was performed at a site near the center, 3.5 mm away from the edge opposite the notch, using a handheld ACTAEON Probe (OsteoBiologics, San Antonio, TX). With each manual trigger, the ACTAEON Probe advanced its nonporous, hemispheric 1-mm–diameter tip to a depth of up to 100 m for a duration of up to 0.25 second, and provided a stiffness measurement on a scale of 0–99. The probe was calibrated to American Society for Testing and Material type A durometer standards (ASTM D2240). On this scale, a relatively low value indicates a relatively soft sample. For each cartilage specimen, a total of 9 stiffness measurements were obtained at the indentation site. To achieve this, the probe tip was applied 3 times to the test site location, triggered consecutively 3 times during each application, and removed from the test site. The stiffness of the sample was calculated as the average of these 9 measurements. This measurement procedure typically required ⬍15 seconds for an individual sample. India ink staining and image analysis. Samples were analyzed by India ink staining and video image analysis to obtain a light reflectance score, as described previously (42). Briefly, the articular surface was swabbed with a solution of India ink in PBS plus PI (1:5), wiped to remove excess ink, and imaged (8-bit, gray scale, 640 ⫻ 480 pixels, 0.1 mm ⫻ 0.1 mm per pixel) using a video acquisition card (LG3; Scion, Frederick, MD) and NIH Image 1.59 software (National Institutes of Health, Bethesda, MD) on a Power Macintosh 7100/80 (Apple Computer, Cupertino, CA). Light reflectance scores from the cartilage surface in the images were normalized to gray-scale calibration targets (Q13; Eastman Kodak, Rochester, NY) that were chosen to approximate the reflectance of light from normal, nonstaining cartilage (gray no. 3 normalized value ⫽ 1) and fibrillated, maximally ink-stained cartilage (gray no. 19 normalized value ⫽ 0). A relatively low reflectance score corresponds to relatively high India ink staining and cartilage degeneration. Cartilage thickness measurement. The thickness of articular cartilage of each core was measured by obtaining and analyzing cartilage images. Each core was positioned to enable imaging at 6 equally spaced circumferential locations. The thickness of cartilage was calculated as the average of the 6 measurements for each sample. Control studies confirmed that this method of cartilage thickness determination provided values similar to those measured on histologic sections obtained from near the center of the core (Pearson’s correlation, slope ⫽ 0.99, R2 ⫽ 0.42). Histopathologic analysis. An osteochondral fragment containing the indentation site was obtained for histopathologic analysis. The fragment was isolated using a 3.5-mm– diameter biopsy punch (part no. 33-33; Miltex Instrument Company, Bethpage, NY), fixed by immersion in 10% neutralbuffered formalin with 1% cetylpyridinium chloride, decalcified, embedded in paraffin, sectioned to 6 m, and stained with Safranin O–fast green (43). Images of the full-thickness cartilage were obtained on an inverted microscope (Nikon Eclipse TE300; Nikon, Melville, NY) using transmitted light microscopy at magnifications of 4⫻ and 10⫻ (field of view [FOV] 2.5 ⫻ 3.9 mm2 and 1.0 ⫻ 1.5 mm2, respectively) and a digital camera (Kodak Microscopy Documentation System 290; Eastman Kodak, New Haven, CT). The 4⫻ images were used to observe overall structural characteristics of the section, while the 10⫻ images were used in the analysis of cellularity. These images were then analyzed for degeneration by determining a histopathology score (29). This histopathology score combined the gross characteristics of articular cartilage with the histologic appearances of surface irregularity, vertical clefts into the transitional zone, vertical clefts into the radial zone, transverse clefts, cloning, hypocellularity, and Safranin O staining. Each characteristic was given a score of 0–2 or 0–3, with 0 representing normal and 2 or 3 representing severe degeneration. The sum of these scores provided the overall histopathology score (scale 0–15), with a relatively high score corresponding to a relatively degenerate cartilage sample. In addition, the surface irregularity component of the histopathology score was added separately to assess surface-specific structural degeneration in each sample. Experiment 1: sources of variation in indentation stiffness. The sources of variability in indentation stiffness measurements were assessed in twenty 10-mm–diameter human osteochondral cores. These samples were chosen at random and thus included a wide range of donor ages and stages of cartilage degeneration. Three observers independently tested the same cores, and both intraobserver and interobserver sources of variability were assessed. Because variability was likely to depend on the extent of degeneration, the samples were categorized into 4 experimental groups, based on their average indentation stiffness score (⬍30, 30–50, 50–70, and ⬎70). The data were analyzed to determine the extent of intraobserver variability, particularly that attributable to “triggering” of the probe (3 consecutive triggers without moving the probe) and “manual repositioning” of the probe (removal of the probe tip from the cartilage surface, reapplication of the probe tip, and measurement). The measures of variability were the standard deviation (SD), the root mean squared coefficient of variation (CV) of repeated measurements (44), and the standardized CV (sCV; CV normalized to population variability) (45,46). Interobserver variability was calculated as the SD, the CV, and the sCV of the sample indentation stiffness (i.e., the average of all 9 measurements on a sample as determined by an individual observer) between the 3 observers. In addition, to check for any systematic changes in indentation stiffness (e.g., those attributable to cumulative creep) during triggering, each of the 3 consecutive measurements (3 trigger- SENSITIVITY OF INDENTATION TESTING TO CARTILAGE DEGENERATION 3385 Figure 1. Representative photographs (A and B) and corresponding histologic micrographs (C and D; Safranin O–fast green stained) of young-age normal and old-age degenerate human cartilage specimens, respectively. The articular surface is intact in the young-age normal sample and is roughened in the old-age degenerate sample. ings) were linearly regressed to values of 1, 2, and 3, respectively, and the resulting slopes were averaged. A slope of 0 indicated zero variability during triggering. Similarly, systematic changes during repositioning were assessed for 3 repositionings in each sample. No cumulative creep was observed during either triggering or repositioning, because the slopes (mean ⫾ SD) of linear fits were 0.3 ⫾ 2.5 and ⫺0.6 ⫾ 3.5, respectively. Experiment 2: indentation stiffness and indices of cartilage degeneration. The relationship between indentation stiffness and indices of cartilage degeneration was analyzed using seventy 