Refining colorectal cancer screening strategies using polygenic risk scores and classical risk factors: A proof-of-concept study in the UK Biobank cohort.

104 Background: While the rising incidence of CRC in younger populations has led to a push for lowering the screening age to 40, current clinical guidelines lack sufficient consideration of individual risk variations, raising concerns about the potential harms of over-diagnosis and false positives. Our proof-of-concept study evaluates the utility of risk stratification for multi-target stool DNA (mt-sDNA) testing by integrating polygenic risk scores and traditional clinical risk factors, addressing gaps in current clinical CRC screening guidelines. Methods: Our study utilized the UK Biobank prospective cohort, consisting of 311,544 unrelated individuals of White British ancestry, aged 40-69, excluding those with any prior malignant cancer diagnosis. We developed a sex-specific Cox proportional hazards model to estimate the risk of developing CRC, with the primary outcome being the first incidence of CRC. Three models were assessed: a polygenic risk score (PRS)-only model, a classical risk factors (RF)-only model, and a combined (PRS + RF) model. The traditional risk factors considered were BMI, smoking status, pack years of smoking, and family history of bowel cancer in first-degree non-adoptive relatives. Model performance was evaluated using the area under the curve (AUC), hazard ratios (HR), and the positive and negative predictive values of mt-sDNA across various risk profiles. Results: The combined model for early CRC screening using mt-sDNA demonstrated superior risk stratification (Table) and improved PPV compared to the risk factors-only model, with enhanced performance even among younger individuals at lower baseline risk. If risk stratification was not used, the average PPV for the test in the general population (ages 40-80) would be 2.13% for males and 1.44% for females. In contrast, individuals in the top 1% of risk would have significantly higher PPVs: 5.62% for males and 3.51% for females. If we limit testing to only the high-risk group, per 100,000 individuals, approximately 3,490 additional cases for males and 2,070 additional cases for females would be detected compared to the average-risk population without stratification. Conclusions: The findings suggest that risk stratification may enhance the effectiveness of screening by identifying high-risk individuals who would derive the greatest benefit. In contrast, broad application of the test across the general population may not achieve an optimal risk-benefit ratio, given the relatively low PPV in average-risk individuals. Model performance of the PRS-only, RF-only, and PRS + RF (Combined) models for females and males, respectively. Hazard Ratio (95% CI) C-index (SE) 5-Year AUC Female PRS-Only 1.50 (1.37 - 1.64) 0.62 (0.01) 0.61 RF + PRS 1.49 (1.36 - 1.63) 0.62 (0.01) 0.61 RF Only 1.09 (0.99 - 1.20) 0.53 (0.02) 0.51 Male PRS-Only 1.49 (1.38 - 1.61) 0.63 (0.01) 0.69 RF + PRS 1.49 (1.38 - 1.61) 0.64 (0.01) 0.71 RF Only 1.23 (1.15 - 1.31) 0.57 (0.01) 0.57