Abstract 6846: Tumor cell metaplastic programs and failure of T cell remodeling drive resistance and progression to frontline chemoimmunotherapy in advanced gastric cancer

Abstract Adding aPD1 to 5-FU/platinum in advanced gastroesophageal adenocarcinomas (GEA) has yielded modest and heterogeneous results. Understanding cooperativity between these two treatment modalities will inform novel combination treatments for GEA. Towards this end, we conducted a trial (n = 47) sequentially adding pembrolizumab to 5-FU/platinum in previously untreated advanced GEA. Using serial biopsy of the primary tumor at baseline, after one cycle of 5-FU/platinum, and after the addition of pembrolizumab we transcriptionally profiled 358,067 single cells and obtained TCR-sequences from single tumor-resident T cells. Stratifying patients into slow- (> 6 months) and fast-progressing (< 6 months) groups based on progression-free survival allowed us to quantify differences in immune, stromal and epithelial subpopulations across treatment time points. Using topic modeling with consensus non-negative matrix factorization (cNMF), we identified gene expression programs in each population at each time point and developed a methodology to verify the stability of those programs across the cohort using cross-validation. Faster clinical progression featured decreased M1-like macrophage repolarization and increased MUC5A and MSLN containing metaplasia programs in tumor cells. We next selected features from cell state gene programs and identified trajectories of tumor cells across a spectrum of treatment sensitive and resistant states. We identified several factors that correlated with these trajectories and regulons that may be influencing them. After pembrolizumab we observed slower progression was associated with further increased CD8 T cell infiltration with increased CXCL13+ T cells. We next calculated covarying programs between patient samples at each time point and performed permutation testing to identify which programs (and respective cell types they were expressed in) were significantly correlated to infer the presence of multicellular tumor communities. We found that the tumor cell interferon program co-varied with the CXCL13 T cell program, and the emergence of a covarying module with tumor-reactive CXCL13 program expression happens as early as after the first dose of chemotherapy. To better understand the behavior of tumor reactive T cells within the tumor microenvironment, we leveraged TCR-sequencing data to study T cell clonality and diversity across treatment time points and between fast and slow progressors. Our data provide an invaluable resource to dissect chemoimmunotherapy response in tumor cells, understand T cell behavior and propose strategies to drive anti-tumor immune hub formation to expand the portion of patients benefiting from immunotherapy approaches. Citation Format: Samuel J. Wright, Minae An, Byung Hoon Min, You Jeong Heo, Milan Parikh, Lynn Bi, Hyuk Lee, Taejun Kim, Song-Yi Lee, Jeonghyeon Moon, Ryan J. Park, Matthew R. Strickland, Woong Yang Park, Won Ki Kang, Kyoung-Mee Kim, Seung Tae Kim, Arnav Mehta, Samuel J. Klempner, Jeeyun Lee. Tumor cell metaplastic programs and failure of T cell remodeling drive resistance and progression to frontline chemoimmunotherapy in advanced gastric cancer [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 6846.