BACKGROUND: Colorectal cancer (CRC) screening guidelines generally recommend healthy lifestyle choices for cancer prevention. However, studies have shown inconsistent associations between various risk factors and advanced neoplasia (AN) development. AIM: To identify potentially modifiable baseline dietary and lifestyle risk factors associated with AN among an asymptomatic Veteran population, while accounting for prior colonoscopic findings and varying surveillance intensity.
METHODS: We used data from a prospective colonoscopy screening study collected by the VA Cooperative Studies Program. From 1994 to 1997, 3,121 asymptomatic Veterans aged 50-75 received a baseline colonoscopy screening, at which time they selfreported dietary and lifestyle information. Veterans were subsequently assigned to colonoscopy surveillance regimens and followed for 10 years. AN was defined as invasive CRC or any adenoma ≥1 cm, or with villous histology, or high-grade dysplasia. To detect associations with AN diagnosis, we utilized a longitudinal joint model with two sub-models. A multivariate logistic regression modeled the longitudinal probability of AN, while a time-to-event process adjusted for survival. Here we focus on the multivariate logistic regression, representing associations of dietary and lifestyle risk factors with the odds of being diagnosed with AN.
RESULTS: Of the 3,121 Veterans, 1,915 received at least one colonoscopy following baseline screening. Among the 1,915, we detected a significant positive association with AN for current daily smokers (odds ratio (OR) 1.43, 95% CI: 1.02-2.01) compared to those with prior or no history. We found a protective effect for each 100 IU of dietary vitamin D consumed (OR 0.95, 95% CI: 0.95-0.99). We did not detect any significant associations with BMI, red meat consumption, or physical activity. We found that African American race had a lower odds of AN compared to Caucasian race (OR 0.57, 95% CI: 0.32-0.97).
CONCLUSIONS: We identified smoking status and vitamin D consumption as potentially modifiable baseline risk factors associated with AN development. While these results suggest possible points of intervention and targeted screening, more evidence is required across more diverse populations. Future efforts should focus on understanding changes in such risk factors on associations with AN for patients over time. Finally, racial differences in AN incidence merit further investigation.