Predictors Of Short Term Death After Stereotactic Radiation For Brain Metastases

Stereotactic radiation (SRT; single or multiple fraction), with or without whole brain radiation (WBRT) is widely used to treat patients with brain metastases. Given the efforts and costs associated with SRT, it would be prudent to be able to identify patients with a likely short survival time post SRT, and in whom such treatment is unlikely to provide a meaningful benefit. We herein review our clinical experience to develop a model that can be used to predict for a short survival time following SRT for brain metastases. As part of an IRB-approved study, we retrospectively reviewed the medical and radiation records of patients with brain metastases treated with SRT at our institution from 2012-2015. Data extracted included: clinical data (e.g., age, gender, smoking status, Karnofsky performance status (KPS), comorbidities), cancer-related data (e.g., size and number of brain metastases, tumor histology, systemic disease status, and prior cranial radiation [prior to SRT], radiotherapy data (doses and fractionation), and survival. A multivariate logistic regression model, based on these clinical/cancer/treatment-related variables, was created to predict overall survival less than 2.5 months, using a backwards stepwise Akaike Information Criterion (AIC) approach. 178 patients were included in the analyses. The overall median survival was 8.7 months. 20% of patients had died within 2.5 months SRT. Via multivariate logistic regression, the following parameters were not predictive for a short survival: gender, comorbidities, smoking status, prior radiation (WBRT or SRT), number of metastases and total treatment volume. The optimal model predictive for a short survival post-SRT included response to systemic therapy (progression vs other), histology (NSCLC vs other), age (continuous) and KPS (continuous). The regression coefficients and p-values can be seen in the attached table.Tabled 1Abstract 2171; Table 1CovariateEstimateOdds Ratiop-valueIntercept-2.36-0.176Age0.0351.040.065Histology (NSCLC)0.0871.090.039Systemic Response (Progression)1.103.000.016KPS0.0141.010.074 Open table in a new tab A relatively simple model, based on readily available clinical data might be able to identify patients with a likely short survival time post SRT, and in whom such treatment is unlikely to provide a meaningful benefit. Additional studies along these lines are needed to make informed treatment decisions.