Machine learning-driven identification of exosome-related biomarkers in head and neck squamous cell carcinoma - Research article

Aug 4, 2026

Upload by Rowan

Head and neck squamous cell carcinoma (HNSCC) remains associated with high morbidity and mortality, while current diagnosis and risk stratification are still limited, particularly in advanced disease. This study leverages public gene-expression data and ComBat batch-effect correction to isolate exosome-related differentially expressed genes, then applies LASSO, SVM-RFE, and random forest to select 10 key candidate biomarkers. Functional enrichment, immune microenvironment profiling, and molecular docking further clarify biological roles and therapeutic implications. Results support accurate prognostic prediction and suggest immune-evasion mechanisms.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages16

Reading Time40 min

AccessFree to download

Background

Exosomes in tumor biology and biomarker potential

Methods

Data sources and batch-effect correction

Gene screening and machine-learning feature selection

Functional enrichment, immune profiling, and molecular docking

Results

Differential expression and biomarker performance

Immune microenvironment associations

Docking-based therapeutic target hints

Conclusion

Exosome-related indicators and clinical value

TYPE Original Research PUBLISHED 22 May 2025 DOI 10.3389/fimmu.2025.1590331 OPEN ACCESS EDITED BY Xingchen Peng, Sichuan University, China REVIEWED BY Deepak Parashar, Medical College of Wisconsin, United States Ran Chen, Anhui Medical University, China Geet Madhukar, University of New Hampshire, United States *CORRESPONDENCE Huan Li [cf250srfmmu@163.com](cf250srfmmu@163.com)[ ](cf250srfmmu@163.com)Jianhua Wei [64265919@qq.com](64265919@qq.com) †These authors have contributed equally to this work RECEIVED 09 March 2025 ACCEPTED 06 May 2025 PUBLISHED 22 May 2025 CITATION He Y, Li Y, Tang J, Wang Y, Zhao Z, Liu R, Yang Z, Li H and Wei J (2025) Machine learning-driven identiﬁcation of exosomerelated biomarkers in head and necksquamous cell carcinoma. Front. Immunol. 16:1590331 . doi: 10.3389/fimmu.2025.1590331 COPYRIGHT © 2025 He, Li, Tang, Wang, Zhao, Liu, Yang, Li and Wei. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY) . The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms. Machine learning-driven identiﬁcation of exosomerelated biomarkers in head and neck squamous cell carcinoma Yaodong He †, Yun Li †, Jiaqi Tang †, Yan Wang, Zhenyan Zhao, Rong Liu, Zihui Yang, Huan Li* and Jianhua Wei* State Key Laboratory of Oral and Maxillofacial Reconstruction and Regeneration, National Clinical Research Center for Oral Diseases, Shaanxi Clinical Research Center for Oral Diseases, Department of Oral and Maxillofacial Surgery, School of Stomatology, The Fourth Military Medical University, Xi’an, China Background: Head and neck squamous cell carcinoma (HNSCC) is a common cancer associated with elevated mortality rates. Exosomes, diminutive extracellular vesicles, signiﬁcantly contribute to tumour development, immunological evasion, and treatment resistance. Identifying exosome-associated biomarkers in HNSCC may improve early diagnosis, treatment targeting, and patient classiﬁcation. Methods: We acquired four publically accessible HNSCC gene expression datasets from the Gene Expression Omnibus (GEO) database and mitigated batch effects utilising the ComBat technique. Differential expression analysis and exosome-related gene screening found a collection of markedly exosome-associated differentially expressed genes (ERDEGs) . Subsequently, 10 key exosome-related genes were further screened by combining three machine learning methods, LASSO regression, SVM-RFE and RF, and a clinical prediction model was constructed. Furthermore, we thoroughly investigated the biological roles of these genes in HNSCC and their prospective treatment implications via functional enrichment analysis, immune microenvironment assessment, and molecular docking conﬁrmation. Results: The study indicated that 10 pivotal exosome-related genes identiﬁed by the machine learning method had considerable differential expression in HNSCC. Clinical prediction models developed from these genes have shown high accuracy in prognostic evaluations of HNSCC patients. Analysis of the immunological microenvironment indicated varying immune cell inﬁltration in HNSCC, and the association with ERDEGs proposed a potential mechanism for immune evasion. Molecular docking validation indicated novel small molecule medicines targeting these genes, establishing a theoretical foundation for pharmacological therapy in HNSCC. Conclusion: This research identiﬁes new exosome-related indicators for HNSCC through machine learning methodologies. The suggested biomarkers, particularly ANGPTL1, exhibit signiﬁcant promise for diagnostic and prognostic uses. The investigation of the immunological microenvir

FAQs

What problem does the study address in head and neck squamous cell carcinoma?

It targets the need for better biomarkers for early diagnosis, prognosis, patient classification, and therapeutic targeting in HNSCC, where outcomes remain unfavorable especially in advanced stages.

How were exosome-related biomarkers identified?

Four public HNSCC gene-expression datasets were collected and batch effects were corrected with ComBat. Exosome-related differentially expressed genes were identified, and three machine-learning approaches (LASSO, SVM-RFE, and random forest) were used to screen 10 key genes for a clinical prediction model.

What evidence supports the clinical and biological relevance of the selected genes?

The selected genes showed distinct differential expression in HNSCC, clinical prediction models demonstrated high prognostic accuracy, immune microenvironment analysis suggested potential immune-evasion mechanisms, and molecular docking indicated small-molecule candidates targeting these genes.

Aug 4, 2026

Upload by Rowan