(486.11) A novel histogram based on the SEER database predicts overall survival in patients with high-grade cerebellar glioma

Poster Board Number: A49

Hubin Duan (The First Hospital of Shanxi Medical University, The First Hospital of Shanxi Medical University, The First Hospital of Shanxi Medical University), Chunyan Hao (The First Hospital of Shanxi Medical University), Yu Zhang (The First Hospital of Shanxi Medical University), Binjun Qiao (The First Hospital of Shanxi Medical University), Zhijie Jing (Shanxi Medical University), Xianghui Chen (The Eleventh Clinical College of Shanxi Medical University), Juan Zheng (The Eleventh Clinical College of Shanxi Medical University), Xiaohong Qiao (The Eleventh Clinical College of Shanxi Medical University)

Objective High-grade cerebellar gliomas are clinically challenging to treat，and there is a relative lack of large sample statistics as a prognostic criterion. In order to compare changes in patient survival under different conditions, we screened the risk factors that affecting survival in patients with high-grade cerebellar gliomas, developed Nomograms and verified their accuracy.

Methods Patients with high-grade cerebellar glioma from 1975 to 2016 were retrieved from the Surveillance, Epidemiology, and End Results (SEER) database, excluding patients with no biopsy or surgery, unknown tissue type, grade, or survival data, then coded patient data for classification and processed, and a prediction group and a validation group were randomly assigned using the CARET procedure in R (4.0.2). We use univariate and multivariate Cox regression models to screen variables that were significantly related to the prognosis of patients with high-grade cerebellar glioma, and these variables were then used to construct Nomogram. Secondly, we use the consistency index (C-index), receiver operating characteristic curve (ROC), calibration chart and decision curve analysis (DCA) to analytically validate the accuracy of the Nomogram model. Finally, Kaplan-Meier survival analysis was used to test the prognostic indicators of Nomogram.

Results A total of 197 patients with high-grade cerebellar glioma were selected from the SEER database, and the patient data were categorized and coded into a prediction group (n = 100) and a validation group (n = 97). Univariate and multivariate Cox regression models were used to obtain four statistically significant indices of age, tumor type, type of surgery, and radiation therapy, which were then used to construct Nomogram with a C-index of 0.830 (95% CI: 0.765-0.895) for the prediction group and 0.789 (95% CI: 0.715-0.863) for the validation group.

Conclusion Nomograms have promising clinical applicability and more accurate predictive outcome. For patients with cerebellar glioma, radiotherapy and surgical resection are definitive in improving prognosis. Older patients tend to die earlier, and the pathological type is also a key factor in prognosis.

This study was supported by the National Natural Science Foundation Youth Fund (30600637), China Postdoctoral Science Foundation Grant , the Shanxi Province Basic Research Program , Scientific research project of Shanxi Provincial Health Commission，Research Project Supported by Shanxi Scholarship Council of China, Science and Technology Project for overseas Students of Shanxi Human Resources and Social Security Department，Lvliang Science and Technology Bureau Scientific Research Program，Foundation of the Eleventh Clinical Medical College of Shanxi Medical University.