Abstract
This study explored the molecular patterns and diagnostic biomarkers associated with cuproptosis in cardioembolic stroke (CES) using bioinformatics tools. GSE58294 expression profile data were downloaded from the Gene Expression Synthesis Database as a training dataset, and cuproptosis-related genes were extracted for analysis. We identified differentially expressed cuproptosis-associated genes (DECAGs) between CES and control samples. In total, 11 DECAGs (MTF1, NFE2L2, DLD, ATP7B, ATP7A, SLC31A1, PDHA1, CDKN2A, DLS, DLAT, PDHB) and activated immune responses differed significantly between CES patients and non-CES controls. Two molecular clusters associated with cuproptosis were discerned in CES. Immunoinfiltration analysis revealed significant immunoheterogeneity among the different molecular clusters, and Cluster1 showed a relatively high level of immunoinfiltration. Weighted gene co-expression network analysis revealed 819 key genes related to CES, 320 key genes related to CES typing, and 37 core intersection genes were identified by Venn analysis to construct a training model. The generalized linear model showed satisfactory performance based on 2 external validation datasets. Within this model, 5 genes (FLT3LG, MAL, TNFRSF25, CASP5, and MAN1C1) were identified as the most significantly associated with CES characteristics. Clinical application columns were established based on the expression levels of 5 CES characteristic genes. Decision curve analysis and correction curves showed that the results had good prediction accuracy. The present findings highlight the crucial role of cuproptosis in the development and diagnosis of CES, indicating its potential as a key factor in understanding and identifying CES.