Abstract 1108: The impact of adjuvant radiotherapy (ART) and survival prediction by machine learning in intracranial hemangiopericytoma (HPC)

Abstract

Abstract Background: Intracranial HPC is a rare brain tumor that accounts for less than 1% of all primary central nervous system malignancies. Surgery is the primary treatment, and while radiation may reduce the risk of local recurrence, its overall effectiveness remains uncertain. Therefore, it is crucial to develop reliable predictions for personalized care and improved management. In this study, we used machine learning (ML) techniques to predict the survival of patients with intracranial HPC. Methods: Data were obtained from the Surveillance, Epidemiology, and End Results (SEER) database (2000-2019). Patients with other malignancies or lack of pathological diagnosis were excluded. T-tests and chi-square tests compared variables, while Kaplan-Meier, log-rank tests, and Cox regression identified prognostic factors for overall survival (OS) and cancer-specific survival (CSS). Patient records were randomly divided into training (80 %) and validation (20 %) sets. Univariate Cox regression and least absolute shrinkage and selection operator were used to identify the significant factors linked to survival, which were used to establish a nomogram for predicting the 3- and 5-year OS and CSS. Results: Of the 367 included patients, 121 underwent surgery and 246 underwent surgery + ART. The median survival time was 54 months. Patients who underwent surgery had 5-year OS rates of 61% and 86% for CSS, while those treated with surgery + ART had 81% and 96% survival rates, respectively. The two treatment groups showed significant differences in the OS and CSS (p<0.022 and p<0.017, respectively). Age ≥ 50 years was a poor prognostic factor for OS, whereas ART, localized stage and regional stage were good prognostic factors. No significant factors were associated with CSS. To construct the OS Nomogram, three predictors, age, stage, and treatment method, were chosen. In the training and validation sets, the C-indices are 0.799 and 0.736, respectively. The receiver operating characteristic curves (AUCs) for the 3- and 5-year OS in the training set were 0.594 and 0.617, respectively, whereas those in the validation set were 0.56 and 0.59. For the CSS nomogram, two factors, age and stage, were utilized. The C-indices for the training and validation sets are 0.814 and 0.774, respectively. In the training set, the AUCs for 3- and 5-year CSS were 0.585 and 0.605, respectively, and 0.563 in the validation set. The most important risk factor was the distant stage, while the most important protective factor was age less than 50 years. Conclusion: This study shows that ART has a significant impact on the treatment of Intracranial HPC and provides better survival outcomes. ML has been used to create prediction models to support personalized patient care. These findings provide valuable insights into the management of this rare brain tumor and the potential of data-driven approaches to improve treatment and outcomes. Citation Format: Sakhr Alshwayyat, Anas Alaa Al-Khalili, Ayah Erjan, Haya Kamal, Tala Abdulsalam, Mustafa Alshwayyat. The impact of adjuvant radiotherapy (ART) and survival prediction by machine learning in intracranial hemangiopericytoma (HPC) [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 1108.