School of Public Health

Laura B. Harrington

The Endogenous Hormonal Milieu Associated with Thrombosis in Postmenopausal Women

Although the use of exogenous hormones, such as hormone therapy (HT) and oral contraceptives (OC), is known to be positively associated with the risk of venous thrombosis (VT), the endogenous hormonal environment associated with thrombotic biomarkers and endpoints is incompletely characterized. This dissertation aimed to elucidate further the relation between endogenous sex hormones as well as events associated with changes in these hormones (vasomotor symptoms [VMS] and hysterectomy and oophorectomy) and thrombotic risk in postmenopausal women.

In the setting of the Seattle Heart and Vascular Health (HVH) study, we evaluated the cross-sectional association between 9 measures of endogenous hormone levels and 8 hemostatic factor levels using separate multiple linear regression analyses among 131 postmenopausal women. Also in the setting of the HVH study, we used multiple logistic regression to estimate VT risk associated with prior hysterectomy/oophorectomy status and current HT use status, compared with no prior hysterectomy and no current HT use, among a population of postmenopausal cases of incident VT (n=1,358) and their matched controls (n=4,112). In the setting of the Women's Health Initiative (WHI), we evaluated the cross-sectional association of 9 measures of hemostatic factor levels with VMS presence and severity among 2,149 postmenopausal women using multiple linear regression. We also evaluated the association between VMS presence, severity, menopausal status at VMS onset, and VMS duration and the risk of VT among postmenopausal participants in the WHI-HT trials (n=24,508) and the WHI-Observational Study (WHI-OS) (n=87,783) using Cox proportional hazards models.

For our first aim, after accounting for multiple comparisons, we found no statistically significant evidence of any hormone-hemostatic factor associations (P-min p=0.10). In analyses without correction for multiple comparisons, there was some suggestion that higher levels of estradiol (pg/mL), estrone (pg/mL), sex hormone binding globulin (nmol/L), total testosterone (ng/dL), and dehydroepiandrosterone (DHEA) (ng/dL) may be associated with lower total protein S levels (%) and that higher estrone levels may be associated with lower antithrombin levels (%) (all p-values <=0.04). There was also some suggestion that higher levels of dehydroepiandrosterone-sulfate (ug/mL) and DHEA (ng/dL) may be associated with lower thrombin generation peak values (nM), lower TG endogenous thrombin potential (nMxMin), and lower nAPCsr levels, and that higher DHEA levels may be associated with lower Factor VII activity levels (%) (all p-values<=0.027). Addressing our second aim, prior hysterectomy with BSO without current HT use was associated with a 30% greater risk of VT than no prior hysterectomy and no current HT use (OR=1.3 [95% CI: 1.0-1.5]; p=0.02) but we found no evidence that prior hysterectomy with BSO with current HT use was associated with VT risk (OR=1.0 [95% CI: 0.78-1.3]). There was no evidence that prior hysterectomy-alone was associated with VT risk with current HT non-use (OR=1.0 [95% CI 0.80-1.3]) or with current HT use (OR=0.76 [95% CI 0.53-1.1]), compared with no prior hysterectomy and no current HT use. For our third aim, after correcting for multiple comparisons (p=0.064), we found no evidence that VMS presence or severity was associated with levels of hemostatic factors among postmenopausal women. There was some suggestion, however, that VMS presence may be associated with a -0.34 adjusted difference in nAPCsr (ratio) compared with no VMS (SE=0.13; p=0.009). For our final aim, we found no evidence of an association between VMS presence, severity, timing or duration and the risk of VT.

Results from this dissertation work suggest that endogenous and exogenous hormones differently impact thrombotic risk. Although the use of exogenous HT is positively associated with the risk of VT, we found little evidence that endogenous hormone levels were associated with hemostatic factors, we found evidence that hysterectomy with BSO and no current HT use (both associated with lower E2 levels) were associated with a slightly greater risk of incident VT than no hysterectomy and no current HT use, and although VMS are hypothesized to be hormonally-related, we found no evidence that VMS are associated with hemostatic factor levels or with the risk of incident VT.

URI: http://hdl.handle.net/1773/27471