Creatinine Kinetic Modeling to Estimate Residual Kidney Creatinine Clearance in Patients Being Hemodialyzed Once or Twice Per Week.

Abstract

Background: Knowledge of residual kidney function is potentially useful in patients receiving hemodialysis for risk stratification, adjusting the dialysis prescription, and early identification of renal function recovery. However, periodic urine collection is problematic. We examined the potential of predicting residual kidney creatinine (water) clearance (KrCrW) without urine collection using a creatinine kinetic model, which allows KrCrW to be estimated based on previously measured or anthropometrically estimated creatinine generation rate (GCr), volume of distribution (VdCr), and measured predialysis serum creatinine.

Methods: Studies were done in 12 patients receiving once weekly hemodialysis and 12 other patients being dialyzed twice a week in whom KrCrW was measured by collection of urine. GCr and VdCr were taken either from the modeling outputs or were estimated from anthropometric values.

Results: The mean modeled GCr was 1091 ± 377 (SD) mg/day, similar to the value predicted by an anthropometric equation suggested by Ix et al. (1198 ± 304). The mean kinetically modeled VdCr was 22.7 ± 2.4 L, somewhat lower than expected. The KrCrW from urine collection was 7.43 ± 4.07 mL/min. Predicted KrCrW from modeled GCr, modeled VdCr, and measured predialysis serum creatinine was similar (7.35 ± 4.01, r² = 0.987) with an average error less than 1%. When anthropometric estimates of GCr and VdCr were used as inputs, the mean modeled KrCrW was somewhat higher (8.66 ± 4.27, y = 1.09x, R² = 0.585) and the mean error was 1.23 ± 2.6 mL/min.

Conclusions: Residual kidney creatinine clearance (KrCrW) can be estimated in patients receiving one or two dialysis treatments weekly based on creatinine kinetic modeling. Using anthropometric estimates of GCr and VdCr in the modeling equations yields similar values of KrCrW to those when modeled GCr and VdCr inputs are used, but with a substantial error. A strategy of using a baseline modeled values of GCr and VdCr for future KrCrW change prediction may be promising, but the stability of GCr over time needs to be confirmed.