Professor of Medicine Indiana University Medical Center Indianapolis, IN
Thomas F. Imperiale, MD1, Joanne K. Daggy, PhD2, Timothy E. Stump, MA2, Laura E. Myers, PhD3; 1Indiana University Medical Center, Indianapolis, IN; 2Indiana University School of Medicine, Indianapolis, IN; 3HSR&D, Roudebush VA Medical Center, Indianapolis, IN
Introduction: Knowing risk for EOCRC would help decide whether to screen persons < age 50. Several risk factors for EOCRC have been identified, but how to use them for decision making is unclear. We describe a process of going from risk factors to risk estimation and whether and how to screen. Methods: We used data from 450 cases of EOCRC among 35-49 year-old male Veterans and 1800 controls matched for site and year of case diagnosis. The top 15 risk factors based on lowest univariable p-values were selected for a multivariable logistic regression model. Model-selection with 10-fold cross-validation determined final model (Table 1). To make the model useful for decision-making, we calculated the relative risk (RR) estimate for 5 risk scenarios: one low-risk scenario (no factors present), 3 intermediate risk (some risk factors present), and 1 high-risk (all risk factors present), and applied these RR estimates to each of 3 SEER age groups (35-39, 40-44, & 45-49 years). For each scenario, we multiplied the baseline SEER CRC incidence rate x (1- Attributable Risk) x RR estimate to calculate a revised CRC incidence rate ("colon age"). When the upper confidence limit of the revised estimate was ≥ current SEER CRC incidence for 50-54 year olds, a recommendation for screening was made; when it was ≥ that for 70-74 year olds, screening with colonoscopy was recommended.
Results: Among 2250 subjects (mean [SD] age 44.3 [4.0]; 65.6% Caucasian), the 12 risk factor model demonstrated good fit (mean squared error = 0.14) and discrimination (c-statistic = 0.74). Table 2 shows 5 risk scenarios for the 3 age groups, revised incidence rates, comparison to the SEER 50-54 / 70-74 age groups, and whether / how to screen. Among 35-to-39-year olds, only the highest risk group (all factors present) would be screened (with colonoscopy). Among 40-44-year olds, some intermediate risk persons would be screened non-invasively, and high-risk persons with colonoscopy. Only low-risk 45-49 year olds could remain unscreened, with non-invasive screening for low-intermediate risk, and colonoscopy for the remaining 3 intermediate / high risk groups. (Table 2) Discussion: By using risk factors for EOCRC among male Veterans and SEER CRC incidence rates to estimate “colon age”, this analysis provides a framework for using these factors for decisions about whether and how to screen for EOCRC. For persons < age 50 whose colon age > 50 years, CRC screening may be recommended. Further study of this model and framework are required.
Disclosures: Thomas Imperiale indicated no relevant financial relationships. Joanne Daggy indicated no relevant financial relationships. Timothy Stump indicated no relevant financial relationships. Laura Myers indicated no relevant financial relationships.