Background: Lung cancer exhibits unique patterns among women including high adenocarcinoma rates among nonsmokers. Inconsistent findings about hormonal factors on risk may reflect incomplete control for confounding, misclassification of exposures, or insufficient attention to variation by histology.
Methods: Among 185,017 women, ages 50 to 71 years, recruited during 1995 and 1996 for the NIH-AARP (American Association of Retired Persons) Diet and Health Study, we identified 3,512 incident lung cancers (including 276 in never smokers) in follow-up through December 2006. Multivariable Cox proportional hazards models estimated relative risks (RR) and 95% CIs for self-reported hormonally related risk factors.
Results: After adjustment for smoking and other confounders, subjects with late menarche were at reduced risk, with the association specific for adenocarcinomas (RR = 0.72 for menarche 15+ vs. <11, P(trend) < 0.01). Subjects with early ages at ovarian cessation (either from natural menopause or bilateral oophorectomy) were at an increased risk for adenocarcinomas and squamous cell tumors, but the associations were strongest for smokers, suggesting either residual confounding or an enhanced effect of menopausally related factors among subjects with decreased endogenous estrogens. In contrast, we saw no relationships of risk with either parity, age at first birth, or exogenous hormone use.
Conclusions: Elevated levels of hormones may adversely affect lung function early in life while assisting with cellular and immunologic responses later in life. Additional attention toward the role of hormonal factors may further our understanding of lung carcinogenesis.
Impact: Our findings provide some support for a role of hormonal factors in the etiology of lung cancer, although the mechanisms appear complicated.