Gastric cancer (GC) is the fifth most diagnosed cancer globally, with limited benefits from anti-tumor therapies, particularly in the perioperative stage. The molecular mechanisms underlying peritoneal metastasis (GCPM), the leading cause of GC-related death, remain poorly understood. GCPM occurs when GC cells find a suitable growth environment in the peritoneum, a complex ecosystem with diverse immune cell populations. During GCPM development, both peritoneal exfoliated tumor cells (PETCs) and immune cells undergo changes, and anti-tumor therapy may induce further evolution, affecting therapeutic sensitivity. A comprehensive understanding of PETCs and peritoneal immune cells at single-cell resolution is crucial for elucidating GCPM mechanisms and developing improved therapies. While previous studies have explored intratumoral heterogeneity in primary GC and metastatic foci, the dynamic heterogeneity of the early-stage peritoneal ecosystem and the therapy-induced evolution, especially concerning immune checkpoints, remains largely unknown. This study aimed to address these gaps by characterizing the peritoneal ecosystem using single-cell RNA sequencing (scRNA-seq) during GCPM progression and treatment.

Several studies have investigated the intratumoral heterogeneity and lineage diversity in primary gastric cancer and peritoneal metastatic foci using single-cell technologies. These studies have revealed the existence of distinct subpopulations of cancer cells and immune cells within the tumor microenvironment. However, the dynamic changes in the composition and function of these cells during GCPM progression and the impact of anti-cancer therapies on the peritoneal ecosystem remain largely unclear. A comprehensive understanding of the cellular heterogeneity of the peritoneal ecosystem is essential for the development of effective therapeutic strategies for GCPM. This study sought to contribute to this understanding by analyzing a large cohort of scRNA-seq data from patients with GCPM.

This study utilized scRNA-seq to profile the cellular composition and heterogeneity of the peritoneal ecosystem in 35 gastric cancer patients. Patients were divided into five groups based on GCPM status and treatment: a control group (benign hysteromyoma), early GC, advanced GC, untreated advanced GC with GCPM, and treated advanced GC with GCPM. Ascites and peritoneal lavage fluid samples were collected, processed to create single-cell suspensions, and subjected to scRNA-seq using the 10x Genomics platform. A total of 191,987 high-quality cells were sequenced and analyzed. Data processing included quality control, doublet removal using DoubletFinder, normalization, integration of multiple datasets using Harmony, dimensionality reduction (PCA, UMAP), and unsupervised clustering. Cell types were annotated using SingleR and marker gene expression. InferCNV was used to identify malignant epithelial cells. Monocle2 was employed for trajectory inference analysis to determine developmental trajectories of cell populations. CellPhoneDB was used to infer cell-cell communication. SCENIC was used to infer transcription factor activity. Patient-derived organoids (PDOs) from ascites samples were cultured and treated with autophagy and mTORC1 inhibitors to assess the effects on cell viability and apoptosis. ELISA and FACS were used to validate findings from the scRNA-seq data. Statistical analyses, including Wilcoxon test, Student's t-test, and ANOVA, were performed to assess significance.

The study identified 12 major cell clusters in the peritoneal ecosystem, predominantly immune cells. During GCPM progression, a decrease in dendritic cells (DCs) and an increase in regulatory CD4 T cells (Tregs) and naive T cells were observed. A subset of monocyte-like DCs showed increased pro-angiogenic capacity and reduced antigen-presenting ability. These monocyte-like DCs were associated with a poor prognosis. A proliferative cycling T cell cluster, exhibiting an exhausted and dysfunctional phenotype, was also identified. Following therapy (chemotherapy and immunotherapy), marked heterogeneity evolution occurred in monocyte-like DCs and cycling T cells, involving cell fate transitions, immune phenotype changes, and metabolic reprogramming. In GC cells, a high-plasticity GC cell cluster exhibited a propensity to transition to a high-proliferative phenotype after therapy, resembling paligenosis. Two autophagy-related genes, MARCKS and TXNIP, were identified as biomarkers of high-plasticity GC and were associated with poor prognosis. In vitro experiments using PDOs showed that autophagy inhibitors significantly induced apoptosis, suggesting a potential therapeutic target. Immunotherapy, compared to chemotherapy, showed an overall reduction in antigen-presenting capacity in myeloid cells and altered immune checkpoint expression patterns, potentially leading to an anti-inflammatory phenotype in immune cells.

This study provides a comprehensive single-cell resolution view of the dynamic changes occurring within the peritoneal ecosystem during GCPM progression and in response to therapy. The findings highlight the critical role of monocyte-like DCs in promoting an immunosuppressive and pro-angiogenic tumor microenvironment. The identification of high-plasticity GC cells and their transition to a highly proliferative state via a paligenosis-like mechanism offers new insights into the processes driving therapy resistance. The observed therapy-induced evolution of both immune and cancer cells demonstrates the complexity of the response to treatment and the need for more sophisticated therapeutic strategies. Targeting autophagy, as demonstrated by the effects of autophagy inhibitors on PDOs, may offer a promising approach for overcoming therapy resistance. The study's findings underscore the importance of considering the intricate interactions between cancer cells and immune cells within the peritoneal ecosystem to develop more effective targeted therapies for GCPM.

This study provides a detailed single-cell transcriptomic analysis of the peritoneal ecosystem in gastric cancer peritoneal metastasis, revealing dynamic changes in immune cells and tumor cells during disease progression and after therapy. The identification of monocyte-like dendritic cells with pro-angiogenic and immunosuppressive properties, and high-plasticity GC cells undergoing paligenosis, provides new therapeutic targets. Autophagy inhibitors demonstrated efficacy in inducing apoptosis in patient-derived organoids, suggesting a potential avenue for future therapeutic intervention. Future studies should focus on validating these findings in larger patient cohorts and exploring the potential of targeting specific pathways involved in paligenosis and immune cell reprogramming to improve outcomes in GCPM.

The study's relatively small sample size, while larger than many similar studies, could limit the generalizability of some findings. The limited number of immunotherapy patients might also affect the conclusions drawn about immunotherapy-induced changes. Further research in larger and more diverse cohorts is needed to validate these findings and explore the impact of different treatment regimens on the peritoneal ecosystem.