Abstract
BackgroundObesity is a well-established risk factor for many cancers, but its association with breast cancer remains inconsistent. Obesity is a highly complex and polygenic condition, driven by multiple biological pathways, some of which have been linked to beneficial health effects, defining a metabolically healthy obese phenotype. We hypothesise that this heterogeneity may partly underlie the variable association with breast cancer, where different biological pathways may play divergent roles in disease risk.MethodsWe applied an approach in two parts to explore the heterogeneous effects of high body mass index (BMI) on breast cancer risk. First, we applied Mendelian randomisation (MR)-based clustering on genetic data (FinnGen, GIANT, BCAC) to identify clusters of BMI-increasing SNPs with differential effects on breast cancer, and the cluster-specific MR results were replicated in the UK Biobank. Second, we integrated proteomic data to estimate the cluster-specific effects of BMI on protein levels and the effect of these proteins on breast cancer risk, using two-sample MR. Mediation analyses were then carried out using the product-of-coefficient method.ResultsWe identified three clusters of SNPs associated with increased BMI while having differential effects on breast cancer risk: one risk-increasing (inverse variance weighted MR estimate: 1.27 [1.10, 1.44]), one medium protective (− 0.75 [− 0.82, − 0.68]), and one highly protective (− 1.92 [− 2.21, − 1.62]), comprising 24, 73, and 7 SNPs respectively, with similar magnitude of effects in the UK Biobank. In the two-sample MR, each cluster was associated with many proteins (70, 369 and 44 respectively, FDR q-value < 0.05), and the mediation analysis identified that three proteins significantly mediate cluster-specific effects on breast cancer risk. Among these proteins, MET is known to be involved in breast cancer, while little is known about the roles of CPM and CST6.ConclusionsThese findings suggest that the effect of obesity on breast cancer is mediated through multiple distinct biological pathways. The identified proteins provide a first insight into the mechanisms underlying these pathways. With further investigation, these results could provide a basis for developing personalised treatment strategies.