SURG-55. Clinical Predictors of Mortality in Brain Tumor Surgery: a 902- patient cohort from a Low-and Middle-Income Country

Abstract BACKGROUND Gliomas and brain metastases represent the most prevalent malignant brain tumors, yet data on surgical outcomes and survival predictors in low- and middle-income countries remain limited. This study aimed to characterize clinical, surgical, and prognostic variables associated with outcomes following resection of primary and metastatic brain tumors in a large tertiary center in Brazil. METHODS A retrospective cohort study was conducted including 902 adult patients who underwent craniotomy for resection of primary or metastatic brain tumors between June 2018 and December 2022. Demographic, clinical, surgical, and molecular data were collected. Outcomes included in-hospital mortality, prolonged hospitalization, and overall survival. Multivariate logistic regression and Cox proportional hazards models were used to identify independent predictors. RESULTS Of 902 patients, 388 (51.9%) had gliomas and 322 (35.7%) had brain metastases. Prolonged hospitalization (>5 days) occurred in 30.5% of cases, and 14.2% developed postoperative neurological or surgical complications. In patients with primary brain tumors, older age (OR = 1.051; p = 0.006) and longer preoperative stay (OR = 1.262; p = 0.031) were associated with increased in-hospital mortality, while IDH mutation was protective (OR = 0.196; p = 0.002). Emergency surgeries and older age were also associated with prolonged hospitalization. Age was the only independent predictor of decreased survival (HR = 1.016; p = 0.004). In the metastatic tumor cohort, awake surgery and resection of a single brain metastasis reduced the risk of death, whereas emergency surgery and motor area involvement increased mortality risk. CONCLUSIONS In this Brazilian cohort, older age and emergency surgical indication were consistently associated with worse short-term and long-term outcomes. IDH mutation status and surgical variables such as approach and tumor location may serve as important predictors in neuro- oncologic care. These findings support the integration of clinical, surgical, and molecular data in guiding brain tumor management in resource-constrained settings.