Individualized prediction of conditional survival for colorectal signet-ring cell carcinoma patients.

ABSTRACT

Background: Conditional survival (CS) considers the time already survived after surgery and may provide additional survival information. The authors sought to construct and validate novel conditional survival nomograms for the prediction of conditional overall survival (OS) and cancer-specific survival (CSS) of colorectal signet-ring cell carcinoma (SRCC) patients. Methods: Patients diagnosed with stage I-III SRCC between 2010 and 2019 were identified from the Surveillance, Epidemiology, and End Results database. The formula calculating CS was CS(x|y) = S(x+y)/S(x), where S(x) represents the survival at x years. CS nomograms were then constructed to predict the 5-year conditional OS and CSS, followed by internal validation. Results: A total of 944 colorectal SRCC patients were finally identified in this study. The 5-year OS and CSS improved gradually with additional survival time. Univariate and multivariate Cox regression analysis conducted in training set revealed that age, race, T stage, LNR, and perineural invasion were independent risk factors for both OS and CSS. Two nomograms with considerable predictive ability were successfully constructed [area under the curve (AUC) for OS 0.788; AUC for CSS 0.847] and validated (AUC for OS 0.773; AUC for CSS 0.799) for the prediction of 5-year OS and CSS, based on the duration of 1-4 years post-surgery survival. Conclusions: The probability of achieving 5-year OS and 5-year CSS in colorectal SRCC patients improved gradually with additional time. Conditional nomograms considering survival time will be more reliable and informative for risk stratification and postoperative follow-up.