Abstract: FR-OR038
The Multi-Phenotype Derived Nephrotic Syndrome Severity (NS2) Score Empowers Genomic Discovery
Session Information
- Genetically-Defined Kidney Diseases: From Variant Calling to Treatment
November 03, 2017 | Location: Room 392, Morial Convention Center
Abstract Time: 05:54 PM - 06:06 PM
Category: Genetic Diseases of the Kidney
- 803 Genetic Epidemiology and Other Genetic Studies of Common Kidney Diseases
Authors
- Gillies, C., University of Michigan, Ann Arbor, Michigan, United States
- Yasutake, K, University of Michigan , Ann Arbor, Michigan, United States
- Wen, Xiaoquan, University of Michigan , Ann Arbor, Michigan, United States
- Sampson, Matt G., University of Michigan, Ann Arbor, Michigan, United States
Group or Team Name
- NEPTUNE
Background
Among patients with nephrotic syndrome (NS), we are interested in discovering genomic factors associated with disease severity over time. This requires creating accurate models of NS biology and a statistical strategy that takes advantage of deep phenotypic data while accounting for modest sample sizes and multiple testing. Thus, we developed a Nephrotic Syndrome Severity (“NS2”) score for patients in the Nephrotic Syndrome Study Network (NEPTUNE).
Methods
NEPTUNE is a prospective, longitudinal study of NS enrolling affected adults and children receiving a clinically indicated biopsy. Rich demographic and clinical data are collected at baseline and over time. Genomic and histologic data are collected at baseline. We used the following parameters to create the multi-phenotype NS2 score: interstitial fibrosis, eGFR, protein/creatinine ratio, eGFR slope, time to complete remission, and time to a composite endpoint. We modeled the relationships between these variables and meaningful covariates using a Bayesian network (BN). The NS2 score represents a latent factor explaining the correlations in the observed data. The BN’s parameters were inferred from 616 patients NEPTUNE participants using Markov chain Monte Carlo.
Results
Compared to existing multi-phenotype methods, NS2 score increased power for discovery without inflating Type I error. With regards to known biomarkers of NS severity, a worse NS2 score was significantly associated with the APOL1 high-risk genotype in black patients (p< 2.2 e-5) and lower tubulointerstitial expression of EGF (p< 5.7 e-10). After FDR control, 1,040 glomerular transcripts were significantly associated with NS2 score. Using geneset enrichment analysis, kidney development genes were among the most enriched NS2-associated glomerular transcripts in adults (p<5.3 e-9), including 15 known Mendelian SRNS genes. In children, TNF alpha induced protein 3's glomerular expression was most associated with NS2 score.
Conclusion
The NS2 score is a robust metric created by capitalizing on extensive clinical data and NS-specific knowledge. As a robust multi-phenotype method, it improved statistical power for discovery without inflating false positives and replicates known genomic associations. Ultimately, using NS2 score as an outcome measure in analyses ranging from gene expression correlation to GWAS may empower genomic discoveries.
Funding
- NIDDK Support