# Cancer‐associated fibroblasts are associated with CD8+ T cell depletion and poor prognosis in colorectal adenocarcinoma: a multi‐omics and machine learning analysis

## TL;DR

This study shows that cancer-associated fibroblasts are linked to reduced T cell presence and worse outcomes in colorectal cancer patients.

## Contribution

The study introduces a machine learning model that improves survival prediction by incorporating aCAF features and multi-omics data.

## Key findings

- aCAFs correlate with advanced tumor stage and reduced CD8+ and CD4+ T cell infiltration.
- aCAF-related genes are involved in immunosuppressive and tumor progression pathways.
- Machine learning models using aCAFs show better prognostic accuracy for survival outcomes.

## Abstract

Fibroblastic proliferation in various tumor microenvironments influences cancer survival through complex interactions with diverse immune responses. This study investigated the impact of histologically unique activated cancer‐associated fibroblasts (aCAFs) on survival outcomes and immune responses and examined their association with various pathophysiological mechanisms. We analyzed a total of 1,024 colorectal adenocarcinoma patients from two cohorts. aCAFs were evaluated based on hematoxylin and eosin‐stained whole‐slide images, and their associations with clinicopathological features, immune cell infiltration, and survival were assessed. We developed a machine learning‐based survival prediction model incorporating aCAFs and clinicopathologic parameters. Additionally, we performed differential gene expression analysis, functional enrichment analyses, and in vitro drug screening of aCAF‐related genes. aCAFs were associated with advanced T stage, lymphovascular invasion, perineural invasion, and decreased CD8+ and CD4+ T cell infiltration. aCAFs were also associated with worse overall and disease‐free survival in both univariate and multivariate analyses. Functional enrichment analysis revealed that aCAF‐related genes were implicated in immunosuppressive signaling, oxidative stress regulation, and tumor progression pathways. Survival prediction models based on machine learning and incorporating aCAFs demonstrated superior prognostic accuracy for overall survival and disease‐free survival compared to models excluding aCAFs. Our analysis of aCAFs' association with immune responses through bioinformatics‐based genomic analysis and machine learning provides a foundation for future research in CRC patients.