全微生物组关联研究鉴定肿瘤相关微生物与人肿瘤微环境之间的串扰

Summary : Using recently refined measures of microbiome abundance in TCGA samples, we use quantitative models, in combination with human RNA-seq and RNA signatures, to identify microbial strains associated with functional modulation of the tumor microenvironment (TME), with implications for prognosis and therapy. Methods: Determining the influence of intra- and peri-tumoral microbes on cancer outcomes is an active area of investigation. Progress has been limited in part by technical challenges to accurate quantitation of microbial species across large clinical cohorts, such as The Cancer Genome Atlas (TCGA). Recent advances in this area have produced refined tumor microbiome estimates across thousands of TCGA samples (Ge et al. 2025). We exploited the refined microbiome information to perform associative modeling between strain abundances and TME signals in human TCGA RNA-seq and -signature data. We reasoned that identifying direct associations between microbe abundances and tumor transcription might add support to a given microbe's importance. Linear models were constructed based on these data and selected technical and clinical covariates across 25 distinct human tumor types. Human RNA abundances and RNA signatures were selected based on potential to identify mechanistic and regulatory elements associated with microbial presence, as well as to infer differences in abundance of human TME cell types. Results : We observed associations between microbe abundances and human transcription, confirming previously reported findings and identifying novel associations. In some cases, these associations were both specific to limited microbial strains and tumor type. For example, the relationships between an alpha-interferon-based RNA signature and microbial strains in serous ovarian carcinoma revealed only a small number of microbes in association, most of which were specific to ovarian cancer. Conclusion : By exploiting advances in microbiome quantitation, a large clinical cancer cohort, and an associative study framework, we identified a subset of microbial species in association with human tumor transcriptional profiles. Although limited by the associative nature of the study, these species are candidates for future efforts that evaluate their mechanistic roles in the TME. The analysis adds to the body of evidence from preclinical models and early clinical observations that microbial elements in patient tumors can modulate tumor and TME function, influencing prognosis and response to therapy.