*Purpose: Graft failure is a major cause of morbidity and mortality in kidney transplant recipients. However, trials aimed to assess the effect of interventions on graft failure are expensive and time-consuming. The purpose of our study is to evaluate the possibility of using post-transplant eGFR, delayed graft function (DGF), and recipient baseline characteristics as surrogate markers of long-term graft failure.

*Methods: Using SRTR data 2006-2016, we identified 110,414 deceased donor kidney transplant (DDKT) recipients and followed them from transplantation to death-censored graft failure (DCGF), death, or administrative censoring on 6/30/2019. Those who died or developed DCGF within 6 months post-transplant were excluded from our study. We used potential surrogate outcomes to predict long-term DCGF using Cox regression and evaluated predictive power of each model using bootstrapped Harrell C-index.

*Results: The median follow-up time was 6.0 years (IQR: 3.9-8.7y). Among the 7 prediction models, post-transplant DGF poorly predicted long-term graft function (Harrell C-index: _0.540.54_0.55). The model based on baseline characteristics had better predictive power (Harrell C-index: _0.650.65_0.66). 6-month eGFR, 1-year, and 2-year eGFR had increased predictive validity (Harrell C-index: _0.650.66_{0.66, 0.69}0.70_0.70, and _0.740.74_0.75, respectively). Adding 6-month and 1-year eGFR to the 2-year eGFR model did not appreciably improve predictive accuracy (Harrell C-index: _0.740.75_0.75; Figure1). A 10-unit decline in 2-year eGFR was associated with 35% increased risk of subsequent DCGF (HR=_0.650.65_0.66, p<0.01; Figure2).

*Conclusions: Among outcomes evaluated, two-year post-transplant eGFR is the best predictor of long-term DCGF risk in DDKT recipients. This can potentially be used in kidney transplant trials as an inexpensive, non-invasive, and reliable surrogate outcome of long-term graft survival.

« Back to 2020 American Transplant Congress