Identification of Shared Signature Genes and Immune Microenvironment Subtypes for Heart Failure and Chronic Kidney Disease Based on Machine Learning

Abstract

Xuefu Wang,1,* Jin Rao,2,* Xiangyu Chen,2,* Zhinong Wang,2 Yufeng Zhang1,2 1School of Health Science and Engineering, University of Shanghai for Science and Technology, Shanghai, People’s Republic of China; 2Department of Cardiothoracic Surgery, Shanghai Changzheng Hospital, Naval Medical University, Shanghai, People’s Republic of China*These authors contributed equally to this workCorrespondence: Zhinong Wang; Yufeng Zhang, Email [email protected]; [email protected]: A complex interrelationship exists between Heart Failure (HF) and chronic kidney disease (CKD). This study aims to clarify the molecular mechanisms of the organ-to-organ interplay between heart failure and CKD, as well as to identify extremely sensitive and specific biomarkers.Methods: Differentially expressed tandem genes were identified from HF and CKD microarray datasets and enrichment analyses of tandem perturbation genes were performed to determine their biological functions. Machine learning algorithms are utilized to identify diagnostic biomarkers and evaluate the model by ROC curves. RT-PCR was employed to validate the accuracy of diagnostic biomarkers. Molecular subtypes were identified based on tandem gene expression profiling, and immune cell infiltration of different subtypes was examined. Finally, the ssGSEA score was used to build the ImmuneScore model and to assess the differentiation between subtypes using ROC curves.Results: Thirty-three crosstalk genes were associated with inflammatory, immune and metabolism-related signaling pathways. The machine-learning algorithm identified 5 hub genes (PHLDA1, ATP1A1, IFIT2, HLTF, and MPP3) as the optimal shared diagnostic biomarkers. The expression levels of tandem genes were negatively correlated with left ventricular ejection fraction and glomerular filtration rate. The CIBERSORT results indicated the presence of severe immune dysregulation in patients with HF and CKD, which was further validated at the single-cell level. Consensus clustering classified HF and CKD patients into immune and metabolic subtypes. Twelve immune genes associated with immune subtypes were screened based on WGCNA analysis, and an ImmuneScore model was constructed for high and low risk. The model accurately predicted different molecular subtypes of HF or CKD.Conclusion: Five crosstalk genes may serve as potential biomarkers for diagnosing HF and CKD and are involved in disease progression. Metabolite disorders causing activation of a large number of immune cells explain the common pathogenesis of HF and CKD.Keywords: heart failure, chronic kidney disease, immunity and inflammation, metabolic disorder, bioinformatics

Keywords

• heart failure
• chronic kidney disease
• immunity and inflammation
• metabolic disorder
• bioinformatics