Abstract Background Microsimulation models use empirical evidence about cancer epidemiology and screening test performance to predict the long-term effectiveness of screening regimens and are essential for developing cancer screening guidelines. Colorectal cancer (CRC) provides a clear example. CRC arises through two pathways, the adenoma-carcinoma pathway and the serrated pathway. Sessile serrated lesions (SSLs) are the primary serrated precursor lesion. SSLs are more difficult to detect and remove than adenomas. Methods We describe version 3.0 of the Colorectal Cancer Simulated Population model for Incidence and Natural history (CRC-SPIN), which adds new information about the serrated pathway and CRC risk in adults under 50, then estimate the effectiveness of decennial colonoscopy from 45 to 75 years old. The model was calibrated using a Bayesian approach to estimate 95% credible intervals (CIs) that reflect uncertainty in predictions. Results The model validated well to studies of the effect of one-time screening and outcomes from surveillance colonoscopy. In the absence of screening, SSLs accounted for 10.6% (95% CI: 3.3-21.6) of CRC, increasing to 23.5% (95% CI: 7.7%-46.0%) with screening due to selective removal of adenomas. Screening was predicted to prevent 93.9% (95% CI: 92.0%-94.3%) of CRC and 95.3% (95% CI: 93.8%-96.5%) of CRC mortality. Conclusions Although SSLs are less common than adenomas, they likely make up a large fraction of CRC that arises in people who participate in screening. This points to the importance of improving the ability to detect SSLs, especially large SSLs, at colonoscopy.
Rutter et al. (Tue,) studied this question.