Multiomics reveals tumor microenvironment remodeling in locally advanced gastric and gastroesophageal junction cancer following neoadjuvant immunotherapy and chemotherapy

Abstract

Background: Perioperative chemotherapy is the standard of care for patients with locally advanced gastric and gastroesophageal junction cancer. Recent evidence demonstrated the addition of programmed cell death protein 1 (PD-1) inhibitors enhanced therapeutic efficacy. However, the mechanisms of response and resistance remain largely undefined. A detailed multiomic investigation is essential to elucidate these mechanisms.

Methods: We performed whole-exome sequencing, whole-transcriptome sequencing, multiplex immunofluorescence and single-cell RNA sequencing on matched pretreatment and post-treatment samples from 30 patients enrolled in an investigator-initiated Phase 2 clinical trial (NCT04908566). All patients received neoadjuvant PD-1 inhibitors in combination with chemotherapy. A major pathologic response (MPR) was defined as the presence of no more than 10% residual viable tumor cells following treatment.

Results: Before treatment, the positive ratio of CD3+T cells in both the tumor parenchyma and stroma was significantly higher in the non-MPR group compared with the MPR group (p=0.042 and p=0.013, respectively). Least absolute shrinkage and selection operator regression was employed for feature gene selection and 13 genes were ultimately used to construct a predictive model for identifying MPR after surgery. The model exhibited a perfect area under curve (AUC) of 1.000 (95% CI: 1.000 to 1.000, p<0.001). Post-treatment analysis revealed a significant increase in CD3+T cells, CD8+T cells and NK cells in the tumor stroma of MPR patients. In the tumor parenchyma, aside from a marked increase in CD8+T cells and NK cells, a notable reduction in macrophage was also observed (all p<0.05). Importantly, forkheadbox protein 3 (FOXP3), the principal marker for regulatory T cells (Treg) cells, showed a significant decrease during treatment in MPR patients. FOXP3 expression in the non-MPR group was significantly higher than in the MPR group (p=0.0056) after treatment. Furthermore, single-cell RNA sequencing analysis confirmed that nearly all Treg cells were derived from the non-MPR group.

Conclusions: Our study highlights the critical role of dynamic changes within the tumor immune microenvironment in predicting the efficacy of neoadjuvant combined immunochemotherapy. We examined the disparities between MPR/non-MPR groups, shedding light on potential mechanisms of immune response and suppression. In addition to bolstering cytotoxic immune responses, specifically targeting Treg cells may be crucial for enhancing treatment outcomes.

Keywords: Gastric Cancer; Immunotherapy; Neoadjuvant; Tumor microenvironment - TME.

Figure 1. The study design. AUC, area under curve; FOXP3, forkheadbox protein 3; MPR, major pathological response; NK, natural killer; PD-1, programmed cell death 1; Tac, activated T cells; Tcm, central memory T cells; Tem, effector memory T cells; Tex, exhausted T cells; TMB, tumor mutation burden; TPM, transcripts per million; Treg, regulatory T cell; Trm, resting memory cells; UMAP, Uniform Manifold Approximation and Projection.

Figure 2. Differential tumor microenvironment before treatment. (A) Box plots showing the differences in the immune cells’ positive ratio between the MPR and non-MPR groups as quantified by mIHC before treatment. The top panel represents the tumor parenchyma, and the bottom panel represents the tumor stroma. (B) Box plots showing the differences in the immune cells’ density between the MPR and non-MPR groups as quantified by mIHC before treatment. The left panel represents the tumor parenchyma, and the right panel represents the tumor stroma. (C) Volcano plot showing DEGs between the MPR and non-MPR groups. Red dots indicate upregulated genes in MPR patients, while blue dots indicate upregulated genes in non-MPR patients. (D) KEGG enrichment analysis for upregulated genes in the MPR group. (E) KEGG enrichment analysis of upregulated genes in the non-MPR group. (F) ROC curve for the prediction of major pathologic responses using the TMB, the positive ratio (mIHC) of PD-1+cells, the positive ratio (mIHC) of CD3+cells, and LASSO model features (DEGs).AUC, area under curve; cAMP,cyclic adenosine monophosphate; DEG, differentially expressed gene; IL, interleukin; KEGG, Kyoto Encyclopedia of Genes and Genomes; LASSO, least absolute shrinkage and selection operator; mIHC, multiplex immunohistochemistry; MPR, major pathological response; NK, natural killer; PD-1, programmed cell death 1; PD-L1, programmed cell death ligand 1; ROC, receiver operating characteristic; TMB, tumor mutation burden; TNF, tumor necrosis factor.

Figure 3. Comparison of immune cell positive ratios in the tumor microenvironment before and after treatment. (A) Line plots showing the variation in the positive ratios of CD8+T cells, NK cells, M1 macrophages, and M2 macrophages within the tumor parenchyma for both MPR and non-MPR groups before and after treatment. (B) Line plots showing the variation in the positive ratios of CD8+T cells, NK cells, M1 macrophages, and CD3+cells within the tumor stroma for both MPR and non-MPR groups before and after treatment. MPR, major pathological response; NK, natural killer.

Figure 4. Changes in immune suppression-related gene expression during treatment. (A) Box plots showing the expression of cytotoxic molecules (CD8A, GZMK, GZMH) before and after treatment in both the MPR and non-MPR groups. (B) Box plots showing the expression levels of chemokine receptors in non-MPR patients before and after treatment. (C) Box plots showing the expression levels of the FOXP3 gene before (Pre) and after (Post) treatment in the MPR and non-MPR groups. FOXP3, forkheadbox protein 3; GZMH, granzyme H; GZMK, granzyme K; MPR, major pathological response; TPM, transcripts per million.

Figure 5. Single-cell analysis of T-cell subtypes and their distribution in the MPR and non-MPR groups. (A) UMAP plot showing the categorization of 1,390 cells into seven major T-cell subtypes: CD8_Trm, CD8_Tem, CD8_Tac, CD4_Trm, CD4_Tcm, CD4_Tex, and Treg. (B) UMAP plot depicting the distribution of cells categorized into the MPR and non-MPR groups. (C) Bar plot illustrating the fraction of each T-cell subtype within the MPR and non-MPR groups. Significant differences between groups are indicated. (D) Proportion of T-cell subtypes in the tumor microenvironment, comparing the MPR and non-MPR groups. (E) Pseudotime trajectory analysis of T cells showing the progression and differentiation states of various T-cell subtypes. (F) Differentiation pathway of T cells using PAGA (partition-based graph abstraction), highlighting the transitions between different T-cell states from CD8_Tac (0) through various intermediates to Tregs (6). (G) Box plots showing the expression levels of the immune checkpoint molecules PDCD1 (PD-1), CTLA-4, and TIGIT in the MPR and non-MPR groups. (H) Box plots illustrating the expression levels of the cytotoxic and activation markers CD8A, NKG7, GZMA, and GZMK in the MPR and non-MPR groups. CTLA-4, cytotoxic T lymphocyte associate protein 4; GZMA, granzyme A; GZMK, granzyme K; MPR, major pathological response; NKG7, natural killer cell granule protein 7; PD-1, programmed cell death protein 1; PDCD1, programmed cell death 1; Tac, activated T cells; Tcm, central memory T cells; Tem, effector memory T cells; Tex, exhausted T cells; TIGIT, T cell immunoreceptor with Ig and ITIM domains; Treg, regulatory T cell; Trm, resting memory cells; UMAP, Uniform Manifold Approximation and Projection.