Machine learning-based radiomics analysis on contrast-enhanced MRI for prediction of cytotoxic T-lymphocyte associated protein 4 expression in patients with breast cancer
Cancer Treatment and Research Communications · 2025
DOI: 10.1016/j.ctarc.2025.101017 ↗Kurz gefasst
Fachartikel in der Zeitschrift Cancer Treatment and Research Communications, 2025. Verfasst von Xiaojie Wang und 6 weiteren Autor:innen. Beteiligte Einrichtung: First Affiliated Hospital of Dalian Medical University, China. Hauptthemen laut MEDLINE: Breast Neoplasms, Magnetic Resonance Imaging, CTLA-4 Antigen, Machine Learning. Die Arbeit nennt 31 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY-NC-ND 4.0).
Abstract (Originalsprache)
BACKGROUND: Cytotoxic T-lymphocyte Associated Protein 4 (CTLA4) has been considered as a well-recognized immune checkpoint. The present study aimed to establish radiomics-based models to predict the CTLA4 expression level in patients with breast cancer (BRCA). METHODS: Clinical, transcription, and magnetic resonance imaging (MRI) data were collected from The Cancer Genome Atlas and The Cancer Imaging Archive databases. Then, the relationship between the expression level of CTLA4 and survival prognosis was explored using differential analysis, survival analysis, multivariate and univariate Cox analysis, difference analysis of immune cell infiltration degree, and enrichment analysis. Afterwards, radiomics-based prediction models for the CTLA4 expression were constructed through feature extraction via the PyRadiomics package, feature selection via the maximum-relevance minimum-redundancy recursive feature elimination algorithm, and model construction via the support vector machine (SVM) and logistic regression (LR) classifiers. RESULTS: The patients were classified into two groups based on the optimal cut-off value for the CTLA4 level: low-expression and high-expression groups. The CTLA4 expression level, as an independent predictor for overall survival (OS), was associated with T cell infiltration (P < 0.001), and was mainly correlated to transcriptional signal activation, the cell cycle and apoptosis in the enrichment analysis. A total of 107 radiomics features were extracted, and five features were selected with importance for model construction. The SVM-based model indicated an area under the receiver operating characteristic curve (AUC) of 0.701 and 0.688 for the training and testing cohorts, respectively, while the LR-based model indicated an AUC of 0.0.834 and 0.738 for the training and testing cohorts, respectively. Breast cancer tissues were recruited to immunohistochemical verification for CTLA4 expression. CONCLUSION: The radiomics-based models constructed on enhanced MRI data to predict the CTLA4 expression exhibited good performance in BRCA patients. These may be employed to instruct individualized immunotherapy before surgery.
Schlagwörter
MeSH-Schlagwörter (MEDLINE)
Stichwörter der Autor:innen
Substanzen
Autor:innen und Institutionen
- First Affiliated Hospital of Dalian Medical University, China.
- First Affiliated Hospital of Dalian Medical University, China. Electronic address: xueer_128@163.com.
- First Affiliated Hospital of Dalian Medical University, China. Electronic address: hjwang@dmu.edu.cn.
Zitieren
APA
Wang, X., Xiao, H., Zhang, L., Ning, N., Qv, H., Gao, X., & Wang, H. (2025). Machine learning-based radiomics analysis on contrast-enhanced MRI for prediction of cytotoxic T-lymphocyte associated protein 4 expression in patients with breast cancer. Cancer Treatment and Research Communications. https://doi.org/10.1016/j.ctarc.2025.101017
Vancouver
Wang X, Xiao H, Zhang L, Ning N, Qv H, Gao X, et al. Machine learning-based radiomics analysis on contrast-enhanced MRI for prediction of cytotoxic T-lymphocyte associated protein 4 expression in patients with breast cancer. Cancer Treatment and Research Communications. 2025. doi:10.1016/j.ctarc.2025.101017
BibTeX
@article{Wang2025,
author = {Wang, Xiaojie and Xiao, Huiyu and Zhang, Lina and Ning, Ning and Qv, Huixian and Gao, Xue and Wang, Hongjiang},
title = {Machine learning-based radiomics analysis on contrast-enhanced MRI for prediction of cytotoxic T-lymphocyte associated protein 4 expression in patients with breast cancer},
journal = {Cancer Treatment and Research Communications},
year = {2025},
doi = {10.1016/j.ctarc.2025.101017}
}
EUVIMED