Background and Objective: Brain stroke (caused by interrupted blood supply to the brain) is one of the most common causes of disability and is the second highest cause of death in the world. Early diagnosis and treatment raises the chance of surviving a stroke, and can result in little or no disability. Treatment is based on the type of stroke and its primary cause; for ischaemic stroke (IS), medication or surgery or both can be recommended, whereas, for haemorrhagic stroke (HS, also known as intracerebral hemorrhage), surgery is recommended. However, there are scarcity of clinical markers which can differentiate IS from HS and further can provide information about severity of cerebral ischemia, endothelial injury, blood–brain barrier disruption, atherosclerosis, thrombus formation, inflammation, and oxidative stress. Metabolomics is a promising approach for identification of metabolic biomarkers and alterations associated with the diseased biology The present study is an effort to identify the diagnostic panel of serum metabolic profiles to differentiate between IS and HS patients and further could help predicting the severity of the cerebral ischemia. Methods: Blood serum samples were collected from suspected brain stroke patients at hospital admission within 6 hours after onset. The serum samples of patients clinically confirmed for IS and HS were only analysed using 800 MHz NMR spectroscopy. For comparative evaluation, the serum samples from 60 age and sex matched normal control (NC) subjects were also obtained for NMR analysis. The partial least square-discriminant analysis (PLS-DA) was performed to confirm the serum metabolic disparity between the study groups and metabolic features of discriminatory relevance were identified making composite use of mean decrease accuracy (MDA) score values estimated for metabolic features employing machine learning random forest (RF) classification analysis. The diagnostic potential of discriminatory metabolites were finally evaluated using receiver operating characteristic (ROC) curve analysis. Results: The serum metabolic profiles of total 107 brain stroke patients (48 IS and 60 HS) were measured and compared with those of 60 age and sex matched normal control (NC) subjects. Compared to NC, the serum levels of various amino acids (histidine, glutamine, threonine, alanine, glycine, dimethylglycine, pyruvate, etc.) were found to be decreased in both HS and IS patients, whereas those of creatine, mannitol, mannose, malonate and organic acids (such as acetate, 3-hydroxybutyrate and 3-hydroxy-isobutyrate) were found to be increased in the sera of brain stroke patients. However, the area under ROC (AUROC) curve values for majority of metabolic features found to be less than 0.8 for the discrimination between IS and HS suggesting that circulatory metabolic features lack sufficient sensitivity and specificity. Conclusions: The study provided primary evidence that various circulatory metabolites (including histidine, glutamine, and alanine) exhibited strong diagnostic potential for distinguishing brain stroke condition from NC subject, however, poor diagnostic accuracy in distinguishing IS from HS. Nevertheless, these metabolic changes can be used to assess the severity of brain stroke condition and monitoring patient response to treatment in emergency settings.

The identified NMR based serum metabolic changes can be used to assess the disease severity and monitoring patient response to treatment in emergency settings.