IV-65 Dan Wright

A population pharmacokinetic model for 51Cr EDTA, an isotopic biomarker for renal function: impact on carboplatin dosing.

Daniel F.B. Wright

School of Pharmacy, University of Otago, Dunedin, New Zealand

Objectives: 

  1. To develop and test a population pharmacokinetic (PK) model for 51Cr EDTA disposition.
  2. To compare the model predictions for renal function (glomerular filtration rate, GFR) to those estimated by conventional methods.
  3. To determine if differences in predicted GFR could change dosing decisions in the clinic using carboplatin as a test case.

Methods: Data from 40 individuals who received 2 mL (~3.7 MBq/mL) of 51Cr EDTA were available for analysis [1]. Four plasma concentrations were measured at approximately 2, 4, 6 and 24 hours after the dose. A population analysis was conducted using NONMEM® v.7.2. Model stability was assessed using a simulation-estimation (sim-est) procedure. Predictions of GFR from the PK model were compared to conventional methods for estimating renal function using mean prediction error (MPE) and root mean square error (RMSE). Creatinine clearance (CLcr) estimated using 24 urine collection was used as a control. Carboplatin dosing was simulated using the Calvert formula [2] for two hypothetical patients. 

Results: A total of 159 51Cr EDTA plasma concentrations from 40 patients were analysed. A two compartment PK model with first-order elimination best fit the data. Significant covariates included creatinine clearance on 51Cr EDTA clearance and weight on central volume. The control, CLcr estimated using a 24-hour urine collection method, provided the closest predictions to the model and was unbiased. However, commonly used eGFR equations (i.e. Cockcroft Gault, MDRD and CKD-Epi) led to negatively biased estimates relative to the model (MPE -19.5 mL/min/1.73m2, -20.6 mL/min/1.73m2 and -16.9 mL/min/1.73m2, respectively). The commonly used ‘slope-intercept’ method for estimating isotopic (51CR EDTA) GFR led to positively biased estimates compared to the model (MPE 15.1 mL/min/1.73m2). Carboplatin doses for a healthy young male representing a typical testicular cancer patient with normal renal function were over-predicted by 116 mg (13%) when GFR was determined using the slope-intercept method versus the model and under-predicted by 134mg (15%), 222mg (25%) and 156mg (18%) when using the eGFR equations, Cockcroft Gault, MDRD and CKD-Epi, respectively.

Conclusions: A population pharmacokinetic model for the disposition of 51Cr EDTA was developed and evaluated. The model provided unbiased estimates of renal function. The biased renal function estimates provided by both eGFR equations and the conventional method for estimating isotopic (51CR EDTA) GFR would lead to clinically important differences in carboplatin doses. 

References:
[1] Putt TL et al. (2014) Eur J Clin Pharmacol 70:1221-6
[2] Calvert AH et al. (1989) J Clin Oncol 7:1748-56

Reference: PAGE 25 (2016) Abstr 5726 [www.page-meeting.org/?abstract=5726]

Poster: Drug/Disease modeling - Other topics