- Email: [email protected]
Serum levels of cartilage oligomeric matrix protein (COMP) correlate with radiographic progression of knee osteoarthritis
Osteoarthritis and Cartilage (2002) 10, 707–713 © 2002 OsteoArthritis Research Society International. Published by Elsevier Science Ltd. All rights reserved. doi:10.1053/joca.2002.0819, available online at http://www.idealibrary.com on
1063–4584/02/$35.00/0
International Cartilage Repair Society
Serum levels of cartilage oligomeric matrix protein (COMP) correlate with radiographic progression of knee osteoarthritis V. Vilı´m*, M. Oleja´rova´*, S. Macha´cˇek*, J. Gatterova´*, V. B. Kraus† and K. Pavelka* *Institute of Rheumatology, Prague, Czech Republic †Department of Medicine, Division of Rheumatology, Allergy & Clinical Immunology, Duke University Medical Center, Durham, NC, U.S.A. Summary Objective: To evaluate the prognostic utility of serum COMP level measured with a new sandwich ELISA, by correlating COMP level with outcome measures of osteoarthritis (OA) progression. Design: Patients (N=48) had symptomatic primary knee OA of Kellgren–Lawrence (K–L) grade I–III and met ACR criteria. These patients were evaluated prospectively as part of a double-blind drug trial of 3 years’ duration and represented the placebo arm of the study. Serum COMP levels were measured by sandwich ELISA with monoclonal antibodies 16-F12 and 17-C10 at baseline and at study end and levels were correlated with changes in (1) joint space width (JSW), (2) K–L grade, (3) Lequesne, and (4) WOMAC indices, over 3 years. Results: The change in JSW over 3 years, summed for both knees, correlated positively with serum COMP level at baseline as well as at study end. Patients were sorted by level of progression based upon a change in K–L grade summed for both knees over 3 years; patients who progressed by two K–L grades were shown to have had significantly higher COMP levels at baseline as well as at study end. Baseline and study end COMP levels did not correlate with the change of Lequesne or WOMAC indices. Baseline COMP levels correlated strongly with end serum COMP levels. Conclusion: Serum COMP has the potential to be a prognostic marker of disease progression. High COMP levels, persisting over the 3-year study period in the patients with radiographic progression, indicated differences in disease activity detectable throughout the entire follow-up interval. © 2002 OsteoArthritis Research Society International. Published by Elsevier Science Ltd. All rights reserved. Key words: Osteoarthritis, Markers, COMP, Serum. Abbreviations: ANOVA, analysis of variance; COMP, cartilage oligomeric matrix protein; CRP, C-reactive protein; ELISA, enzyme linked immunosorbent assay; HA, hyaluronan; JSW, joint space width; K–L, Kellgren and Lawrence; mAb, monoclonal antibody; OA, osteoarthritis.
the time joint tissue degeneration is already advanced. Therefore, much attention has been focused on developing assays for ‘molecular markers’, namely cartilage-derived macromolecules or their fragments whose release into the circulation from the joint may reflect disturbances in joint tissue turnover9,10. A particular need has been to develop markers to sensitively predict initiation and progression of OA. Besides COMP, hyaluronan (HA) is the only other serologic marker reported to have predictive value for the progression of knee OA11. Baseline serum HA levels were significantly increased over 5 years, when progressors were compared with non-progressors (P<0.05). The association of HA level to OA progression was even more significant (P<0.001) when end serum HA levels were examined. The previously published studies evaluating serum COMP as a potential prognostic marker of OA progression utilized polyclonal antibodies12–16. We have developed a panel of monoclonal antibodies (mAb) to human COMP17. Our previous study of serum COMP, using mAb 17-C10 in an inhibition ELISA, distinguished groups of OA-affected and unaffected subjects, and also subgroups sorted by severity of OA and number of large joints involved18, suggesting that COMP level could be used as a surrogate measure of disease status. Our current aim was to evaluate
Introduction Cartilage oligomeric matrix protein (COMP) is a member of a thrombospondin family of extracellular proteins that was initially isolated from cartilage1. COMP is a calcium-binding protein2 of high molecular weight (>500 kDa), composed of five identical subunits3. The carboxy-terminal globular domain of native COMP binds to collagens I, II, and IX4–6. Mutations in domains of type 3 calcium-binding repeats or the carboxy-terminal globular domain, that affect binding of calcium or collagen type IX, respectively, cause two types of skeletal dysplasias, pseudoachondroplasia and some forms of multiple epiphyseal dysplasia5,6. Osteoarthritis (OA) is a common disease causing pain and disability in a significant proportion of the adult population7. It is characterized by the progressive destruction of articular cartilage and concomitant changes in subchondral bone8. A diagnosis of knee OA, traditionally based upon weight bearing radiographs and pain, is usually made by Received 5 October 2001; accepted 19 April 2002. Supported by grants NK/4887-3 and NK/6813-3 from IGA MZ C { R and grants R29AG15108 and P60AG11268 from NIH (VBK). Address correspondence to: Vladimı´r Vilı´m, Institute of Rheumatology, Na Slupi 4, 128 50 Prague 2, Czech Republic. Tel.: (420)-(2)-24915234; Fax: (420)-(2)-24914451; E-mail: [email protected]
707
708
V. Vilı´m et al.: Serum COMP and progression of OA Table I Characteristics of the 48 patients at entry in the study Age, years, mean (range) Gender, male/female, number (%) OA duration, years, mean (S.D.) BMI*, kg/m2, mean (S.D.) ISK†, mean (S.D.) Knee OA, unilateral/bilateral, number (%) Kellgren–Lawrence grade‡, I/II/III, number (%) Knee synovitis, total/unilateral/bilateral, number (%) OA of other large joints (total/hip/spine/hip+spine), number (%)
62.8 (48–74) 14/34 (29%/71%) 11.1 (6.2) 26.4 (1.7) 8.5 (2.4) 7/41 (15%/85%) 8/22/18 (17%/46%/37%) 32/26/6 (67%/54%/13%) 15/8/6/1 (31%/17%/13%/2%)
*BMI=body mass index. †ISK=The Index of Severity for Knee OA according to Lequesne et al.21. ‡A patient was classified according to the grade of the worst knee in cases of bilateral knee OA.
the prognostic utility of serum COMP level measured with a new sandwich ELISA, by correlating COMP level with outcome measures of OA progression.
Patients and methods PATIENTS
Selected patients (N=48) had symptomatic primary knee OA of Kellgren–Lawrence (K–L)19 grade I–III and met ACR criteria20. These patients were evaluated prospectively as part of a double-blind drug trial of 3 years’ duration and represented the patients randomized to the placebo arm of the study. In addition to placebo, these patients were treated with paracetamol, up to 3 g daily. Patients were predominantly female with bilateral knee OA (Table I). The study was approved by the local research ethics committee. Fasting sera were obtained at entry and at the end of the study, and stored at −70°C. Standing anteroposterior weight-bearing radiographs of both knees were obtained simultaneously with sera, and patients were interviewed to obtain two multidimensional measures of disease severity, the Lequesne algofunctional index (ISK)21, and the WOMAC index22. The signal knee was defined without knowledge of the radiographic data as the worst of the two knees on the basis of subjective complaints and objective clinical findings. At entry, patients were also examined by a single rheumatologist, who clinically assessed three signs of knee joint inflammation (tenderness, non-bony swelling, and warmth), on which basis a determination of synovitis was made (with synovitis defined as the presence of joint tenderness of grade at least 1 plus swelling and/or warmth of grade at least 1 for the same knee)23. RADIOGRAPHIC ASSESSMENT AND READING PROCEDURES
Joint space width (JSW) measurements and K–L grading were performed on both the baseline and end radiographs at the same time, exactly as described24, by two independent readers blinded to the patient’s name, date, and chronology of the radiographs. COMMERCIALLY AVAILABLE IMMUNOASSAYS
C-reactive protein (CRP) levels were measured by sandwich ELISA assay supplied as a kit (UBI MAGIWELL CRP QUANTITATIVE AD-401) by UBI, Mountain View, CA.
