Background/Purpose: Laboratory and animal studies suggest that lower sex hormone levels promote production and activity of pro-inflammatory cytokines that trigger bone resorption. This paradigm is a hypothesized mechanism for accelerated bone loss following menopause, and may influence bone loss in men and premenopausal women as cytokine levels may vary across the range of endogenous hormones within these groups. Confirmation of these hypotheses in epidemiologic studies could inform the development of interventions to prevent bone loss. We examined whether interleukin-6 (IL-6) concentration is a mediator of the cross-sectional relation between total estradiol concentration and hip bone mineral density (BMD), separately among men, premenopausal women and postmenopausal women in the Framingham Osteoporosis Study. We hypothesized that a direct association between total estradiol and BMD would be attenuated after adjusting for IL-6, suggesting IL-6 as a potential mediator.

Methods: Fasting blood samples and BMD measures were obtained (1996-2001) from 1,230 men, 215 premenopausal and 1,300 women. Serum total estradiol (pg/mL) and IL-6 (pg/mL) concentrations were measured via LC-MS/MS and ELISA, respectively. Femoral neck BMD (g/cm²) was measured with a Lunar DPX-L. Separately for men, premenopausal and postmenopausal women, total estradiol was categorized into quartiles and analysis of covariance was used to calculate least squares-adjusted mean BMD for each quartile and test for a linear trend, adjusting for age, BMI, height, physical activity, current smoking  and in postmenopausal women, use of hormone replacement therapy (HRT). To examine potential effect modification by HRT in postmenopausal women analyses were also stratified by HRT use. Models were additionally adjusted for IL-6 and if an association between total estradiol and BMD was consequently attenuated, IL-6 was considered a mediator.

Results: Mean age was 61 y (range 29-86 y). There was a statistically significant positive association between total estradiol and BMD among men and a similar, though not statistically significant, relation in postmenopausal women (Table). There was no association in premenopausal women. In postmenopausal women the relation was similar for HRT users and non-users (data not shown). Associations were nearly identical after adjusting for IL-6, suggesting no mediation by IL-6.

Conclusion: These cross-sectional findings suggest that while lower estrogen status was associated with lower hip BMD in men and postmenopausal women, these relations are not influenced by IL-6. Subsequent analyses should examine additional markers of estrogen status (free estradiol, estrone, sex hormone-binding globulin) and pro-inflammatory cytokines (CRP, TNF-α), the role of hormonal contraceptive use and phase of menstrual cycle in premenopausal women, and prospective changes in BMD.