# Tumor microbiome-transcriptome crosstalk identifies Prevotella as an immunotherapeutic predictor in NSCLC

## TL;DR

This study finds that the presence of Prevotella bacteria in lung cancer tissue is linked to better outcomes and response to immunotherapy in non-small cell lung cancer patients.

## Contribution

The study identifies Prevotella as a novel predictor of immunotherapy response in NSCLC by analyzing microbiome-transcriptome interactions.

## Key findings

- PT-resident Prevotella is associated with improved prognosis and immune checkpoint inhibitor response in NSCLC.
- A diagnostic model combining Prevotella abundance and blood indicators outperforms PD-L1 as a biomarker for immunotherapy response.
- Prevotella is linked to immune cell populations like CD8+ T cells and activated dendritic cells in NSCLC.

## Abstract

Background: The tumor-resident microbiome plays a pivotal role in shaping the tumor immune microenvironment; however, its relationship with the host transcriptome and the response to immune checkpoint inhibitors (ICIs) remains largely uncharacterized in non-small cell lung cancer (NSCLC). This study aimed to elucidate the relationship between tissue-resident microbiota, host transcriptomic alterations, and immunotherapy response in NSCLC.

Methods: Paired tumor (T) and paracancerous tissue (PT) samples from patients with NSCLC were analyzed using 2bRAD-M and bulk RNA sequencing to generate comprehensive microbiome and transcriptome profiles. The conditional mutual information algorithm was employed to systematically investigate intratumoral microbe-host interactions. Associations between key microbes and patient prognosis, ICI response, and response to epidermal growth factor receptor (EGFR)-targeted therapy were assessed across four independent local clinical cohorts.

Results: Higher microbial richness, α-diversity, and β-diversity were observed in PT samples than in T samples. Specifically, PT-resident Bradyrhizobium and Prevotella were identified as key bacterial taxa significantly associated with immune cell populations, including CD8+ T cells, natural killer cells, and activated dendritic cells. Among these, PT-resident Prevotella, but not Bradyrhizobium, was independently associated with improved prognosis of patients with NSCLC and ICI response in both local clinical sets and public datasets. Furthermore, a combined diagnostic model integrating PT-resident Prevotella abundance with routine clinical blood indicators demonstrated markedly superior predictive performance for ICI response compared with the conventional biomarker PD-L1. By contrast, PT-resident Prevotella exhibited no association with treatment response in the EGFR-targeted therapy cohort.

Conclusion: PT-resident Prevotella is strongly associated with the prognosis and ICI response in patients with NSCLC. Moreover, integration of PT-resident Prevotella with routine clinical blood indicators holds promise as a potential auxiliary diagnostic tool to facilitate personalized immunotherapy in NSCLC.