Abstract: FR-OR49
Implications of Follow-Up Time on the Optimal Weighting of the Acute and Chronic Slopes to Predict Treatment Effects on Clinical End Points
Session Information
- Interventions to Reduce CKD Progression
November 03, 2023 | Location: Room 119, Pennsylvania Convention Center
Abstract Time: 05:42 PM - 05:51 PM
Category: CKD (Non-Dialysis)
- 2302 CKD (Non-Dialysis): Clinical, Outcomes, and Trials
Authors
- Greene, Tom, University of Utah Health, Salt Lake City, Utah, United States
- Inker, Lesley Ann, Tufts Medical Center, Boston, Massachusetts, United States
- Heerspink, Hiddo Jan L., Universiteit Groningen Afdeling Gezondheidswetenschappen, Groningen, Groningen, Netherlands
Group or Team Name
- CKD-EPI(CT).
Background
We have shown in a meta-analysis of 66 randomized treatment comparisons (RTCs) that treatment effects on the established clinical endpoint (CE) based on doubling of serum creatinine (SCR), GFR ≤ 15 ml/min/1.73m2 or kidney failure are in aggregate predicted accurately by the mean GFR slope over 3 years. We evaluate implications of follow-up time for the optimal weighting of the acute and chronic GFR slopes to predict treatment effects on the CE.
Methods
For each RTC, we used a mixed effects model to estimate treatment effects on the acute (evaluated from baseline to 3 months) and chronic (evaluated after 3 months) GFR slopes, and Cox regression to estimate treatment effects on the CE. We used an extended multivariable Bayesian meta-regression model to relate the treatment effects on the CE jointly to those on the acute and chronic slopes. The extended model expresses the optimal weighted average of the acute and chronic slopes as α×[Acute Slope]+(1-α)×[Chronic Slope], and allows α to depend on the median follow-up time of each RTC.
Results
The multivariable model accurately predicted the treatment effect on the CE (median R2 = 0.97 for a RTC with 3 years follow-up). The optimal weight α for the acute slope relative to the chronic slope had a strong inverse relationship with the median follow-up time, decreasing from a median (95% Bayesian credible interval) of 0.105 (0.065, 0.154) at 2 years by 34% to 0.069 (0.049,0.091) at 3 years and by 65% to 0.037 (0.013,0.075) at 4.5 years.
Conclusion
The optimal weighting of the acute and chronic slopes for predicting the treatment effect on the CE assigns a major role to the acute effect when the CE is evaluated over a relatively short follow-up time typical of some CKD trials, but deemphasizes the acute effect and more closely approximates the chronic slope when the CE is evaluated over longer follow-up times which are relevant to patients.
Effect of Follow-up Time on Optimal Weighted Average of Acute and Chronic Slopes
| Term | Median | 95% Bayesian CI |
| Intercept | -0.02 | (-0.08, 0.04) |
| % hazard reduction in the CE per 0.75 ml/min/1.73m2/year greater effect on optimal weighted average of the acute and chronic GFR slopes | 33% | (27%,38%) |
| Optimal α when evaluating CE over 2 years | 0.105 | (0.065,0.154) |
| Optimal α when evaluating CE over 3 years | 0.069 | (0.049, 0.091) |
| Optimal α when evaluating CE over 4.5 years | 0.037 | (0.013,0.075) |
| Trial level R2 with follow-up standardized to 3 years | 0.97 | (0.82, 1.00) |
Funding
- Private Foundation Support