Background/Purpose:

Osteoarthritis (OA) involves the whole joint, with alterations of cartilage, synovium and bone. Whether bone changes are a cause or consequence of the disease remains controversial. As there is no curative treatment for OA, identifying causal factors is of major importance. Thus, our aim was to compare observational and causal associations between BMD and different OA phenotypes.

Methods:

Individual level clinical and genetic data on 384 838 unrelated participants of European ancestry from the UK Biobank were analysed. As DXA was only available for 1% of the participants, we first tested estimated BMD of the heel (eBMD) as a risk factor for OA at all sites (n= 48 431 cases), at the knee (n= 19 727) or at the hip (n= 11 875), respectively, using hospital diagnoses. The analyses were replicated using knee and hip joint replacement therapy as definitions for severe OA. To assess causality, mendelian randomization (MR) analyses were performed using genetic instruments as proxies for femoral neck (FN) BMD and lumbar spine (LS) BMD. As they are randomly assigned at birth, they are not affected by confounders. Four MR methods were performed with sensitivity analyses excluding genetic instruments associated with body mass index (BMI), to preclude pleiotropy with a known causal confounder. The level of significance after Bonferroni correction was set at 0.05/6 : P < 0.0083. All participants gave their written consent to take part in the UK Biobank settings. This project was approved by the UK Biobank committee and has ethical permission.

Results:

Linear regression models adjusted for age, sex and BMI revealed that an increase in eBMD by 1 SD was associated with increased risk of all OA (OR= 1.03 [95% CI 1.02 to 1.04]), knee OA (OR= 1.07 [95% CI 1.05 to 1.09]) and hip OA (OR= 1.09 [95% CI 1.07 to 1.11]). Results were similar when considering only incident OA cases or severe OA.

MR analyses by the inverse-variance weighted method demonstrated a causal effect of genetically determined FN BMD, that was associated with all OA (per 1 SD increase, OR=1.15 [95% CI: 1.06 to 1.25]) and both knee (OR=1.20 [95% CI: 1.05 to 1.37]) and hip OA (OR=1.23 [95% CI: 1.12 to 1.36]). Similar findings were demonstrated with joint replacement therapy definitions. The weighted median, the penalized weighted median and Egger’s regression method provided similar results. LS BMD was only causally associated with knee OA.

Conclusion:

This study demonstrates that high BMD is a risk factor for knee and hip OA and for severe OA, independently of age, sex and BMI. To our knowledge, this is the first evidence of a causal effect of FN BMD. Increasing BMD may not be beneficial for the prevention or the treatment of OA.

ACR Meeting Abstracts - https://acrabstracts.org/abstract/bone-mineral-density-is-a-causal-risk-factor-for-knee-and-hip-osteoarthritis-a-population-based-and-mendelian-randomization-study-in-the-uk-biobank/