Nai-Fang Chi1, Tzu-Hao Chang2, Chen-Yang Lee3, Yu-Wei Wu3, Ting-An Shen3, Lung Chan4, Yih-Ru Chen5, Hung-Yi Chiou5, Chung Y Hsu6, Chaur-Jong Hu7. 1. Department of Neurology, Shuang Ho Hospital, Taipei Medical University, New Taipei City, Taiwan; Department of Neurology, School of Medicine, College of Medicine, Taipei Medical University, Taipei, Taiwan; Department of Neurology, Faculty of Medicine, School of Medicine, National Yang-Ming University, Taipei, Taiwan; Department of Neurology, Neurological Institute, Taipei Veterans General Hospital, Taipei, Taiwan. 2. Graduate Institute of Biomedical Informatics, Taipei Medical University, Taipei, Taiwan; Clinical Big Data Research Center, Taipei Medical University Hospital, Taipei, Taiwan. 3. Graduate Institute of Biomedical Informatics, Taipei Medical University, Taipei, Taiwan. 4. Department of Neurology, Shuang Ho Hospital, Taipei Medical University, New Taipei City, Taiwan; Department of Neurology, School of Medicine, College of Medicine, Taipei Medical University, Taipei, Taiwan. 5. School of Public Health, College of Public Health, Taipei Medical University, Taipei, Taiwan. 6. Department of Neurology, China Medical University Hospital, Taichung, Taiwan; Graduate Institute of Clinical Medical Science, China Medical University, Taichung, Taiwan. 7. Department of Neurology, Shuang Ho Hospital, Taipei Medical University, New Taipei City, Taiwan; Department of Neurology, School of Medicine, College of Medicine, Taipei Medical University, Taipei, Taiwan; The Ph.D. Program for Neural Regenerative Medicine, College of Medical Science and Technology, Taipei Medical University, Taipei, Taiwan; Taipei Neuroscience Institute, Taipei Medical University, Taipei, Taiwan. Electronic address: chaurjongh@tmu.edu.tw.
Abstract
BACKGROUND/ PURPOSE: Metabolites in blood have been found associated with the occurrence of vascular diseases, but its role in the functional recovery of stroke is unclear. The aim of this study is to investigate whether the untargeted metabolomics at the acute stage of ischemic stroke is able to predict functional recovery. METHODS: One hundred and fifty patients with acute ischemic stroke were recruited and followed up for 3 months. Fasting blood samples within 7 days of stroke were obtained, liquid chromatography and mass spectrometry were applied to identify outcome-associated metabolites. The patients' clinical characteristics and identified metabolites were included for constructing the outcome prediction model using machine learning approaches. RESULTS: By using multivariate analysis, 220 differentially expressed metabolites (DEMs) were discovered between patients with favorable outcomes (modified Rankin Scale, mRS ≤ 2 at 3 months, n = 77) and unfavorable outcomes (mRS ≥ 3 at 3 months, n = 73). After feature selection, 63 DEMs were chosen for constructing the outcome prediction model. The predictive accuracy was below 0.65 when including patients' clinical characteristics, and could reach 0.80 when including patients' clinical characteristics and 63 selected DEMs. The functional enrichment analysis identified platelet activating factor (PAF) as the strongest outcome-associated metabolite, which involved in proinflammatory mediators release, arachidonic acid metabolism, eosinophil degranulation, and production of reactive oxygen species. CONCLUSION: Metabolomics is a potential method to explore the blood biomarkers of acute ischemic stroke. The patients with unfavorable outcomes had a lower PAF level compared to those with favorable outcomes.
BACKGROUND/ PURPOSE: Metabolites in blood have been found associated with the occurrence of vascular diseases, but its role in the functional recovery of stroke is unclear. The aim of this study is to investigate whether the untargeted metabolomics at the acute stage of ischemic stroke is able to predict functional recovery. METHODS: One hundred and fifty patients with acute ischemic stroke were recruited and followed up for 3 months. Fasting blood samples within 7 days of stroke were obtained, liquid chromatography and mass spectrometry were applied to identify outcome-associated metabolites. The patients' clinical characteristics and identified metabolites were included for constructing the outcome prediction model using machine learning approaches. RESULTS: By using multivariate analysis, 220 differentially expressed metabolites (DEMs) were discovered between patients with favorable outcomes (modified Rankin Scale, mRS ≤ 2 at 3 months, n = 77) and unfavorable outcomes (mRS ≥ 3 at 3 months, n = 73). After feature selection, 63 DEMs were chosen for constructing the outcome prediction model. The predictive accuracy was below 0.65 when including patients' clinical characteristics, and could reach 0.80 when including patients' clinical characteristics and 63 selected DEMs. The functional enrichment analysis identified platelet activating factor (PAF) as the strongest outcome-associated metabolite, which involved in proinflammatory mediators release, arachidonic acid metabolism, eosinophil degranulation, and production of reactive oxygen species. CONCLUSION: Metabolomics is a potential method to explore the blood biomarkers of acute ischemic stroke. The patients with unfavorable outcomes had a lower PAF level compared to those with favorable outcomes.