10-mm–diameter osteochondral cores. These samples were selected based on a macroscopic articular surface appearance that was either smooth (normal, n ⫽ 58) (Figure 1A) or roughened (degenerate, n ⫽ 12) (Figure 1B). In addition, samples were categorized into 3 adult age groups: young (20–39 years, n ⫽ 18), middle (40–59 years, n ⫽ 24), and old (ⱖ60 years, n ⫽ 28). This provided sample groups of young-age normal (YN), middle-age normal (MN), middle-age degenerate (MD), old-age normal (ON), and old-age degenerate (OD). Insufficient numbers of samples were obtained for a young-age degenerate group. Together, these samples were obtained from 38 human donors (17 women and 21 men), with 36 from the lateral femoral condyle (LFC) and 34 from the medial femoral condyle (MFC). Samples were tested by indentation testing, India ink staining and image analysis, cartilage thickness measurements, and histopathologic analysis, as described above. The resultant data were analyzed statistically (as follows), using SYSTAT 9 software (Systat, Richmond, CA). Effects of location, aging, and degeneration. Several statistical methods were used to determine the effects of anatomic location (LFC and MFC) and experimental group (YN, MN, MD, ON, and OD) on indentation stiffness, the reflectance score, cartilage thickness, and histopathology scores (overall score as well as scores for individual characteristics). First, to investigate the effect of LFC and MFC 3386 locations, t-tests were performed to compare LFC and MFC samples for each of the 5 experimental groups, with ␣ ⬍ 0.01 (Bonferroni adjustment) used as the significance level to reflect multiple planned comparisons. The effect of each experimental group (YN, MN, MD, ON, and OD) within each of the LFC and MFC locations was analyzed separately by one-way analysis of variance (ANOVA). Using Tukey’s test, the effect of aging was determined by comparing the normal-sample groups (YN, MN, and ON), while the effect of degeneration was assessed by comparing the YN (young adult control tissue) against all degenerate-sample groups (MD and OD) as well as by comparing normal and degenerate samples in age-matched groups (i.e., MN versus MD, and ON versus OD). (Because many of the LFC and MFC samples were obtained from different donors, the alternative statistical approach of repeatedmeasures ANOVA and nonparametric Wilcoxon’s tests was not used, because such an approach would have disregarded a large portion of the data.) Analogously, for the nonparametric measures (i.e., the histopathology score), the effects of LFC and MFC locations were analyzed using the Mann-Whitney U test (with adjusted significance level), and the effect of the experimental group within each of the LFC and MFC locations was analyzed with the Kruskal-Wallis test and Dunn’s test. Data are presented as the mean ⫾ SEM, unless indicated otherwise. Relationships between indentation stiffness and indices of degeneration. To determine which of the many parameters were closely related to each other, a factor analysis using the method of principal components was performed. Indentation stiffness, reflectance score, cartilage thickness, overall histopathology score, and donor age were the parameters included for this analysis. The principal components matrix was computed, and the loading coefficients (describing how each parameter contributed to each factor) were obtained. For each factor, the dominant contributors were identified as those whose coefficients (ranging from ⫺1 to ⫹1) had an absolute value of ⬎0.5. As many factors as were necessary were calculated, in order to account for each parameter as a dominant contributor. In addition, analyses were conducted to determine whether indentation stiffness was related to cartilage thickness or surface damage. Indentation stiffness was correlated to the reflectance score and cartilage thickness by parametric univariate linear regression analysis (47). Indentation stiffness was correlated to the histopathology scores by the nonparametric Spearman’s rank method (47) (because the subjective histopathology score was an ordinal variable rather than a continuous variable). Sensitivity and specificity of indentation stiffness to indices of degeneration. To determine the accuracy of the indentation stiffness measurements in distinguishing various traditional indices of cartilage degeneration, sensitivity and specificity values were determined. The sensitivity of indentation stiffness to macroscopic degeneration was determined as the proportion of the number of data points from macroscopically degenerate samples (by indentation stiffness), of the total number of data points from macroscopically degenerate samples (true positives/total positives), at a particular indentation stiffness cutoff value. For example, at an indentation stiffness cutoff value of 50, there were 12 data points from macroscopically degenerate samples (total positive), of which 11 had indentation stiffness values of ⬍50 (and therefore were true BAE ET AL positives for macroscopic degeneration). Thus, the sensitivity at the indentation stiffness cutoff value of 50 was 0.92 (11 of 12). For specificity, at the indentation stiffness cutoff value of 50, there were 58 data points from macroscopically normal samples (total negative), of which 55 had indentation stiffness values of ⬎50 (true negative). Thus, the specificity was 0.95 (55 of 58). Using these methods, sensitivity and specificity values were calculated for a range of indentation stiffness cutoff values. For the calculation of the sensitivity and specificity of indentation stiffness to other indices of degeneration (i.e., the reflectance score and the overall histopathology score), threshold values for each index were needed in order to classify the sample as apparently normal or degenerate, based on the value of the index. This allowed segregating the scatter plot of data (the indentation stiffness score versus the index of degeneration) into 4 quadrants of true positive, true negative, false positive, and false negative. The status of being true or false would be determined by the value of the index, and the status of being positive or negative would be determined by the value of the indentation stiffness score. Examples chosen to illustrate the effects of varying threshold values were determined as a) the average value for all macroscopically normal samples, b) the average for all macroscopically degenerate samples, and c) the midpoint value between the two, to segregate samples that are roughly normal from those that are roughly degenerate. For reflectance scores, values below the threshold value indicated degenerate samples. For histopathology scores, values above the threshold value indicated degenerate samples. Once the data were segregated in this way, sensitivity and specificity were computed for a range of indentation stiffness cutoff values. To describe the tradeoff between sensitivity to and specificity for each of the above indices of