Abstract: FR-PO1126
Identifying Functional Subclasses of Nephrotic Syndrome by Consensus Non-Negative Matrix Factorization Clustering of Glomerular Transcriptome
Session Information
- Glomerular Diseases: Clinical, Outcomes, Trials - II
October 26, 2018 | Location: Exhibit Hall, San Diego Convention Center
Abstract Time: 10:00 AM - 12:00 PM
Category: Glomerular Diseases
- 1203 Glomerular Diseases: Clinical, Outcomes, and Trials
Authors
- Hamidi, Habib, University of Michigan, Ann Arbor, Michigan, United States
- Eddy, Sean, University of Michigan, Ann Arbor, Michigan, United States
- Kretzler, Matthias, University of Michigan, Ann Arbor, Michigan, United States
- Sedor, John R., Cleveland Clinic, Cleveland, Ohio, United States
- Holzman, Lawrence B., University of Pennsylvania, Philadelphia, Pennsylvania, United States
- O'Toole, John F., Cleveland Clinic, Cleveland, Ohio, United States
- Mariani, Laura H., University of Michigan, Ann Arbor, Michigan, United States
Background
Defining nephrotic syndrome in mechanistic terms is a prerequisite to develop targeted therapies. Consensus non-negative matrix factorization (NMF) is a clustering approach used with tumor specimen transcriptomics to identify functionally relevant subtypes.
Methods
NMF clustering was applied to glomerular mRNA expression from nephrotic syndrome patient biopsies (NEPTUNE cohort). Individual gene expression levels from the glomerular compartment were normalized to mean expression to maximize individual patient differences over shared disease & tissue expression. Cox proportional hazards models were fit for complete proteinuria remission (CR, UPCR <0.3 mg/mg) and ESRD/40% eGFR decline. Significance analysis of microarray identified cluster specific differentially expressed genes that were used in pathway enrichment analysis to determine functional relevance.
Results
NMF separated 138 patients into 4 clusters which did not differ by diagnosis (MCD, FSGS, MN and IgAN, p 0.25), age (p 0.72), sex (p 0.25) or UPCR (p 0.28). Cluster 1 had lower mean eGFR (63 mL/min vs 81, 77 and 85; p 0.04) and greater black race (54% vs 10%, 24%, 19%, p <0.01). In unadjusted models, cluster 1 had least CR (HR 0.51, 95%CI 0.3–1.0, p-value 0.06) and greatest loss of eGFR (HR 5.4 95%CI 2.5–12.0, p-value <0.01). The eGFR association persisted after adjustment for baseline eGFR and race. Pathway enrichment of cluster-specific genes demonstrated unique processes: cluster 1-targetable signaling pathways (e.g. integrin, EGF-R, TLR), cluster 2-pentose phosphate metabolism, cluster 3-metabolic pathways (e.g.Vit D, pyruvate, galactose), cluster 4-cadherin signaling.
Conclusion
NMF clustering of glomerular kidney tissue mRNA expression levels revealed 4 clusters which were distinct from conventional classification. Pathway enrichment of gene signatures for each cluster revealed distinct molecular processes which may help to inform future treatment strategies.
Funding
- NIDDK Support