Investigating causal relationships between sleep characteristics and risk of breast cancer: a Mendelian randomization study


Year:

Session type:

Theme:

Rebecca Richmond1,Emma Anderson1,Hassan Dashti2,Samuel Jones3,Jacqueline Lane2,Caroline Relton1,Marcus Munafò4,Debbie Lawlor1,Martin Rutter5,Richa Saxena2,Mike Weedon3,Richard Martin1,George Davey Smith1

1University of Bristol,2Massachusetts General Hospital, Boston,3University of Exeter,4University of Bristol, UK,5University of Manchester

Abstract

Background

Previous observational studies have identified associations between sleep characteristics and breast cancer. The carcinogenic effect of sleep patterns has also been investigated among night shift workers, with inconsistent findings. Here we conducted a Mendelian randomization (MR) analysis using genetic variants robustly associated with chronotype (morning/evening preference), insomnia and sleep duration to investigate causal links with breast cancer.

Method

Observational associations between sleep characteristics and breast cancer in UK Biobank (9,599 cases, 170,616 controls) were first assessed using logistic regression with adjustment for age, assessment centre, ancestry and several socio-economic, lifestyle and reproductive factors. For the MR analysis, breast cancer status was regressed against predicted values of sleep characteristics based on genotype in a logistic regression model, with adjustment for age, assessment centre, ancestry and genotyping chip. Two-sample MR was also conducted using summary data from a large genome-wide association study of breast cancer (122,977 cases, 105,974 controls) conducted by the Breast Cancer Association Consortium (BCAC).

Results

In observational analysis, morning preference was inversely associated with breast cancer risk (OR:0.89; 95%CI:0.85, 0.93), insomnia was positively associated (OR:1.21;1.14,1.29) and there was little evidence for an association with sleep duration (OR:1.01;0.99,1.03). Using 351 genetic variants associated with chronotype, 57 with insomnia and 91 with sleep duration, in one-sample MR analysis there was some evidence for a causal effect of morning preference (OR 0.77;0.56,1.05) and weaker evidence for insomnia (OR:1.11;0.70,1.76) and sleep duration (OR:1.14;0.91,1.43). Findings for a protective effect of morning preference (OR 0.93;0.89,0.96) and adverse effect of sleep duration (OR:1.20;1.01,1.41) were supported by two-sample MR using data from BCAC, while there was limited evidence for insomnia (OR:0.97;0.84,1.11).

Conclusion

Consistent findings for a protective effect of morning preference on breast cancer risk in both observational and Mendelian randomization analysis support hypotheses around carcinogenic light-at-night, while evidence for a causal effect of sleep disruption is more limited.