degeneration, ROC plots were generated. For an ideal test, all of the normal and diseased samples would be segregated perfectly (e.g., by using a single cutoff value). In such a case, the ROC plot would rise vertically and reach a plateau after achieving maximum sensitivity, with a resulting area under the curve of 1. Conversely, for the worst possible test, the ROC plot would be a straight line with a slope of 1, resulting in an area under the curve of 0.5. Thus, the ability of indentation testing to distinguish between normal and degenerate samples, as indicated by various indices of degeneration, was determined as the area under the ROC curve. RESULTS Experiment 1: sources of variation in indentation stiffness. The variation in measurements of indentation stiffness was attributable to both intrasample and intersample factors. For an individual sample, the variation due to triggering was generally less than that due to repositioning, in all groups (Table 1). The variation in the indentation stiffness score due to triggering (SDtrigger) ranged from 2.4 to 3.4 for the different experimental groups, with the corresponding CVtrigger ranging from 3.5% to 13.1%. The variation due to repositioning (SDreposition) ranged from 2.9 to 5.6 (CVreposition 6.7– SENSITIVITY OF INDENTATION TESTING TO CARTILAGE DEGENERATION 3387 Table 1. Intersample and intrasample indentation stiffness variability in experiment 1* Intrasample variability Intraobserver Intersample variability Group, indentation score† Avg SD ⬍30 30–50 50–70 ⬎70 24 37 63 77 All 52 Interobserver Repositioning Triggering SD/Avg No. of samples SD CV sCV SD CV sCV SD CV sCV 2.2 4.5 4.1 5.5 9.4 12.0 6.6 7.2 4 4 9 3 4.4 8.3 5.3 8.1 21.4 24.3 9.6 12.5 13.7 15.5 6.1 8.0 3.3 5.6 3.4 2.9 16.4 18.2 7.6 6.7 10.4 11.6 4.9 4.3 2.6 3.4 2.9 2.4 13.1 10.8 5.6 3.5 8.4 6.9 3.6 2.3 19.4 37.4 20 6.2 16.6 10.6 3.7 11.9 7.6 2.0 9.0 5.7 * Values for SD/Avg (standard deviation/average), CV (coefficient of variation), and sCV (standardized CV) are expressed as percentages. For each group, there were 3 observers, 3 repositionings, and 3 triggerings. † Samples were analyzed as 4 experimental groups, based on their average indentation stiffness score. 18.2%). In comparison, the variation between observers was slightly greater, with the SDobserver ranging from 4.4 to 8.3 (CVobserver 9.6–24.3%). However, the interobserver variation in stiffness measurement for a particular sample was smaller than the overall variation between cartilage samples (SD 19, CV 37.4%) (Table 1). Experiment 2: indentation stiffness and indices of cartilage degeneration. Effects of aging and degeneration. Indentation stiffness. The major source of variation in experiment 2, like that in experiment 1, was between samples. The intrasample variations for repositioning (SDreposition 2.5–4.1, CVreposition 5.8–16.0%) (Table 2) and triggering (SDtrigger 2.4–3.2, CVtrigger 4.9–12.0%) (Table 2) in experiment 2 were generally similar to those in experiment 1 (Table 1), especially considering groups with similar (average) stiffness values. The interobserver variations in experiment 1 (Table 1) were also generally smaller than the intersample variations within the groups of experiment 2 (SD 4.2–13.9, CV 4.2–37%) (Table 2). Thus, the indentation test protocol provided measurements that were reasonably precise for an indi- vidual sample, relative to the variation between samples of the same experimental group. Indentation stiffness was slightly different between the MFC and LFC locations for certain experimental groups, and was markedly different among the experimental groups within the LFC and MFC locations (Figures 2A and B). YN samples were stiffer at the LFC location (P ⬍ 0.05) than at the MFC location (difference in indentation stiffness score ⫽ 9). Other experimental groups (MN, MD, ON, and OD) did not show statistically significant differences between the LFC and MFC locations. In contrast, at both the LFC and MFC locations, indentation stiffness varied markedly (P ⬍ 0.001) among the sample groups, particularly between the normal-sample and degenerate-sample groups. For both the LFC and MFC locations, the normal-sample age groups (YN, MN, and ON) had stiffness values that were indistinguishable (P ⬎ 0.8). In contrast, the degeneratesample groups (OD and MD) had stiffness values that were significantly lower than those of the YN group (P ⬍ 0.001 in each case) as well as the age-matched Table 2. Indentation stiffness variability in experiment 2* Intrasample variability Intersample variability Repositioning Triggering Group Avg SD SD/Avg No. of samples YN MN MD ON OD 68 65 30 65 37 7.1 10.2 4.2 10.0 13.9 10.4 15.7 14.0 15.5 37.3 18 20 4 20 8 3.2 2.5 2.9 4.1 4.0 5.8 6.3 9.9 9.6 16.0 5.6 6.0 9.5 9.2 15.4 2.6 2.8 2.4 2.8 3.2 4.9 5.6 10.5 5.7 12.0 4.7 5.4 10.1 5.5 11.5 All 60 15.2 25.2 70 3.4 9.0 8.6 2.6 5.5 5.2 SD CV sCV SD CV sCV * Values for SD/Avg (standard deviation/average), CV (coefficient of variation), and sCV (standardized CV) are expressed as percentages. For each group, there were 3 repositionings and 3 triggerings. YN ⫽ young-age normal; MN ⫽ middle-age normal; MD ⫽ middle-age degenerate; ON ⫽ old-age normal; OD ⫽ old-age degenerate. 3388 Figure 2. Indentation stiffness score (A and B), reflectance score (C and D), cartilage thickness (E and F), and histopathology scores of overall (G and H) and surface irregularity characteristics (I and J), grouped according to adult age and macroscopic grade, as well as anatomic location. Surf. irreg. ⫽ surface irregularity; YN ⫽ young-age normal, MN ⫽ middle-age normal, MD ⫽ middle-age degenerate, ON ⫽ old-age normal, OD ⫽ old-age degenerate; LFC ⫽ lateral femoral condyle; MFC ⫽ medial femoral condyle. Values are the mean and SEM. ⴱ ⫽ P ⬍ 0.05; ⴱⴱ ⫽ P ⬍ 0.01; ⴱⴱⴱ ⫽ P ⬍ 0.001. normal-sample groups (ON and MN, respectively; P ⬍ 0.05 in each case). The difference in indentation stiffness between the normal-sample and degenerate-sample groups was ⬃30, which was much greater than the difference between the LFC and MFC locations for each experimental group. Reflectance score. The reflectance score was markedly different between the MFC and LFC locations for certain experimental groups, and among the experi- BAE ET AL mental groups within the LFC and MFC locations (Figures 2C and D). Both the MN and ON groups had significantly (P ⬍ 0.05) greater reflectance scores (by ⬃0.2) at the LFC location than at the MFC location, while the other groups (YN, MD, and OD) had similar (P ⫽ 1.0) reflectance scores at both locations. Similar to the pattern for indentation stiffness, normal and degenerate samples showed large differences in the reflectance score at the LFC location. Although normalsample groups (YN, ON, and MN) had reflectance scores that were not statistically different (P ⬎ 0.08 in each case), degenerate-sample groups (MD and OD) had reflectance scores that were generally lower than those of the YN group (by ⬃0.3–0.5) as well as the corresponding age-matched normal-sample