Abstract: SA-PO920
Evaluating GFR Slope as a Surrogate End Point Across Diseases
Session Information
- CKD: Observational Research and Patient-Oriented Interventions
November 05, 2022 | Location: Exhibit Hall, Orange County Convention Center‚ West Building
Abstract Time: 10:00 AM - 12:00 PM
Category: CKD (Non-Dialysis)
- 2202 CKD (Non-Dialysis): Clinical‚ Outcomes‚ and Trials
Authors
- Collier, Willem H., University of Utah Health, Salt Lake City, Utah, United States
- Haaland, Benjamin, University of Utah Health, Salt Lake City, Utah, United States
- Inker, Lesley Ann, Tufts Medical Center, Boston, Massachusetts, United States
- L Heerspink, Hiddo Jan, Universitair Medisch Centrum Groningen Afdeling Cardiologie, Groningen, Groningen, Netherlands
- Greene, Tom, University of Utah Health, Salt Lake City, Utah, United States
Group or Team Name
- Chronic Kidney Disease Epidemiology Collaboration
Background
We have previously demonstrated that treatment effects on GFR slope can be used to accurately predict treatment effects on time-to-kidney failure (KFRT) in randomized CKD trials. It has been hypothesized that accuracy can vary by sub-categories of disease. Yet, low statistical power in subgroups can challenge the interpretation of subgroup-to-subgroup variability in GFR slope as a surrogate.
Methods
We performed analyses on N=47 previously conducted CKD trials. For each trial, we estimated treatment effects on time-to-KFRT using proportional hazards models and treatment effects on GFR chronic slope using mixed-effects models. Meta-regression methods were used to assess the strength of the association between treatment effects on the separate endpoints within three disease-defined subgroups: Diabetes (DN), glomerular diseases (GN), and CKD without diabetes (CKD). We used a novel Bayesian modeling approach (partial-pooling) to facilitate improved precision in our estimation of model parameters. The results were contrasted with those obtained using an earlier no-pooling approach.
Results
Relative to no-pooling, the partial-pooling model improved precision in parameter estimates key to the interpretation of the surrogate (see the 95% credible intervals in Figure 1 legends). The partial-pooling model also improved prediction of treatment effects on time-to-KFRT (see narrower confidence bands on the left of Figure 1).
Conclusion
On a collection of CKD trials, use of a novel meta-regression approach for surrogate evaluation indicated homogeneity in the quality of GFR chronic slope across subgroups of trials defined by disease.
α: Meta-regression intercept β: Meta-regression slope. We display meta-regression lines as well as the upper and lower confidence bands in grey (partial-poling) and red (no-pooling). Grey circles are trials not in the featured disease-defined subgroup.
Funding
- Private Foundation Support