Characterizing the Relationship between Expression Quantitative Trait Loci (eQTLs), DNA Methylation Quantitative Trait Loci (mQTLs), and Breast Cancer Risk Variants

Purpose: To assess the association of a polygenic risk score (PRS) for functional genetic variants with the risk of developing breast cancer. Methods: Summary data-based Mendelian randomization (SMR) and heterogeneity in dependent instruments (HEIDI) were used to identify breast cancer risk variants associated with gene expression and DNA methylation levels. A new SMR-based PRS was computed from the identified variants (functional PRS) and compared to an established 313-variant breast cancer PRS (GWAS PRS). The two scores were evaluated in 3560 breast cancer cases and 3383 non-cancer controls and also in a prospective study (n = 10,213) comprising 418 cases. Results: We identified 149 variants showing pleiotropic association with breast cancer risk (eQTL(HEIDI) > 0.05 = 9, mQTL(HEIDI) > 0.05 = 165). The discriminatory ability of the functional PRS (AUC(continuous) [95% CI]: 0.540 [0.526 to 0.553]) was found to be lower than that of the GWAS PRS (AUC(continuous) [95% CI]: 0.609 [0.596 to 0.622]). Even when utilizing 457 distinct variants from both the functional and GWAS PRS, the combined discriminatory performance remained below that of the GWAS PRS (AUC(continuous), combined [95% CI]: 0.561 [0.548 to 0.575]). A binary high/low-risk classification based on the 80th centile PRS in controls revealed a 6% increase in cases using the GWAS PRS compared to the functional PRS. The functional PRS identified an additional 12% of high-risk cases but also led to a 13% increase in high-risk classification among controls. Similar findings were observed in the SCHS prospective cohort, where the GWAS PRS outperformed the functional PRS, and the highest-performing PRS, a combined model, did not significantly improve over the GWAS PRS. Conclusions: While this study identified potentially functional variants associated with breast cancer risk, their inclusion did not substantially enhance the predictive accuracy of the GWAS PRS.