Objectives Colorectal cancer (CRC) is one of the most common cancers worldwide. Stage II/III patients undergo curative-intent surgery yet still face the recurrence risk. Detecting recurrences early provides the best opportunity for optimal treatment. We aimed to develop a microsimulation model to evaluate CRC management-associated outcomes based on current guidelines, including the performance of guideline-recommended surveillance in detecting recurrences. Design Two separate individual-level state transition (microsimulation) models for colon and rectal cancer were built with a lifetime horizon using monthly cycles. The models integrated treatment and surveillance strategies per current guidelines. Setting The currently recommended surveillance modalities by National Comprehensive Cancer Network guidelines for surveilling patients with CRC after curative-intent surgery. Participants 65-year-old patients with stage II and stage III CRC who underwent curative-intent surgery in the USA. Outcome measures Cumulative recurrences, detected recurrences, detection rate, overall survival and recurrence-free survival in a 5-year horizon, as well as average life expectancy, were the outcome measures used. Results Over 5 years, disease recurrence was observed in 9.5% of patients with stage II–III colon cancer and in 38.0% of patients with stage II–III rectal cancer. Of these, 82.5% and 85.5% were detected via surveillance, respectively, within 5 years. The predicted 5-year overall survival was 86.0% for colon cancer and 69.3% for rectal cancer, with corresponding recurrence-free survival rates of 78.9% and 53.8%. Based on current guidelines-recommended surveillance, detecting one colon cancer recurrence requires 148 carcinoembryonic antigen (CEA) tests, 37 CT scans and 21 colonoscopies. In contrast, detecting one rectal cancer recurrence requires 31 CEA tests, 8 CT scans and 4 colonoscopies. Conclusions Our validated model suggests that relative to an optimal benchmark in which all recurrences are detected, recurrence detection under current guidelines may be suboptimal, indicating room for improvement. As new tests emerge, this model could be a valuable tool for evaluating existing clinical practices and the potential of new tests to enhance patient outcomes.
Samur et al. (Sun,) studied this question.