0824 Using Mendelian Randomization To Understand How Chronotype Influences Breast Cancer Risk

Abstract

Introduction: Circadian rhythms have been implicated in breast cancer development. Individuals with a morning chronotype have been shown to have reduced risk of breast cancer (OR 0.88, 95% CI 0.82, 0.93 per category increase) using genetic data in Mendelian randomization (MR), which minimizes risk of reverse causation and confounding. However, univariable MR may yield biased estimates if the genetic variants are not specific (e.g. if they have a direct influence on other lifestyle factors) and additional MR techniques have been developed to evaluate this. We aimed to establish the extent to which the observed causal effect of chronotype might be explained by other lifestyle factors associated with breast cancer (body mass index (BMI), alcoholic drinks per week (DPW) and age at first birth (AFB)) using bidirectional and multivariable MR. Methods: We investigated whether chronotype and other lifestyle factors (BMI, DPW and AFB) are causally related by performing bidirectional MR in a two‐sample framework, using genomewide association study (GWAS) summary statistics for chronotype (UKBiobank, n= 449,734), BMI (GIANT, n=339,205), DPW (GSCAN, n=226,223) and AFB (SSGAC, n=251,151). We next estimated the effect of chronotype on breast cancer controlling for BMI, DPW and AFB, using GWAS summary statistics for breast cancer from BCAC (n=228,951) in a multivariable MR approach. Results: Bidirectional MR analysis revealed evidence for a causal effect of increased BMI on morning chronotype (beta=0.09 category increase per SD (95% CI=0.01,0.17; p=0.025)) and of increased DPW on evening chronotype (beta=0.37 category increase per SD (95% CI=0.10,0.64; p=0.006)). In multivariable MR, there was robust evidence for an independent, causal effect of chronotype on breast cancer after controlling for BMI, DPW and AFB, although the magnitude of the causal effect was 13% lower. Conclusion: The protective effect of morning chronotype on breast cancer observed in univariable MR analysis can in part be explained by other lifestyle factors, including adiposity and alcohol. However, evidence of an independent effect of chronotype on breast cancer support hypotheses around a direct effect of circadian timing on mammary oncogenesis which requires further investigation.