Abstract
IntroductionRectal cancer, characterized by a high global incidence rate, is a significant malignancy. This study aimed to develop a competing risk model utilizing the SEER database to estimate the risk of cancer-specific mortality in patients diagnosed with locally advanced rectal cancer (LARC).MethodsThis research employed a retrospective cohort design, selecting patients diagnosed with LARC (stages III-IV) from the SEER database spanning 2010 to 2015. An external validation cohort comprising 203 patients was identified from Tianjin People's Hospital. A competing risk model was developed to estimate the likelihood of cancer-specific mortality among the patients. The predictive accuracy was assessed using the area under the receiver operating characteristic (ROC) curve and the C-statistic at various time points, while calibration was evaluated using a calibration curve.ResultsA total of 7608 patients with LARC were analyzed, and they were randomly allocated to either the training cohort (N = 5700) or the validation cohort (N = 1908). A competing risk model was developed to predict cancer-specific survival in these patients. Increased risk of non-cancer death was observed among older patients, black patients, and single patients. In the training cohort, the model demonstrated area under the curve (AUC) values of 0.795, 0.78, and 0.783 for 1-year, 3-year, and 5-year predictions, respectively; in the validation cohort, the AUC values were 0.791, 0.792, and 0.779 for the same time points, reflecting good discriminative ability. The C-statistic was 0.775 for the training set and 0.794 for the validation set, confirming the model's strong predictive performance for cause-specific mortality.ConclusionWith high accuracy and reliability, the model aids physicians and patients in formulating clinical decisions and follow-up strategies. Furthermore, our findings revealed that the log odds of positive lymph nodes (LODDS) was inadequate as a standalone predictor of LARC-specific mortality.