groups (MN and ON, respectively, by ⬃0.3). At the MFC, the reflectance score was also higher in the YN group than in all other groups. However, there was no difference (P ⬎ 0.90 in each case) between normal-sample and degenerate-sample groups of either middle or old age (i.e., MN versus MD, and ON versus OD). These results indicate that after India ink staining, the reflectance score varied with normal aging and degeneration in a location-dependent manner. Cartilage thickness. Cartilage thickness was similar for the LFC and MFC locations and between all experimental groups. Planned comparisons between the MFC and LFC for each experimental group did not reveal any effect of anatomic location (P ⫽ 1.0 for each group). In addition, the experimental groups showed negligible effects of normal aging and degeneration at the LFC (P ⫽ 0.6) (Figure 2E) and MFC (P ⫽ 1.0) (Figure 2F) locations. Cartilage thickness averaged 2.19 ⫾ 0.07 mm for LFC samples and 2.13 ⫾ 0.07 mm for MFC samples. Histopathology score. The overall histopathology scores were similar between LFC and MFC locations but varied markedly among experimental groups within the LFC and MFC locations (Figures 2G and H). The histopathology scores were similar for LFC and MFC locations in each experimental group (P ⬎ 0.4). For the LFC location (Figure 2G), the normal-sample groups (YN, MN, ON) had histopathology scores (average 2.1 ⫾ 1.3) that were indistinguishable (P ⫽ 1.0). Here, the degenerate-sample groups had histopathology scores that either were significantly higher than those of the YN group as well as the age-matched normal-sample groups or that exhibited strong trends for such differences. For the MFC location (Figure 2H), the normalsample groups showed an age-related change, with a significant (P ⬍ 0.01) increase in histopathology score from YN (mean ⫾ SD 1.6 ⫾ 0.2) to ON (3.9 ⫾ 0.3). In SENSITIVITY OF INDENTATION TESTING TO CARTILAGE DEGENERATION Table 3. Principal components analysis* Parameter Reflectance score Histopathology overall score Indentation stiffness score Age Cartilage thickness % of total variance explained by the factor Factor 1 Factor 2 ⴚ0.858 0.843 ⴚ0.804 0.726 0.090 ⫺0.048 0.012 0.079 0.292 0.985 53.3 20.0 * Dominant parameters are shown in boldface. Except where indicated otherwise, values are the loading coefficient of variation. addition, at the MFC location, the degenerate samples had histopathology scores that were significantly higher than those of the YN group (P ⬍ 0.05 in each case) and showed trends toward being higher than those of their respective age-matched normal-sample groups. Among histopathologic features, the surface irregularity characteristic was the one that varied significantly between groups and contributed to the variation in the overall score (Figures 2I and J). The representative micrographs of YN (Figure 1C) and OD (Figure 1D) illustrate the typical surface irregularity and vertical and horizontal clefts into the transitional zone in the degenerate-sample groups. Both the overall histopathology score and the surface irregularity characteristic of the histopathology score showed a pattern (Figures 2G–J) that was fairly similar (inversely related) to that of the indentation stiffness values (Figures 2A and B). Relationship between indentation stiffness and indices of degeneration. Principal components analysis. A total of 2 principal components was needed to account for the major variations in sample parameters (Table 3). Factor 1, which accounted for 53% of the total variance, was determined by 4 dominant (loading coefficient of ⬎0.5) parameters: indentation stiffness, the reflectance score, the histopathology score, and donor age. Factor 2, which accounted for 20% of the total variance, comprised only one dominant parameter, cartilage thickness. The association of the dominant parameters (all parameters listed in Table 3 except cartilage thickness) contributing to factor 1 indicates generally strong correlations between any 2 individual dominant parameters. In contrast, cartilage thickness was the only nondominant parameter contributing to factor 1 while being the only dominant parameter contributing to factor 2. This indicates that cartilage thickness is weakly correlated with or unrelated to any of the other parameters. Linear regression. The indentation stiffness values correlated with certain indices of degeneration. A decrease in indentation stiffness was related to a decrease in the reflectance score (R2 ⫽ 0.35) (Figure 3A), as well as an increase in the overall histopathology score 3389 (2 ⫽ 0.44) (Figure 3C). A decrease in indentation stiffness was also related to an increase in the surface irregularity characteristic of the histopathology score (2 ⫽ 0.34) (Figure 3D). However, indentation stiffness showed no correlation with cartilage thickness (R2 ⫽ 0.01) (Figure 3B). Sensitivity and specificity of indentation stiffness to the indices of degeneration. As expected, the sensitivity of indentation stiffness to macroscopic degeneration increased with an increasing cutoff value for the indentation stiffness score, while specificity exhibited the opposite trend. Sensitivity to macroscopic degeneration (Figure 4A) increased with indentation stiffness cutoff values ranging from 15 to 50, reaching 90% at the indentation stiffness cutoff value of 48. The specificity for macroscopic degeneration (Figure 4A) was maximum at relatively low indentation stiffness cutoff values, and decreased to 90% specificity at the indentation stiffness cutoff of 52. Thus, the indentation stiffness cutoff range of 48–52 provided an area of Figure 3. Relationship between indentation stiffness and the indices of degeneration. A, Reflectance score. B, Cartilage thickness. C, Overall histopathology score. D, Surface irregularity histopathology score. Experimental groups include young-age normal (䊐), middle-age normal (‚), middle-age degenerate (Œ), old-age normal (E), and old-age degenerate (F). Lines represent the linear regression fits of the data and are shown only to indicate trends. For A and B, parametric linear regression analysis was performed to determine P and R2 values. For C and D, Spearman’s rank correlation was used to determine P and 2 values. 