BACKGROUND: This study aimed to explore hub circadian rhythm-related genes (CRRGs) associated with sepsis-associated acute kidney injury (saAKI) using machine learning algorithms to provide novel ideas for the diagnosis and treatment of this disease.
MATERIALS: Two AKI datasets (GSE30718 and GSE139061) were obtained from the Gene Expression Omnibus (GEO) database. We initially identified differentially expressed genes (DEGs) and performed WGCNA to identify candidate CRRGs. Following this, diagnostic CRRGs were identified on the basis of machine learning algorithms. The ssGSEA method was used to measure the link between hub CRRGs and immune cells. GSEA and GSVA were used to understand the relevant functions and pathways. The hub CRRGs were validated using the GSE255281 dataset and RT‒qPCR of clinical blood samples.
RESULTS: On the basis of the DEGs and WGCNA results, a total of 21 candidate CRRGs were obtained for saAKI. Seven diagnostic CRRGs were identified via the Lasso, random forest (RF) and SVM-RFE algorithms. We then constructed a predictive nomogram using the diagnostic CRRGs, and the calibration, decision and ROC curves revealed that it had high diagnostic efficiency. Through the evaluation of 2 subdatasets, 4 hub CRRGs (DEFB1, EGF, REN and PTPRD) were ultimately identified. Immune cell infiltration analysis indicated that there might be immune dysregulation in saAKI. GSEA and GSVA revealed different biological functions and pathways related to immunity, inflammation, metabolism and material transport that might exist in different subgroups of hub CRRGs. The validation dataset GSE255281 and RT‒qPCR analysis verified the expression of EGF and PTPRD.
CONCLUSION: Our research utilized comprehensive machine learning methods to construct a CRRG-based diagnostic model for saAKI. The identified CRRGs may be potential biomarkers of saAKI, and this study may provide new strategies for the diagnosis and treatment of saAKI.
