Identification and validation of the diagnostic signature associated with immune microenvironment of acute kidney injury based on ferroptosis-related genes through integrated bioinformatics analysis and machine learning

Abstract

Background: Acute kidney injury (AKI) is a common and severe disease, which poses a global health burden with high morbidity and mortality. In recent years, ferroptosis has been recognized as being deeply related to Acute kidney injury. Our aim is to develop a diagnostic signature for Acute kidney injury based on ferroptosis-related genes (FRGs) through integrated bioinformatics analysis and machine learning. Methods: Our previously uploaded mouse Acute kidney injury dataset GSE192883 and another dataset, GSE153625, were downloaded to identify commonly expressed differentially expressed genes (coDEGs) through bioinformatic analysis. The FRGs were then overlapped with the coDEGs to identify differentially expressed FRGs (deFRGs). Immune cell infiltration was used to investigate immune cell dysregulation in Acute kidney injury. Functional enrichment analysis and protein-protein interaction network analysis were applied to identify candidate hub genes for Acute kidney injury. Then, receiver operator characteristic curve analysis and machine learning analysis (Lasso) were used to screen for diagnostic markers in two human datasets. Finally, these potential biomarkers were validated by quantitative real-time PCR in an Acute kidney injury model and across multiple datasets. Results: A total of 885 coDEGs and 33 deFRGs were commonly identified as differentially expressed in both GSE192883 and GSE153625 datasets. In cluster 1 of the coDEGs PPI network, we found a group of 20 genes clustered together with deFRGs, resulting in a total of 48 upregulated hub genes being identified. After ROC analysis, we discovered that 25 hub genes had an area under the curve (AUC) greater than 0.7; Lcn2, Plin2, and Atf3 all had AUCs over than this threshold in both human datasets GSE217427 and GSE139061. Through Lasso analysis, four hub genes (Lcn2, Atf3, Pir, and Mcm3) were screened for building a nomogram and evaluating diagnostic value. Finally, the expression of these four genes was validated in Acute kidney injury datasets and laboratory investigations, revealing that they may serve as ideal ferroptosis markers for Acute kidney injury. Conclusion: Four hub genes (Lcn2, Atf3, Pir, and Mcm3) were identified. After verification, the signature's versatility was confirmed and a nomogram model based on these four genes effectively distinguished Acute kidney injury samples. Our findings provide critical insight into the progression of Acute kidney injury and can guide individualized diagnosis and treatment.

Keywords: Acute kidney injury; diagnostic signature; ferroptosis-related genes; immune microenvironment; machine learning.