COMP ELISA
COMP was analysed with a new sandwich ELISA, employing two previously described mAbs, 16-F12 and 17-C1017, whose epitopes have been mapped to the pentameric helical NH2-terminal domain and the EGF-like domain of the COMP subunit, respectively (Vilim, unpublished results). These mAbs did not compete for binding to COMP antigen, when tested in steric competition experiments set up in the format of antibody capture assays (Vilim, unpublished results). MAbs were purified from ascitic fluid by chromatography on a column of immobilized Protein G and concentrated by ultrafiltration. MAb 17-C10 was labeled by LC-biotin using EZ-Link NHS-LC-Biotin (Pierce) to yield a molar ratio IgG:biotin of approximately 1:50. MAb 16-F12 was used as the first (capture) mAb, and biotinylated mAb 17-C10 was used as the second (detection) mAb in the assay. Nunc-Immuno (MaxiSorp) plates (Nunc) were coated with 100 g/ml of mAb 16-F12 in PBS/0.02% sodium azide, 50 l/well. Plates were covered with sealing tape and put on an orbital shaker for 2 h and then left standing at least overnight (up to 3 weeks) at 4°C. Coating solution could be used repeatedly. Coated plates were washed twice with PBS and blocked with 3% BSA/ PBS/0.02% sodium azide, [120 l/well, 2 h on the orbital shaker and than standing overnight (or up to 1 week) at 4°C]. Blocked plates were washed twice with PBS immediately before use. COMP standard, purified as previously described17 from human articular cartilage, was diluted to 10 g/ml with 0.05% Tween-20/10 mM EDTA/0.01% Thiomerosal/1% BSA, aliquoted and kept frozen at −70°C. Once thawed, this diluted standard was either used or discarded. Prior to ELISA, samples/standards were diluted in 0.01% BSA in PBS on low-binding Nunc MicroWell plates. Standard was titrated from the initial concentration of 400 ng/ml to six wells (400, 200, 100, 50, 25, 12.5 ng/ml). Serum samples were diluted 1/20. Samples and standards were mixed and transferred (50 l from each well) onto coated, blocked, and washed MaxiSorp plate and incubated for 2 h on orbital shaker at room temperature. Plates were washed five times with PBS and the labeled second mAb (biotinylated 17-C10) was added (50 l/well). The second mAb was diluted from a concentrated stock to 2 g/ml in 0.01% BSA/PBS. Plates were incubated for 2 h on the orbital shaker at room temperature and then washed five times with PBS. Streptavidin-horseradish peroxidase conjugate (Amersham Pharmacia Biotech) was diluted 500× by 0.01% BSA/PBS and added at 50 l/well. Plates
Osteoarthritis and Cartilage Vol. 10, No. 9 were incubated for 30 min on an orbital shaker at room temperature and then washed five times with PBS/0.05% Tween 20. Peroxidase substrate (o-phenylenediamine plus H2O2), 100 l/well was added for 30 min at room temperature. The production of chromophore was stopped by addition of 50 l/well of 2 M H2SO4. Plates were read at a wavelength of 492 nm. COMP concentrations in samples were calculated from a standard curve and results were expressed as equivalents of the standard human antigenic COMP (g/ml). Samples, which fell below/above the standard range, were measured repeatedly at lower/higher dilution. The order of the samples was then randomized, and the measurements repeated a second time. These two independent measurements had to fall within the limits of interassay variance of the assay. Each plate contained four identical samples of ‘control’ serum that was stored in aliquots at −70°C. Each day a fresh aliquot of the control serum was thawed and used on each plate to calculate intra- and interassay variance25; these were found to be less than 8% and less than 10%, respectively. All samples were assayed simultaneously with the same set of reagents. Analyses were performed without knowledge of the radiographic and clinical data.
709
Fig. 1. Correlation between baseline (a) and end (b) serum COMP levels, and radiographic OA progression, expressed as a change in JSW for the signal knee over 3 years. The linear regression line is shown.
STATISTICAL METHODS
An association between continuous variables was assessed using the Pearson test and linear regression. Associations between continuous and qualitative variables were evaluated by analysis of variance (ANOVA). Comparisons of COMP levels in two groups of patients were done by the non-parametric Mann–Whitney test. Baseline and end COMP levels within one group of patients were compared using a paired t-test. A P-value (two-tailed) <0.05 was considered significant.
Fig. 2. Correlation between baseline (a) and end (b) serum COMP levels, and radiographic OA progression, expressed as a change in JSW summed for both knees over 3 years. The linear regression line is shown.
Results Only age significantly affected COMP levels in this study. Other potential confounding variables [gender, body mass index, OA duration, CRP level at entry, K–L grade at entry, the presence of knee synovitis at entry, the presence of OA of other large joints at entry, and the number of large joints (knee, hip, spine) affected at entry] were without a significant effect. Unadjusted COMP levels are demonstrated in the Figures, since adjusting for age did not change the overall results. An association between a change in JSW over 3 years and serum COMP level was assessed. We found a significant correlation of serum COMP level with a change in JSW for the signal knee over 3 years. The association was significant for COMP level at the study end (P<0.001), but not for COMP level at baseline (Fig. 1). When summed for both knees, a change in JSW over 3 years correlated positively with serum COMP level at baseline (P<0.01) as well as at study end (P<0.001) (Fig. 2). The correlation remained statistically significant after adjusting COMP levels for age. When the threshold used to designate a patient as a ‘progressor’ was defined as a decrease in JSW over 3 years >0.5 mm for either knee, a total of 10 of the 48 patients progressed, and 38 of the patients did not progress. Although serum COMP levels in these two groups overlapped (Fig. 3), the mean COMP level was higher in the progressors (Table II) and the difference in
Fig. 3. Baseline (a) and study end (b) serum COMP levels in patients sorted according to severity of OA progression, expressed as the decrease in JSW for either knee over 3 years. Horizontal lines show mean COMP levels. Patients showing a decrease of JSW <0.5 mm were considered to be non-progressors (N=38); patients showing a decrease of JSW >0.5 mm were considered to be progressors (N=10). *=P<0.05 and **=P<0.01 comparing progressors with non-progressors (Mann–Whitney test).
COMP level between the two groups was significant using a non-parametric Mann–Whitney test (P<0.05 for COMP level at baseline and P<0.01 for COMP level at the study end). Patients were further sorted by level of progression based upon a change in K–L grade over 3 years. The signal knee showed either no change in K–L grade (group 0), or an increase by one K–L grade (group 1). No patient progressed more than one K–L grade, when only the signal knee was taken into account. Although serum COMP levels
710
V. Vilı´m et al.: Serum COMP and progression of OA Table II Serum COMP levels for groups of patients by radiographic progression Serum COMP (g/ml)
All patients (N=48) Mean± S.D. Median (range) Decrease of JSW, either knee (mm) Mean± S.D. Median (range) Increase in K–L grade, signal knee Mean± S.D. Median (range) Increase in K–L grade, both knees† Mean± S.D. Median (range)
Baseline
Study end
4.16±1.03 3.92 (2.2–6.3)
4.47±1.37 4.24 (1.9–8.4)
<0.5 (N=38) 3.96±0.94 3.88 (2.2–6.3)
>0.5 (N=10) 4.92±1.05 5.02 (3.3–6.3)*
<0.5 (N=38) 4.11±0.90 4.16 (2.3–6.1)
>0.5 (N=10) 5.85±1.96 6.35 (1.9–8.4)**
0 (N=33) 3.96±1.01 3.88 (2.2–6.3)
1 (N=15) 4.60±0.96 4.37 (3.4–6.3)*
0 (N=33) 4.25±1.24 4.14 (1.9–7.8)
1 (N=15) 4.97±1.56 4.36 (2.3–8.4)
0 (N=29) 3.91±0.97 3.88 (2.2–6.1)
2 (N=9) 5.11±1.01** 5.32 (3.5–6.3)
0 (N=29) 4.12±1.11 4.18 (1.9–6.3)
2 (N=9) 5.79±1.71** 5.04 (3.9–8.4)
Decrease of JSW, either knee: *P<0.05 and **P<0.01 compared with group <0.5 (Mann–Whitney test). Increase in K–L grade, signal knee: *P<0.05 compared with group 0 (Mann–Whitney test). Increase in K–L grade, both knees: **P<0.01 compared with group 0 (ANOVA, Bonferroni post test). †Serum COMP levels in the group of ten patients who progressed by only one K–L grade were not significantly different from levels in the group of patients who did not progress (data not shown).
