Immune-related potential biomarkers and therapeutic targets in coronary artery disease

Abstract

Background: Coronary artery disease (CAD) is a complex illness with unknown pathophysiology. Peripheral biomarkers are a non-invasive method required to track the onset and progression of CAD and have unbeatable benefits in terms of early identification, prognostic assessment, and categorization of the diagnosis. This study aimed to identify and validate the diagnostic and therapeutic potential of differentially expressed immune-related genes (DE-IRGs) in CAD, which will aid in improving our knowledge on the etiology of CAD and in forming genetic predictions.

Methods: First, we searched coronary heart disease in the Gene Expression Omnibus (GEO) database and identified GSE20680 (CAD = 87, Normal = 52) as the trial set and GSE20681 (CAD = 99, Normal = 99) as the validation set. Functional enrichment analysis using protein-protein interactions (PPIs), Gene Ontology (GO), and Kyoto Encyclopedia of Genes and Genomes (KEGG) was carried out on the identified differentially expressed genes. Optimal feature genes (OFGs) were generated using the support vector machine recursive feature elimination algorithm and the least absolute shrinkage and selection operator (LASSO) algorithm. Furthermore, immune infiltration in CAD patients and healthy controls was compared using CIBERSORT, and the relationship between immune cells and OFGs was examined. In addition, we constructed potential targeted drugs for this model through the Drug-Gene Interaction database (DGIdb) database. Finally, we verify the expression of S100A8-dominated OFGs in the GSE20681 dataset to confirm the universality of our study.

Results: We identified the ten best OFGs for CAD from the DE-IRGs. Functional enrichment analysis showed that these marker genes are crucial for receptor-ligand activity, signaling receptor activator activity, and positive control of the response to stimuli from the outside world. Additionally, CIBERSORT revealed that S100A8 could be connected to alterations in the immune microenvironment in CAD patients. Furthermore, with the help of DGIdb and Cytoscape, a total of 64 medicines that target five marker genes were subsequently discovered. Finally, we verified the expression of the OFGs genes in the GSE20681 dataset between CAD patients and normal patients and found that there was also a significant difference in the expression of S100A8.

Conclusion: We created a 10-gene immune-related prognostic model for CAD and confirmed its validity. The model can identify potential biomarkers for CAD prediction and more accurately gauge the progression of the disease.

Keywords: Gene Expression Omnibus (GEO); bioinformatics analysis; coronary artery disease (CAD); immune-related genes (IRGs); optimal feature genes (OFGs).

FIGURE 1

The network of this article.

FIGURE 2

Analysis of DE-IRGs. (A) Volcano plot of the DE-IRGs in CAD. (B) PPI network. (C) GO enrichment analysis. (D) KEGG enrichment analysis.

FIGURE 3

Ten DE-IRGs were shown to be CAD diagnostic genes. (A,B) Using the LASSO logistic regression algorithm, with penalty parameter tuning conducted by 10-fold cross-validation, 22 CAD-related features were selected. (C,D) The SVM-RFE algorithm was used to filter the 10 DE-IRGs to identify the optimal combination of feature genes. Finally, 16 genes (maximal accuracy = 0.726, minimal RMSE = 0.274) were identified as the OFGs. (E) The 10 marker genes obtained from the LASSO and SVM-RFE models. (F) The AUC for disease samples was determined using a logistic regression model. (G) ROC curves for the 10 marker genes.

FIGURE 4

Single-gene GSVA. (A) OFGs model of GSVA results. (B–K) Expression levels of single marker genes in the GSVA.

FIGURE 5

Analysis of the immunological landscape. (A) Immune cell expression levels in samples from healthy individuals and CAD patients. (B) Comparison of immunological microenvironments in CAD patients and healthy samples. (C) Differential study of the immunological microenvironments caused by single genes in OFGs. *p < 0.05, **p < 0.01, ***p < 0.001.

FIGURE 6

Prediction of marker gene-targeted drugs.

FIGURE 7

Expression of S100A8 gene in the test set and validation set. (A) S100A8 expression in GSE20680. (B) S100A8 expression in GSE20681.