Validation of a Clinical Calculator Predicting Freedom From Colon Cancer Recurrence After Surgery on the Basis of Molecular and Clinical Variables

BACKGROUND: The Memorial Sloan Kettering clinical calculator for estimating the likelihood of freedom from colon cancer recurrence on the basis of clinical and molecular variables was developed at a time when testing for microsatellite instability was performed selectively, based on patient age, family history, and histologic features. Microsatellite stability was assumed if no testing was done. OBJECTIVE: This study aimed to validate the calculator in a cohort of patients who had all been tested for microsatellite instability. DESIGN: Retrospective cohort analysis. SETTINGS: Comprehensive cancer center. PATIENTS: This study included consecutive patients who underwent curative resection for stage I, II, or III colon cancer between 2017 and 2019. INTERVENTION: Universal testing of mircrosatellite phenotype in all cases. MAIN OUTCOME MEASURES: The calculator’s predictive accuracy was assessed using the concordance index and a calibration plot of predicted versus actual freedom from recurrence at 3 years after surgery. For a secondary sensitivity analysis, the presence of a tumor deposit(s) (disease category N1c) was considered equivalent to one positive lymph node (category N1a). RESULTS: With a median follow-up of 32 months among survivors, the concordance index for the 745 patients in the cohort was 0.748 (95% CI, 0.693–0.801), and a plot of predicted versus observed recurrences approached the 45° diagonal, indicating good discrimination and calibration. In the secondary sensitivity analysis for tumor deposits, the concordance index was 0.755 (95% CI, 0.700–0.806). LIMITATIONS: This study was limited by its retrospective, single-institution design. CONCLUSIONS: These results, based on inclusion of actual rather than imputed microsatellite stability status and presence of tumor deposits, confirm the predictive accuracy and reliability of the calculator. See Video Abstract.