A Nomogram Model for Predicting Prognosis of Patients with Medulloblastoma.

Abstract

To identify the prognostic factors associated with cancer-specific survival in medulloblastoma (MB), and to use them for establishing a nomogram model to predict cancer-specific survival. In total, 268 patients with MB were included; they were rigorously respectively screened from the Surveillance, Epidemiology, and End Results database from 1988 to 2015 and statistically analyzed in R language. This study focused on cancer-specific death and used the cox regression analysis for variable filtering. The model was calibrated using C-index, area under the curve (AUC), and calibration curve. As per our findings, it was determined that extension (localized: hazard ratio [HR]=0.5899, p=0.00963; further extension: indicator) and treatment modality (radiation after surgery chemotherapy sequence unknown: HR=0.3646, p=0.00192; no surgery: indicator) were statistically significant in the prognosis of MB and were finally utilized to construct a nomogram model for predicting the condition. The AUC values were 0.649, 0.629, and 0.64 at 2, 3, and 5 years, respectively. Tumor extension and treatment modality were independent prognostic factors for MB.