Personalizing age‐specific survival prediction and risk stratification in intracranial grade II/III ependymoma

Background Models for estimation of survival rates of patients with intracranial grade II/III ependymoma (EPN) are scarce. Considering the heterogeneity in prognostic factors between pediatric and adult patients, we aimed to develop age‐specific nomograms for predicting 3‐, 5‐, and 8‐year survival for these patients. Methods A total of 1390 cases (667 children; 723 adults) of intracranial grade II/III EPNs diagnosed between 1988 and 2015 were extracted from the Surveillance, Epidemiology, and End Results (SEER) database for our study. Univariable and multivariable Cox analyses were employed to identify independent prognostic predictors. Age‐specific nomograms were developed based on the results of multivariate Cox analyses. We also evaluated the performance of these predictive models by concordance index, calibration curves, time‐dependent receiver operating characteristic curves, and decision curve analyses. Results Considerable heterogeneity in prognostic factors was highlighted between pediatric and adult patients. Age, sex, tumor grade, surgery treatment and radiotherapy were identified as significant predictors of overall survival for children, and age, tumor grade, tumor size, surgery treatment, and marital status for adult. Based on these factors, age‐specific nomogram models were established and internally validated. These models exhibited favorable discrimination and calibration characteristics. Nomogram‐based risk classification systems were also constructed to facilitate risk stratification in EPNs for optimization of clinical management. Conclusions We developed the first nomograms and corresponding risk classification systems for predicting survival in patients with intracranial grade II/III EPN. These easily used tools can assist oncologists in making accurate survival evaluation. Considerable heterogeneity in prognostic factors was highlighted between pediatric and adult patients with intracranial grade II/III ependymoma in our study. And the present study is, to our knowledge, the first study to develop nomograms for estimation of survival in ependymomas. We have demonstrated excellent discrimination and calibration of these predictive models. Besides, our nomogram models showed superiority in the clinical utility as well. Moreover, the novel proposed nomogram‐based risk classification systems can also help neurosurgeons to classify risk more objectively.