3123 Background: In the KEYNOTE-059 study, the anti-PD-1 checkpoint inhibitor pembrolizumab was shown to have a modest overall response of 11.6%. Common predictors of response including, high microsatellite instability (MSI-H), PD-L1 expression, tumor mutational burden (TMB) and tumor inflammation signature (TIS), were not individually sufficient for patient selection. Recent pancancer studies have highlighted a unique population of cancer patients whose tumors appear to be driven by oncogene amplifications on extrachromosomal DNA (ecDNA). These ecDNA-driven tumors are aggressive and characterized by high levels of genomic instability. We sought to understand if tumors that possess ecDNA may represent a subset of the patient group that is non-responsive to anti-PD-1 therapy. Methods: We determined the ecDNA status of gastric cancer patients (N = 108) using whole genome sequencing (WGS) from the TCGA Pan-cancer dataset These patients had been previously subtyped for EBV status, genomic stability (GS), microsatellite instability (MSI), and chromosomal instability (CIN). Patients that were ecDNA+ were re-classified into a set regardless of gastric subtype. Additionally, TMB, TIS, and PD-L1 expression levels were collected. Results: 32% of gastric cancer patients were positive for ecDNA signatures and mutually exclusive from the 23% of MSI-H patients. We found that ecDNA positive tumors had statistically significantly lower TIS than all other groups (p-value < 0.05) except CIN tumors (p-value = 0.09). The ecDNA positive tumors also had lower PD-L1 expression than all but GS tumors. Only MSI-H showed statistically significantly higher TMB scores compared to every other group (p-value < 0.001), no difference in TMB scores were observed between every other pair of groups. Conclusions: Patients whose tumors are ecDNA positive represent a unique population that display signatures for non-response to checkpoint inhibitor therapy, including MSS, low TIS, and PD-L1 expression. Thus, the determination of tumor ecDNA status may have utility as an additional patient selection strategy for checkpoint inhibitor therapy. As ecDNA are not limited to gastric cancers, this study highlights the importance of the development of a clinical diagnostic test for ecDNA status and the need for further research on ecDNA biology, its impact on immunotherapy response, and potential ecDNA-directed therapeutics