The Impact of the Timing of PD-1 Inhibition on Disease Control for Brain Metastases Treated with Stereotactic Radiosurgery

Abstract

PD-1 inhibitors have shown single-agent activity against brain metastases, but their synergy with radiosurgery is unknown. The purpose of this study was to evaluate whether the timing of anti-PD-1 therapy when combined with stereotactic radiosurgery (SRS) for brain metastases may affect treatment outcomes. For this retrospective study, an institutional database of patients with intracranial metastases treated with SRS between 2014 and 2016 was reviewed for individuals who had received immunotherapy (IT) with the PD-1 inhibitors Nivolumab and Pembrolizumab. Patients were divided into two cohorts based on the timing of IT. Individuals were classified as having received concurrent IT if they received anti-PD-1 agents prior to or within within three weeks of SRS. Adjuvant IT was defined as administration of PD-1 inhibitors more than three weeks after SRS. The student's t-test aFisher’s exact test were utilized to compare baseline clinical and demographic covariates. The Kaplan-Meier method and log-rank test were then used to estimate and compare local control (LC), distant brain control (DBC), and overall survival (OS) between these two cohorts. Cox proportional hazards models were used to identify predictors of these outcome metrics. Fifty patients were eligible for this analysis, and the majority had been diagnosed with either metastatic non-small cell lung cancer (NSCLC, n=24) or melanoma (n=22). Twenty-seven (54%) received concurrent IT, and 23 (46%) received adjuvant IT (median time between SRS and adjuvant IT: 3.1 months, range 0.9 months – 11.9 months). Concurrent CTLA-4 was administered along with PD-1 blockade in only 2 patients (4%). Twelve patients (24%) had previously undergone whole brain radiation therapy (WBRT), and 37 patients (74%) had received prior chemotherapy. Mean follow-up time was 7.1 months. Baseline clinical and demographic characteristics were comparable between the two groups. At six months, LC was 100% for patients receiving concurrent IT and 76% in the adjuvant IT group (p=0.04). Similarly, DBC was 71% and 41%, respectively ( p=0.02). OS was not significantly different between the two groups (p = 0.36). On multivariate analysis, controlling for age, performance status, and prior WBRT, concurrent IT was predictive of both improved LC (HR = 0.07, p = 0.049) and improved DBC (HR = 0.30, p = 0.02). Subset analysis demonstrates that improved DBC with the administration of concurrent IT was seen in patients with both melanoma (p=0.02) and NSCLC (p=0.04). In this heavily pre-treated cohort undergoing radiosurgery for brain metastases, concurrent PD-1 IT administered within three weeks of SRS appears to improve LC and DBC when compared to adjuvant PD-1 blockade.