Integrated omics analysis identifies GSTA1 and ALDH1A1 as candidate biomarkers for low-level albuminuric diabetic kidney disease: a mixed prospective-retrospective cross-sectional study

Abstract

Diagnosing a non-albuminuric diabetic kidney disease (DKD) remains challenging, primarily attributable to the absence of specific biomarkers. This study identifies potential biomarkers via integrated proteomic-metabolomic analysis, to provide evidence for optimizing DKD management. This combined retrospective and prospective observational study were conducted at the Department of Endocrinology. All patients with diabetes mellitus (DM) between 2023 and 2024 were retrospectively reviewed and classified into normo-, low-, high-, and very-high albuminuric DKD (subdivided NADKD, LADKD, HADKD, and VADKD) and simple DM for prevalence analysis. Newly diagnosed subdivided NADKD, LADKD, HADKD, and VADKD patients from January to March 2025 were prospectively enrolled to explore differential proteins via integrated proteomic-metabolomic analysis. Finally, 270 simple DM, 270 LADKD and 270 HADKD/VADKD patients were randomly enrolled to preliminarily verify the diagnostic performance of candidate protein. The retrospective investigation revealed that non-albuminuric DKD patients are predominantly of advanced age, and the subdivided NADKD is rare, accounting for 0.7% of all DM patients and 1.6% of all DKD patients. In the prospective omics cohort, LADKD differed metabolically from simple DM and albuminuric DKD across 33 metabolic pathways involving 29 differential proteins. These proteins were primarily involved in extracellular exosome, cytosol, nucleus, membrane, extracellular space, and cytoplasm. In the KEGG enrichment network, they primarily participate in key pathways including Metabolic pathways, Biosynthesis of amino acids, Ribosome, One carbon pool by folate, Coronavirus disease, and Carbon metabolism. After screening, seven candidate proteins obtained an excellent classification performance and achieved a strong 5-fold cross-validation mean AUC exceeding 0.8 in the XGBoost classifier. Finally, the SHAP analysis confirmed that GSTA1 and ALDH1A1 ranked higher in feature importance across on both the training set (SHAP ± SE: 0.219 ± 0.033, 0.203 ± 0.037) and testing set (SHAP ± SE: 0.229 ± 0.050, 0.174 ± 0.038). This study proposes a potential disease trajectory of DM→LADKD→ albuminuric DKD, which appears to contrast with the older age profile observed in non-albuminuric DKD patients. Glutathione S-transferase A1 and aldehyde dehydrogenase 1 alpha 1 emerge as the most biologically significant candidate biomarkers, collectively implicating tubular injury as a potential trajectory in LADKD.