Machine learning and transcriptomic analysis identify tubular injury biomarkers in patients with chronic kidney disease.

Abstract

Purpose: Chronic Kidney Disease (CKD) is emerging as a major public health problem, with a lack of precise diagnostic biomarkers in clinical settings. The primary objective is to discover biomarkers for early clinical detection of CKD and to gain a deeper understanding of its underlying pathophysiological processes.

Methods: Samples from renal tubules of CKD patients and healthy controls were subjected to differential expression analysis. Weighted Gene Co-expression Network Analysis (WGCNA) was utilized to detect genes associated with renal tubular damage in CKD. Subsequently, Support Vector Machine Recursive Feature Elimination (SVM-RFE) and Least Absolute Shrinkage and Selection Operator (LASSO) algorithms were employed to identify and validate potential biomarker candidates.

Results: Four key renal biomarkers, namely DUSP1, GADD45A, TSC22D3, and ZFAND5, were successfully identified. Receiver Operating Characteristic (ROC) curve analysis and nomogram construction demonstrated their remarkable diagnostic capabilities. These biomarkers were also found to affect the degree of immune cell infiltration in CKD and exhibited a notable correlation with Glomerular Filtration Rate (GFR) and serum creatinine (SCr) levels.

Conclusion: These four identified biomarkers for renal tubular injury play important roles in immune function and inflammatory responses in CKD, potentially providing a theoretical foundation for dissecting molecular mechanisms and developing therapeutic strategies in CKD.