| Background. Hormone therapy (HT) along with calcium and vitamin D (CaD) supplementation prevent bone loss and reduce fracture risk in postmenopausal women. While the association of estrogen and systemic bone loss has been evaluated in many studies, few studies have explored the association between estrogen and oral bone loss. Past studies have compared HT users with non-users in relation to oral bone loss, and not with actual serum estrogen level. To understand the role of estrogen in oral bone loss, concomitant analysis of both serum estrogen level and exogenous HT use is necessary.;Materials and Methods. In Aim 1 and 2, we constructed two weighted genetic risk scores (GRSs) based on 16 fracture-associated SNPs (Fx-GRS) and 50 bone mineral density (BMD) SNPs (BMD-GRS) identified from the published meta-genome-wide association study (GWAS), and evaluated the interaction of these GRSs with HT (Aim 1) and CaD (Aim 2) use on fracture risk using Cox proportional hazards regression and the case-only approach. We analyzed data from 9,922 white postmenopausal women from the Women's Health Initiative HT randomized clinical trials (CT) in Aim 1. Data from a subset of 5,823 women who also participated in CaD CT were analyzed in Aim 2. We tested the presence of multiplicative interaction using an interaction term in the Cox regression models. Interaction on the multiplicative scale was also evaluated in a case-only analysis using logistic regression. We calculated the relative excess risk due to interaction (RERI) as a measure of additive interaction. For this analysis, we used both continuous scale and quartile categorized GRSs (lowest, middle two and highest).;In Aim 3, we evaluated the associations of oral alveolar crest heights (ACH) with serum estradiol concentration (E2) and HT use in 613 postmenopausal women from the Buffalo OsteoPerio study, an ancillary study to the WHI observational Study. We compared baseline measure as well as 5-year changes in whole mouth mean and worst site ACH levels among serum E2 level groups (undetectable, 5.00--18.00, 18.00--46.07, and >46.07 pg/ml), and between HT never and ever users using analysis of variance and analysis of covariance (ANCOVA). We analyzed the association of ACH loss with serum E2 and HT use using logistic regression.;Results. We did not find any evidence for the presence of multiplicative interaction between HT and either of the GRSs on total fracture incidence (p for interaction in Cox regression model, Fx-GRS: 0.49, BMD-GRS: 0.63). The results were similar in case-only analyses when analyzed with 1,608 fracture cases. However, we observed a significant positive additive interaction in both GRS. Placebo assigned women in the highest quartile of both GRSs had excess fracture risk, with RERI (95% CI) of 0.35 (0.01, 0.69; p=0.047) and 0.38 (0.01, 0.75; p=0.046) for Fx-GRS and BMD-GRS, respectively. The relative risk reduction across Fx-GRS quartile were 21%, 26% and 34% for lowest, middle two and highest quartile, respectively. We observed similar risk reductions in BMD-GRS strata. However, the heterogeneity of HT effect on fracture was not statistically significant (Fx-GRS p for heterogeneity = 0.40, BMD-GRS p for heterogeneity = 0.40).;In the analysis of CaD, we did not observe any significant interaction of CaD with Fx-GRS on either of the scale examined. However, we found significant multiplicative (p for interaction = 0.01) and negative additive interactions between BMD-GRS and CaD (RERI (95% CI, p-value), mid two quartiles: -0.61 (-1.18, -0.05, p=0.03), highest quartile: -0.76 (-1.43, -0.09, p=0.03)). In a stratified analysis, we observed a significant fracture risk reduction with CaD intervention in women with low genetic predisposition to low BMD (lowest quartile of BMD-GRS) (HR=0.60, 95 % CI=0.44, 0.81), but not in those with higher BMD-GRS (p for heterogeneity = 0.01). No significant heterogeneity of CaD effect was found across Fx-GRS strata (p for heterogeneity = 0.90).;We found no association between serum E2 and oral ACH. However, we observed that women who had never used HT had more whole mouth mean ACH loss (adjusted mean +/- standard error of the mean, 2.49 mm +/- 0.05 vs. 2.32 mm +/- 0.03, ANCOVA p-value=0.01) and worst site of ACH loss (4.73 mm +/- 0.12 vs. 4.42 mm +/- 0.08, ANCOVA p-value=0.03) compared to ever users. When analyzed with ACH severity categories, we observed that the odds of having severe ACH loss for those who never used HT were significantly higher than ever users (OR=2.00; 95% CI: 1.11--3.62). We did not find any association between five-year progression in ACH and baseline serum E2 measure or HT. (Abstract shortened by UMI.). |
PDF Full Text Request