Differential Metabolomic Biomarkers Analysis of Apheresis Platelet Storage Based on Machine Learning Models

ABSTRACT:

Platelet storage lesion (PSL) is a critical factor limiting the efficacy of platelet transfusion. Although hypothermic (4°C) storage is considered beneficial for maintaining platelet function, the underlying mechanisms of systemic metabolic reprogramming remain incompletely understood. This study aims to systematically characterize the metabolic evolution of platelets during storage at different temperatures and screen for key quality biomarkers using untargeted metabolomics combined with machine learning. Platelet samples were stored at 4°C and 22°C, respectively, and collected at various time points (Days 0–7). Untargeted metabolomics analysis was performed using liquid chromatography-mass spectrometry (LC-MS). Principal component analysis (PCA) was used for quality control. Pathway enrichment analysis was conducted using KEGG, HMDB, and Lipidmaps databases. Twenty-two machine learning algorithms were employed to construct classification models to distinguish storage conditions, with robustness verified by ablation experiments. Storage temperature significantly reshapes the purine, energy and lipid metabolic trajectories of platelets. Storage at 22°C induces metabolic homeostasis imbalance, whereas 4°C hypothermia helps maintain platelet metabolic homeostasis by inhibiting excessive metabolic consumption and the accumulation of oxidative stress precursors. This study not only confirms the advantages of hypothermic storage from a metabolic perspective but also identifies potential biomarkers for clinical platelet quality assessment.

INSTRUMENT(S): Liquid Chromatography MS - negative - reverse-phase, Liquid Chromatography MS - positive - reverse-phase

PROVIDER: MTBLS15165 | MetaboLights | 2026-07-27

REPOSITORIES: MetaboLights