Subtyping strokes using blood‐based protein biomarkers: A high‐throughput proteomics and machine learning approach
Shubham Misra,Praveen Singh,Shantanu Sengupta,Manoj Kushwaha,Zuhaibur Rahman,Divya Bhalla,Pumanshi Talwar,Manabesh Nath,Rahul Chakraborty,Pradeep Kumar,Amit Kumar,Praveen Aggarwal,Achal K. Srivastava,Awadh K. Pandit,Dheeraj Mohania,Kameshwar Prasad,Nishant K. Mishra,Deepti Vibha
DOI: https://doi.org/10.1111/eci.14372
2024-12-12
European Journal of Clinical Investigation
Abstract:In this study we employed high‐throughput proteomics and machine learning approaches on serum samples obtained from 340 stroke patients and identified 20 proteins that correctly classified IS from ICH, of which three protein biomarkers including GFAP, MMP‐9 and APOC‐1 along with clinical variables differentiated IS from ICH within 24 h with an accuracy of 92% and a very high negative predictive value of 94%. Background Rapid diagnosis of stroke and its subtypes is critical in early stages. We aimed to discover and validate blood‐based protein biomarkers to differentiate ischemic stroke (IS) from intracerebral haemorrhage (ICH) using high‐throughput proteomics. Methods We collected serum samples within 24 h from acute stroke (IS & ICH) and mimics patients. In the discovery phase, SWATH‐MS proteomics identified differentially expressed proteins, which were validated using targeted proteomics in the validation phase. We conducted interaction network and pathway analyses using Cytoscape 3.10.0. We determined cut‐off points using the Youden Index. We developed three prediction models using multivariable logistic regression analyses. We assessed the model performance using statistical tests. Results We included 20 IS and 20 ICH in the discovery phase and 150 IS, 150 ICH, and six stroke mimics in the validation phase. We quantified 375 proteins using SWATH‐MS. Between IS and ICH, we discovered 20 differentially expressed proteins. In the validation phase, the combined prediction model including three biomarkers: GFAP (aOR 0.04; 95%CI .02–.11), MMP‐9 (aOR .09; .03–.28), APO‐C1 (aOR 5.76; 2.66–12.47) and clinical variables independently differentiated IS from ICH (accuracy: 92%, negative predictive value: 94%). Adding biomarkers to clinical variables improved discrimination by 26% (p
medicine, general & internal, research & experimental