Abstract: FR-PO849
Development of Novel Algorithms to Characterize Lupus Nephritis (LN) Renal Activity and Chronicity Using Urine Proteomics
Session Information
- Glomerular Diseases: Immunology, Inflammation - I
November 08, 2019 | Location: Exhibit Hall, Walter E. Washington Convention Center
Abstract Time: 10:00 AM - 12:00 PM
Category: Glomerular Diseases
- 1202 Glomerular Diseases: Immunology and Inflammation
Authors
- Akhgar, Ahmad, AstraZeneca, Boyds, Maryland, United States
- Zeng, Lingmin, AstraZeneca, Boyds, Maryland, United States
- Farris, Alton Brad, Emory University, Atlanta, Georgia, United States
- Yu, Binbing, AstraZeneca, Boyds, Maryland, United States
- Sinibaldi, Dominic, AstraZeneca, Boyds, Maryland, United States
- Cobb, Jason, Emory University School of Medicine, Atlanta, Georgia, United States
- Battle, Monica, Emory University, Atlanta, Georgia, United States
- Illei, Gabor, Viela Bio, Gaithersburg, Maryland, United States
- Lim, S. Sam, Emory University, Atlanta, Georgia, United States
- White, Wendy I., AstraZeneca, Boyds, Maryland, United States
Background
We assessed the utilization of the widely available, high throughput platform for multiplex analysis of urine proteomics to define limited sets of baseline markers correlating with LN histopathology results assessed by the NIH Activity (NIH-AI) and Chronicity (NIH-CI) Indices. Novel algorithms were developed to characterize patients into high and low subgroups. Longitudinal urine sample were used as surrogates to monitor the disease indices over time.
Methods
Baseline urine samples were collected from 42 LN patients. Baseline LN biopsies were interpreted by an expert pathologist. Additional LN samples were collected over a year. Urine samples were tested on a large Luminex multiplex platform. Stepwise regression and single variable regression results were combined to limit the possible candidate urine biomarkers. A multivariate logistic regression model (MLRM) was applied to the remaining markers. Selection criteria for biomarkers were p values < 0.05, high area under the curve, and a low misclassification rate. Receiver operating characteristic curves based on cross-validations evaluated the predictiveness of selected biomarkers.
Results
Data from 288 markers were reduced to 177 by assessing assay robustness. Four markers (CD163, ferritin, KIM-1, and antileukoproteinase) were identified (P<0.05) in the MLRM, which predicted 93% of the NIH-AI high patients with a false positive rate (FPR) of 11%. The predicted probability of high activity patients decreased over time relative to baseline as serum albumin increased and proteinuria decreased, reflecting decreased activity. The markers identified for NIH-CI were hepatocyte growth factor, Eotaxin-2, IL-6R β, and ITAC. The model predicted 81% of patients with high NIH-CI with a FPR near 0%.
Conclusion
This study supports the continued evaluation of urine biomarkers with a multiplex immunoassay platform validated to clinical laboratory standards and could support wide distribution as a Luminex-based test. Novel combinations of markers were associated with renal histopathologic activity and chronicity (high and low categories) in LN. Treatment led to alterations in activity and chronicity longitudinally.
Funding
- Commercial Support –