Lung cancer is the most common and deadliest cancer in the world. Tobacco smoking is a leading risk factor of lung cancer. Female hormones, especially estrogen, have been suggested to be associated with the risk of lung cancer among women, compared to men, when adjusting for potential confounding factors, including smoking and environmental tobacco smoking. The effects of estrogen on lung cancer are believed to occur via estrogen receptors (ERs), which either regulate the cell proliferation or interact with lung cancer susceptibility genes. Hormonal factors (menstrual characteristics, reproductive history, and exogenous hormone use) represent the cumulative lifetime exposure to estrogen and have been studied in association with the risk of lung cancer. The current epidemiologic studies showed inconsistent results on the association between menstrual and reproductive factors and the risk of lung cancer. Population stratification, small sample size and limited power, residual confounding, and reverse causality could possibly explain the inconsistency. We conducted this study and tried to overcome the limitations of previous studies, in order to explore potential impact of hormonal factors on the development of lung cancer.
Aim 1 of this doctoral dissertation focused on the association between menstrual/reproductive hormonal factors and the occurrence of lung cancer in the Jiangsu Four Cancers Study. We also investigated the potential gene-environment interactions between parity and genetic susceptibilities (microRNA genes, stem cell regulation genes, NF-κB pathway genes and HIF-1α pathway genes). Multivariable unconditional logistic regression models were employed in the analyses while adjusting for age, income, education, county of residence, body mass index, smoking status, pack-years of smoking, and family history of lung cancer. Among 680 female lung cancer cases and 1,808 female controls, later menopause (at >54 vs <46 years old) was associated with the development of lung cancer (semi-Bayes adjusted odds ratio, sbOR=1.61, 95% CI=1.10-2.36). More pregnancies (2 or 3 vs 0 or 1) was inversely associated with lung cancer (sbOR=0.71, 95% CI=0.53, 0.95). Ever being a smoker and having two or fewer pregnancies in one’s lifetime could jointly increase the odds of lung cancer (relative excess risk due to interaction, RERI=1.71, 95% CI=0.03, 3.38). An increased number of ovulatory cycles was associated with increased probability of lung cancer (sbOR per 13 ovulatory cycles=1.02, 95% CI=1.00+, 1.04). Rs197412, a genetic variant in microRNA-related genes, showed significant different distributions between people with parity<=3 and people with parity>=4. Relative excess risk due to interaction (RERI) of rs197412 TT genotype and parity<=3 was -1.60(95% CI=-3.10, -0.10); ratio of odds ratio (ROR) was 0.34(95% CI=0.16, 0.72), indicating a sub-additive and sub-multiplicative effect modification by lower parity in the association between rs197412 (TT genotype) and the occurrence of lung cancer.
The aim 2 pooled analysis included a total of 2,456 Asian female lung cancer cases and 5,342 Asian female controls. Despite detected statistical heterogeneity, study-specific analysis showed no evident distinctions between different study designs. Therefore, a random effect of study site was integrated into the multivariable logistic regressions to allow for inter-study heterogeneities. Age, smoking status, comprehensive smoking index, and family history of lung cancer were adjusted for in the regression models. We found that late onset of menarche conferred elevated odds of lung cancer (adjusted OR= 1.16, 95% CI=1.01, 1.33 for 15-16 years old and adjusted OR=1.24, 95% CI=1.05, 1.45 for 17 years or older, compared with 14 years or younger). Late onset of menopause at 55 or older was associated with lung cancer with OR=1.24 and 95% CI=1.02, 1.51. Non-natural menopause was associated with an OR of 1.39 (95% CI=1.13, 1.71). More live births showed reversed association with lung cancer (OR of 3-4 live births=0.82, 95% CI= 0.72, 0.94, OR of 5 or more live births: 0.71 (95% CI:=0.60, 0.84), compared with 0-2 live births (Ptrend<0.001). A later first child delivery seemed associated with an increased susceptibility: OR of 21-25 years old= 1.23 (95% CI=1.06, 1.40), OR of 26 or older=1.27 (95% CI=1.06, 1.52, Ptrend=0.010). Oral contraceptives use appeared to be protective with an OR of 0.69 (95% CI=0.57, 0.83). Observed associations were stronger for adenocarcinoma than squamous cell carcinoma, and for published studies than unpublished ones. These relationships were not clearly modified by tobacco smoking status, probably because of lower prevalence of smoking among Asian women. This was a first and the largest pooling study of lung cancer among Asian women and the observed associations suggested potential roles of hormone-related pathways in the etiology of lung cancer.
In aim 3, in order to clarify the inconsistent results on the roles of age at menarche and age at menopause in the development of lung cancer, two-sample Mendelian randomization (MR) studies and polygenetic risk score-based analyses were carried out using published genome-wide association studies and the raw data from Female Lung Cancer Consortium in Asia. The results indicated no evident causal effect of ages at menarche (inverse-variance weighted OR per year increase=1.03, 95% CI=0.87, 1.21) and menopause (OR per year increase=1.02, 95% CI= 0.87, 1.21) on the occurrence of lung cancer, regardless of histological types. Polygenetic risk score-based analyses showed similar results.
This dissertation provided epidemiological evidence on the associations between hormonal factors and the susceptibility of lung cancer in Asian women. It confirmed the observed associations between later menarche and later menopause and increased probability of lung cancer, as well as the association between increased number of live birth and decreased probability of lung cancer, among Asian women. It also confirmed the associations between lung cancer and its susceptible genes (including microRNA genes, stem cell regulation genes, NF-κB pathway genes and HIF-1α pathway genes) in postmenopausal Asian women. In the MR analyses, ages at menarche and menopause were not showing a causal effect on lung cancer, as both ages are complicated representations of genetic and environmental factors. Parity seemed to modify the effects of lung cancer susceptible genes. These findings implied roles of hormonal factors in causal inference and risk prediction and could provide practical applications for risk population identification and personalized medicine for lung cancer.