3390 BAE ET AL score threshold values of 0.59 (average of normal samples), 0.35 (average of degenerate samples), and 0.47 (midpoint), the area under the ROC curve was 0.82, 0.78, and 0.76, respectively. For the overall histopathology score, the threshold values were 2.3 (average of normal samples), 7.3 (average of degenerate samples), and 4.8 (midpoint). The sensitivity and specificity curves also right-shifted as the histopathology score threshold value decreased (i.e., became increasingly normal) (Figure 4E). However, the amount of shift was greater for sensitivity than for specificity. Consequently, ROC plots (Figure 4F) were distinct for the 3 threshold values. For the histopathology threshold values of 2.3, 7.3, and 4.8, the area under the ROC plots was 0.79, 0.95, and 0.91, respectively. These results indicate that indentation stiffness can distinguish between samples of distinct histopathologic grades. Figure 4. Sensitivity analysis. Left column, Sensitivity (– –, —, – – –) and specificity (‚, }, and E) of indentation stiffness to macroscopic degeneration (A), the reflectance score (C), and the overall histopathology score (E). Right column, Receiver operating characteristic plots (‚, }, and E) of indentation cutoff sensitivity and specificity to macroscopic degeneration (B), the reflectance score (D), and the overall histopathology score (F). Different symbols and lines (C–F) represent the threshold values for the index of degeneration, which were calculated as the average of all macroscopically normal samples (E and – –), the average of all macroscopically degenerate samples (‚ and – – –), and the average of the 2 values (} and —). DGN ⫽ degenerate; NL ⫽ normal. high sensitivity and specificity, with each parameter being ⬎90%. The corresponding ROC curve (Figure 4B) rose rapidly and reached maximum sensitivity at a low 1 ⫺ sensitivity value. When sensitivity increased to 0.9, 1 ⫺ specificity had already decreased to 0.04. The area under the ROC curve was 0.96, which was close to the ideal test value of 1.0. This ROC curve analysis indicates that indentation testing is highly accurate in discriminating between macroscopically normal and degenerate samples. Indentation stiffness was also sensitive to degeneration, as judged by the reflectance and histopathology scores. For the reflectance score, the threshold values, calculated as a) the average of macroscopically normal samples, b) the average of macroscopically degenerate samples, and c) the midpoint between a and b, were 0.59, 0.35, and 0.47, respectively. The sensitivity and specificity curves right-shifted as the reflectance score threshold value increased (i.e., became increasingly normal) (Figure 4C), without marked differences between the ROC plots (Figure 4D). For the reflectance DISCUSSION This study provides a detailed examination of the short-duration indentation stiffness of human articular cartilage, gross morphologic and structural indices of early cartilage degeneration in the same samples, and relationships between indentation stiffness and indices of degeneration. The results indicate that indentation stiffness generally decreases as articular cartilage degeneration progresses to a mild stage, as assessed by gross morphology (Figures 2A and B), India ink staining (Figures 2C and D), and histopathology (Figures 2G–J). The variation in indentation stiffness correlated well with parameters that are indicative of early changes in the articular surface, including India ink staining (Figure 3A) and the surface irregularity characteristic (Figure 3D) of the histopathology score (Figure 3C). For these samples that ranged from normal to an early stage of degeneration (without demonstrating cartilage thinning), indentation stiffness did not depend detectably on cartilage thickness (Figure 3B). For macroscopically normal samples, the indentation stiffness score did not depend on age groups (Figures 2A and B). The indentation stiffness values were diagnostically sensitive to (and specific for) macroscopic degeneration (Figures 4A and B), as well as for degeneration as described by reflectance scores after India ink staining (Figures 4C and D) and histopathology scores (Figures 4E and F). Because variations in indentation stiffness will depend markedly on the types of samples studied, the samples were chosen in accordance with the specific objective of determining whether early stages of degeneration could be discerned by indentation stiffness measurement. Samples from donors ages 28–91 years were SENSITIVITY OF INDENTATION TESTING TO CARTILAGE DEGENERATION chosen to represent a wide range of adult (skeletally mature) samples. In addition, only macroscopically normal and mildly degenerate samples were selected for all age groups, because more severe erosion can be readily identified by macroscopic inspection. It should be noted that such samples are not representative of randomly selected samples from the particular age group, because the frequency and severity of degeneration increase with age (48). Both the LFC and the MFC were of interest to examine, because both sites are prone to age-associated degeneration, with the MFC showing such degeneration at an earlier age (48). Other studies of indentation of human articular cartilage samples have tested these same sites (2,4,5,39) as well as sites on the patellofemoral groove (2,4,5,39), patellae (2,4), tibial plateau (2,4,39), and femoral head (7,17). However, those samples were not chosen according to both age and degeneration grade criteria. The repeatability of measurements from any instrument, when used by one or many individuals, is important to characterize. This is especially true for a handheld mechanical test instrument. The variability in indentation stiffness measurements, expressed as the standard deviation of intraobserver and interobserver measurements (SD 2.4–5.6 and 4.4–8.3, respectively) (Table 1), was small relative to the differences between the normal-sample and degenerate-sample groups (difference in indentation stiffness scores of ⬃30) (Table 2 and Figures 2A and B). Therefore, the probe was able to distinguish differences associated with degeneration. The variability in repeated measurements has not been reported often but would likely be affected by the exact setting in which measurements are made and by the type of samples that are analyzed. Additional studies in clinical situations, or in arthroscopic simulations on cadaveric knee joints, may be useful to characterize the probe further. The intraobserver variation of stiffness measurements obtained by successively placing the probe tip at the “same” site was generally the highest for moderately soft samples. This suggests that spatial gradients in indentation stiffness exist, and that the indentation stiffness measurement is also sensitive to very local variations in cartilage structure. Nevertheless, the standard protocol used here, 9 measurements from 3 tip applications, yielded a precise measure. It is possible, however, that fewer measurements (i.e., only 1 triggering) can be made with more tip applications (i.e., ⬎3 repositionings) to increase the speed of measurement at a particular site. The strong dependence of indentation stiffness on certain indices of degeneration is generally consistent 3391 with the findings of previous studies, whereas the weak variation among the different age groups indicates that age is not a strong confounding variable. Previous studies showed that cartilage indentation stiffness decreased with osteoarthritic degeneration (17,18). Negligible effects of normal aging on indentation stiffness have also been reported previously, although in those studies only gross morphology was used to distinguish between normal and degenerate