Immune checkpoint inhibition in EBV-associated gastric cancer: A multi-center international retrospective analysis.

4043 Background: Epstein-Barr virus associated gastric cancer (EBVaGC) is found in ~9% of GC and is associated with a unique immunophenotype. Prior studies suggest 15-100% objective response rates (ORR) to immune checkpoint inhibition (ICI), but survival outcomes are limited by small cohort size. This study sought to clarify outcomes of ICI in EBVaGC across 9 global tertiary cancer centers, evaluate if next-generation sequencing (NGS) can replace EBER ISH as the gold standard for detection, and identify molecular features of ICI-responsive EBVaGC. Methods: Retrospective data were collected on patients (pts) with metastatic EBVaGC and stratified into patients receiving ICI alone or with chemotherapy, and by 1L vs. 2L+. Progression-free survival (PFS) and overall survival (OS) were analysed using the Kaplan-Meier method and were calculated from the start of treatment to the date of progression or death for PFS, and last follow up or death for OS. Cox regression model stratified by regions was used to examine factors associated with PFS. EBV viral reads were detected by NGS using independent cohorts by MSK-IMPACT and Caris Life Sciences and concordance with EBER ISH was evaluated in each cohort. Additional phenotypic classification was performed using clinicogenomic data from the Caris Precision Oncology Alliance. Results: A total of 91 pts who received ICI in the metastatic setting were included. Median age was 62 years (range 30-86) and 91% were male with pts from US (n = 15), Europe (n = 5), and Asia (n = 71). ECOG PS at the time of ICI was ≤1 in 97%. PD-L1 CPS score was ≥5 in 69%. Pts who received first-line chemotherapy + ICI (n = 42) achieved a 49% investigator-assessed ORR (CR/PR), median PFS (mPFS) 8.3 months (mo), and median OS (mOS) 38 mo compared with mOS 18 mo in pts receiving first-line chemotherapy alone (n = 35). Pts who received ICI-alone achieved a similar 49% ORR with 1L and later line mPFS 6 mo and 3.2 mo, respectively, which increase to 10.0 mo and 6.6 mo, when limiting to PD-L1 CPS >5. Univariate analysis for PFS among all pts demonstrated improved outcomes in those where PD-L1 CPS score was ≥5 (HR = 0.57, 95%CI:0.34,0.97). EBV detection by MSK-IMPACT and Caris achieved over 98% positive agreement and 99.9% negative agreement. Additional EBVaGC transcriptomic immune phenotyping data will be presented. Conclusions: This global EBVaGC experience demonstrated that EBVaGC can reliably be identified from NGS, negating the need for costly EBER ISH. Furthermore, pts with EBVaGC achieve a high ORR on ICI with or without concurrent chemotherapy. While 1L PFS was similar to non-EBVaGC, prolonged mOS was achieved, potentially reflecting frequent conversion surgeries and regional differences. Even among EBVaGC, immune heterogeneity exists, and PD-L1 CPS >5 identified pts who derived ICI benefit. Further research is needed to examine the patterns of disease progression in EBVaGC treated with ICI.