Evaluation of a combined model of Polygenic Risk Score and mismatch repair genes in the association of colorectal cancer for Norwegian cohort

Abstract

Background and aims: Recent studies have shown that combining polygenic risk score (PRS) and carrier status for germline pathogenic variants in colorectal cancer (CRC) susceptibility genes (e.g. MLH1, MSH2, MSH6, PMS2) may increase the success of predicting CRC. This study aims to examine the prediction performance of CRC in Norwegian data using the status of pathogenic variants in the mismatch repair (MMR) genes with the available PRS models in the literature. Methods: Our Norwegian cohort included 805 CRC cases, 86 of which carried a pathogenic variant in one of the MMR genes. As a control group, we included 8856 individuals without a cancer diagnosis, of which 179 were carriers for a pathogenic MMR variant. We first conducted a broad experiment to determine the best-performing PRS model for the Norwegian cohort. Afterwards, we established a combined analysis with the PRS model and the status of MMR genes. Results: Among 10 PRS models tested, the best-performing PRS model for the Norwegian cohort included 204 single nucleotide polymorphisms (SNPs) (AUC=0.604). We also observed that the combined model of PRS and the status of MMR significantly improved the prediction performance. Conclusion: The findings suggest that a combined model of a PRS and the status of MMR genes improves the prediction performance of CRC in Norwegian data.