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 shijinbao00721@aliyun.com

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