Comprehensive analysis of molecular characteristics and tumor immune microenvironment in stage II and III gastric carcinoma

Abstract

Tumor microenvironment immune type (TMIT) is the novel classification scheme based on both the expression of PD-L1 and density of CD8-positive tumor infiltrating lymphocytes. We aimed to apply this classification in stage II and III gastric cancer (GC) patients and assess the prognostic and molecular genetic implications of this classification. A total of 392 Stage II and III GC patients who were treated by curative surgical resection followed by 5-fluorouracil based adjuvant chemotherapy in Seoul National University Bundang Hospital were included in this study. Tissue microarrays were constructed from the formalin fixed paraffin embedded tissue samples, and the clinical information were collected retrospectively. Based on the immunohistochemistry (IHC) results of PD-L1 and CD8, TMIT classification of GC was performed as follows: type I (PD-L1+/CD8High), type II (PD-L1-/CD8Low), type III (PD-L1+/CD8Low), type IV (PD-L1-/CD8High). The clinicopathologic features including overall survival according to these four types were analyzed for the evaluation of prognostic performance of TMIT. For the comprehensive assessment of molecular characteristics of GC in immuno-oncology related perspective, IHC for tumor infiltrating immune cell markers (CD8, Foxp3), markers for epithelial-mesenchymal transition (E-cadherin, vimentin), markers representing cancer stem cells (CD44, Sox2, CD133, OCT3/4), as well as EBV in situ hybridization and microsatellite instability testings were performed. To elucidate the possible relationship between mutational profiles of GC and immune microenvironment, we analyzed gene expression data and clinical information from two publicly available transcriptome database. In addition, we performed deep targeted sequencing on 80 selected cases from all four TMITs, using the targeted sequencing panel of 170 recurrently mutated genes in various types of solid tumors. I have found that EBV+ and MSI-H GCs are distinct subtypes that are tightly associated with TMIT I (PD-L1+/CD8High), and OS within the CD8High group differs according to PD-L1 expression. Therefore, I conclude that co-assessment of PD-L1 and CD8+ TILs is clinically relevant, has a possible prognostic role, and warrants further investigation as a predictive marker for immune checkpoint blockade. Moreover, I have found an inverse association between EMT phenotype and PD-L1 expression, and close association between EMT features and TMIT II in GCs, which are the opposite results compared to other types of solid tumors. Additional TMIT-associated tumor characteristics include cancer stemess: I have found a tight association between CD44 positivity, a cancer stem cell marker, and TMIT I phenotype, which is consistent with recent findings that CD44+ tumor cells play important roles on cancer progression by expressing PD-L1. Finally, by performing deep targeted sequencing on selected GC tissue samples, I have found that TMIT I tumors have more numbers of somatic mutations compared to other groups and are enriched with somatic mutations of major cancer related genes including PIK3CA. TMIT II tumors were enriched with mutations of RUNX1 gene, and NTRK3 mutations were relatively specific to TMIT IV. TMIT III had unique somatic mutational profile, harbouring mutations of genes such as APC, TSC1, JAK1, MET, HRAS and RHEB. Clustering analysis based on somatic mutational profiles have identified two groups, one with higher mutational burden (cluster 1) and the other with lower (cluster 2)

cluster 1 had significant association with MSI-H GCs and showed the slight tendency of shorter overall survival. Recent advances of immunotherapy in solid tumors have facilitated the search for valuable predictive factor for favorable treatment outcome. TMIT was developed for better understanding of immune microenvironment and more effective immune treatment strategy. Based on the findings from this study, we conclude that application of TMIT classification in GC would be helpful for selecting the patients who would have favorable response to immunotherapy, and that this classification could be utilized as the significant prognostic indicator in stage II and III GC. By clarifying the relationship between molecular profile and microenvironment of GC, we expect to have clues for deeper understanding of the pathogenesis of GC as well as the oncogenesis and progression of other types of solid tumor.