Ethan Lim, Hui Lin, Preeti K. Parhar, Chengyu Shi, Maria Chan1
18:15 - 20:15 | Mon 5 Mar | Caribbean ABC | MoPO.76
Stereotactic radiosurgery (SRS) has been rapidly rising as an optimal brain metastasis treatment. We report patient characteristics and survival outcomes for patients treated at a regional center of a large institution. We retrospectively collected data on 83 patients (a total of 122 brain metastasis), who underwent single-fraction brain SRS from 2011 to 2017. A total of 40 features were identified and analyzed for their potential impact on treatment outcomes including: dosimetry parameters (PTV size, dose to tumor, dose to normal tissue), number of lesions, primary tumor type, patient's age, sex, race, pre/post-SRS seizures, use of whole-brain radiation treatment, chemotherapy, steroid, survival rates, recurrence rates, post-treatment necrosis, etc. The data were processed using machine learning. Nearly half of the patients in this study are currently still alive. One-year and 2-year survival rates are 59% and 53%, and six patients survived more than 5-years. The results show that the corresponding features impacting outcome as selected by the machine learning include treatment sites, tumor sizes, normal brain volume receiving >12Gy. Based on our data, the features influencing survival the most appear to be tumor size, number of lesions, and patient age. However, with so many additional factors and with multiple crossover factors that can interact and influence the outcome simple statistical measurement results in less robust data analysis than machine learning algorithms.