specimens (18,39). The variation in India ink staining among normal samples at the MFC location (Figure 2D) is consistent with the general increase in ink staining with increasing age in randomly selected samples, at various cartilage surfaces in the human knee joint (20), as well as in osteoarthritis (28). The slight increase in histopathology score, and in particular surface irregularity, also confirms the selection of relatively normal samples but also of slight age- and site-related variations in the normalsample groups. In addition, the lack of significant variation in thickness of the samples in the 5 experimental groups used in this study also indicates that the samples used in this study were appropriately chosen to be in early stages of degeneration without deep erosion, because cartilage is known to become thinner with the more extensive degenerative changes of aging (49,50) and osteoarthritis (49,51). The difference in indentation stiffness scores, determined by ACTAEON probe, between youngnormal LFC and young-normal MFC samples (mean ⫾ SD of 73 ⫾ 6 and 64 ⫾ 7, respectively) (Figures 2A and B) is generally consistent with biomechanical measures in similar human cartilage samples. Another shortduration indentation study using a handheld instrument showed mean ⫾ SD indenter forces of 5.6 ⫾ 1.2 Newtons and 4.9 ⫾ 1.2 Newtons (4) for the LFC and MFC sites, respectively. A long-duration indentation study showed similar results, with a mean ⫾ SD aggregate modulus of 0.70 ⫾ 0.23 MPa for the LFC and 0.59 ⫾ 0.11 MPa for the MFC (13). The tensile strength of the superficial layer of cartilage is another measure that may be closely related to indentation stiffness (33). In a recent study, the mean ⫾ SD tensile strength for LFC and MFC cartilage was found to be 19.6 ⫾ 4.8 MPa and 7.8 ⫾ 7.4 MPa, respectively (52), which also is in agreement with the site-associated difference seen in the present study. However, it is not necessarily straightforward to compare a rapid, clinically applicable indentation test result, as determined here, with biomechanical measures that are commonly derived from laboratory techniques. Factors such as inhomogeneity (32) and anisotropy (53) of articular cartilage affect the manner in which a cartilage 3392 sample behaves under different loading configurations. In a confined compression test, for example, induced strains are mostly in the axial (loading) direction (32). In comparison, short-term indentation loading induces complex intra-tissue strain that consists of axial compression, transverse tension, and shear (30). In addition, comparisons among various indentation methods may even be difficult, because slight differences in indenter geometry and indentation loading protocol significantly affect the extent and magnitude of strains in the human cartilage samples (30). Results of this study are consistent with and extend results of a past study that examined how indentation stiffness varies with cartilage degeneration (5), by defining in more detail the relationship between stiffness and specific features of cartilage degeneration. The significant negative correlation between indentation stiffness and histopathology score (Figure 3C) confirms the findings of a recent study that used a different indentation probe and the Mankin histopathology scoring system (5,28). Because histopathologic grading schemes involve many components of structural and cellular organization of articular cartilage (28,29), it was not clear whether specific components of the degeneration process were responsible for the correlation with indentation stiffness. The detailed histopathologic analyses in this study indicated a particularly strong correlation of indentation stiffness with surface disruption, as indicated by the reflectance score (Figure 3A) and the surface irregularity characteristic of the histopathology score (Figure 3D). This result seems quite reasonable, because indentation causes extension and shear of cartilage near and underneath the indenter and would be predicted to be sensitive to alteration of the properties of the cartilage tissue near and at the articular surface. One of the factors that affected the strength of correlation may be the proximity of the portions of the individual cartilage samples that were used for the various analyses. Indentation testing would be expected to be affected the most by the properties of tissue near the indenter (54). The reflectance score was measured for the whole articular surface of the sample, although additional analyses of more localized regions near the indentation site did not markedly alter the extent of correlation between indentation stiffness and the reflectance score (data not shown). In addition, the histopathology grading examines cartilage that is only a thin vertical tissue section, which represents only a small fraction of the volume of cartilage that is likely to affect the indentation measurements. Nevertheless, the correlations observed suggest that such sample analysis and BAE ET AL comparison provide useful information on the relationships among the different tissue properties. The relationship between indentation stiffness and cartilage thickness is an important factor to consider for clinical application of indentation stiffness. In theory (30,31,55,56), as the cartilage becomes thin, the apparent stiffness measured by the indentation device will increase as the device starts to be influenced by the bone. If cartilage is modeled as a homogeneous, isotropic, and incompressible layer bonded to a rigid base (30), for a sphere-ended indentation depth of 100 m, there is an increase of ⬃20% in the apparent stiffness (defined as load/displacement) as the thickness decreases from 3.5 to 1.5 mm, which encompasses the thickness of cartilage sample in this study. This phenomenon becomes prominent when the cartilage thickness is less than or similar to the indenter characteristic size (i.e., radius). Because it is not always possible in vivo to determine the thickness of the cartilage sample being indented, it is preferred to have an indentation probe that outputs similar stiffness measurements irrespective of the cartilage thickness. In this regard, the lack of correlation between indentation stiffness and cartilage thickness (Figure 3B) indicates that for the normal and mildly degenerate sample populations used in this study, the variation in thickness has no marked effects on indentation stiffness measured by the probe. As cartilage becomes thinner, as it will in more advanced stages of degeneration, the sample thickness may begin to influence the indentation stiffness measurements considerably. However, at these later stages of degeneration, it is likely that the diagnosis of severe degeneration can be made without the aid of an indentation instrument. The indentation stiffness scores can help a user interpret the results of the indentation testing in a number of ways. The values for YN cartilage samples (73 ⫾ 6 and 