Fig. 4. Baseline (a) and study end (b) serum COMP levels in patients sorted according to severity of OA progression, expressed as the change in K–L grade for the signal knee over 3 years. Horizontal lines show mean COMP levels. Group 0 nonprogressors (N=33); group 1 increased one K–L grade (N=15). *=P<0.05 and ns=not significant when compared with Group 0 (Mann-Whitney test).
in these two groups overlapped (Fig. 4), the mean COMP level was higher in group 1 patients and the difference in baseline COMP level between the two groups was significant using a non-parametric Mann–Whitney test (P<0.05), the difference in COMP level at the study end was not significant. Subjects were also categorized according to the sum of the change that occurred in K–L grade for both knees: group 0 no change, group 1 increased one K–L grade, group 2 increased 2 K–L grades. The mean COMP levels in these groups differed significantly at baseline as well as at study end (ANOVA, P<0.01 before adjusting and P<0.05 after adjusting for age). Group 2 had higher mean COMP levels than (1) group 0 non-progressors (P<0.01 before adjusting and P<0.05 after adjusting for age) and (2) group 1 that progressed by one K–L grade (P<0.05, both before and after adjusting for age) at baseline as well as at study end (Bonferroni multiple comparisons post hoc test)
Fig. 5. Baseline (a) and study end (b) serum COMP levels in patients sorted according to severity of OA progression, expressed as the change in K–L grade summed for both knees over 3 years. Horizontal lines show mean COMP levels. Group 0 nonprogressors (N=29); group 1 increased one K–L grade (N=10); group 2 increased two K–L grades (N=9). **=P<0.01 when compared with Group 0; *=P<0.05 when compared with Group 1 (ANOVA and Bonferroni post test).
(Fig. 5). Mean and median COMP levels by radiographic progression status are summarized in Table II. Baseline serum COMP levels correlated strongly with study end serum COMP levels (P<0.0001) (Fig. 6). Since two former studies12,13 have shown a correlation between progression status and the change in serum COMP level over time, we also evaluated our data in this manner. Progression was defined as a change in JSW of >0.5 mm for either knee over 3 years, or as a 2 grade increase in K–L grade summed for both knees over 3 years. Although we noticed an increase in serum COMP level over the 3-year follow-up period in the whole cohort (Table II), this increase was similar for all patients. The change in COMP (from baseline to study end) was not significantly different in the groups separated according to progression status using a paired t-test.
Osteoarthritis and Cartilage Vol. 10, No. 9
Fig. 6. Correlation between baseline and study end serum COMP levels. The linear regression line is shown.
Baseline and study end COMP levels did not correlate with changes of ISK or WOMAC indices.
Discussion We have developed a sandwich ELISA with two monoclonal antibodies to human COMP. The sandwich ELISA format offers the advantage of higher specificity and sensitivity over other ELISA methods of biomarker quantification. We performed this study with the goal of relating serum COMP levels to the progression of knee OA. We demonstrated that baseline serum COMP levels correlated with the magnitude of OA radiographic progression, thus serum COMP has the potential to be a prognostic marker of disease progression. In previous studies, we have observed an association between serum COMP level and knee OA disease severity (K–L grade), serum COMP level and the number of OA affected large joints18, and serum COMP level and synovitis23. In this much smaller study we observed the same trends, which however, did not achieve statistical significance, we presume due to the lack of power. Age was the only other factor that we have identified in previous studies, whose effect on serum COMP levels remained statistically significant even in the relatively small group analysed here. However, adjusting COMP levels for age did not affect conclusions reached by this study. Several previous studies have evaluated serum COMP and OA progression. Sharif et al.12 and Petersson et al.13 found no correlation between baseline serum COMP levels and subsequent OA progression during 5 years of follow-up of patients with established knee OA and during 3 years of follow-up of patients with early knee OA, respectively. They found, however, a significant positive correlation between the change (the increase) in serum COMP level during the first year12 and during the 3-year interval13 respectively, and radiographic progression during the whole follow-up interval. Conrozier et al.14 found a significant correlation between baseline serum COMP levels and 1-year radiographic progression of patients with established hip OA. They found no correlation between the change in serum COMP level over time and yearly mean narrowing of joint space width. In agreement with the hip OA study, we observed a similar correlation between serum COMP level at baseline and radiographic progression in a knee OA cohort. In contrast to the two previous studies on knee OA, we detected no association between the change in serum COMP level and progression status. Instead, we found that serum COMP levels at baseline and at study end correlated strongly, indicating that serum COMP level distinguished
711 progressors from non-progressors throughout the follow-up interval. Although it is expressed by other tissues within a joint26, COMP prevails in the articular cartilage and it has been demonstrated that serum levels are representative of cartilage catabolism10. Thus, the correlation of serum COMP level with joint space narrowing, a radiographic indicator of cartilage catabolism, is plausible. The recent report that COMP is expressed by vascular smooth muscle cells27 hampers straightforward interpretation of serum COMP levels in the setting of joint disease. While the potential contribution of these cells to serum COMP level is difficult to predict, it seems reasonable to expect that production of COMP by non-cartilaginous tissues is responsible for a certain background or constitutive serum level, and that pathological degradation of cartilage adds significantly to this background level. HA is the only other biomarker, which has been shown to predict knee OA progression11. Like COMP, HA is not tissue specific but serum levels increase significantly in response to activity of pathological joints28,29. Thus, it is indeed possible for a tissue specific process to measurably contribute to the level of a biomarker in the circulation. In general, we found the best correlation between COMP level and radiographic progression when both knees were included in the analysis. This result is expected for a marker, such as COMP, whose potential release from all joints in the body contributes to the serum level. It is difficult to decide whether measurement of JSW is superior to composite scoring systems such as K–L grading for assessing radiographic progression of OA, thus the use of both approaches has been recommended30. It has became obvious recently, that precise measurement of the interbone distance on weight-bearing anteroposterior radiographs may be obscured by the fact that even relatively minor changes in a patient’s position (usually due to different degrees of flexion) can result in large changes in the measured apparent joint space width. This methodological shortcoming impacts practically all longitudinal studies, including ours, initiated prior to 1995, when the influence of joint positioning on the standardization and reproducibility of tibiofemoral JSW measurement on sequential OA knee joint radiographs was published31. A recent retrospective assessment of several former longitudinal OA studies has elucidated the overall effect of lower limb positioning on JSW measurements32,33. In total, correct alignment of the medial tibial plateau with the X-ray beam, and standardization of knee rotation on sequential radiographs ranged from 9 to 19% of all examined pairs of radiographs in four retrospectively assessed cohorts of OA patients32. Significant widening of joint space in the medial compartment was found in 23% of all examined knees. We are aware of this potential shortcoming, the presence of which in our study is indicated by the fact that we recorded widening of the knee joint space in 17 patients, (35%) out of 48 who received placebo, over the period of 3 years. Although, hypothetically, there may be several reasons for apparent radiographic improvement33, it is very probable that at least some of these apparent ‘improvements’ can be ascribed to misalignment of sequential radiographs. However, a correlation of serum COMP level with radiographic OA progression was not based on JSW measurements alone in this study. We also observed a correlation of serum COMP level with K–L grade worsening. Although the K–L grading system is considered to be relatively insensitive, results do not depend on JSW alone, are less affected by joint positioning and standardization of grading is based
712 upon atlases. Of note, we recorded no cases of ‘improvement’ over the 3-year interval using the K–L grading system. The fact that serum COMP and OA progression are correlated on the basis of both JSW measurements and K–L grading argues strongly, in our opinion, that this finding is valid. The test does not meet the criterion of discriminant validity, as its ability to detect small changes seems to be rather limited; however, a similar limitation could be said to apply to the currently accepted ‘gold standard’ for OA, namely joint radiography. Furthermore, at present, serum COMP level cannot be used to prognosticate the course of OA in an individual patient due to the high variability and the overlap of COMP values in affected and unaffected groups. However, COMP level may aid in identifying groups of patients at risk of faster disease progression and in differentiating groups of patients during clinical studies and drug trials that are analysed using statistical methods.
