Analysis of the role of tumor microenvironment on clinical outcomes of chemotherapy in patients with advanced gastric cancer using high plex digital spatial profiling.

Abstract

456 Background: Tumor consists of not only cancer cells but also various immune cells or stromal cells. And tumor microenvironment (TME) is known to be associated with response and resistance to treatment. However, the role of TME in the response to chemotherapy in advanced gastric cancer (AGC) remains unclear. We evaluated whether TME can affect the survival in patients with AGC who received first-line chemotherapy using high plex digital spatial profiling (DSP). Methods: Patients with AGC who received first-line chemotherapy were registered. Archival tumor tissue before first-line chemotherapy were obtained and tissue microarrays (TMAs) were made. Regions of interest (ROIs) were selected on each TMA and high plex DSP including RNA transcript was performed in all ROIs using GeoMx DSP (Whole Transcriptome Atlas) after masking by cytokeratin (CK) immunofluorescence. Differentially expressed gene (DEG), geneset enrichment analysis (GSEA), gene ontology (GO) and cell subsets were analyzed separately in CK-positive and CK-negative areas, using EdgeR, mSigDb, DAVID and CIBERSORT. Results: Thirty-three tumor tissues from 33 patients and 5 normal tissues were analyzed. All patients received fluoropyrimidine and platinum doublet chemotherapy. HER2+ was 15.2% and all with HER2+ AGC received trastuzumab plus chemotherapy. Five received nivolumab plus chemotherapy. Response rate was 48.5% and progression-free survival (PFS) was 9.8 months (95% CI, 6.85-NA). In CK-positive area, high expression of PIGR (polymeric Immunoglobulin receptor gene) (FDR q=0.013) and low expression of IFNE (FDR q=0.027) were associated with long PFS. Response to interferon beta (FDR q=0.044), cytokine-cytokine receptor interaction (FDR q=0.014) and antigen processing and presentation pathway (FDR q=0.015) were less activated in patients with long PFS in GO and GSEA. Although the density of CD8+ T cell did not affect PFS (6.9 vs. 10.8, p=0.25), high density of CX3CR1+memory CD8+T cells (Tmem) showed longer PFS (18.3 vs. 6.6 months, p=0.01), compared with low density of CX3CR1+CD8+Tmem. In CK-negative area, low density of CD8+ T cells were associated with longer PFS (6.3 vs. 18.3 months, p= 0.008). Conclusions: TME particularly in CK-positive area might affect survival for chemotherapy in AGC. PIGR and CX3CR1+CD8+Tmem were suggested as good candidate markers for predicting clinical outcomes of chemotherapy, indicating further analysis of TME will be needed especially in patients with AGC who received immune checkpoint inhibitors.