The Association Between Sleep Traits and the Risk of Lung Cancer: UK Biobank Cohort

Background: Lung cancer has the highest incidence of any cancer and is the leading cause of cancer-related death worldwide. As smoking rates have declined over the last few decades, the proportion of individuals diagnosed with lung cancer among those who have never smoked increases. Thus, it remains important to identify other factors involved in the etiology of the disease. Sleep traits have been hypothesized as potential modifiable risk factors for lung cancer. This dissertation aimed to 1) comprehensively examine the associations between sleep traits and lung cancer risk using traditional epidemiologic methods (e.g., Cox regression) and 2) to assess potential causal associations between sleep duration (per hour increase) and lung cancer risk, insomnia (per category increase) and lung cancer risk, and chronotype (per category increase) and lung cancer risk using Mendelian randomization (MR) analyses. Methods: Utilizing the United Kingdom Biobank Cohort I examined the association between sleep traits (sleep duration, chronotype, insomnia) and lung cancer risk. Cox Hazards regression was used to estimate Hazards Ratios (HRs) and 95% Confidence Intervals (CIs) of associations between these three sleep traits and lung cancer risk. Joint effects analyses were also conducted, and non-linearity of sleep duration and lung cancer risk associations was examined. MR analyses were conducted to estimate causal HRs for the associations between sleep duration, chronotype, insomnia, and lung cancer risk. Analyses were stratified by smoking status to examine associations unconfounded by smoking (i.e., among never smokers). Furthermore, analyses were stratified by biologic sex and smoking to examine potential effect measure modification of associations between sleep traits and lung cancer risk. Results: Results of this study suggested potential associations between sleep traits and lung cancer risk. In main effects analysis, long sleep, when compared to short sleep duration was positively associated with lung cancer risk. Usually experiencing insomnia symptoms, when compared to never/rarely experiencing insomnia symptoms, was positively associated with lung cancer risk. No associations between chronotype and sleep duration were evident in overall analysis. Evidence from both aims suggested positive associations between the presence of insomnia symptoms and lung cancer risk. However, among never smokers, no statistically significant associations were observed in either aim. Two-sample MR revealed minute positive associations between insomnia and lung cancer risk. Discussion: In this dissertation the association between insomnia and lung cancer risk may have been residually confounded by smoking status; among never smokers no evidence was found linking insomnia and lung cancer risk. In one-sample MR analysis, the strong positive association between insomnia and lung cancer risk may have resulted from violations of the independence assumption. In stratified one-sample MR, no association between insomnia and lung cancer was observed among the neversmoker strata. It remains unclear to what extent the observed association between insomnia and lung cancer risk in two-sample MR was impacted by smoking status. Future research should focus on examining associations between insomnia and lung cancer among a larger cohort of never smoking individuals. In conclusion, the associations observed between insomnia and lung cancer were likely impacted by smoking status, and future research is needed to tease apart the impact of smoking on lung cancer from that of insomnia on lung cancer.