Fecal immunochemical tests (FITs) are increasingly used as noninvasive screening tests in colorectal cancer-screening programs. Polygenic risk scores (PRS) are increasingly propagated for risk stratification in colorectal cancer screening. We aimed to assess the potential of combining FIT results and PRS to enhance diagnostic accuracy of detecting advanced neoplasia (AN) compared with using FIT results alone. Of 10,362 participants of screening colonoscopy in Southern Germany who conducted either one of two quantitative FITs, genotyping was done in all participants with AN (colorectal cancer or advanced adenoma) and a random subset of controls. Among 5,306 individuals, a PRS was calculated on the basis of the number of risk alleles in 140 SNPs. Partial areas under the receiver operating characteristics (ROC) curves (pAUCs) were computed for FIT and PRS alone and combined, focusing on a specificity range of 100%-80%. Both FITs showed similar performance characteristics with pAUCs of 0.661 (95% confidence interval (CI), 0.625-0.698; Ridascreen Hemoglobin) and 0.682 (95% CI, 0.661-0.701; FOB Gold) for AN detection. PRS alone reached a pAUC of 0.524 (95% CI, 0.499-0.550) and 0.530 (95% CI, 0.516-0.545), respectively, and its addition to FIT did not improve pAUCs (0.659; 95% CI, 0.622-0.697) and 0.667 (95% CI, 0.650-0.687), respectively. This finding was confirmed by investigating sensitivities at fixed specificities at 85%, 90%, and 95%. Partial AUCs also did not improve when adding the weighted PRS to FIT instead of the unweighted PRS. In summary, the combination with PRS did not improve diagnostic accuracy of FIT-based screening in a large asymptomatic colorectal cancer screening population from South-Western Germany.
Prevention relevance: In our study, combining polygenic risk score with fecal immunochemical test (FIT) did not improve diagnostic accuracy for advanced colorectal neoplasia detection compared with FIT alone. So far, such a combination cannot be recommended because it would come at extra costs and effort despite no relevant gain in neoplasia detection.
©2022 American Association for Cancer Research.