Background: Pulmonary hypertension (PH) is marked by elevated pulmonary artery pressures due to various causes, impacting right heart function and survival. Disulfidptosis, a newly recognized cell death mechanism, may play a role in PH, but its associated genes (DiGs) are not well understood in this context. This study aims to define the diagnostic relevance of DiGs in PH.
Methods: Using GSE11726 data, we analyzed DiGs and their immune characteristics to identify core genes influencing PH progression. Various machine learning models, including RF, SVM, GLM, and XGB, were compared to determine the most effective diagnostic model. Validation used datasets GSE57345 and GSE48166. Additionally, a CeRNA network was established, and a hypoxia-induced PH rat model was used for experimental validation with Western blot analysis.
Results: 12 DiGs significantly associated with PH were identified. The XGB model excelled in diagnostic accuracy (AUC = 0.958), identifying core genes DSTN, NDUFS1, RPN1, TLN1, and MYH10. Validation datasets confirmed the model's effectiveness. A CeRNA network involving these genes, 40 miRNAs, and 115 lncRNAs was constructed. Drug prediction suggested therapeutic potential for folic acid, supported by strong molecular docking results. Experimental validation in a rat model aligned with these findings.
Conclusion: We uncovered the distinct expression patterns of DiGs in PH, identified core genes utilizing an XGB machine-learning model, and established a CeRNA network. Drugs targeting the core genes were predicted and subjected to molecular docking. Experimental validation was also conducted for these core genes.
Keywords: Diagnostic model; Disulfidptosis-related genes; Hypoxic pulmonary hypertension; Immune cell infiltration; Machine learning; Pulmonary hypertension.
© 2024. The Author(s).