Breast cancer risk factors and their effects on survival: a Mendelian randomisation study

Abstract

Background: Observational studies have investigated the association of risk factors with breast cancer prognosis. However, the results have been conflicting and it has been challenging to establish causality due to potential residual confounding. Using a Mendelian randomisation (MR) approach, we aimed to examine the potential causal association between breast cancer-specific survival and nine established risk factors for breast cancer: alcohol consumption, body mass index, height, physical activity, mammographic density, age at menarche or menopause, smoking, and type 2 diabetes mellitus (T2DM).

Methods: We conducted a two-sample MR analysis on data from the Breast Cancer Association Consortium (BCAC) and risk factor summary estimates from the GWAS Catalog. The BCAC data included 86,627 female patients of European ancestry with 7054 breast cancer-specific deaths during 15 years of follow-up. Of these, 59,378 were estrogen receptor (ER)-positive and 13,692 were ER-negative breast cancer patients. For the significant association, we used sensitivity analyses and a multivariable MR model. All risk factor associations were also examined in a model adjusted by other prognostic factors.

Results: Increased genetic liability to T2DM was significantly associated with worse breast cancer-specific survival (hazard ratio [HR] = 1.10, 95% confidence interval [CI] = 1.03-1.17, P value [P] = 0.003). There were no significant associations after multiple testing correction for any of the risk factors in the ER-status subtypes. For the reported significant association with T2DM, the sensitivity analyses did not show evidence for violation of the MR assumptions nor that the association was due to increased BMI. The association remained significant when adjusting by other prognostic factors.

Conclusions: This extensive MR analysis suggests that T2DM may be causally associated with worse breast cancer-specific survival and therefore that treating T2DM may improve prognosis.

Keywords: Breast cancer risk factors; Breast cancer survival; GWAS Catalog; Lifestyle; Mendelian randomisation.

Fig. 1

Effect of the nine breast cancer risk factors on breast cancer-specific survival in all breast cancers. The y-axis shows the −log₁₀(P value) effect for the association. The x-axis corresponds to log (hazard ratio) effect for each of the traits on breast cancer survival. The risk factors with false discovery rate (FDR) < 0.05 are coloured in red; the size of the circle is proportional to the −log₁₀(FDR)

Fig. 2

Plot showing the effect sizes of the SNP effects on breast cancer-specific survival for all breast cancers (y-axes) and the SNP effects on T2DM (x-axes) with 95% confidence intervals. Each dot represents one of the 95 SNPs used in the T2DM genetic instrument. The slopes indicate the estimate for each of the five different MR tests