Which risk factors determine cartilage thickness and composition change in radiographically normal knees? - Data from the Osteoarthritis Initiative.

Objective: Therapy for osteoarthritis ideally aims at preserving structure before radiographic change occurs. This study tests: a) whether longitudinal deterioration in cartilage thickness and composition (transverse relaxation-time T2) are greater in radiographically normal knees "at risk" of incident osteoarthritis than in those without risk factors; and b) which risk factors may be associated with these deteriorations.

Design: 755 knees from the Osteoarthritis Initiative were studied; all were bilaterally Kellgren Lawrence grade [KLG] 0 initially, and had magnetic resonance images available at 12- and 48-month follow-up. 678 knees were "at risk", whereas 77 were not (i.e., non-exposed reference). Cartilage thickness and composition change was determined in 16 femorotibial subregions, with deep and superficial T2 being analyzed in a subset (n ​= ​59/52). Subregion values were used to compute location-independent change scores.

Results: In KLG0 knees "at risk", the femorotibial cartilage thinning score (-634 ​± ​516 ​μm) over 3 years exceeded the thickening score by approximately 20%, and was 27% greater (p ​< ​0.01; Cohen D -0.27) than the thinning score in "non-exposed" knees (-501 ​± ​319 ​μm). Superficial and deep cartilage T2 change, however, did not differ significantly between both groups (p ​≥ ​0.38). Age, sex, body mass index, knee trauma/surgery history, family history of joint replacement, presence of Heberden's nodes, repetitive knee bending were not significantly associated with cartilage thinning (r<1%), with only knee pain reaching statistical significance.

Conclusions: Knees "at risk" of incident knee OA displayed greater cartilage thinning scores than those "non-exposed". Except for knee pain, the greater cartilage loss was not significantly associated with demographic or clinical risk factors.