Immune Microenvironment Alterations and Identification of Key Diagnostic Biomarkers in Chronic Kidney Disease Using Integrated Bioinformatics and Machine Learning
Jinbao Shi,Aliang Xu,Liuying Huang,Shaojie Liu,Binxuan Wu,Zuhong Zhang
DOI: https://doi.org/10.2147/pgpm.s488143
2024-11-27
Pharmacogenomics and Personalized Medicine
Abstract:Jinbao Shi, Aliang Xu, Liuying Huang, Shaojie Liu, Binxuan Wu, Zuhong Zhang Department of Nephrology, Ningde Hospital of Traditional Chinese Medicine, Ningde, Fujian, People's Republic of China Correspondence: Jinbao Shi, Department of Nephrology, Ningde Hospital of Traditional Chinese Medicine, No. 16 Donghu Road, Ningde, Fujian, People's Republic of China, Email Background: Chronic kidney disease (CKD) involves complex immune dysregulation and altered gene expression profiles. This study investigates immune cell infiltration, differential gene expression, and pathway enrichment in CKD patients to identify key diagnostic biomarkers through machine learning methods. Methods: We assessed immune cell infiltration and immune microenvironment scores using the xCell algorithm. Differentially expressed genes (DEGs) were identified using the limma package. Gene Set Enrichment Analysis (GSEA) and Gene Set Variation Analysis (GSVA) were performed to evaluate pathway enrichment. Machine learning techniques (LASSO and Random Forest) pinpointed diagnostic genes. A nomogram model was constructed and validated for diagnostic prediction. Spearman correlation explored associations between key genes and immune cell recruitment. Results: The CKD group exhibited significantly altered immune cell infiltration and increased immune microenvironment scores compared to the normal group. We identified 2335 DEGs, including 124 differentially expressed immune-related genes. GSEA highlighted significant enrichment of inflammatory and immune pathways in the high immune score (HIS) subgroup, while GSVA indicated upregulation of immune responses and metabolic processes in HIS. Machine learning identified four key diagnostic genes: RGS1, IL4I1, NR4A3, and SOCS3. Validation in an independent dataset (GSE96804) and clinical samples confirmed their diagnostic potential. The nomogram model integrating these genes demonstrated high predictive accuracy. Spearman correlation revealed positive associations between the key genes and various immune cells, indicating their roles in immune modulation and CKD pathogenesis. Conclusion: This study provides a comprehensive analysis of immune alterations and gene expression profiles in CKD. The identified diagnostic genes and the constructed nomogram model offer potent tools for CKD diagnosis. The immunomodulatory roles of RGS1, IL4I1, NR4A3, and SOCS3 warrant further investigation as potential therapeutic targets in CKD. Keywords: chronic kidney disease, diagnostic biomarkers, immune microenvironment, GSEA, GSVA, machine learning Chronic kidney disease (CKD) is a progressive condition that poses a significant public health challenge worldwide, affecting millions of people and leading to high morbidity and mortality rates. 1 CKD is characterized by the gradual loss of kidney function over time, which can eventually lead to end-stage renal disease, necessitating dialysis or kidney transplantation. 2 The pathogenesis of CKD involves a complex interplay of genetic, environmental, and immunological factors. 3 Recent studies have highlighted the crucial role of immune dysregulation and altered gene expression profiles in the progression of CKD. 4 However, the specific mechanisms underlying these changes and their impact on disease progression remain poorly understood. Numerous studies have explored the role of immune cells in CKD, demonstrating that immune cell infiltration and activation are key features of CKD pathology. 5,6 For instance, T cells, and macrophages have been implicated in promoting inflammation and fibrosis in the kidney, contributing to disease progression. 7 Despite these advances, there remains a gap in understanding the comprehensive landscape of immune cell infiltration and its relationship with gene expression changes in CKD. Furthermore, the integration of bioinformatics and machine learning methodologies remains underutilized in identifying reliable diagnostic biomarkers in CKD. Prior studies have frequently suffered from inadequate utilization of bioinformatics and machine learning methodologies for the identification of reliable diagnostic biomarkers. Additionally, recent advancements in unsupervised deep learning approaches have shown promise in uncovering complex patterns in biomedical data. 8–10 Machine learning techniques, such as the least absolute shrinkage and selection operator (LASSO) and Random Forest, have demonstrated significant potential in identifying key diagnostic genes and constructing predictive models in various diseases. 11,12 Given the complex nature of CKD and the involvement of immune dysregulation, there is a pressing need for comprehensive studies that integrate bioinformatics and machine lear -Abstract Truncated-
pharmacology & pharmacy