A plasma metabolomics workflow for breast cancer detection using quantitative GC/MS and machine learning
- 1MS Business Unit
Abstract
Blood-based metabolomic profiling has been widely investigated for breast cancer (BC) detection; however, clinical implementation remains limited due to variability in sample handling, analytical reproducibility, and overfitting during statistical analysis. We established a plasma GC/MS metabolomics workflow for discriminating BC from healthy controls (HC) using conventional machine-learning algorithms. Plasma samples (n = 360; BC = 180, HC = 180) were collected prospectively under standardized preanalytical conditions before surgery and the initiation of systematic anticancer therapy and analyzed using a quantitative GC/MS platform with automated derivatization. Feature selection and model development were conducted using three machine-learning (ML) algorithms (Lasso logistic regression (LR), random forest classifier (RFC), and support vector machine (SVM)). A total of 45 metabolite candidate biomarkers were identified, and the optimal number of metabolite features for each algorithm was estimated by a recursive feature elimination (RFE)-based strategy. The best-performing models achieved area under the ROC curve values (AUC) of 0.910 (LR), 0.893 (RFC), and 0.843 (SVM). We selected prioritizing candidate biomarkers consistently expressed across the multi-algorithm pipeline. A bagging ensemble model improved stability (AUC = 0.911) and reduced false-positive predictions in the independent HC dataset. In addition, model stability with respect to false-positive predictions was assessed using an independent HC cohort (n = 15) that was collected at a separate institution. These results indicate that a plasma metabolomics workflow combined with conventional multi-algorithm ML, algorithm-specific feature selection, and independent assessment provides stable discrimination between BC and HC in a moderately sized cohort.
Related
- Exploratory Integration of Metabolomics and Machine Learning Identifies Candidate Signatures of Spontaneous Preterm Birth
- Metabolic Profiling, Antioxidant Activity, and Mechanistic Insights of Chicory from Northern and Southern China: Integrating Metabolomics, Machine Learning, Molecular Docking, Molecular Dynamics Simulations, and In Vitro Assays
- Beyond Traditional Covariates: An Interpretable Machine Learning Workflow for Improved Hybrid Pharmacometric Modeling
- Short-Term Molecular Assessment of Peripheral Blood Transcriptomics and Urinary Metabolomics After Switching From Cigarettes to Heated Tobacco Products Using Machine Learning and Bayesian Network Modeling
- Machine learning quantitative OCT derived fibrosis indicators for corneal haze evaluation
BibTeX
@misc{ojima2026plasma,
title = {{A plasma metabolomics workflow for breast cancer detection using quantitative GC/MS and machine learning}},
author = {N. Ojima and T. Yokoyama and S. Nishiumi and S. Kikuchi and K. Yoneda and M. Kimura and K. Tanaka and M. Morimoto and A. Miyazaki and T. Tomita and M. Ohyanagi and M. Nagahashi and Y. Miyoshi and M. Ikeda},
howpublished = {bioRxiv},
year = {2026},
month = sep,
doi = {10.64898/2026.09.23.753710},
url = {https://www.biorxiv.org/content/10.64898/2026.09.23.753710v1},
note = {CC BY},
}
RIS
TY - UNPB
TI - A plasma metabolomics workflow for breast cancer detection using quantitative GC/MS and machine learning
AU - N. Ojima
AU - T. Yokoyama
AU - S. Nishiumi
AU - S. Kikuchi
AU - K. Yoneda
AU - M. Kimura
AU - K. Tanaka
AU - M. Morimoto
AU - A. Miyazaki
AU - T. Tomita
AU - M. Ohyanagi
AU - M. Nagahashi
AU - Y. Miyoshi
AU - M. Ikeda
PY - 2026
DA - 2026/09/23
PB - bioRxiv
DO - 10.64898/2026.09.23.753710
UR - https://www.biorxiv.org/content/10.64898/2026.09.23.753710v1
AB - Blood-based metabolomic profiling has been widely investigated for breast cancer (BC) detection; however, clinical implementation remains limited due to variability in sample handling, analytical reproducibility, and overfitting during statistical analysis. We established a plasma GC/MS metabolomics workflow for discriminating BC from healthy controls (HC) using conventional machine-learning algorithms. Plasma samples (n = 360; BC = 180, HC = 180) were collected prospectively under standardized preanalytical conditions before surgery and the initiation of systematic anticancer therapy and analyzed using a quantitative GC/MS platform with automated derivatization. Feature selection and model development were conducted using three machine-learning (ML) algorithms (Lasso logistic regression (LR), random forest classifier (RFC), and support vector machine (SVM)). A total of 45 metabolite candidate biomarkers were identified, and the optimal number of metabolite features for each algorithm was estimated by a recursive feature elimination (RFE)-based strategy. The best-performing models achieved area under the ROC curve values (AUC) of 0.910 (LR), 0.893 (RFC), and 0.843 (SVM). We selected prioritizing candidate biomarkers consistently expressed across the multi-algorithm pipeline. A bagging ensemble model improved stability (AUC = 0.911) and reduced false-positive predictions in the independent HC dataset. In addition, model stability with respect to false-positive predictions was assessed using an independent HC cohort (n = 15) that was collected at a separate institution. These results indicate that a plasma metabolomics workflow combined with conventional multi-algorithm ML, algorithm-specific feature selection, and independent assessment provides stable discrimination between BC and HC in a moderately sized cohort.
ER -