Background The actual value of biochemical markers (biomarkers) for osteoarthritis (OA) remains doubtful, partly because available studies are mostly small and focus on late rather than the more relevant early OA stages.

Objectives To investigate the ability of (combinations of) biomarkers to reflect OA burden and 5-year progression in a large cohort of (very) early knee and hip OA subjects.

Methods Ten biomarkers (Cartilage: uCTX-II, sCOMP, sPIIANP, sCS846. Bone: uCTX-I, uNTX-I, sOC, sPINP. Synovial tissue: sHA, sPIIINP) were assessed by ELISA or RIA in baseline samples from CHECK (Cohort Hip & Cohort Knee), a prospective cohort of 1002 individuals with symptoms of (very) early knee and/or hip OA.

Knee and hip radiographs were obtained at baseline and after 2 and 5 years. OA burden was expressed as the summed Kellgren & Lawrence (K&L) score for knees and hips at baseline. OA progression was expressed as the area under the curve (AUC) of 5-year summed K&L score progression.

Associations were investigated through multiple linear regression. Biomarkers were used individually as well as in combinations of standardized scores (Z-scores).

Results Data were complete for approximately 800 participants at baseline and 725 participants at all time points. Baseline summed K&L grades 0, 1, 2, 3, and 4 were present in 33.9, 26.8, 24.0, 9.6, and 5.7% of participants, respectively (max K&L grade=1 for individual joints). AUC of summed K&L grades between baseline and 5 years were 0, 1.5-3.0, 3.0-5.5, and 5.5-17.5 for 39.0, 18.3, 22.5, and 20.1% of participants, respectively.

At baseline, uCTX-II, sCOMP, sPIIANP, sHA, and sPIIINP showed statistically significant associations with the summed K&L score (max standardized betas 0.183, P=0.000 and 0.161, P=0.000 for sHA and uCTX-II, respectively), that persisted for sHA and uCTX-II after adjustment for age, gender and BMI (stand betas 0.113, P=0.003 and 0.129, P=0.000, respectively). Only cartilage marker combinations showed stronger associations with baseline summed K&L grades than individual biomarkers (stand beta 0.150, P=0.000 for uCTX-II+sCOMP+PIIANP+CS846 combination, after adjustment).

Associations with AUC of the summed K&L score were statistically significant for uCTX-II, sCOMP, sPIIANP, and sCS846 (stand betas 0.089-0.133, P≤0.018) and persisted after adjustment for age, gender, BMI, and baseline summed K&L score for sCOMP only (stand beta 0.105, P=0.005). Stronger associations were observed for cartilage and synovial combinations (stand beta 0.128, P=0.001 for uCTXII+sCOMP+sPIIANP+sCS846, and stand beta 0.110, P=0.003 for sCOMP+sPIIINP).

Conclusions Associations with burden and progression of (very) early OA were observed for cartilage and synovial biomarkers. The low grade of the associations may be due to limitations of biomarkers as well as radiography in reflecting disease in OA, especially in early-stage disease. Associations may be stronger at 10-year follow-up and/or when using more sensitive measures of individual radiographic OA features (e.g. osteophytes, joint space narrowing). Nevertheless, together with literature data these results indicate that the search for biomarkers for (very) early OA should be primarily aimed at (combinations of) cartilage and synovial metabolism.

This study was funded by CHECK, an initiative of the Dutch Arthritis Association.

Disclosure of Interest None Declared