Integrative analysis of GEO data on the microbial community in colorectal cancer tissues and its impact on the tumor immune microenvironment

Abstract

Objective Colorectal cancer (CRC) development is closely associated with gut microbiota dysbiosis and disruption of the tumor immune microenvironment. This study aimed to characterize and independently validate the cross-dataset associations of the “microbiome-immune axis” in CRC through multi-omics integration analysis and independent large-scale cohort validation methods. Methods A multi-stage strategy was applied. The microbiome cohort GSE163366 was analyzed to identify gut dysbiosis in CRC. Single-cell transcriptomic data from GSE132465 characterized tumor immune remodeling. Cross-dataset association analysis linked microbial abundances (GSE163366) with immune cell proportions (GSE132465), and bulk transcriptomics explored underlying pathways. Independent validation cohort GSE237523 rigorously tested the robustness of core bacterial genera. Results CRC patients exhibited significantly reduced α diversity (Shannon index, P = 1.29×10-7) and altered community structure (PERMANOVA, P = 0.001). The discovery set identified seven key differentially abundant genera, including pathogenic bacteria such as Fusobacterium nucleatum and Bacteroides fragilis, which were enriched in CRC, whereas beneficial bacteria such as Akkermansia muciniphila and Roseburia intestinalis were depleted. The tumor microenvironment exhibited features of reduced adaptive immune infiltration, including a significant reduction in B-cell proportion (11.5% decrease). Cross-dataset association analysis revealed that pathogenic bacteria showed positive associations with myeloid cells but negative associations with B-cells, whereas beneficial bacteria exhibited the opposite pattern. KEGG pathway enrichment analysis demonstrated that differentially expressed genes were significantly enriched in key immune signaling pathways, including the Toll-like receptor and NF-κB signaling pathways. Within the independent validation dataset GSE237523, all seven core bacterial genera identified in the discovery cohort were detected with significant abundance changes. Among these, six genera (85.7%) exhibited completely consistent directions of abundance change in the independent cohort. Notably, core pathogenic bacteria, including Fusobacterium nucleatum and Bacteroides fragilis, demonstrated high cross-cohort consistency. Conclusion This study provides systematic characterization and independent validation of gut microbiota dysbiosis and its cross-dataset association with the immunosuppressive microenvironment of CRC. The reproducible abundance changes of six out of seven core microbial genera (85.7%) confirm cross-cohort robustness, offering a validated reference for understanding the microbiota-immune axis in tumor progression and developing microbiome-based diagnostic strategies.