Abstract
The interaction between intratumoral microbiome and the tumor microenvironment (TME) has furthered our understanding of tumor ecology. Yet, the implications of their interaction for lung cancer management remain unclear. In the current work, we collected host transcriptome samples and matched intratumoral microbiome samples, as well as detailed clinical metadata from The Cancer Genome Atlas (TCGA) of 478 patients with lung adenocarcinoma (LUAD). Utilizing the multiomics integration approach, we comprehensively investigated the crosstalk between the TME and intratumoral microbiome in patients with LUAD. First, we developed a prognostic model based on the TME signatures (TMEindex) that clearly distinguished clinical, survival, and response to immunotherapy of patients with LUAD. Additionally, we found profound differences in intratumoral microbiota signatures, including alpha- and beta-diversity, among patients with different survival risks based on the TME signatures. In depth, we detected that genera Luteibacter and Chryseobacterium were strongly negatively and positively associated with patients' survival risk, respectively, suggesting their opposing roles in cancer progression. Moreover, we developed a model that fused intratumoral microbial abundance information with TME signatures, called intratumoral microbiome-modified TMEindex (IMTMEindex), leading in predicting patient overall survival at 1-, 3-, and 5-years. Future clinical profiling of the specific intratumoral microbes in the TME could improve prognosis, inform immunotherapy, and facilitate the development of novel therapeutics for LUAD.