References 1. Hedbom E, Antonsson P, Hjerpe A, Aeschlimann D, Paulsson M, Rosa-Pimentel E, et al. Cartilage matrix proteins. An acidic oligomeric protein (COMP) detected only in cartilage. J Biol Chem 1992;267: 6132–6. 2. Chen H, Deere M, Hecht JT, Lawler J. Cartilage oligomeric matrix protein is a calcium-binding protein, and a mutation in its type 3 repeats causes conformational changes. J Biol Chem 2000;275:26538–44. doi:10.1074/jbc.M909780199. 3. Oldberg A, Antonsson P, Lindblom K, Heinegard D. COMP (cartilage oligomeric matrix protein) is structurally related to the thrombospondins. J Biol Chem 1992;267:22346–50. 4. Rosenberg K, Olsson H, Morgelin M, Heinegard D. Cartilage oligomeric matrix protein shows high affinity zinc-dependent interaction with triple helical collagen. J Biol Chem 1998;273:20397–403. 5. Thur J, Rosenberg K, Nitsche DP, Pihlajamaa T, Ala-Kokko L, Heinegard D, et al. Mutations in cartilage oligomeric matrix protein causing pseudoachondroplasis and multiple epiphyseal displasia affect binding of calcium and collagen I, II, and IX. J Biol Chem 2001;276:6083–92. doi:10.1074/ jbc.M009512200. 6. Holden P, Meadows RS, Chapman KI, Grant ME, Kadler KE, Briggs MD. Cartilage oligomeric matrix protein interacts with type IX collagen, and disruptions to these interactions identify a pathogenetic mechanism in a bone dysplasia family. J Biol Chem 2001;276:6046–55. doi:10.1074/jbc.M009507200. 7. Lawrence RC, Helmick CG, Arnett FC, Deyo RA, Felson DT, Giannini EH, et al. Estimates of the prevalence of arthritis and selected musculoskeletal disorders in the United States. Arthritis Rheum 1998;41:778–99. 8. Brandt KD. The pathogenesis of osteoarthritis. Rheumatol Rev 1991;1:3–11. 9. Garnero P, Rousseau J-C, Delmas PD. Molecular basis and clinical use of biochemical markers of bone, cartilage, and synovium in joint diseases. Arthritis Rheum 2000;43:953–68. 10. Otterness IG, Swindell AC, Zimmerer RO, Poole AR, Ionescu M, Weiner E. An analysis of 14 molecular markers for monitoring osteoarthritis: segregation of
V. Vilı´m et al.: Serum COMP and progression of OA
11.
12.
13.
14.
15.
16.
17.
18.
19. 20.
21.
22.
23.
24.
the markers into clusters and distinguishing osteoarthritis at baseline. Osteoarthritis Cart 2000;8: 180–5. doi:10.1053/joca.1999.0288. Sharif M, George E, Shepstone L, Knudson W, Thonar EJ-MA, Cushnaghan J, et al. Serum hyaluronic acid level as a predictor of disease progression in osteoarthritis of the knee. Arthritis Rheum 1995;38:760–7. Sharif M, Saxne T, Shepstone L, Kirwan JR, Elson CJ, Heinegard D, et al. Relationship between serum cartilage oligomeric matrix protein levels and disease progression in osteoarthritis of the knee joint. Br J Rheumatol 1995;34:306–10. Petersson IF, Boegard B, Svensson D, Heinegard D, Saxne T. Changes in cartilage and bone metabolism identified by serum markers in early osteoarthritis of the knee joint. Br J Rheumatol 1998;37:46–50. Conrozier T, Saxne T, Shan Sei Fan C, Mathieu P, Tron A-M, Heinegard D, et al. Serum concentrations of cartilage oligomeric matrix protein and bone sialoprotein in hip osteoarthritis: A one year prospective study. Ann Rheum Dis 1998;57:527–32. Georges C, Vigneron H, Ayral X, Listrat V, Ravaud P, Dougados M, et al. Serum biologic markers as predictors of disease progression in osteoarthritis of the knee. Arthritis Rheum 1997;40:590–1. Petersson I, Boegard T, Dahlstrom J, Svensson B, Heinegard D, Saxne T. Bone scan and serum markers of bone and cartilage in patients with knee pain and osteoarthritis. Osteoarthritis Cart 1998;6: 33–9. Vilı´m V, Lenz ME, Vyta´sˇek R, Masuda K, Pavelka K, Kuettner KE, et al. Characterization of monoclonal antibodies recognizing different fragments of cartilage oligomeric matrix protein in human body fluids. Arch Biochem Biophys 1997;341:8–16. Clark AG, Jordan JM, Vilı´m V, Renner JB, Dragomir A, Luta G, et al. Serum cartilage oligomeric matrix protein reflects osteoarthritis presence and severity: The Johnston county osteoarthritis project. Arthritis Rheum 1999;42:2356–64. Kellgren JH, Lawrence JS. Radiological assessment of osteo-arthrosis. Ann Rheum Dis 1957;16:494–502. Altman R, Asch E, Bloch D, Bole D, Bornstein D, Brandt K, et al. Development of criteria for the classification and reporting of osteoarthritis. Classification of osteoarthritis of the knee. Arthritis Rheum 1986;29:1039–49. Lequesne MG, Mery C, Samson M, Gerard P. Indexes of severity for osteoarthritis of the hip and knee. Validation-value in comparison with other assessment tests. Scand J Rheumatol 1987;65(Suppl.): 85–9. Bellamy N, Buchanan WW, Goldsmith CH, Campbell J, Stitt LW. Validation study of WOMAC: a health status instrument for measuring clinically important patient relevant outcomes to antirheumatic therapy in patients with osteoarthritis of the hip or knee. J Rheumatol 1988;15:1833–40. Vilı´m V, Vyta´sˇek R, Oleja´rova´ M, Macha´cˇek S, Gatterova´ J, Procha´zka B, et al. Serum cartilage oligomeric matrix protein reflects the presence of clinically diagnosed synovitis in patients with knee osteoarthritis. Osteoarthritis Cart 2001;9:612–18. doi:10.1053/joca.2001.0434. Pavelka K, Gatterova´ J, Gollova´ V, Urbanova´ Z, Sedla´cˇkova´M, Altman RD. A 5-year randomized controlled, double-blind study of glycosaminoglycan
Osteoarthritis and Cartilage Vol. 10, No. 9
25. 26.
27.
28.
29.
polysulphuric acid complex (Rumalon®) as a structure modifying therapy in osteoarthritis of the hip and knee. Osteoarthritis Cart 2000;8:335–42. doi:10.1053/joca.1999.0307. Rodbard D. Statistical quality control and routine data processing for radioimmunoassays and immunometric assays. Clin Chem 1974;20:1255–70. Muller G, Michel A, Altenburg E. COMP (cartilage oligomeric matrix protein) is synthesized in ligament, tendon, meniscus, and articular cartilage. Conn Tissue Res 1998;39:233–244. Reissen R, Fenchel M, Chen H, Axel DI, Karsch KR, Lawler J. Cartilage oligomeric matrix protein (thrombospondin-5) is expressed by human vascular smooth muscle cells. Arterioscler Thromb Vasc Biol 2001;21:47–54. Engstrom-Laurent A, Hallgren R. Circulating hyaluronic acid levels vary with physical activity in healthy subjects and in rheumatoid arthritis patients. Relationship to synovitis mass and morning stiffness. Arthritis Rheum 1987;30:1333–8. Manicourt DH, Poilvache P, Nzeusseu A, VanEgeren A, Devogelaer J-P, Lenz ME, et al. Serum levels of
713
30. 31.
32.
33.
hyaluronan, antigenic keratan sulfate, matrix metalloproteinese 3, and tissue inhibitor of metalloproteinases 1 change predictably in rheumatoid arthritis patients who have begun activity after a night of bed rest. Arthritis Rheum 1999;42:1861–9. Dieppe PA. Recommended methodology for assessing the progression of osteoarthritis of the hip and knee joints. Osteoarthritis Cart 1995;3:73–7. Buckland-Wright JC, MacFarlane DG, Williams SA, Waed RJ. Accuracy and precision of joint space width measurements in standard and macroradiographs of osteoarthritic knees. Ann Rheum Dis 1995;54: 872–80. Mazzuca SA, Brandt KD, Dieppe PA, Doherty M, Katz BP, Lane KA. Effect of alignment of the medial tibial plateau and x-ray beam on apparent progression of knee osteoarthritis in the standing anteroposterior radiograph. Arthritis Rheum 2001;44:1786–94. Mazzuca SA, Brandt KD, Dieppe PA, Doherty M, Sowers M. Reasons for apparent widening of medial joint space in conventional standing anteroposterior knee radiographs of patients with osteoarthritis (Abstract). Trans 47th Ann Meet ORS 2001;26:657.
