Background: Although plasma estrogen and androgen levels are significantly positively associated with risk of postmenopausal breast cancer, it is unknown how long a single blood hormone measure can predict subsequent breast cancer risk as most prior studies have included less than 10 years of follow-up. If including hormone levels into current breast cancer risk prediction models improves discriminative ability, evaluating how the hormone/breast cancer association varies by years since blood collection can help determine the most relevant time period for hormone measures. In addition, whether the associations vary by tumor hormone receptor status is unclear.
Methods: We conducted a prospective nested case-control analysis within the Nurses' Health Study cohort. Blood samples were collected in 1989-1990 and then in 2000-2002. Among eligible postmenopausal women not using postmenopausal hormones (PMH) at blood collection, 796 cases were diagnosed through June, 2010. Two control subjects (n=1,583) were matched to cases on age and time of blood collection. We used unconditional logistic regression analyses to estimate the relative risks (RRs, 95%CIs) by controlling for breast cancer risk factors.
Results: The intra-class correlation coefficients for the two blood measures collected 10 years apart were 0.69 for estradiol, 0.71 for testosterone, 0.54 for dehydroepiandrosterone sulfate (DHEAS), and 0.74 for sex hormone-binding globulin (SHBG). Overall, women in the top 25% of levels of estradiol, free estradiol, testosterone, free testosterone, and DHEAS (compared to the lowest 25% of levels) were at a 50-107% higher risk of breast cancer (all p-trends<0.004). Relative risks were similar when comparing cases diagnosed 1-10 vs. 11-20 years (and 16-20 years) after blood collection (all p-values for interaction by follow-up period>0.15). For example, comparing the highest with lowest (reference) quartile of hormones measured 10 year or less before breast cancer diagnosis, the multivariable RRs (95%CIs) for overall breast cancer were 2.0 (1.4-2.7; p-trend<0.001) for estradiol, 1.4 (1.1-1.9; p-trend=0.001) for testosterone, and 0.63 (0.47-0.85; p-trend=0.006) for SHBG. The comparable RRs (95%CIs) for blood samples collected 11-20 years before diagnosis were 2.0 (1.3-3.1; p-trend=0.002) for estradiol, 1.6 (1.0-2.4; p-trend=0.11) for testosterone, and 0.68 (0.44-1.04; p-trend=0.04) for SHBG. Importantly, with the exception of DHEAS, the associations varied significantly by hormone receptor status of the tumor. For example, the multivariable RRs (95%CIs) comparing the highest vs. lowest quartile were 2.8 (2.0-4.0; p-trend<0.001) for ER+/PR+ tumors (n=347) vs. 1.1 (0.6-2.1; p-trend=0.98) for ER-/PR- tumors (n=80) for estradiol (p-heterogeneity<0.01), 1.8 (1.3-2.5; p-trend<0.001) vs. 0.6 (0.3-1.2; p-trend=0.35) for testosterone (p-heterogeneity<0.01), and 0.46 (0.31-0.67; p-trend=0.01) vs. 1.07 (0.56-2.06; p-trend=0.67) for SHBG (p-heterogeneity=0.01).
Conclusions: A single blood sex hormone measurement in postmenopausal women can predict risk of hormone receptor positive breast cancer for up to 20 years.
Citation Format: Xuehong Zhang, Shelley S. Tworoger, A. Heather Eliassen, Susan E. Hankinson. Postmenopausal plasma sex hormone levels and subsequent risk of breast cancer over 20 years of follow up. [abstract]. In: Proceedings of the Eleventh Annual AACR International Conference on Frontiers in Cancer Prevention Research; 2012 Oct 16-19; Anaheim, CA. Philadelphia (PA): AACR; Cancer Prev Res 2012;5(11 Suppl):Abstract nr B105.