Constructing and Validating a Prognosis Predictive Nomogram for Cancer-Specific Survival in Rectal Cancer Patients Receiving Preoperative Radiotherapy

Abstract

Background A predictive tool is required to identify the cancer-specific survival in rectal cancer (RC) patients who have opted to receive preoperative radiotherapy. Methods A database containing the data on RC patients’ records of Surveillance, Epidemiology, and End Results (SEER) receiving surgery during 2000–2014 was selected. All patients received neoadjuvant radiotherapy (NR). The correlation of clinicopathological parameters was analyzed using the Chi-square test and the survival risk factors were analyzed using the Cox proportional hazards analysis (univariate and multivariate). Finally, the nomogram was developed and validated to visually represent an accurate prediction of the probability of 3- and 5-year cancer-specific survival (CSS) based on the screened variables of the cohort. Results 11,499 rectal cancer patients were included in our cohort. Patients’ records were randomly allocated to either the development or validation cohorts based on an equal ratio (1:1). Performing the multivariate Cox regression analysis incorporating these variables in the development cohort determined 11 independent prognostic factors. Statistically significant differences were recorded among subgroups using log-rank tests, which confirmed the appropriateness and acceptability of factor stratifications. Then, the nomogram was constructed and its concordance index (C-index) values in the development cohort (0.720) and validation cohort (0.717) were evaluated to be higher (P<0.05) than those of the AJCC stage (0.631 and 0.633 respectively). Also, the 3-year AUC values of this nomogram were higher than those of the AJCC stage in both the development cohort (0.746 vs. 0.631) and the validation cohort (0.745 vs. 0.640). Using DCA curves, the predictive potential of the currently developed nomogram outperformed the conventional AJCC staging system. Conclusion The nomogram model might be a more reliable tool to predict prognosis accurately in rectal cancer patients receiving preoperative radiotherapy.