Background
Predicting prognosis in refractory metastatic colorectal cancer (mCRC) patients is needed to guide decision making. The Colon Life nomogram was developed to predict 12-week mortality in refractory mCRC patients. The
Conclusion
We demonstrated evident miscalibration of the Colon Life nomogram upon external validation, which hampers its use in clinical practice. We recommend conducting a study with a sufficiently large sample size to update the Colon Life nomogram or to develop a new model including QoL.
Methods
The validation cohort consists of 150 QUALITAS study patients, an observational substudy of the Prospective Dutch CRC cohort, who were treated with FTD/TPI between 2016 and 2019. Model performance was assessed on discrimination, calibration, and clinical usefulness. The additional prognostic value of baseline quality of life (QoL) and thymidine kinase (TK1) expression in tissue was explored.
Results
Of the 150 patients, 25 (16.7%) died within 12 weeks of starting FTD/TPI treatment. The C-statistic was 0.63 (95% C.I. 0.56-0.70). The observed/expected ratio was 0.52 (0.37-0.73). The calibration intercept and slope were -1.06 (-1.53 to -0.58) and 0.41 (0.01-0.81), respectively, which indicated overestimation of 12-week mortality by the nomogram. Decision curve analysis showed the nomogram did not yield a positive net benefit at clinically meaningful thresholds for predicted 12-week mortality. Addition of QoL to the nomogram improved the C-statistic to 0.85 (0.81-0.89). TK1 expression was associated with progression-free survival but not with overall survival.