Abstract: PO0966
Joint Model of eGFR Slope: Data from LEADER in Patients with Type 2 Diabetes and High Risk of Cardiovascular Events
Session Information
- Diabetic Kidney Disease: Clinical - 1
October 22, 2020 | Location: On-Demand
Abstract Time: 10:00 AM - 12:00 PM
Category: Diabetic Kidney Disease
- 602 Diabetic Kidney Disease: Clinical
Authors
- Mann, Johannes F., KfH Kidney Centre, Munich, Germany
- Bosch-Traberg, Heidrun, Novo Nordisk A/S, Søborg, Denmark
- Rasmussen, Soren, Novo Nordisk A/S, Søborg, Denmark
- Wolthers, Benjamin, Novo Nordisk A/S, Søborg, Denmark
- L Heerspink, Hiddo Jan, University of Groningen, University Medical Center Groningen, Groningen, Netherlands
Background
The LEADER cardiovascular (CV) outcome trial (NCT01179048) suggested liraglutide provides renal benefits vs placebo in patients with type 2 diabetes and high CV risk. Aiming to improve the modeling of eGFR slope (surrogate marker of renal outcomes), this post hoc analysis compared a joint model with the usual random slope model.
Methods
Two models were applied: 1) random slope model for eGFR using an effect modifier for treatment (liraglutide vs placebo) in change from baseline; 2) joint model using two processes (the same random slope model and a hazard model for time to a composite endpoint [all-cause death or ESRD]). These processes were correlated using predicted patient-level effects of eGFR across time, adjusting eGFR slopes for dropouts due to the composite. Those with renal impairment (eGFR <60 mL/min/1.73 m2 and UACR ≥30 mg/g) at baseline were also analyzed.
Results
Comparing the joint vs random slope models, average eGFR slopes were decreased for liraglutide (-1.77 vs -1.70 mL/min/1.73 m2/year) and placebo (-2.03 vs -1.96 mL/min/1.73 m2/year) for all participants (Table). For those with renal impairment, the eGFR slopes estimated using the two models were markedly different for both liraglutide (-2.49 vs -2.08 mL/min/1.73 m2/year) and placebo (-3.50 vs -3.04 mL/min/1.73 m2/year). A 1 mL/min/1.73 m2 lower eGFR value at baseline increased risk of the composite by 3% and change from baseline of -1 mL/min/1.73 m2 increased it by 6% for all participants. Respective values for those with renal impairment were 6% and 13%.
Conclusion
Liraglutide reduced eGFR slope vs placebo in both models. Joint modeling, which utilized dropout due to death or ESRD, altered the estimation of eGFR slope in LEADER, with a more marked decrease in the high risk group vs the random slope model. It did not change the liraglutide vs placebo treatment effect. Joint modeling may be useful in analyzing future trial data.
Funding
- Commercial Support –