Li Q, Tian Y, Niu J, Guo E, Lu Y, Dang C, Feng L, Li L, Wang L. Identification of diagnostic signatures for ischemic stroke by machine learning algorithm.
J Stroke Cerebrovasc Dis 2024;
33:107564. [PMID:
38215553 DOI:
10.1016/j.jstrokecerebrovasdis.2024.107564]
[Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [MESH Headings] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 09/14/2023] [Revised: 12/25/2023] [Accepted: 01/07/2024] [Indexed: 01/14/2024] Open
Abstract
OBJECTIVE
Ischemic stroke (IS) is one of the major diseases threatening human health and survival and a leading cause of acquired mortality and disability in adults. The aim of this study was to screen diagnostic features of IS and to explore the characteristics of immune cell infiltration in IS pathogenesis.
METHODS
The microarray data of IS (GSE16561, GSE58294, GSE37587, and GSE124026) in the GEO database were merged after removing the batch effect. Then integrated bioinformatic analysis and machine-learning strategies were adopted to analyze the functional correlation and select diagnostic signatures. The WGCNA was used to identify the co-expression modules related to IS. The CIBERSORT algorithm was performed to assess the inflammatory state of IS and to investigate the correlation between diagnostic signatures and infiltrating immune cells.
RESULTS
Functional analysis of dysregulated genes showed that immune response-regulating signaling pathway and pattern recognition receptor activity were enriched in the pathophysiology of IS. The turquoise module was identified as the significant module with IS. By using Lasso and SVM-RFE learning methods, we finally obtained four diagnostic genes, including LAMP2, CR1, CLEC4E, and F5. The corresponding results of AUC of ROC prediction model in training and validation cohort were 0.954 and 0.862, respectively. The immune cell infiltration analysis suggested that plasma cells, resting and activated NK cells, activated dendritic cells, memory B cells, CD8+ T cells, naïve CD4+ T cells, and resting mast cells may be involved in the development of IS. Additionally, these diagnostic signatures might be correlated with multiple immune cells in varying degrees.
CONCLUSION
We identified four biologically relevant genes (LAMP2, CR1, CLEC4E, and F5) with diagnostic effects for IS, our results further provide novel insights regarding molecular mechanisms associated with various immune cells that related to IS for future investigations.
Collapse