Xiong P, Huang Q, Mao Y, Qian H, Yang Y, Mou Z, Deng X, Wang G, He B, You Z. Identification of an immune-related gene panel for the diagnosis of pulmonary arterial hypertension using bioinformatics and machine learning.
Int Immunopharmacol 2025;
144:113694. [PMID:
39616855 DOI:
10.1016/j.intimp.2024.113694]
[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: 04/07/2024] [Revised: 11/03/2024] [Accepted: 11/20/2024] [Indexed: 12/15/2024]
Abstract
OBJECTIVE
This study aimed to screen an immune-related gene (IRG) panel and develop a novel approach for diagnosing pulmonary arterial hypertension (PAH) utilizing bioinformatics and machine learning (ML).
METHODS
Gene expression profiles were retrieved from the Gene Expression Omnibus (GEO) database to identify differentially expressed immune-related genes (IRG-DEGs). We employed five machine learning algorithms-LASSO, random forest (RF), boosted regression trees (BRT), XGBoost, and support vector machine recursive feature elimination (SVM-RFE) to identify biomarkers derived from IRG-DEGs associated with the diagnosis of PAH, incorporating them into the IRG-DEGs panel. Validation of these biomarker levels in lung tissue was conducted in a hypoxia-induced mouse model of PAH, investigating the correlation between AIMP1, IL-15, GLRX, SOD1, Fulton's index (RVHI), and the ratio of pulmonary artery medial thickness to external diameter (MT%). Subsequently, we developed a nomogram model based on the IRG-DEGs panel in lung tissue for diagnosing PAH. The expression, distribution, and pseudotime analysis of these biomarkers across various immune cell types were assessed using single-cell sequencing datasets. Finally, we evaluated the diagnostic utility of the nomogram model based on the IRG-DEGs panel in peripheral blood mononuclear cells (PBMCs) for diagnosing PAH.
RESULTS
A total of 36 upregulated and 17 downregulated IRG-DEGs were identified in lung tissue from patients with PAH. AIMP1, IL-15, GLRX, and SOD1 were subsequently selected as novel immune-related biomarkers for PAH through the aforementioned machine learning algorithms and incorporated into the IRG-DEGs panel. Experimental results from mice with PAH validated that the expression levels of AIMP1, IL-15, and GLRX in lung tissue were elevated, while SOD1 expression was significantly reduced. Additionally, GLRX and AIMP1 exhibited positive correlations with Fulton's index (RVHI). The expression levels of GLRX, IL-15, and AIMP1 showed positive correlations with MT%, whereas SOD1 exhibited negative correlations with MT%. Analysis of single-cell sequencing data further revealed that the levels of IRG-DEG panel members gradually increased during the pseudotime trajectory from PBMCs to macrophages, correlating with macrophage activation. The area under the curve (AUC) for diagnosing PAH using a nomogram model based on the IRG-DEGs panel derived from lung tissue samples and PBMCs was ≥0.969 and 0.900, respectively.
CONCLUSIONS
We developed an IRG-DEGs panel containing AIMP1, IL-15, GLRX, and SOD1, which may facilitate the diagnosis of pulmonary arterial hypertension (PAH). These findings provide novel insights that may enhance diagnostic and therapeutic approaches for PAH.
Collapse