Early Objective Response as a Prognosticator for Final Objective Response and Survival following SRS for Brain Metastases Treated with Concurrent Systemic Therapy

Abstract

Increasingly, patients with brain metastasis are treated with combinations of stereotactic radiosurgery (SRS) and CNS-penetrating systemic therapies. Early identification of treatment response or failure may be useful for choice of early salvage therapies. We hypothesized that first follow-up objective response after SRS is prognostic for final response and survival in two cohorts treated with CNS-penetrating systemic therapies. A retrospective institutional review board-approved cohort study of patients diagnosed with brain metastasis from 2006-2017 at a single academic institution was conducted. To study the impact of concurrent systemic therapy, patients treated with PD-1/PD-L1 inhibitors or lapatinib were eligible for inclusion. The primary outcome was overall survival (OS), while final objective response was a secondary outcome. Objective response was defined as the percentage reduction in lesion maximum diameter, and categorized according to RECIST. Immediate concurrent therapy was defined as SRS delivered within one half-life of systemic therapy. During the study period, 235 patients were eligible for inclusion and underwent SRS to 1,490 lesions. Among these, 367 (25%) and 119 (8%) lesions were treated either with immediate concurrent immunotherapy or lapatinib, respectively. Median radiographic follow-up was 11 months. First radiographic follow-up (FFU) occurred at a median of 1.6 months after SRS, at which time median objective response was 33%, with 206 (14%) complete responses (CR), 657 (44%) partial responses (PR), and 47 (3%) lesions with progressive disease (PD). After full follow-up, median best objective response was 67%, with 598 (40%) CR, 525 (35%) PR, and 151 (10%) PD. The negative predictive value of FFU CR and PR for ruling out PD were 98% and 94%, respectively. Among lesions with FFU CR or PR, the PPV for eventual CR was 77%. Median FFU objective response (40% vs. 33%, p<0.01) and best objective response (100% vs. 63%, p<0.01) were superior with immediate concurrent therapy. Median OS was 27.3 months (95% CI 23.7-31.7). Among all patients, FFU response was not prognostic for adjusted OS (HR 1.01 per 10% response, p=0.84) in a multivariate proportional hazards model. However, FFU response was significantly associated with adjusted OS among patients treated with immediate concurrent therapy (HR 0.90, p=0.01). This effect was specifically driven by patients receiving immunotherapy (HR 0.87, p<0.01), rather than lapatinib (HR 0.96, p=0.62). In sensitivity analysis, this corresponded to a 12-month OS of 86% vs. 77% (p<0.01) among patients achieving or not achieving at least a PR at FFU. After SRS, FFU objective response was predictive of final objective response and PD, and was prognostic for survival among patients treated with concurrent immunotherapy but not lapatinib. Further prospective analyses are warranted to assess the validity of FFU objective response as a biomarker for early immunotherapy response or failure.