PulseExploreJournal ClubDebatesTrendingResearchersJournals
Instagram
HomeExploreJournal ClubTrending
Synapse
⌘+K
Synapse
March 5, 2026Frontiers in Immunology0 citationsOpen Access

Occupational silica exposure drives systemic immune dysregulation and tumor microenvironment susceptibility: evidence from a real-world study

HHHan HaoZZZaitian ZhangHZHui Zhang

Key Points

  • This study aims to uncover how occupational silica exposure affects immune functions and tumor microenvironment alterations that could promote cancer.
  • Analyzed health data from 5,482 industrial workers to assess silica exposure effects.
  • Constructed machine learning models, utilizing immune parameters to predict carcinoembryonic antigen positivity.
  • Validated findings through silica-stimulated monocytes and colorectal cancer cell lines.
  • Integrated publicly available gene data to build interaction networks and identify key genes.
  • Conducted single-cell RNA sequencing to explore specific immune cell expression.
  • CatBoost algorithm showed the best prediction for CEA positivity based on immune and exposure data.
  • Monocyte-to-lymphocyte ratio and silica exposure were significant predictors identified by SHAP analysis.
  • Silica exposure led to NF-κB activation, enhancing IL-6 secretion and CEA expression in colorectal cancer cells.
  • A multi-omics approach revealed 42 genes linking silica exposure to colorectal cancer pathways.
  • Higher expression of inflammation-related genes was observed in tumor-associated macrophages.

Abstract

Background Occupational exposure to carcinogenic dusts such as silica is a well-established risk factor for cancer. However, the molecular mechanisms linking early exposure to tumor-promoting microenvironmental changes remain poorly defined. Emerging evidence suggests that chronic immune dysregulation and remodeling of the tumor microenvironment (TME) may serve as critical intermediates. Methods We analyzed occupational health data from 5,482 industrial workers in Anhui Province, China. Explainable machine learning models were constructed using exposure profiles and hematological immune parameters to predict carcinoembryonic antigen (CEA) positivity, with feature contributions interpreted via SHAP values. Experimental validation involved silica-stimulated THP-1 monocytes and colorectal cancer (CRC) cell lines to assess inflammatory activation and paracrine regulation of CEA. Silica- and CRC-associated genes were integrated from public databases to construct protein–protein interaction networks, identify hub genes, and evaluate prognostic significance using TCGA and GSE39582 datasets. Single-cell RNA sequencing (scRNA-seq) analysis was used to resolve cell type–specific expression patterns. Results Among 14 algorithms tested, CatBoost exhibited the highest predictive performance for CEA positivity. SHAP analysis highlighted the monocyte-to-lymphocyte ratio and silica exposure as dominant contributors. Mediation analysis confirmed that systemic inflammation partially mediated the silica–CEA association. In vitro , silica activated NF-κB–dependent IL-6 secretion in THP-1 cells, and conditioned media dose-dependently upregulated CEA expression in CRC cells—an effect attenuated by NF-κB inhibition or IL-6 neutralization. Multi-omics analysis identified 42 overlapping genes linking silica exposure to CRC, with enrichment in cytokine signaling, adhesion, and matrix remodeling pathways. A hub gene–based risk score was significantly associated with overall survival. scRNA-seq analysis revealed elevated expression of inflammation- and adhesion-related genes in tumor-associated macrophages. Conclusions Occupational silica exposure induces macrophage-driven inflammatory signaling that promotes early CEA elevation and TME remodeling. Integrating machine learning with experimental and multi-omics validation provides a translational framework for identifying exposure-responsive biomarkers and immune-related cancer risk in occupational settings.

Ask AI
Helpful
Bookmark
Share
View Full Paper

Cite This Study

Hao et al. (2026) studied this question.

synapsesocial.com/papers/69a91d55d6127c7a504c00dehttps://doi.org/10.3389/fimmu.2026.1775236
Ask AI
Helpful
Bookmark
Share
View Full Paper