1063–4584/02/$35.00/0
International Cartilage Repair Society
Serum levels of cartilage oligomeric matrix protein (COMP) correlate with radiographic progression of knee osteoarthritis V. Vilı´m*, M. Oleja´rova´*, S. Macha´cˇek*, J. Gatterova´*, V. B. Kraus† and K. Pavelka* *Institute of Rheumatology, Prague, Czech Republic †Department of Medicine, Division of Rheumatology, Allergy & Clinical Immunology, Duke University Medical Center, Durham, NC, U.S.A. Summary Objective: To evaluate the prognostic utility of serum COMP level measured with a new sandwich ELISA, by correlating COMP level with outcome measures of osteoarthritis (OA) progression. Design: Patients (N=48) had symptomatic primary knee OA of Kellgren–Lawrence (K–L) grade I–III and met ACR criteria. These patients were evaluated prospectively as part of a double-blind drug trial of 3 years’ duration and represented the placebo arm of the study. Serum COMP levels were measured by sandwich ELISA with monoclonal antibodies 16-F12 and 17-C10 at baseline and at study end and levels were correlated with changes in (1) joint space width (JSW), (2) K–L grade, (3) Lequesne, and (4) WOMAC indices, over 3 years. Results: The change in JSW over 3 years, summed for both knees, correlated positively with serum COMP level at baseline as well as at study end. Patients were sorted by level of progression based upon a change in K–L grade summed for both knees over 3 years; patients who progressed by two K–L grades were shown to have had significantly higher COMP levels at baseline as well as at study end. Baseline and study end COMP levels did not correlate with the change of Lequesne or WOMAC indices. Baseline COMP levels correlated strongly with end serum COMP levels. Conclusion: Serum COMP has the potential to be a prognostic marker of disease progression. High COMP levels, persisting over the 3-year study period in the patients with radiographic progression, indicated differences in disease activity detectable throughout the entire follow-up interval. © 2002 OsteoArthritis Research Society International. Published by Elsevier Science Ltd. All rights reserved. Key words: Osteoarthritis, Markers, COMP, Serum. Abbreviations: ANOVA, analysis of variance; COMP, cartilage oligomeric matrix protein; CRP, C-reactive protein; ELISA, enzyme linked immunosorbent assay; HA, hyaluronan; JSW, joint space width; K–L, Kellgren and Lawrence; mAb, monoclonal antibody; OA, osteoarthritis.
the time joint tissue degeneration is already advanced. Therefore, much attention has been focused on developing assays for ‘molecular markers’, namely cartilage-derived macromolecules or their fragments whose release into the circulation from the joint may reflect disturbances in joint tissue turnover9,10. A particular need has been to develop markers to sensitively predict initiation and progression of OA. Besides COMP, hyaluronan (HA) is the only other serologic marker reported to have predictive value for the progression of knee OA11. Baseline serum HA levels were significantly increased over 5 years, when progressors were compared with non-progressors (P<0.05). The association of HA level to OA progression was even more significant (P<0.001) when end serum HA levels were examined. The previously published studies evaluating serum COMP as a potential prognostic marker of OA progression utilized polyclonal antibodies12–16. We have developed a panel of monoclonal antibodies (mAb) to human COMP17. Our previous study of serum COMP, using mAb 17-C10 in an inhibition ELISA, distinguished groups of OA-affected and unaffected subjects, and also subgroups sorted by severity of OA and number of large joints involved18, suggesting that COMP level could be used as a surrogate measure of disease status. Our current aim was to evaluate
Introduction Cartilage oligomeric matrix protein (COMP) is a member of a thrombospondin family of extracellular proteins that was initially isolated from cartilage1. COMP is a calcium-binding protein2 of high molecular weight (>500 kDa), composed of five identical subunits3. The carboxy-terminal globular domain of native COMP binds to collagens I, II, and IX4–6. Mutations in domains of type 3 calcium-binding repeats or the carboxy-terminal globular domain, that affect binding of calcium or collagen type IX, respectively, cause two types of skeletal dysplasias, pseudoachondroplasia and some forms of multiple epiphyseal dysplasia5,6. Osteoarthritis (OA) is a common disease causing pain and disability in a significant proportion of the adult population7. It is characterized by the progressive destruction of articular cartilage and concomitant changes in subchondral bone8. A diagnosis of knee OA, traditionally based upon weight bearing radiographs and pain, is usually made by Received 5 October 2001; accepted 19 April 2002. Supported by grants NK/4887-3 and NK/6813-3 from IGA MZ C { R and grants R29AG15108 and P60AG11268 from NIH (VBK). Address correspondence to: Vladimı´r Vilı´m, Institute of Rheumatology, Na Slupi 4, 128 50 Prague 2, Czech Republic. Tel.: (420)-(2)-24915234; Fax: (420)-(2)-24914451; E-mail: [email protected]
707
708
V. Vilı´m et al.: Serum COMP and progression of OA Table I Characteristics of the 48 patients at entry in the study Age, years, mean (range) Gender, male/female, number (%) OA duration, years, mean (S.D.) BMI*, kg/m2, mean (S.D.) ISK†, mean (S.D.) Knee OA, unilateral/bilateral, number (%) Kellgren–Lawrence grade‡, I/II/III, number (%) Knee synovitis, total/unilateral/bilateral, number (%) OA of other large joints (total/hip/spine/hip+spine), number (%)
62.8 (48–74) 14/34 (29%/71%) 11.1 (6.2) 26.4 (1.7) 8.5 (2.4) 7/41 (15%/85%) 8/22/18 (17%/46%/37%) 32/26/6 (67%/54%/13%) 15/8/6/1 (31%/17%/13%/2%)
*BMI=body mass index. †ISK=The Index of Severity for Knee OA according to Lequesne et al.21. ‡A patient was classified according to the grade of the worst knee in cases of bilateral knee OA.
the prognostic utility of serum COMP level measured with a new sandwich ELISA, by correlating COMP level with outcome measures of OA progression.
Patients and methods PATIENTS
Selected patients (N=48) had symptomatic primary knee OA of Kellgren–Lawrence (K–L)19 grade I–III and met ACR criteria20. These patients were evaluated prospectively as part of a double-blind drug trial of 3 years’ duration and represented the patients randomized to the placebo arm of the study. In addition to placebo, these patients were treated with paracetamol, up to 3 g daily. Patients were predominantly female with bilateral knee OA (Table I). The study was approved by the local research ethics committee. Fasting sera were obtained at entry and at the end of the study, and stored at −70°C. Standing anteroposterior weight-bearing radiographs of both knees were obtained simultaneously with sera, and patients were interviewed to obtain two multidimensional measures of disease severity, the Lequesne algofunctional index (ISK)21, and the WOMAC index22. The signal knee was defined without knowledge of the radiographic data as the worst of the two knees on the basis of subjective complaints and objective clinical findings. At entry, patients were also examined by a single rheumatologist, who clinically assessed three signs of knee joint inflammation (tenderness, non-bony swelling, and warmth), on which basis a determination of synovitis was made (with synovitis defined as the presence of joint tenderness of grade at least 1 plus swelling and/or warmth of grade at least 1 for the same knee)23. RADIOGRAPHIC ASSESSMENT AND READING PROCEDURES
Joint space width (JSW) measurements and K–L grading were performed on both the baseline and end radiographs at the same time, exactly as described24, by two independent readers blinded to the patient’s name, date, and chronology of the radiographs. COMMERCIALLY AVAILABLE IMMUNOASSAYS
C-reactive protein (CRP) levels were measured by sandwich ELISA assay supplied as a kit (UBI MAGIWELL CRP QUANTITATIVE AD-401) by UBI, Mountain View, CA.
