A thirty percent chronic decline in inverse serum creatinine is an excellent predictor of late renal allograft failure

• We previously reported that the percentage of change in inverse serum creatinine (Δ1/Cr) was the best of several time-dependent serum creatinine- derived predictors of renal allograft failure in patients not administered cyclosporine (CsA). To further validate the utility of Δ1/Cr, we collected creatinine levels (mean, 90.7 ± 45.2 creatinine measurements) in 100 patients treated for 6.7 ± 5.9 years with CsA. We also validated Δ1/Cr using a limited creatinine-sampling strategy, then performed multivariate Cox proportional hazards analysis of 1,663 transplantations. A time-dependent covariate determined by the date of first chronic decline (excluding creatinine levels from periods of acute rejection) in Δ1/Cr to less than -30% of baseline similarly was predictive of graft failure in 101 patients treated without CsA (relative risk, 5.04;95% confidence interval, 2.18 to 11.6; P = 0.0002) and 100 patients treated with CsA (relative risk, 5.02;95% confidence interval, 2.50 to 10.1; P < 0.0001). A limited creatinine-sampling strategy (measured at 1 week, 1,3,6,12,18,24, and 36 months, and each year thereafter) reduced the ability of Δ1/Cr less than -30% to predict graft failure. In 1,663 patients, Δ1/Cr less than -30% first occurred a median of 1.0 years posttransplantation (n = 792 of 1,663 patients) and 3.0 years before graft failure (n = 478 of 897 patients with graft failure). In a multivariate model (n = 1,663) that included baseline function, acute rejection, and other covariates, Δ1/Cr less than -30% was a strong independent predictor of graft failure (relative risk, 2.56; 95% confidence interval, 2.12 to 3.09; P < 0.0001). Thus, Δ1/Cr less than -30% is an excellent predictor of graft failure that is similarly predictive in patients treated with and without CsA. A limited sampling strategy for creatinine diminishes, but does not negate, the usefulness of Δ1/Cr less than -30%.