Validation of the Score Index for Radiosurgery (SIR) in Predicting Survival of Patients With Brain Metastases Submitted to Radiosurgery

Abstract

To validate the effectiveness of the SIR to predict overall survival of brain-metastatic patients treated with stereotactic radiosurgery (SRS). We conducted a retrospective analysis for SIR validation in patients treated between April 2010 and July 2012 at a radiosurgery cohort. A comparison with the other 3 major reported scores was performed: The RTOG Recursive Partitioning Analysis (RPA), the Basic Score for Brain Metastases (BSBM) and the Graded Prognostic Assessment (GPA). Each score was categorized into 3 groups according to similar survival, group difference within indexes were compared by log-rank test and differences between indexes by individual Cox regression likelihood-ratios (LR) and hazard-ratios (HR). Independent prognostic factors were compared by multivariate Cox regression model. Overall survival (OS) was considered from the procedure date and survival curves were estimated by Kaplan-Meier method. There were 110 patients with 238 treated lesions. Median follow-up for patients alive was 19.1 months (range, 5.4 - 44.0 months). The median dose was 18 Gy (12 - 22 Gy) and the most common primary sites were lung (40,4%) and breast (17,6%). Median OS was 9.9 months. Median survival for indexes were: grouped SIR from 1 to 3, 4 to 6, and 7 to 10 was 1.7, 7.3 and 19.9 months, respectively (p < 0.0001 / LR = 35.85 - HR = 3.69, 95% CI, 2.36-5.77). RPA classes I, II, and III was 2.1, 9.7 and 32.1 months, respectively (p = 0.0011 / LR = 13,47- HR = 2.17, 95% CI, 1.43-3.30). BSBM from 0, 1 or 2, and 3 was 17.4, 10.4 and 2.5 months, respectively (p < 0.0001 / LR = 17.82- HR = 2.58, 95% CI, 1.64-4.06) and grouped GPA from 0 to 1.5, 2.0 to 3.0, and 3.5 to 4.0 was 5.0, 12.4 months and not reached, respectively (p < 0.0001 / LR = 20.97 - HR = 2.31, 95% CI, 1.60-3.32). On multivariate analysis of prognostic factors, significance was observed for age (p = 0.004 - HR = 1.54, 95% CI, 1.152-2.067), KPS (p < 0.001 - HR = 2.83, 95% CI, 1.874-4.284), extracranial disease status (p = 0.004 - HR = 1.60, 95% CI, 1.162-2.202), volume of the largest lesion (p = 0.004 - HR = 1.84, 95% CI, 1.217-2.792), time to CNS metastases onset from the primary diagnostic date (p = 0.024 - HR = 1.66, 95% CI, 1.217-2.792) and breast as primary site (p = 0,039 - HR = 0.456, 95% CI, 0.217-0.959). No association with survival was observed for number of treated lesions, previous WBRT and enhancement pattern. As previous reports, SIR proved to be a good predictor of overall survival on patients treated with SRS. In our population, this model showed to be the most accurate between the available scores. Prognostic factors such as age, KPS, extracranial disease status, volume of the largest lesion and time to onset of the brain metastases are reliable and should be considered at the time of the patient evaluation.