Recalibrating the Genetics and Epidemiology of Colorectal Cancer Consortium (GECCO) Environmental Risk Score for Use in US Veterans

Abstract

Background: Risk for colorectal cancer (CRC) may accumulate through multiple environmental factors. Understanding their effects, along with genetics, age and family history, could allow improvements in clinical decisions for screening protocols. We aimed to extend previous work by recalibrating an environmental risk score (e-Score) for CRC among a sample of US Veteran participants of the Million Veteran Program (MVP). Methods: Demographic, lifestyle, and CRC data from 2011-2022 were abstracted from survey responses and health records of 227,504 male MVP participants. Weighting for each environmental factor’s effect size was recalculated using VA training data to create a recalibrated e-Score. This recalibrated score was compared with the original weighted e-Score in a validation sample of 113,752 (n cases=590). Nested multiple logistic regression models tested associations between quintiles for recalibrated and original e-Scores. Likelihood Ratio Tests were used to compare model performance. Results: Age (p<0.0001), education (p<0.0001), diabetes (p<0.0001), physical activity (p<0.0001), smoking (p<0.0001), NSAID use (p<0.0001), calcium (p=0.015), folate (p=0.020), and fruit consumption (p=0.019) were significantly different between CRC case and control groups. In the validation sample, the recalibrated e-Score model significantly improved the base model performance (p<0.001), but the original e-Score model did not (p=0.07). The recalibrated e-Score model quintile 5 was associated with significantly higher odds for CRC compared with quintile 1 (Q5 vs Q1: 1.79, 95% CI: 1.38-2.33). Conclusions: Multiple environmental factors, and the recalibrated e-Score quintiles were significantly associated with CRC cases. Impact: A recalibrated, Veteran-specific e-Score could be used to help personalize CRC screening and prevention strategies.