Identification of biomarkers related to iron death in diabetic kidney disease based on machine learning algorithms

Abstract

Background: While ferroptosis has been recognised for its key role in tumour development, its involvement in DKD is not well understood. Identifying differentially expressed ferroptosis-related genes (DEIRGs) could help improve early diagnosis and treatment strategies for DKD.

Aim: Diabetic kidney disease (DKD) is a complication of diabetes that can progress to end-stage renal disease. Early diagnosis and identification of biomarkers related to its pathogenesis are crucial. This study aims to investigate the role of ferroptosis, a type of programmed cell death, in DKD, which remains largely unexplored.

Objective: The objective of this study was to screen for diagnosis-related DEIRGs (DDEIRGs) in DKD and construct a diagnostic model with high accuracy.

Method: We intersected differentially expressed genes in the DKD dataset with ferroptosis-related genes to obtain DEIRGs. Gene importance was ranked using the random forest and Adaboost algorithms, and DDEIRGs were identified by intersecting results. A diagnostic model was constructed using logistic regression, and its accuracy was evaluated. Additionally, the immune landscape of DDEIRGs was analysed, and RT-qPCR was used to validate gene expression levels.

Results: The diagnostic model constructed with logistic regression demonstrated high diagnostic accuracy for DKD. Immune landscape analysis of DDEIRGs provided further insights into their potential roles. RT-qPCR confirmed the differential expression of diagnosis-related genes.

Conclusion: This study successfully identified diagnosis-related ferroptosis genes in DKD and constructed an accurate diagnostic model. These findings enhance our understanding of the role of ferroptosis in DKD and may contribute to the development of new diagnostic and therapeutic approaches.

Keywords: Diabetic kidney disease; RT-qPCR; diagnostic model; iron death; machine learning.