COMP ELISA
COMP was analysed with a new sandwich ELISA, employing two previously described mAbs, 16-F12 and 17-C1017, whose epitopes have been mapped to the pentameric helical NH2-terminal domain and the EGF-like domain of the COMP subunit, respectively (Vilim, unpublished results). These mAbs did not compete for binding to COMP antigen, when tested in steric competition experiments set up in the format of antibody capture assays (Vilim, unpublished results). MAbs were purified from ascitic fluid by chromatography on a column of immobilized Protein G and concentrated by ultrafiltration. MAb 17-C10 was labeled by LC-biotin using EZ-Link NHS-LC-Biotin (Pierce) to yield a molar ratio IgG:biotin of approximately 1:50. MAb 16-F12 was used as the first (capture) mAb, and biotinylated mAb 17-C10 was used as the second (detection) mAb in the assay. Nunc-Immuno (MaxiSorp) plates (Nunc) were coated with 100 g/ml of mAb 16-F12 in PBS/0.02% sodium azide, 50 l/well. Plates were covered with sealing tape and put on an orbital shaker for 2 h and then left standing at least overnight (up to 3 weeks) at 4°C. Coating solution could be used repeatedly. Coated plates were washed twice with PBS and blocked with 3% BSA/ PBS/0.02% sodium azide, [120 l/well, 2 h on the orbital shaker and than standing overnight (or up to 1 week) at 4°C]. Blocked plates were washed twice with PBS immediately before use. COMP standard, purified as previously described17 from human articular cartilage, was diluted to 10 g/ml with 0.05% Tween-20/10 mM EDTA/0.01% Thiomerosal/1% BSA, aliquoted and kept frozen at −70°C. Once thawed, this diluted standard was either used or discarded. Prior to ELISA, samples/standards were diluted in 0.01% BSA in PBS on low-binding Nunc MicroWell plates. Standard was titrated from the initial concentration of 400 ng/ml to six wells (400, 200, 100, 50, 25, 12.5 ng/ml). Serum samples were diluted 1/20. Samples and standards were mixed and transferred (50 l from each well) onto coated, blocked, and washed MaxiSorp plate and incubated for 2 h on orbital shaker at room temperature. Plates were washed five times with PBS and the labeled second mAb (biotinylated 17-C10) was added (50 l/well). The second mAb was diluted from a concentrated stock to 2 g/ml in 0.01% BSA/PBS. Plates were incubated for 2 h on the orbital shaker at room temperature and then washed five times with PBS. Streptavidin-horseradish peroxidase conjugate (Amersham Pharmacia Biotech) was diluted 500× by 0.01% BSA/PBS and added at 50 l/well. Plates
Osteoarthritis and Cartilage Vol. 10, No. 9 were incubated for 30 min on an orbital shaker at room temperature and then washed five times with PBS/0.05% Tween 20. Peroxidase substrate (o-phenylenediamine plus H2O2), 100 l/well was added for 30 min at room temperature. The production of chromophore was stopped by addition of 50 l/well of 2 M H2SO4. Plates were read at a wavelength of 492 nm. COMP concentrations in samples were calculated from a standard curve and results were expressed as equivalents of the standard human antigenic COMP (g/ml). Samples, which fell below/above the standard range, were measured repeatedly at lower/higher dilution. The order of the samples was then randomized, and the measurements repeated a second time. These two independent measurements had to fall within the limits of interassay variance of the assay. Each plate contained four identical samples of ‘control’ serum that was stored in aliquots at −70°C. Each day a fresh aliquot of the control serum was thawed and used on each plate to calculate intra- and interassay variance25; these were found to be less than 8% and less than 10%, respectively. All samples were assayed simultaneously with the same set of reagents. Analyses were performed without knowledge of the radiographic and clinical data.
709
Fig. 1. Correlation between baseline (a) and end (b) serum COMP levels, and radiographic OA progression, expressed as a change in JSW for the signal knee over 3 years. The linear regression line is shown.
STATISTICAL METHODS
An association between continuous variables was assessed using the Pearson test and linear regression. Associations between continuous and qualitative variables were evaluated by analysis of variance (ANOVA). Comparisons of COMP levels in two groups of patients were done by the non-parametric Mann–Whitney test. Baseline and end COMP levels within one group of patients were compared using a paired t-test. A P-value (two-tailed) <0.05 was considered significant.
Fig. 2. Correlation between baseline (a) and end (b) serum COMP levels, and radiographic OA progression, expressed as a change in JSW summed for both knees over 3 years. The linear regression line is shown.
Results Only age significantly affected COMP levels in this study. Other potential confounding variables [gender, body mass index, OA duration, CRP level at entry, K–L grade at entry, the presence of knee synovitis at entry, the presence of OA of other large joints at entry, and the number of large joints (knee, hip, spine) affected at entry] were without a significant effect. Unadjusted COMP levels are demonstrated in the Figures, since adjusting for age did not change the overall results. An association between a change in JSW over 3 years and serum COMP level was assessed. We found a significant correlation of serum COMP level with a change in JSW for the signal knee over 3 years. The association was significant for COMP level at the study end (P<0.001), but not for COMP level at baseline (Fig. 1). When summed for both knees, a change in JSW over 3 years correlated positively with serum COMP level at baseline (P<0.01) as well as at study end (P<0.001) (Fig. 2). The correlation remained statistically significant after adjusting COMP levels for age. When the threshold used to designate a patient as a ‘progressor’ was defined as a decrease in JSW over 3 years >0.5 mm for either knee, a total of 10 of the 48 patients progressed, and 38 of the patients did not progress. Although serum COMP levels in these two groups overlapped (Fig. 3), the mean COMP level was higher in the progressors (Table II) and the difference in
Fig. 3. Baseline (a) and study end (b) serum COMP levels in patients sorted according to severity of OA progression, expressed as the decrease in JSW for either knee over 3 years. Horizontal lines show mean COMP levels. Patients showing a decrease of JSW <0.5 mm were considered to be non-progressors (N=38); patients showing a decrease of JSW >0.5 mm were considered to be progressors (N=10). *=P<0.05 and **=P<0.01 comparing progressors with non-progressors (Mann–Whitney test).
COMP level between the two groups was significant using a non-parametric Mann–Whitney test (P<0.05 for COMP level at baseline and P<0.01 for COMP level at the study end). Patients were further sorted by level of progression based upon a change in K–L grade over 3 years. The signal knee showed either no change in K–L grade (group 0), or an increase by one K–L grade (group 1). No patient progressed more than one K–L grade, when only the signal knee was taken into account. Although serum COMP levels
710
V. Vilı´m et al.: Serum COMP and progression of OA Table II Serum COMP levels for groups of patients by radiographic progression Serum COMP (g/ml)
All patients (N=48) Mean± S.D. Median (range) Decrease of JSW, either knee (mm) Mean± S.D. Median (range) Increase in K–L grade, signal knee Mean± S.D. Median (range) Increase in K–L grade, both knees† Mean± S.D. Median (range)
Baseline
Study end
4.16±1.03 3.92 (2.2–6.3)
4.47±1.37 4.24 (1.9–8.4)
<0.5 (N=38) 3.96±0.94 3.88 (2.2–6.3)
>0.5 (N=10) 4.92±1.05 5.02 (3.3–6.3)*
<0.5 (N=38) 4.11±0.90 4.16 (2.3–6.1)
>0.5 (N=10) 5.85±1.96 6.35 (1.9–8.4)**
0 (N=33) 3.96±1.01 3.88 (2.2–6.3)
1 (N=15) 4.60±0.96 4.37 (3.4–6.3)*
0 (N=33) 4.25±1.24 4.14 (1.9–7.8)
1 (N=15) 4.97±1.56 4.36 (2.3–8.4)
0 (N=29) 3.91±0.97 3.88 (2.2–6.1)
2 (N=9) 5.11±1.01** 5.32 (3.5–6.3)
0 (N=29) 4.12±1.11 4.18 (1.9–6.3)
2 (N=9) 5.79±1.71** 5.04 (3.9–8.4)
Decrease of JSW, either knee: *P<0.05 and **P<0.01 compared with group <0.5 (Mann–Whitney test). Increase in K–L grade, signal knee: *P<0.05 compared with group 0 (Mann–Whitney test). Increase in K–L grade, both knees: **P<0.01 compared with group 0 (ANOVA, Bonferroni post test). †Serum COMP levels in the group of ten patients who progressed by only one K–L grade were not significantly different from levels in the group of patients who did not progress (data not shown).
