Utility of serum beta-trace protein as a tool for estimating residual kidney function in peritoneal dialysis patients

Abstract

Background: Beta-trace protein (BTP) is a novel marker for residual kidney function (RKF) without need for urinary collection. We aimed to examine its utility as a tool for estimating RKF in incident peritoneal dialysis (PD) patients. Methods: This was a post hoc analysis of incident PD patients from the balANZ trial cohort. The outcomes evaluated were trends of serum BTP concentration with time, factors associated with change in BTP using mixed-effect multilevel linear regression and correlation of BTP with mean urinary urea and creatinine clearances (measured glomerular filtration rate (GFR)). Performances of two BTP-derived equations (Shafi-Eqn and Steubl-Eqn) to estimate GFR were evaluated by reporting bias (median difference between estimated and measured GFR), precision (interquartile range of median bias), accuracy (±2 mL/min of measured GFR) and P30 (percentage estimates within 30% of measured GFR) with confidence intervals (CIs) generated by bootstrapping 2000 replicates. The agreement between BTP-estimated GFR and measured GFR was also plotted graphically on Bland–Altman analysis. Results: The study included 161 PD patients. BTP concentration increased with dialysis vintage and was inversely correlated with measured GFR ( r = −0.64). Larger increases in BTP were associated with longer PD vintage and higher dialysate glucose exposure. Biases of BTP-estimated GFRs (Shafi-Eqn and Steubl-Eqn) were 1.2 mL/min/1.73 m2 (95% CI 1.0–1.3 mL/min/1.73 m2) and 0.4 mL/min/1.73 m2 (95% CI 0.2–0.6 mL/min/1.73 m2), respectively. Both BTP-estimated GFRs had poor precision (3.2 mL/min/1.73 m2 (95% CI 2.9–3.5 mL/min/1.73 m2) and 2.8 mL/min/1.73 m2 (95% CI 2.5–3.2 mL/min/1.73 m2), respectively) and accuracy of estimates (55% (95% CI 52–60%) and 59% (95% CI 55–63%), respectively). The mean difference of BTP-estimated GFR (Shafi-Eqn and Steubl-Eqn) and measured GFR were −1.14 mL/min/1.73 m2 and −0.42 mL/min/1.73 m2, respectively, with large limit of agreement on Bland–Altman plot. Conclusions: Serum BTP level was inversely related to RKF but neither BTP-estimated GFR equations were sufficiently accurate for routine use in PD patients.