Background Associations between adiposity and osteoarthritis (OA) in non-weight-bearing joints suggest that besides mechanical factors also systemic influences contribute to OA. Systemically active substances secreted by adipose tissue, including leptin and adiponectin, are hypothesized to play a role in OA.
Objectives To examine whether leptin and adiponectin mediate the association between body mass index (BMI) and hand and knee OA.
Methods Cross-sectional data of a population-based study were used. Participants completed questionnaires and underwent standardized physical examination of hands and knees to define OA according to clinical American College of Rheumatology criteria. Fasting serum leptin and adiponectin were measured with immunoassays. Potential mediation was investigated using the Baron and Kenny framework. Four assumptions were investigated: associations between (1) BMI and OA (pathway C), (2) BMI and adipokines (pathway A), (3) adipokines and OA (pathway B), and (4) attenuation of the total association between BMI and OA after including adipokines (pathway C'). No exposure-mediator interaction and mediator-outcome confounding was assumed. Assumptions were investigated using logistic and linear regression analyses as appropriate. Odds Ratios (ORs) were calculated per standard deviation (SD) difference in BMI, and per ten and five units difference in leptin and adiponectin, respectively. Percentage mediation with 95% confidence intervals (CIs) was estimated, only when all four assumptions were fulfilled. Models were adjusted for age, ethnicity and education, and stratified by sex.
Results In 6462 participants (56% women, median age 56 years (range 45–65), mean BMI 26.3 kg/m2), prevalence of hand OA, knee OA and combined hand and knee OA were 8%, 10% and 4%, respectively. Median leptin and adiponectin concentrations were 7.1 ug/L (range 0.9–60.9) and 6.0 mg/L (0.5–23.7) in men, and 19.1 ug/L (0.5–262.0) and 10.5 mg/L (0.5–98.6) in women. BMI was positively associated with OA presence and serum leptin in both men and women (Table). A negative association was observed between BMI and serum adiponectin (-0.73 mg/L per SD BMI, 95% CI -0.55;-0.91). Leptin was positively associated with most OA types, except knee OA in men. Leptin partially mediated the association of BMI with hand OA in men (9% mediation, 95% CI 5;17) and women (30%, 13;198), and the association of BMI with knee OA in women (15%, 12;21). Similar analyses for adiponectin revealed a negative association of adiponectin with hand OA in men and partial mediation of the association of BMI with hand OA in men (19%, 12;37), whereas mediation was absent in other subgroups.
Conclusions Leptin partially mediated the association of BMI and hand OA in both men and women, as did adiponectin in men. In addition, mediation by leptin for the association of BMI and knee OA was demonstrated in women. These findings suggest that systemic mediators contribute to hand OA, and to a lesser extent to knee OA.
Disclosure of Interest None declared