Fig. 4. Baseline (a) and study end (b) serum COMP levels in patients sorted according to severity of OA progression, expressed as the change in K–L grade for the signal knee over 3 years. Horizontal lines show mean COMP levels. Group 0 nonprogressors (N=33); group 1 increased one K–L grade (N=15). *=P<0.05 and ns=not significant when compared with Group 0 (Mann-Whitney test).
in these two groups overlapped (Fig. 4), the mean COMP level was higher in group 1 patients and the difference in baseline COMP level between the two groups was significant using a non-parametric Mann–Whitney test (P<0.05), the difference in COMP level at the study end was not significant. Subjects were also categorized according to the sum of the change that occurred in K–L grade for both knees: group 0 no change, group 1 increased one K–L grade, group 2 increased 2 K–L grades. The mean COMP levels in these groups differed significantly at baseline as well as at study end (ANOVA, P<0.01 before adjusting and P<0.05 after adjusting for age). Group 2 had higher mean COMP levels than (1) group 0 non-progressors (P<0.01 before adjusting and P<0.05 after adjusting for age) and (2) group 1 that progressed by one K–L grade (P<0.05, both before and after adjusting for age) at baseline as well as at study end (Bonferroni multiple comparisons post hoc test)
Fig. 5. Baseline (a) and study end (b) serum COMP levels in patients sorted according to severity of OA progression, expressed as the change in K–L grade summed for both knees over 3 years. Horizontal lines show mean COMP levels. Group 0 nonprogressors (N=29); group 1 increased one K–L grade (N=10); group 2 increased two K–L grades (N=9). **=P<0.01 when compared with Group 0; *=P<0.05 when compared with Group 1 (ANOVA and Bonferroni post test).
(Fig. 5). Mean and median COMP levels by radiographic progression status are summarized in Table II. Baseline serum COMP levels correlated strongly with study end serum COMP levels (P<0.0001) (Fig. 6). Since two former studies12,13 have shown a correlation between progression status and the change in serum COMP level over time, we also evaluated our data in this manner. Progression was defined as a change in JSW of >0.5 mm for either knee over 3 years, or as a 2 grade increase in K–L grade summed for both knees over 3 years. Although we noticed an increase in serum COMP level over the 3-year follow-up period in the whole cohort (Table II), this increase was similar for all patients. The change in COMP (from baseline to study end) was not significantly different in the groups separated according to progression status using a paired t-test.
Osteoarthritis and Cartilage Vol. 10, No. 9
Fig. 6. Correlation between baseline and study end serum COMP levels. The linear regression line is shown.
Baseline and study end COMP levels did not correlate with changes of ISK or WOMAC indices.
Discussion We have developed a sandwich ELISA with two monoclonal antibodies to human COMP. The sandwich ELISA format offers the advantage of higher specificity and sensitivity over other ELISA methods of biomarker quantification. We performed this study with the goal of relating serum COMP levels to the progression of knee OA. We demonstrated that baseline serum COMP levels correlated with the magnitude of OA radiographic progression, thus serum COMP has the potential to be a prognostic marker of disease progression. In previous studies, we have observed an association between serum COMP level and knee OA disease severity (K–L grade), serum COMP level and the number of OA affected large joints18, and serum COMP level and synovitis23. In this much smaller study we observed the same trends, which however, did not achieve statistical significance, we presume due to the lack of power. Age was the only other factor that we have identified in previous studies, whose effect on serum COMP levels remained statistically significant even in the relatively small group analysed here. However, adjusting COMP levels for age did not affect conclusions reached by this study. Several previous studies have evaluated serum COMP and OA progression. Sharif et al.12 and Petersson et al.13 found no correlation between baseline serum COMP levels and subsequent OA progression during 5 years of follow-up of patients with established knee OA and during 3 years of follow-up of patients with early knee OA, respectively. They found, however, a significant positive correlation between the change (the increase) in serum COMP level during the first year12 and during the 3-year interval13 respectively, and radiographic progression during the whole follow-up interval. Conrozier et al.14 found a significant correlation between baseline serum COMP levels and 1-year radiographic progression of patients with established hip OA. They found no correlation between the change in serum COMP level over time and yearly mean narrowing of joint space width. In agreement with the hip OA study, we observed a similar correlation between serum COMP level at baseline and radiographic progression in a knee OA cohort. In contrast to the two previous studies on knee OA, we detected no association between the change in serum COMP level and progression status. Instead, we found that serum COMP levels at baseline and at study end correlated strongly, indicating that serum COMP level distinguished
711 progressors from non-progressors throughout the follow-up interval. Although it is expressed by other tissues within a joint26, COMP prevails in the articular cartilage and it has been demonstrated that serum levels are representative of cartilage catabolism10. Thus, the correlation of serum COMP level with joint space narrowing, a radiographic indicator of cartilage catabolism, is plausible. The recent report that COMP is expressed by vascular smooth muscle cells27 hampers straightforward interpretation of serum COMP levels in the setting of joint disease. While the potential contribution of these cells to serum COMP level is difficult to predict, it seems reasonable to expect that production of COMP by non-cartilaginous tissues is responsible for a certain background or constitutive serum level, and that pathological degradation of cartilage adds significantly to this background level. HA is the only other biomarker, which has been shown to predict knee OA progression11. Like COMP, HA is not tissue specific but serum levels increase significantly in response to activity of pathological joints28,29. Thus, it is indeed possible for a tissue specific process to measurably contribute to the level of a biomarker in the circulation. In general, we found the best correlation between COMP level and radiographic progression when both knees were included in the analysis. This result is expected for a marker, such as COMP, whose potential release from all joints in the body contributes to the serum level. It is difficult to decide whether measurement of JSW is superior to composite scoring systems such as K–L grading for assessing radiographic progression of OA, thus the use of both approaches has been recommended30. It has became obvious recently, that precise measurement of the interbone distance on weight-bearing anteroposterior radiographs may be obscured by the fact that even relatively minor changes in a patient’s position (usually due to different degrees of flexion) can result in large changes in the measured apparent joint space width. This methodological shortcoming impacts practically all longitudinal studies, including ours, initiated prior to 1995, when the influence of joint positioning on the standardization and reproducibility of tibiofemoral JSW measurement on sequential OA knee joint radiographs was published31. A recent retrospective assessment of several former longitudinal OA studies has elucidated the overall effect of lower limb positioning on JSW measurements32,33. In total, correct alignment of the medial tibial plateau with the X-ray beam, and standardization of knee rotation on sequential radiographs ranged from 9 to 19% of all examined pairs of radiographs in four retrospectively assessed cohorts of OA patients32. Significant widening of joint space in the medial compartment was found in 23% of all examined knees. We are aware of this potential shortcoming, the presence of which in our study is indicated by the fact that we recorded widening of the knee joint space in 17 patients, (35%) out of 48 who received placebo, over the period of 3 years. Although, hypothetically, there may be several reasons for apparent radiographic improvement33, it is very probable that at least some of these apparent ‘improvements’ can be ascribed to misalignment of sequential radiographs. However, a correlation of serum COMP level with radiographic OA progression was not based on JSW measurements alone in this study. We also observed a correlation of serum COMP level with K–L grade worsening. Although the K–L grading system is considered to be relatively insensitive, results do not depend on JSW alone, are less affected by joint positioning and standardization of grading is based
712 upon atlases. Of note, we recorded no cases of ‘improvement’ over the 3-year interval using the K–L grading system. The fact that serum COMP and OA progression are correlated on the basis of both JSW measurements and K–L grading argues strongly, in our opinion, that this finding is valid. The test does not meet the criterion of discriminant validity, as its ability to detect small changes seems to be rather limited; however, a similar limitation could be said to apply to the currently accepted ‘gold standard’ for OA, namely joint radiography. Furthermore, at present, serum COMP level cannot be used to prognosticate the course of OA in an individual patient due to the high variability and the overlap of COMP values in affected and unaffected groups. However, COMP level may aid in identifying groups of patients at risk of faster disease progression and in differentiating groups of patients during clinical studies and drug trials that are analysed using statistical methods.
