19 [2.21, 4.62],

p < 0.001 for KL ≥ 2 in HBM cases vs. controls). Our data indicate an increased prevalence of radiographic knee OA in HBM individuals compared with controls, consistent with existing epidemiological evidence that increased BMD is a risk factor for OA at this joint [2], [5], [6], [8] and [36]. As we hypothesized, associations with HBM were stronger for the osteophyte variables compared with both semi-quantitative and measured JSN, particularly in models adjusted for BMI, and the stronger association we observed CHIR-99021 cell line between HBM and knee OA defined as KL ≥ 2 (osteophytosis) versus KL ≥ 3 (osteophytosis plus JSN) is likely to be a further reflection of this. However, we found little evidence of an association between HBM and subchondral sclerosis, possibly explained by the very low prevalence of this feature and/or the difficulty of assessing its presence or absence on simple visual inspection

of radiographs. A positive association between HBM and chondrocalcinosis PCI-32765 clinical trial was also seen; however, while chondrocalcinosis was also associated with radiographic knee OA, it did not explain the HBM–OA association observed. Adjusting for BMI attenuated the HBM–knee OA association by approximately 50%, suggesting that the HBM–OA association at the knee is partly mediated through increased BMI. We previously reported that HBM is associated with a metabolic phenotype comprising greater BMI [9], and increased fat mass in women [13]; similar body composition changes have been observed in association with OA [37] and [38].

The primary mechanism by which weight/BMI contributes to OA in load-bearing joints has not been fully established; in particular the relative contribution of increased joint loading due to greater body weight [14] and [15] versus the effects of circulating medroxyprogesterone metabolic factors such as adipokines [39] remains to be determined. It is interesting that, in our total body DXA analyses, adjusting for fat mass led to greater attenuation of the HBM–OA association compared with lean mass adjustment. This is consistent with some previous studies suggesting that increased fat mass relative to lean mass may be particularly associated with OA at the knee [40] and [41], possibly indicating a role for metabolic factors over and above body weight in determining OA risk. There may be gender differences in these relationships, for example a recent study suggested that fat mass and lean mass may be more important determinants of knee OA in women and men respectively [42]; this observation may therefore reflect the greater proportion of women in our study. Unfortunately numbers of males with total body DXA data did not permit gender-stratified analysis. It should be noted that these analyses were restricted to a subgroup of HBM cases and family controls only, resulting in limited statistical power.