Abstract: SU-OR16
Proteomic Risk Assessment of CKD Progression in the Chronic Renal Insufficiency Cohort
Session Information
- Emerging Translational Research to Improve CKD Outcomes
October 25, 2020 | Location: Simulive
Abstract Time: 05:00 PM - 07:00 PM
Category: CKD (Non-Dialysis)
- 2101 CKD (Non-Dialysis): Epidemiology, Risk Factors, and Prevention
Authors
- Dubin, Ruth F., University of California San Francisco, San Francisco, California, United States
- Deo, Rajat, University of Pennsylvania, Philadelphia, Pennsylvania, United States
- Zheng, Zihe, University of Pennsylvania, Philadelphia, Pennsylvania, United States
- Shou, Haochang, University of Pennsylvania, Philadelphia, Pennsylvania, United States
- Ren, Yue, University of Pennsylvania, Philadelphia, Pennsylvania, United States
- Li, Hongzhe, University of Pennsylvania, Philadelphia, Pennsylvania, United States
- Segal, Mark, University of California San Francisco, San Francisco, California, United States
- Feldman, Harold I., University of Pennsylvania, Philadelphia, Pennsylvania, United States
- Ganz, Peter, University of California San Francisco, San Francisco, California, United States
Background
Quantification of thousands of plasma proteins simultaneously is now feasible in large cohorts using the SomaScan aptamer assay. In this study, we applied large-scale proteomics to patients with chronic kidney disease (CKD) to discover new biomarkers and risk models of CKD progression.
Methods
We measured 4638 unique plasma proteins among 3249 participants in the Chronic Renal Insufficiency Cohort(CRIC), with follow-up to 13 years. Mean age was 59 years, mean estimated glomerular filtration rate (eGFR) 42 ml/min/1.73m2, and 50% were diabetic. The study outcome was 10-year risk of 50% decline in eGFR, end-stage renal disease or renal transplant (n=1171 events). Associations of individual proteins with the composite outcome were analyzed in Cox survival models adjusted for demographics, comorbidities, eGFR and proteinuria. Protein-only risk models were constructed using elastic net regression and compared to the 4-variable Kidney Failure Risk Equation (KFRE). KFRE variables (age, gender, eGFR and proteinuria) were refit to CRIC. For risk modeling, the cohort was split into 80% derivation/20% validation.
Results
Among the 4638 assayed proteins, after adjustment for eGFR, 1535 proteins were associated with CKD progression at FDR <0.05; 529 were significant at Bonferroni p<1x10-5. After full adjustment, 459 proteins met FDR significance and 77 proteins met Bonferroni significance. A 58 protein risk model for 10-year CKD progression derived by elastic net achieved a c-statistic of 0.860 (95% CI: 0.834, 0.885) in the validation set, equal to the refit KFRE c-statistic of 0.857 (95% CI 0.831, 0.884). The c-statistic for the proteomic model was not enhanced by addition of clinical risk factors. Additionally, we were able to identify protein biomarkers that are unique to progression of diabetic vs. non-diabetic CKD.
Conclusion
Through large-scale proteomics, we discovered numerous novel biomarkers that predict the risk of CKD progression. The proteomic risk model has excellent discrimination, equal to the refit clinical model. Ongoing analyses of the biological functions of the newly discovered biomarkers may identify new therapeutic targets to slow CKD progression.
Funding
- NIDDK Support