OBJECTIVES: Colon cancer (CC) constitutes approximately 6% of global cancer incidence. Post-curative surgery, current guidelines recommend adjuvant chemotherapy for high-risk patients, as defined by clinicopathologic factors, followed by routine surveillance and monitoring. However, some patients face ongoing risk of recurrence. We developed a novel comprehensive microsimulation model for this population within the framework of current guidelines, providing a nuanced understanding of CC management.

METHODS: A microsimulation model, replicating disease progression and guideline recommendations, was developed using a lifetime horizon with monthly cycles. The starting age of the cohort was 65 years. Treatment decisions post-curative-surgery, post-recurrence, and surveillance algorithms were incorporated into the model following the current guidelines. Disease progression parameters and performance characteristics of the recommended testing modalities were derived from the published literature. The model was validated using the National Cancer Database records following ISPOR validation guidelines.

RESULTS: Our model predicted a 10-year recurrence-free survival of 71.5% for high-risk Stage II and 62.8% for Stage III patients. The 10-year overall survival rate was 73.7% for high-risk Stage II and 68% for Stage III survivors, with post-surgery average life expectancy of 17.1 years and 16.0 years, respectively. The model predicted 810 recurrences per 10,000 high-risk Stage II survivors (250 locoregional and 560 distant) and, 1,820 recurrences for Stage III survivors (570 locoregional and 1,250 distant) over 10 years. However, only 510 (63%) and 1,370 (75%) of recurrences were detected by the current surveillance modalities.

CONCLUSIONS: Our clinically validated model enables the evaluation of long-term clinical outcomes in high-risk Stage II and Stage III CC survivors. Model predictions highlighted suboptimal recurrence detection rates under the current screening modalities. Given the newer emerging molecular residual disease tests, this model could serve as a foundational tool for evaluating both current clinical practices and the potential of these new tests to improve early detection.