Brain metastases in upper gastrointestinal tumours: A multi-centre, retrospective analysis of clinical predictors and outcomes in a multi-ethnic Asian cohort.

e16079 Background: Gastric and esophageal cancers are the 5th and 11th most common worldwide, but brain metastasis (BM) in upper gastrointestinal (UGI) cancers is rare, with a reported 1–3% incidence. Identifying clinicopathological predictors is critical to improve both surveillance and management given this poor prognosis. We aim to evaluate risk factors for BM and determinants of overall survival (OS) and progression-free survival (PFS) in patients with metastatic UGI cancers. Methods: We conducted a multi-centre retrospective analysis of 803 patients with Stage IV Gastric, Esophageal and Gastro-Esophageal Junction cancers diagnosed in 2000–2024 across four tertiary hospitals in Singapore. Patients with de novo metastatic and recurrent disease were included. Logistic regression was used to identify predictors of BM development. OS and PFS were analysed using Cox proportional hazards models. OS was defined as time from diagnosis of BM to death from any cause. PFS was defined as time from diagnosis of BM to first documented disease progression. Results: Median age at cancer diagnosis was 65.3 years. Most patients were male (65.7%), Chinese (77.7%) with ECOG 0 or 1 (68.8%). Primary tumour sites were gastric (66.8%), esophageal (17.8%) and GEJ (14.7%). BM occurred in 77 patients (9.6%). Histological subtypes were predominantly adenocarcinoma (55.8%) and SCC (15.6%). On univariate analysis, age > 65 years (OR 0.54, p = 0.03) and diagnosis after 2010 (OR 0.34, p = 0.02) were associated with lower odds of BM, while T4 disease increased BM risk (OR 2.37, p = 0.03). On multivariable analysis, only age remained significant. HER2 and PDL-1 status were not associated with BM development. Median OS following BM diagnosis was 8.2 months (95% CI 5.4-12.1). Median Follow Up Time was 65.4 months (95% CI 49.8-∞). Among patients with BM, multivariable analysis demonstrated poorer OS with esophageal primary (v. GEJ, HR 2.76, p < 0.01), PDL CPS < 1 (HR 4.95, p < 0.01), HER2+ (HR 5.81, p < 0.01), leptomeningeal disease (HR 6.27, p < 0.01), multifocal BM (HR 1.96, p = 0.02), symptomatic BM (HR 4.05, p = 0.03) and absence of prior systemic therapy (HR 3.43, p < 0.01). Surgical resection of BM (HR 0.49, p = 0.05) and N0 disease (HR 0.20, p < 0.01) were associated with improved OS. On univariate analysis, ECOG > 1 (HR 2.14, p = 0.04), T4 disease (HR 2.41, p = 0.01) and multifocal BM (HR 1.89, p = 0.02) were associated with a poorer PFS, while N0 disease was associated with improved PFS (HR 0.36, p < 0.01). On multivariate analysis, HER2+ (HR 5.21, p < 0.01) and absence of previous systemic therapy (HR 1.72, p = 0.04) were also associated with poor PFS. Conclusions: In our cohort, the higher incidence of BM (9.6%) as compared to published series deserve further validation and investigation. Multimodality treatment - systemic therapy and surgical resection of BM - are associated with improved outcomes.