64 ⫾ 7 for the LFC and MFC, respectively) (Figures 2A and B) may be used as reference (normal) values against which other values measured by the probe on a patient can be compared. In addition, it may be possible to monitor progression of cartilage disease or repair by indentation testing. Based on the values of standard deviation of repeated measurements for indentation stiffness scores (⬃3.5 for repositioning over all of the test populations) (Tables 1 and 2), and based on performing 3 successive measurements as was done here, the 95% confidence interval for a site measurement is ⫾5 (47). This is much less than the difference (⬃30) in indentation stiffness scores between normalsample and degenerate-sample groups. In contrast, variability of measurements between sites and between SENSITIVITY OF INDENTATION TESTING TO CARTILAGE DEGENERATION operators is a factor that would complicate use of indentation stiffness probes, in general. ROC analyses defined the sensitivity and specificity of the indentation stiffness measurement, as well as the accuracy of indentation stiffness as a diagnostic test. The observed area under the ROC plots of 0.76– 0.95 (Figures 4D and F) compares favorably with many of the typical values (of the area under the ROC curve) found for medical diagnostic tests, in particular imaging methods. In a study examining the relationship between radiographic osteoarthritis and patient-reported knee pain, the area under the ROC plots was at best 0.79 (57). In another study comparing the ability of various magnetic resonance imaging field strengths and pulse sequences to detect relatively large (diameter of 4–12 mm) experimental cartilage lesions, the area under the ROC plots ranged from 0.59 to 0.96 for 10 different protocols, with a mean ⫾ SD of 0.76 ⫾ 0.13 (58). The sensitivity and specificity, along with the ROC plots, can help a user interpret the results of the indentation testing by providing objective bases for the accuracy of indentation stiffness measurements for detecting cartilage degeneration. This type of analysis is particularly useful in that the user can adopt it for a wide range of diagnostic situations. For example, for screening situations in which one would be willing to sacrifice specificity to achieve high sensitivity, the user would select an indentation cutoff value that is relatively high (thus more samples are diagnosed as degenerate), and know exactly what the trade-off would be. Although this study focuses on the use of indentation testing as a clinical diagnostic device, there are many other possible uses for indentation testing. In the past, the selection criterion for distinguishing between normal and degenerate cartilage has been gross macroscopic observation. It would be possible to use indentation stiffness alone as a categorization tool, or to use indentation stiffness along with the relationships between indentation stiffness and other indices of degeneration, found in this study, to target sample groups with certain values for these other degeneration indices. In conclusion, the marked changes in indentation stiffness, the reflectance score, and the histopathology score were observed to be primarily associated with cartilage degeneration and not with aging or cartilage thickness. The indentation stiffness appeared to be particularly sensitive to the articular surface disruption evident in degenerate specimens, but insensitive to the range of cartilage thickness seen in this study, which makes it useful for assessing the early-stage degeneration as well as a repaired or regenerated tissue after a 3393 therapeutic intervention. The ROC analyses gave a guideline for diagnosis and a quantitative measure of comparing the accuracy of this device with other means of diagnosing the health of articular cartilage. In principle, indentation stiffness could be used to identify local areas of cartilage degeneration and to instigate appropriate interventions designed to prevent more severe deterioration. Although it is ultimately the responsibility of the user to interpret and apply the indentation stiffness results, these guidelines provide detailed and objective bases for clinical applications. ACKNOWLEDGMENTS We thank Karen Bowden for preparation of the histologic slides, and Arthrex Inc. for providing the Osteochondral Autograft Transfer System. REFERENCES 1. Dashefsky JH. Arthroscopic measurement of chondromalacia of patella cartilage using a microminiature pressure transducer. Arthroscopy 1987;3:80–5. 2. Lyyra T, Jurvelin J, Pitkanen P, Vaatainen U, Kiviranta I. Indentation instrument for the measurement of cartilage stiffness under arthroscopic control. Med Eng Phys 1995;17:395–9. 3. Niederauer MQ, Cristante S, Niederauer GM, Wilkes RP, Singh SM, Messina DF, et al. A novel instrument for quantitatively measuring the stiffness of articular cartilage. Trans Orthop Res Soc 1998;23:905. 4. Lyyra T, Kiviranta I, Vaatainen U, Helminen HJ, Jurvelin JS. In vivo characterization of indentation stiffness of articular cartilage in the normal human knee. J Biomed Mater Res 1999;48:482–7. 5. Franz T, Hasler EM, Hagg R, Weiler C, Jakob RP, Mainil-Varlet P. In situ compressive stiffness, biochemical composition, and structural integrity of articular cartilage of the human knee joint. Osteoarthritis Cartilage 2001;9:582–92. 6. Appleyard RC, Swain MV, Khanna S, Murrell GA. The accuracy and reliability of a novel handheld dynamic indentation probe for analysing articular cartilage. Phys Med Biol 2001;46:541–50. 7. Kempson GE, Freeman MAR, Swanson SAV. The determination of a creep modulus for articular cartilage by indentation tests of the human femoral head. J Biomech 1971;4:239–50. 8. Coletti JM, Akeson WH, Woo SL-Y. A comparison of the physical behavior of normal articular cartilage and the arthroplasty surface. J Bone Joint Surg 1972;54:147–60. 9. Parsons JR, Black J. The viscoelastic shear behavior of normal rabbit articular cartilage. J Biomech 1977;10:21–9. 10. Jurvelin J, Kiviranta I, Tammi M, Helminen HJ. Effect of physical exercise on indentation stiffness of articular cartilage in the canine knee. Int J Sports Med 1986;7:106–10. 11. Mow VC, Gibbs MC, Lai WM, Zhu WB, Athanasiou KA. Biphasic indentation of articular cartilage. II. A numerical algorithm and an experimental study. J Biomech 1989;22:853–61. 12. Jurvelin J, Kiviranta I, Saamanen A-M, Tammi M, Helminen HJ. Indentation stiffness of young canine knee articular cartilage: influence of strenuous joint loading. J Biomech 1990;23:1239–46. 13. Athanasiou KA, Rosenwasser MP, Buckwalter JA, Malinin TI, Mow VC. Interspecies comparisons of in situ intrinsic mechanical properties of distal femoral cartilage. J Orthop Res 1991;9:330–40. 14. Sokoloff L. Elasticity of aging cartilage. Proc Fed Am Soc Exp Biol 1966;25:1089–95. 3394 15. Kempson GE. Mechanical properties of articular cartilage. In: Freeman MAR, editor. Adult articular cartilage. 2nd ed. Tunbridge Wells (England): Pitman Medical; 1979. p. 333–414. 