TLSB-16 GENOMIC AND CLINICAL PREDICTORS OF SURVIVAL FOLLOWING SURGICAL RESECTION OF BRAIN METASTASES
Abstract
Abstract Survival following surgical resection of brain metastases remains highly heterogeneous, and molecular determinants of postoperative prognosis are poorly characterized. We investigated clinical and molecular predictors of overall survival in a surgically treated cohort. A retrospective analysis of 228 patients who underwent brain metastasis resection at the University of Pittsburgh Medical Center was performed. Molecular profiling using the UPMC Oncomine Comprehensive Assay was performed in 138 patients (60.5%). A three-stage analytical framework was employed: bootstrap forest analysis for variable importance ranking, univariable Cox regression with false discovery rate correction, and multivariable Cox proportional hazards modeling integrating clinical and molecular predictors. The cohort predominantly comprised lung (53.1%), breast (11.8%), genitourinary (11.8%), gastrointestinal (11%), and melanoma (9.6%) primaries. Median overall survival was 7.3 months (95% CI: 6.3–9.2). The most frequent genetic alterations were TP53 mutations (61.6%), KRAS mutations (31.2%), and CDKN2A/B copy number alterations (24.6%, 23.2%). Bootstrap forest analysis identified BRCA2, NF2, TP53, FANCA, and RAD51 alterations as the top five genetic predictors of mortality. On univariable analysis, NF2 copy number alterations (HR 2.95, p = 0.006), MYC copy number alterations (HR 2.7, p = 0.022), KRAS copy number alterations (HR 2.14, p = 0.034), and ATM mutations (HR 2.6, p = 0.026) were associated with worse survival. Patients with good performance status (ECOG 0–1) demonstrated longer median OS compared to those with poor status (15.5 vs. 4.7 months, p < 0.0001). On multivariable analysis, independent predictors of survival included preoperative ECOG performance status (HR 1.98, p < 0.001), KRAS copy number alterations (HR 2.74, p = 0.02), and postoperative systemic therapy (HR 0.23, p < 0.001). ECOG performance status, KRAS copy number alterations, and postoperative systemic therapy are independently associated with survival following brain metastasis resection. Integrating molecular profiling with clinical factors provides a framework for postoperative survival risk stratification