Background/Purpose: Lupus nephritis (LN) and diabetic kidney disease (DKD) are leading causes of chronic kidney disease (CKD) that differ by age, sex, and underlying mechanisms. Both would benefit from earlier detection, but in both cases conventional markers such as serum creatinine or estimated glomerular filtration rate (eGFR) are insensitive for detection of early renal damage. We hypothesized that a panel of kidney-enriched cell type specific plasma proteins could provide earlier detection of LN and DKD and reflect disease-specific biological pathways.

Methods: Four kidney-specific biomarkers—uromodulin (UMOD), myo-inositol oxygenase (MIOX), heparanase (HPSE), and apoptosis-resistant E3 ubiquitin ligase 1 (AREL1)—were measured in plasma specimens from apparently healthy volunteers (AHV), patients with DKD, and patients with biopsy-confirmed LN. Cohorts included type 1 and type 2 diabetes CKD (combined n=99) and LN (n=31), with age and sex matched AHV controls (total n=130). Biomarker levels were quantified by ELISA. A multi-class logistic regression model was trained to classify AHV, DKD, and LN based on the four biomarkers. Performance was evaluated by area under the ROC curve (AUC) for each class (one-vs-rest), as well as confusion matrix and predicted probability distributions (Figure 1). Pathway enrichment (KEGG and Reactome) as well as protein–protein interaction (STRING) analyses were performed to interpret functional roles of the biomarkers. Data from PERL and FIND were provided by NIDDK CR, a program of the National Institute of Diabetes and Digestive and Kidney Diseases.

Results: LN patients had markedly lower UMOD and MIOX but higher HPSE and AREL1 compared to AHV and DKD (p< 0.001). DKD patients exhibited reduced HPSE and elevated MIOX relative to AHV. The panel yielded AUCs of 0.99 for LN, 0.89 for DKD, and 0.88 for AHV, indicating strong discriminatory performance across all three groups. Confusion matrix and probability plots confirmed minimal misclassification and clear separation between classes, underscoring the model’s robustness. Notably, the biomarker panel vastly outperformed creatinine and eGFR in detecting LN. Pathway enrichment analysis linked HPSE to glycosaminoglycan metabolism (extracellular matrix remodeling), MIOX to inositol phosphate metabolism (tubular function), and AREL1 to MHC class I antigen processing and proteasomal degradation pathways (Figure 2, Panel A), consistent with its known role as an E3 ubiquitin ligase. Consistently, STRING network analysis showed UMOD and MIOX clustering in a renal tubular transport network, HPSE in an extracellular matrix network, and AREL1 in a protein degradation network, consistent with its role in ubiquitin-mediated proteostasis (Figure 2, Panel B).

Conclusion: A four-protein biomarker panel reliably detects early LN and DKD and differentiates between them, when compared to healthy states, providing both high diagnostic accuracy and insight into disease-specific mechanisms. Its strong performance—especially the near-perfect discrimination for LN (AUC = 0.99)—suggests this panel can greatly enhance early detection of lupus nephritis and diabetic kidney disease.

Figure 1. Demographic and clinical characteristics of study cohort

Figure 1. Multi-class classification performance of the four-biomarker panel

Figure 2. Distinct pathway enrichment and interaction networks for kidney biomarkers

« Back to ACR Convergence 2025

ACR Meeting Abstracts - https://acrabstracts.org/abstract/aa-novel-kidney-specific-biomarker-panel-accurately-differentiates-lupus-nephritis-from-diabetic-kidney-disease-through-distinct-molecular-pathways-novel-kidney-specific-biomarker-panel-accurately-dif/