16. Hori RY, Mockros LF. Indentation tests of human articular cartilage. J Biomech 1976;9:259–68. 17. Roberts S, Weightman B, Urban J, Chappell D. Mechanical and biochemical properties of human articular cartilage in osteoarthritic femoral heads and in autopsy specimens. J Bone Joint Surg 1986;68:278–88. 18. Kempson GE, Spivey CJ, Swanson SA, Freeman MA. Patterns of cartilage stiffness on normal and degenerate human femoral heads. J Biomech 1971;4:597–609. 19. Collins DH. The pathology of articular and spinal disease. London: Edward Arnold; 1949. p. 74–115. 20. Meachim G, Emery IH. Quantitative aspects of patello-femoral cartilage fibrillation in Liverpool necropsies. Ann Rheum Dis 1974;33:39–47. 21. Madsen SJ, Patterson MS, Wilson BC. The use of India ink as an optical absorber in tissue-simulating phantoms. Phys Med Biol 1992;37:985–93. 22. Torok A. Physical properties of India ink. In: Director, Research and Development, Sanford Faber Corporation, Newark, NJ; 1995. 23. Maroudas A. Physico-chemical properties of articular cartilage. In: Freeman MAR, editor. Adult articular cartilage. 2nd ed. Tunbridge Wells (England): Pitman Medical; 1979. p. 215–90. 24. Meachim G. Light microscopy of Indian ink preparations of fibrillated cartilage. Ann Rheum Dis 1972;31:457–64. 25. McCluney WR. Introduction to radiometry and photometry. Boston: Artech House; 1994. 26. Chang DG, Iverson EP, Schinagl RM, Sonoda M, Amiel D, Coutts RD, et al. Quantitation and localization of cartilage degeneration following the induction of osteoarthritis in the rabbit knee. Osteoarthritis Cartilage 1997;5:357–72. 27. Meachim G, Ghadially FN, Collins DH. Regressive changes in the superficial layer of human articular cartilage. Ann Rheum Dis 1965;24:23–30. 28. Mankin HJ, Lipiello L. Biochemical and metabolic abnormalities in articular cartilage from osteoarthritic human hips. J Bone Joint Surg 1970;52:424–34. 29. Shapiro F, Glimcher MJ. Induction of osteoarthrosis in the rabbit knee joint: histologic changes following menisectomy and meniscal lesions. Clin Orthop 1980;147:287–95. 30. Hayes WC, Keer LM, Herrmann KG, Mockros LF. A mathematical analysis for indentation tests of articular cartilage. J Biomech 1972;5:541–51. 31. Sakamoto M, Li G, Hara T, Chao EYS. A new method for theoretical analysis of static indentation test. J Biomech 1996;29: 679–85. 32. Schinagl RM, Gurskis D, Chen AC, Sah RL. Depth-dependent confined compression modulus of full-thickness bovine articular cartilage. J Orthop Res 1997;15:499–506. 33. Mow VC, Good PM, Gardner TR. A new method to determine the tensile properties of articular cartilage using the indentation test. Trans Orthop Res Soc 2000;25:103. 34. Kempson GE. Relationship between the tensile properties of articular cartilage from the human knee and age. Ann Rheum Dis 1982;41:508–11. 35. Beck JR, Shultz EK. The use of relative operating characteristic (ROC) curves in test performance evaluation. Arch Pathol Lab Med 1986;110:13–20. 36. Zweig MH, Campbell G. Receiver-operating characteristic (ROC) plots: a fundamental evaluation tool in clinical medicine. Clin Chem 1993;39:561–77. 37. Swets JA. Measuring the accuracy of diagnostic systems. Science 1988;240:1285–93. 38. Frank EH, Grodzinsky AJ, Koob TJ, Eyre DR. Streaming poten- BAE ET AL 39. 40. 41. 42. 43. 44. 45. 46. 47. 48. 49. 50. 51. 52. 53. 54. 55. 56. 57. 58. tials: a sensitive index of enzymatic degradation in articular cartilage. J Orthop Res 1987;5:497–508. Swann AC, Seedhom BB. The stiffness of normal articular cartilage and the predominant acting stress levels: implications for the aetiology of osteoarthrosis. Br J Rheum 1993;32:16–25. Black J, Shadle CA, Parsons JR, Brighton CT. Articular cartilage preservation and storage. II. Mechanical indentation testing of viable, stored articular cartilage. Arthritis Rheum 1979;22:1102–8. Kiefer GN, Sunbdy K, McAllister D, Shrive NG, Frank CB, Lam T, et al. The effect of cryopreservation on the biomechanical behavior of bovine articular cartilage. J Orthop Res 1989;7: 494–501. Chang DG, Iverson EP, Schinagl RM, Sonoda M, Amiel D, Coutts RD, et al. Video imaging of articular cartilage degeneration in the rabbit knee after ACL transection. Trans Orthop Res Soc 1997; 22:629. Yoshioka M, Shimizu C, Harwood FL, Coutts RD, Amiel D. The effects of hyaluronan during the development of osteoarthritis. Osteoarthritis Cartilage 1997;5:251–60. Gluer CC, Blake G, Lu Y, Blunt BA, Jergas M, Genant HK. Accurate assessment of precision errors: how to measure the reproducibility of bone densitometry techniques. Osteoporos Int 1995;5:262–70. Njeh CF, Hans D, Li J, Fan B, Fuerst T, He YQ, et al. Comparison of six calcaneal quantitative ultrasound devices: precision and hip fracture discrimination. Osteoporos Int 2000;11:1051–62. Orgee JM, Foster H, McCloskey EV, Khan S, Coombes G, Kanis JA. A precise method for the assessment of tibial ultrasound velocity. Osteoporos Int 1996;6:1–7. Glantz SA. Primer of biostatistics. 3rd ed. San Francisco: McGrawHill; 1992. Bennett GA, Waine H, Bauer W. Changes in the knee joint at various ages with particular reference to the nature and development of degenerative joint disease. New York: The Commonwealth Fund; 1942. Karvonen RL, Negendank WG, Teitge RA, Reed AH, Miller PR, Fernandez-Madrid F. Factors affecting articular cartilage thickness in osteoarthritis and aging. J Rheumatol 1994;21:1310–8. Hudelmaier M, Glaser C, Hohe J, Englmeier K-H, Reiser M, Putz R, et al. Age-related changes in the morphology and deformational behavior of knee joint cartilage. Arthritis Rheum 2001;44: 2556–61. Burgkart R, Glaser C, Hyhlik-Dürr A, Englmeier K-H, Reiser M, Eckstein F. Magnetic resonance imaging–based assessment of cartilage loss in severe osteoarthritis: accuracy, precision, and diagnostic value. Arthritis Rheum 2001;44:2072–7. Temple MM, Bae WC, Rivard KL, Sah RL. Age- and siteassociated biomechanical weakening of human articular cartilage of the femoral condyle: relationship to cellularity and wear. Trans Orthop Res Soc 2002;27:84. Wang CC, Chahine NO, Hung CT, Ateshian GA. Optical determination of anisotropic material properties of bovine articular cartilage in compression. J Biomech 2003;36:339–53. Bae WC, Lewis CW, Sah RL. Intra-tissue strain distribution in normal human articular cartilage during clinical indentation testing. Trans Orthop Res Soc 2003;28:254. Haider MA, Holmes MH. A mathematical approximation for the solution of a static indentation test. J Biomech 1997;30:747–51. Mak AF, Lai WM, Mow VC. Biphasic indentation of articular cartilage. I. Theoretical analysis. J Biomechanics 1987;20:703–14. Lachance L, Sowers M, Jamadar D, Jannausch M, Hochberg M, Crutchfield M. The experience of pain and emergent osteoarthritis of the knee. Osteoarthritis Cartilage 2001;9:527–32. Woertler K, Strothmann M, Tombach B, Reimer P. Detection of articular cartilage lesions: experimental evaluation of low- and high-field-strength MR imaging at 0.18 and 1.0 T. J Magn Reson Imaging 2000;11:678–85.