Affiliation: Gillings School of Global Public Health, Department of Epidemiology
Purpose: Smoking is a suspected risk factor for breast cancer, with hypothesized links to estrogen-mediated, genotoxic, and growth-factor dependent mechanisms. Each mechanism can be modeled by overexpression of ER, p53, and EGFR, respectively. This dissertation examines associations between smoking and biomarkers for each mechanistic pathway. Methods: Our population-based study included 1,970 women diagnosed with invasive breast cancer in central and eastern North Carolina. Single and multigene biomarker outcomes were characterized as binary (+/-) or continuous measures for protein or mRNA. Single gene measures included ER/ESR1, p53/MDM2, and EGFR. Multigene mRNA signatures included a luminal score (LS); a p53 signature used to describe wild-type (Wt) or mutant (Mut) activity; and an algorithm-based proliferation score (PS). We used logistic and linear regression models to estimate associations between smoking and biomarker outcomes. Results: (Aim 1) When compared with never smokers, the odds of ER+, ESR1+, and LS+ tumors were nearly doubled among current smokers, those who smoked 20 or more years, and those who smoked within 5 years of diagnosis. Quantitative levels of ESR1 mRNA were highest among current smokers compared to never smokers overall and among women with ER+ breast cancer; however, we did not observe associations between smoking and continuous ER protein expression. (Aim 2) ER- cases with a history of ever smoking were at increased odds of having breast tumors with the p53 IHC+ molecular phenotype. In addition, long smoking duration was also associated with higher quantitative levels of p53 protein among ER- breast tumors but not ER+ breast tumors. The EGFR IHC+ phenotype was inconsistently linked to smoking for both ER+ and ER- tumors. With respect to our multigene mRNA signatures, smoking was not linked to either the p53 Wt or p53 Mut subtype; however, with respect to the proliferation score, smoking metrics were consistently linked to lower odds for the PS+ subtype. Conclusions: Both single and multigene measures for each mechanistic pathway captured tumor changes associated with smoking. Findings from our study have implications for understanding potential mechanisms underlying smoking and breast cancer risk.