Purpose: There is consistent evidence of the high impact of increased BMI on knee OA and knee pain. There is also some evidence that hormone levels are associated to OA, though this relationship is influenced by BMI. We hypothesized that the decreasing hormone levels may play a role in the causal pathway that leads to a higher prevalence and incidence of knee OA and pain at least in women. Therefore, we tested this hypothesis by using Structural Equation Modeling (SEM) in men and women independently. The aim of this study was to explore the causal relationships between sex hormones, inflammation factors and BMI in radiographic knee OA (RKOA) and chronic knee pain (CKP). Methods: We used the Rotterdam Study cohort, a population based prospective study with participants aged 45 and older. The participants underwent an interview on pain questionnaire, blood and radiographic measurements at baseline as well as follow-up radiographic measurements. In our study for prevalent radiographic knee OA (RKOA) we included a total of 12,546 observed knees of which 606 were knee OA cases. For incident radiographic knee OA, from the 11,940 knees without OA, 4,090 were lost to follow-up, so we included 7,850 knees without knee OA at baseline from which 533 developed knee OA during a median follow-up time of 5 years. RKOA was defined as a KL score of 2 or higher. Participants provided answers during a home interview about current musculoskeletal pain in the knees, hips, hands and lower back. If present, the duration was also recorded. We included 5,897 participants in our analysis for prevalent chronic knee pain (CKP). There were 2,349 participants with CKP and 3,548 with no pain in either of their knee joints. CKP was defined as having pain in the knee joint for more than 3 months. We estimated the direct and indirect effects of BMI, CRP, estradiol and sex hormone binding globulin (SHBG) on prevalence and incidence of RKOA and CKP by using SEM. SEM is a multivariate statistical analysis technique that is used to analyze structural relationships. Results: Our results from the SEM analysis on the prevalence of RKOA (Figure 1), show that BMI has a significant association with an effect estimate of 0.1 on RKOA in women. The indirect effects from BMI through estradiol or through SHBG to RKOA were also present only in women but with a much smaller effect estimate (estimate=0.004 through estradiol, estimate=0.01 through SHBG) compared to the direct effect of BMI. In contrast, we found an indirect effect from BMI through CRP to RKOA (estimate=-0.01) only in men. In case of CKP, we did not find indirect effects of BMI though the sex hormones or CRP in neither women nor men. BMI has a strong direct effect on CKP in both women (estimate = 0.18) and men (estimate = 0.15) as shown in Figure 2. The SEM models showed good indices of fit: CFI (Comparative Fit Index) and the RMSEA (Root Mean Square Error of Approximation). CFI should be above 0.95 and RMSEA below 0.05. For the RKOA model: CFI=0.98 and RMSEA=0.046. For the CKP model: CFI=0.98 and RMSEA=0.038. Conclusions: We found indeed indirect effects of BMI on RKOA through estradiol and SHBG in women, but not in men. However, the contribution of the path through sex hormones is negligible to the total effect of BMI on RKOA. In men we found a (small) indirect effect of BMI on RKOA through CRP, a well-known inflammation factor. Curiously, this effect was protective.View Large Image Figure ViewerDownload Hi-res image Download (PPT)
Read full abstract