Machine learning algorithms develop a tumor-educated platelets-related gene signature to predict colorectal cancer prognosis and therapy response

  • iScience. 2026 Jun 17;29(7):116229. doi: 10.1016/j.isci.2026.116229.
Yueqiong Lao  1  2  3 ,  Jie Yang  1  2  4 ,  Minli Hu  3  1 ,  Jing Li  3  1 ,  Weikang Xu  3  1 ,  Jiahui Xu  3  1 ,  Huan Wang  3  1 ,  Hechenhao Jiang  3  1 ,  Zhihao Pei  3  1 ,  Xinyi Qiu  1 ,  Kunyuan Wang  3  1 ,  Xuan Li  5  6 ,  Hui Yang  3  1
Affiliations
  • 1. Department of Gastroenterology, The Second Affiliated Hospital of Guangzhou Medical University, Guangzhou 510245, China.
  • 2. Department of Oncology, Xiangya Hospital, Central South University, Changsha 410008, China.
  • 3. Guangzhou Institute of Cardiovascular Disease, The Second Affiliated Hospital of Guangzhou Medical University, Guangzhou 510245, China.
  • 4. Department of Dermatology, Xiangya Hospital, Central South University, Changsha 410008, China.
  • 5. Guangzhou Key Laboratory for Research and Development of Nano-Biomedical Technology for Diagnosis and Therapy&Guangdong Provincial Education Department Key Laboratory of Nano-Immunoregulation Tumour Microenvironment, Department of Oncology, The Second Affiliated Hospital of Guangzhou Medical University, Guangzhou 510245, China.
  • 6. Department of Oncology, The First Affiliated Hospital, Hengyang Medical School, University of South China, Hengyang, Hunan 421001, China.
PMID: 42375537 DOI: 10.1016/j.isci.2026.116229
Abstract

Tumor-educated platelets (TEPs) have recently emerged as an important component of liquid biopsy, yet the clinical relevance in colorectal Cancer (CRC) remains unclear. Here, we employed 10 machine learning algorithms to develop a stable, accurate TEP-related gene signature (TEPGS) to explore its links to tumor-associated macrophages (TAMs) and spatial platelet abundance. TEPGS correlated strongly with poor prognosis and outperformed 71 published gene signatures in predicting CRC overall survival. Multi-omics analysis displayed that high TEPGs were marked by increased TP53 mutations, copy number alterations, diminished immune features, enrichment of pro-tumor SPP1+/FCN1+ TAMs, and elevated spatial platelet abundance. Patients with high TEPGS exhibited resistance to immunotherapy but responded to a BRAF V600E inhibitor, while TEPGS showed tentative value for predicting cetuximab response and preliminary utility for bevacizumab. Functional assays confirmed ARPC1B as an oncogene. Our findings establish TEPGS as a valuable biomarker for prognostic stratification and tailored therapy selection in CRC.

Keywords
Cancer; Computing methodology; Immunology; Machine learning.
Products