Population Pharmacokinetic-Pharmacodynamic Modeling of Granulocyte Colony-Stimulating Factor to Optimize Dosing and Timing for CD34+ Cell Harvesting

Abstract

Granulocyte colony-stimulating factor (G-CSF) mobilizes peripheral blood (PB) progenitor cells from bone marrow (BM) into circulation for PB stem cell transplantation (PBSCT). This study aimed to develop a population pharmacokinetic-pharmacodynamic (PK-PD) model of filgrastim in healthy subjects to optimize PB CD34⁺ cell collection. Plasma filgrastim concentrations and CD34⁺ cell count data were obtained from a clinical study involving healthy Korean subjects. A total of 1378 plasma concentration measurements and 982 CD34⁺ cell count data collected from 53 subjects were used in the PK-PD model. Filgrastim PKs were adequately described by a one-compartment linear disposition model with an additional transit compartment for absorption. Log-transformed body weight was the only significant covariate affecting the volume of distribution and clearance. CD34⁺ cell mobilization was best captured by a modified Friberg model, assuming continual entry of proliferating BM stem cells into PB via a single transit compartment. Simulation results suggested that the 5 μg/kg twice-daily dosing regimen may yield higher CD34⁺ cell counts compared to the 10 μg/kg once-daily regimen for achieving target CD34⁺ cell counts of 20/μL and 50/μL. We successfully developed a robust PK-PD model of G-CSF that optimizes the yield of CD34⁺ cells during allogeneic PBSCT. This model can guide the efficient determination of optimal G-CSF dosing regimens and CD34⁺ cell harvesting strategies.

Keywords: CD34+; filgrastim; granulocyte colony‐stimulating factor; population pharmacokinetic‐pharmacodynamic modeling.

FIGURE 1

Population pharmacokinetic‐pharmacodynamic model structure of the final model. Circ₀, baseline circulating cell count; Circ(t), circulating cell count at time t; CL, clearance; Cp, plasma G‐CSF concentration; Emax, maximum drug effect; EC50, half maximum effective concentration; gamma, self‐replication rate when circulating cell levels decrease compared with baseline value Circ₀; ka, absorption rate constant; kel, elimination rate constant from circulating pool; kprol, proliferation rate constant; ktr_pk, transit rate constant of G‐CSF; ktr_pd, transfer rate constant of CD34⁺.

FIGURE 2

Goodness‐of‐fit plots of the final population pharmacokinetic‐pharmacodynamic model for G‐CSF (a–d) and CD34⁺ (e–h). (a, e) Population predictions against observations; (b, f) individual predictions against observations; (c, g) time against population‐weighted residuals (PWRES); (d, h) population predictions against PWRES.

FIGURE 3

Linear scale (a, c) and semi‐log scale (b, d) of visual predictive checks (VPCs) (500 simulations) of the final model for G‐CSF (a, b) and CD34⁺ (c, d). The observed concentrations and CD34⁺ cell counts are depicted by dots. Solid blue lines indicate 95th, 50th, and 5th percentiles of the observed concentrations. Black dashed lines indicate the predicted medians. Blue shaded regions indicate 95% confidence intervals for the predicted 5th and 95th percentiles, whereas pink shaded regions indicate 95% confidence intervals for the predicted 50th percentiles. Red shaded regions indicate the outlined areas.

FIGURE 4

Time‐median CD34⁺ cell count profiles of 5 μg/kg twice‐daily and 10 μg/kg once‐daily dosing regimens after 4 days (a), 5 days (b), and 6 days (c) in 1000 virtual individuals. The shaded regions indicate the 5th to 95th percentiles.