A visual model for prognostic estimation in patients with primary diffuse large B-cell lymphoma of small intestine and colon: analysis of 1,613 cases from the SEER database.

Abstract

BackgroundTreatment modalities for primary diffuse large B-cell lymphoma of Small intestine and colon (PIC-DLBCL) have changed significantly during the past decades. However, limited information on the trends of clinical outcome of PIC-DLBCL patients has been reported, and the influence of marital status and medical insurance on prognosis is ignored.MethodsThis was a retrospective analysis the survival of PIC-DLBCL patients using the Surveillance, Epidemiology, and End Results (SEER) database between 2002 and 2016. The patients were divided into the training and validation cohort. In the training cohort, univariate and multivariable Cox regression analysis, Log-rank test and the Kaplan-Meier method were used to find out the independent prognostic factors, from which the visual prognostic model (nomogram and graphical web page) was established. C-index and calibration plots were used to evaluate the prediction accuracy of the model. In the validation cohort, both Decision curve analysis (DCA) and Receiver operating characteristic (ROC) curve was performed to compare the model with the International Prognostic Index (IPI) scoring model which is universally used to estimate prognosis of PIC-DLBCL.ResultsA total of 1,613 patients were collected, and the 5-year overall survival of all cases was 64.5%. Age at diagnosis (HR =2.58, 95% CI: 2.29–2.91), Ann Arbo stage (HR =1.34, 95% CI: 1.24–1.44), Divorced or Separated (HR=1.21, 95% CI: 1.06–1.38), Uninsured (HR =1.32, 95% CI: 1.19–1.45) and Primary colon (HR =1.23, 95% CI: 1.08–1.40) were associated with prognosis and were used to build up the visual model (nomogram and graphical web page). Both DCA and ROC curve showed that the model had better authentication capability than the IPI scoring model (AUC 0.820 vs. 0.714). The calibration plots showed that the model could accurately predict patient prognosis.ConclusionsThe visual model could output individual estimate prognosis simply and correctly, including marital status and medical insurance for the first time. Consideration of both medical and social factors, this study provided a new way to explore the improving prognosis of PIC-DLBCL.