Patient-Reported Cognitive Outcomes and Survival after Stereotactic Radiosurgery for 15 or More Brain Metastases

Abstract

Overall survival (OS) in metastatic cancer patients has been improving with continued advancements, necessitating greater attention to treatment related toxicities. Patients with ≥ 15 brain metastases (BM) continue to be treated with whole brain radiation therapy (WBRT), despite poor neurocognitive outcomes. We analyzed our institutional experience of treating these patients with stereotactic radiosurgery (SRS), with the aim of evaluating safety, cognitive outcomes, and survival metrics. Patients who received SRS for ≥ 15 BMs in 1-5 fractions from 2014-2022 using the previously described single isocenter multi-target technique were included. Cognitive outcomes were objectively evaluated using serial Patient-Reported Outcome Measurement Information System (PROMIS) scores. Kaplan-Meier method was used for survival analysis and log-rank test was used for intergroup comparisons. A total of 118 patients underwent 124 courses of LINAC-based SRS. The mean and median number of lesions treated per patient was 24.8 and 20, respectively (range 15 - 94). Most common primary histologies were lung (47.6%), melanoma (21.0%), and breast (14.5%). The median SRS dose was 24 Gy (range 18 - 30 Gy) with 87.9% receiving 3 fraction SRS. At the time of SRS, 19.4% patients had received prior WBRT and 24.2% had received at least one prior SRS course. The rate of any grade radiation necrosis (RN) and ≥ grade 3 RN were 15.3% and 3.2% respectively. New onset seizures were seen in 2.4%, alopecia in 2.4%, and subjective cognitive decline in 4% of patients. Median follow-up by reverse Kaplan-Meier method was 17.8 months (m). Cognitive data was available for 38 patients. Mean PROMIS scores at baseline, 3m, 6m and 9m after SRS were 32.0, 31.6, 30.4 and 28.7 out of 40, respectively. When longitudinal trends were available, 25 of 31 patients had a stable or improved PROMIS score. The 12m local control was 97.6%. Cumulative incidence of distant intracranial failure was 79.2%. One year freedom from neurological death, leptomeningeal disease, and salvage WBRT were 85.6%, 87.6% and 71.7% respectively. Median OS from brain metastases diagnosis was 11.3m (95% CI 7.2m - 15.3m) overall, while it was 9.2m (6.5 - 11.8m) after excluding patients who had prior WBRT or SRS (n = 79). One-year survival was 40% from diagnosis and 38.7% from SRS. On Cox regression analysis, prior WBRT (p = 0.002), higher KPS (p = 0.050), systemic therapy after SRS (p<0.001), and controlled extracranial disease (p = 0.007) predicted for improved OS. We present here the largest study evaluating SRS for patients with ≥ 15 BMs. We found that SRS was safe, had favorable cognitive outcomes, and comparable survival outcomes to contemporary studies evaluating WBRT in this patient population. Treatment-naïve patients had a median survival of > 6 months, long enough to benefit from cognitive sparing with SRS. Our study supports further randomized studies comparing SRS and memory avoidance WBRT approaches for patients with ≥ 15 BMs.