Distinguish ing Intracerebral Hemorrhage from Acute Cerebral Infarction through Metabolomics

Xuxin Zhang; Li Yanzhao; Yan Liang; Pitong Sun; Xue Wu; Jinhui Song; Xiaoyu Sun; Mo Hong; Peng Gao; Dongfeng Deng

Ayuda

Distinguish ing Intracerebral Hemorrhage from Acute Cerebral Infarction through Metabolomics

Xuxin Zhang ; Li Yanzhao ; Yan Liang ; Pitong Sun ; Xue Wu ^[1] ; Jinhui Song ; Xiaoyu Sun ^[1] ; Mo Hong ^[1] ; Peng Gao ^[1] ; Dongfeng Deng
1. [1] First Affiliated Hospital of Dalian Medical University
  
  First Affiliated Hospital of Dalian Medical University
  
  China
Localización: Revista de investigación clínica, ISSN 0034-8376, ISSN-e 2564-8896, Vol. 69, Nº. 4 6, 2017, págs. 319-328
Idioma: inglés
Enlaces
- Texto completo (pdf)
Resumen
- Background: Acute cerebral infarction (ACI) and intracerebral hemorrhage (ICH) are potentially lethal cerebrovascular diseases that seriously impact public health. ACI and ICH share several common clinical manifestations but have totally divergent therapeutic strategies. A poor diagnosis can affect stroke treatment. Objective: To screen for biomarkers to differentiate ICH from ACI, we enrolled 129 ACI and 128 ICH patients and 65 healthy individuals as controls. Methods: Patients with stroke were diagnosed by computed tomography/magnetic resonance imaging, and their blood samples were obtained by fingertip puncture within 2-12 h after stroke initiation. We compared changes in metabolites between ACI and ICH usingdried blood spot-based direct infusion mass spectrometry technology for differentiating ICH from ACI. Results: Through multivariate statistical approaches, 11 biomarkers including 3-hydroxylbutyrylcarnitine, glutarylcarnitine (C5DC), myristoylcarnitine, 3-hydroxypalmitoylcarnitine, tyrosine/citrulline (Cit), valine/phenylalanine, C5DC/3-hydroxyisovalerylcarnitine, C5DC/palmitoylcarnitine, hydroxystearoylcarnitine, ratio of sum of C0, C2, C3, C16, and C18:1 to Cit, and propionylcarnitine/methionine were screened. An artificial neural network model was constructed based on these parameters. A training set was evaluated by cross-validation method. The accuracy of this model was checked by an external test set showing a sensitivity of 0.8400 (95% confidence interval [CI], 0.7394-0.9406) and specificity of 0.7692 (95% CI, 0.6536-0.8848). Conclusion: This study confirmed that metabolomic analysis is a promising tool for rapid and timely stroke differentiation and prediction based on differential metabolites.