References 1. Hedbom E, Antonsson P, Hjerpe A, Aeschlimann D, Paulsson M, Rosa-Pimentel E, et al. Cartilage matrix proteins. An acidic oligomeric protein (COMP) detected only in cartilage. J Biol Chem 1992;267: 6132–6. 2. Chen H, Deere M, Hecht JT, Lawler J. Cartilage oligomeric matrix protein is a calcium-binding protein, and a mutation in its type 3 repeats causes conformational changes. J Biol Chem 2000;275:26538–44. doi:10.1074/jbc.M909780199. 3. Oldberg A, Antonsson P, Lindblom K, Heinegard D. COMP (cartilage oligomeric matrix protein) is structurally related to the thrombospondins. J Biol Chem 1992;267:22346–50. 4. Rosenberg K, Olsson H, Morgelin M, Heinegard D. Cartilage oligomeric matrix protein shows high affinity zinc-dependent interaction with triple helical collagen. J Biol Chem 1998;273:20397–403. 5. Thur J, Rosenberg K, Nitsche DP, Pihlajamaa T, Ala-Kokko L, Heinegard D, et al. Mutations in cartilage oligomeric matrix protein causing pseudoachondroplasis and multiple epiphyseal displasia affect binding of calcium and collagen I, II, and IX. J Biol Chem 2001;276:6083–92. doi:10.1074/ jbc.M009512200. 6. Holden P, Meadows RS, Chapman KI, Grant ME, Kadler KE, Briggs MD. Cartilage oligomeric matrix protein interacts with type IX collagen, and disruptions to these interactions identify a pathogenetic mechanism in a bone dysplasia family. J Biol Chem 2001;276:6046–55. doi:10.1074/jbc.M009507200. 7. Lawrence RC, Helmick CG, Arnett FC, Deyo RA, Felson DT, Giannini EH, et al. Estimates of the prevalence of arthritis and selected musculoskeletal disorders in the United States. Arthritis Rheum 1998;41:778–99. 8. Brandt KD. The pathogenesis of osteoarthritis. Rheumatol Rev 1991;1:3–11. 9. Garnero P, Rousseau J-C, Delmas PD. Molecular basis and clinical use of biochemical markers of bone, cartilage, and synovium in joint diseases. Arthritis Rheum 2000;43:953–68. 10. Otterness IG, Swindell AC, Zimmerer RO, Poole AR, Ionescu M, Weiner E. An analysis of 14 molecular markers for monitoring osteoarthritis: segregation of
V. Vilı´m et al.: Serum COMP and progression of OA
11.
12.
13.
14.
15.
16.
17.
18.
19. 20.
21.
22.
23.
24.
the markers into clusters and distinguishing osteoarthritis at baseline. Osteoarthritis Cart 2000;8: 180–5. doi:10.1053/joca.1999.0288. Sharif M, George E, Shepstone L, Knudson W, Thonar EJ-MA, Cushnaghan J, et al. Serum hyaluronic acid level as a predictor of disease progression in osteoarthritis of the knee. Arthritis Rheum 1995;38:760–7. Sharif M, Saxne T, Shepstone L, Kirwan JR, Elson CJ, Heinegard D, et al. Relationship between serum cartilage oligomeric matrix protein levels and disease progression in osteoarthritis of the knee joint. Br J Rheumatol 1995;34:306–10. Petersson IF, Boegard B, Svensson D, Heinegard D, Saxne T. Changes in cartilage and bone metabolism identified by serum markers in early osteoarthritis of the knee joint. Br J Rheumatol 1998;37:46–50. Conrozier T, Saxne T, Shan Sei Fan C, Mathieu P, Tron A-M, Heinegard D, et al. Serum concentrations of cartilage oligomeric matrix protein and bone sialoprotein in hip osteoarthritis: A one year prospective study. Ann Rheum Dis 1998;57:527–32. Georges C, Vigneron H, Ayral X, Listrat V, Ravaud P, Dougados M, et al. Serum biologic markers as predictors of disease progression in osteoarthritis of the knee. Arthritis Rheum 1997;40:590–1. Petersson I, Boegard T, Dahlstrom J, Svensson B, Heinegard D, Saxne T. Bone scan and serum markers of bone and cartilage in patients with knee pain and osteoarthritis. Osteoarthritis Cart 1998;6: 33–9. Vilı´m V, Lenz ME, Vyta´sˇek R, Masuda K, Pavelka K, Kuettner KE, et al. Characterization of monoclonal antibodies recognizing different fragments of cartilage oligomeric matrix protein in human body fluids. Arch Biochem Biophys 1997;341:8–16. Clark AG, Jordan JM, Vilı´m V, Renner JB, Dragomir A, Luta G, et al. Serum cartilage oligomeric matrix protein reflects osteoarthritis presence and severity: The Johnston county osteoarthritis project. Arthritis Rheum 1999;42:2356–64. Kellgren JH, Lawrence JS. Radiological assessment of osteo-arthrosis. Ann Rheum Dis 1957;16:494–502. Altman R, Asch E, Bloch D, Bole D, Bornstein D, Brandt K, et al. Development of criteria for the classification and reporting of osteoarthritis. Classification of osteoarthritis of the knee. Arthritis Rheum 1986;29:1039–49. Lequesne MG, Mery C, Samson M, Gerard P. Indexes of severity for osteoarthritis of the hip and knee. Validation-value in comparison with other assessment tests. Scand J Rheumatol 1987;65(Suppl.): 85–9. Bellamy N, Buchanan WW, Goldsmith CH, Campbell J, Stitt LW. Validation study of WOMAC: a health status instrument for measuring clinically important patient relevant outcomes to antirheumatic therapy in patients with osteoarthritis of the hip or knee. J Rheumatol 1988;15:1833–40. Vilı´m V, Vyta´sˇek R, Oleja´rova´ M, Macha´cˇek S, Gatterova´ J, Procha´zka B, et al. Serum cartilage oligomeric matrix protein reflects the presence of clinically diagnosed synovitis in patients with knee osteoarthritis. Osteoarthritis Cart 2001;9:612–18. doi:10.1053/joca.2001.0434. Pavelka K, Gatterova´ J, Gollova´ V, Urbanova´ Z, Sedla´cˇkova´M, Altman RD. A 5-year randomized controlled, double-blind study of glycosaminoglycan
Osteoarthritis and Cartilage Vol. 10, No. 9
25. 26.
27.
28.
29.
polysulphuric acid complex (Rumalon®) as a structure modifying therapy in osteoarthritis of the hip and knee. Osteoarthritis Cart 2000;8:335–42. doi:10.1053/joca.1999.0307. Rodbard D. Statistical quality control and routine data processing for radioimmunoassays and immunometric assays. Clin Chem 1974;20:1255–70. Muller G, Michel A, Altenburg E. COMP (cartilage oligomeric matrix protein) is synthesized in ligament, tendon, meniscus, and articular cartilage. Conn Tissue Res 1998;39:233–244. Reissen R, Fenchel M, Chen H, Axel DI, Karsch KR, Lawler J. Cartilage oligomeric matrix protein (thrombospondin-5) is expressed by human vascular smooth muscle cells. Arterioscler Thromb Vasc Biol 2001;21:47–54. Engstrom-Laurent A, Hallgren R. Circulating hyaluronic acid levels vary with physical activity in healthy subjects and in rheumatoid arthritis patients. Relationship to synovitis mass and morning stiffness. Arthritis Rheum 1987;30:1333–8. Manicourt DH, Poilvache P, Nzeusseu A, VanEgeren A, Devogelaer J-P, Lenz ME, et al. Serum levels of
713
30. 31.
32.
33.
hyaluronan, antigenic keratan sulfate, matrix metalloproteinese 3, and tissue inhibitor of metalloproteinases 1 change predictably in rheumatoid arthritis patients who have begun activity after a night of bed rest. Arthritis Rheum 1999;42:1861–9. Dieppe PA. Recommended methodology for assessing the progression of osteoarthritis of the hip and knee joints. Osteoarthritis Cart 1995;3:73–7. Buckland-Wright JC, MacFarlane DG, Williams SA, Waed RJ. Accuracy and precision of joint space width measurements in standard and macroradiographs of osteoarthritic knees. Ann Rheum Dis 1995;54: 872–80. Mazzuca SA, Brandt KD, Dieppe PA, Doherty M, Katz BP, Lane KA. Effect of alignment of the medial tibial plateau and x-ray beam on apparent progression of knee osteoarthritis in the standing anteroposterior radiograph. Arthritis Rheum 2001;44:1786–94. Mazzuca SA, Brandt KD, Dieppe PA, Doherty M, Sowers M. Reasons for apparent widening of medial joint space in conventional standing anteroposterior knee radiographs of patients with osteoarthritis (Abstract). Trans 47th Ann Meet ORS 2001;26:657.