Comparing the prognostic performance of iBOX and biopsy-proven acute rejection for long-term kidney graft survival

Biopsy-proven acute rejection (BPAR) occurs in approximately 10% of kidney transplant recipients in the first year making superiority trials unfeasible. IBOX, a quantitative composite of eGFR, proteinuria, anti-HLA DSA, and + full/- abbreviated kidney histopathology, is a new proposed surrogate endpoint. BPAR's prognostic ability was compared to iBOX in a pooled cohort of 1534 kidney transplant recipients from four datasets, including two prospective randomized controlled trials. Discrimination analyses showed mean c-statistic differences between both iBOX compared to BPAR of 0.25 (95%CI: 0.17, 0.32) for full iBOX and 0.24 (95%CI: 0.16, 0.32) for abbreviated iBOX, indicating statistically significantly higher c-statistic values for the iBOX prognosis of death-censored graft survival. Mean (standard error) c-statistics were 0.81±0.03 for full iBOX, 0.80±0.03 for abbreviated iBOX, and 0.57±0.03 for BPAR. In calibration analyses, predicted graft loss events from both iBOX models were not significantly different from those observed. However, for BPAR, the predicted events were significantly (p<0.01) different (observed 64; predicted: 70 full iBOX; 76 abbreviated iBOX; 173 BPAR). IBOX at one-year post-transplant is superior to BPAR in the first year post-transplant in graft loss prognostic performance, providing valuable additional information facilitating the demonstration of superiority of novel immunosuppressive regimens.