Page 77 - Read Online

Page 28 Pandey et al. J Transl Genet Genom 2021;5:22-36 I http://dx.doi.org/10.20517/jtgg.2020.45

where θ and θ are the V and CL of a 70-kg patient. The IIV was high for V, CL, k , absorption rate
c
c
a
CL
Vc
constant, and k , the rate constant for transit from central to peripheral compartment. The IIV in PK
cp
parameters was assumed to be an exponential function of η, as shown in Equation (2). The model
performance was evaluated using a simulation-based diagnostic tool, visual predictive check (VPC). Based
[33]
on the VPC results, the model gave good fits with the observed PK data .
[34]
Similarly, Wiczling et al. developed a population PK model to capture the variation of HU concentration
in plasma and urine with time. In this study, a one-compartment model with first-order elimination
through renal and non-renal pathways provided good fits to the data. Additionally, a gamma-distributed
absorption rate (transit absorption model) was used owing to the delay in absorption observed in several
patients. In the transit absorption model, the absorption compartment consists of N transit compartments,
t
and the input to the central compartment, u(t), was given by the following equation :
[34]
(7)

where F is the bioavailability, D is the drug dose, and k is the transit rate in between compartments. k is
tr
tr
given by (N + 1)/MTT, where MTT is the mean transit time. The proportional residual error model given
t
by Equation (4) was used to account for RV . The following equation gave the individual parameters:
[34]
(8)

th
where θ and θ median are the PK parameters for the i individual and median covariate, COV is the
i
i
th
continuous covariate of i individual, COV median is the median value for a particular covariate, and θ COV is
the regression coefficient. The weight was identified as a significant covariate for the apparent volume of
distribution, V/F, and CL/F due to the metabolic pathway. The model performed well, as evaluated from
the goodness of fits, VPC plots, and the individual patient fits obtained for the variation of HU plasma
[34]
concentration and HU urine amount with time .
In another study, Estepp et al. performed a population PK study in children with SCD. The model
[35]
[34]
was similar to the model developed earlier by Wiczling et al. . However, elimination through only one
pathway was considered. The study participants received HU in capsule and liquid forms. Since the drug
was administered on two occasions, the model expressed the PK parameters as a function of two random
variables to account for inter-individual (η) and inter-occasion variability (k) .
[35]
(9)

where θ is the set of PK parameters for i individual and k occasion. V and CL were expressed as
th
th
c
ik
functions of bodyweight using Equation (5). They showed that their model adequately described the data
based on the goodness-of-fit plots and VPC plots for liquid and capsule formulations. The individual
patients fit for PK data in patients receiving the liquid and capsule formulations showed that the model
simulation matched the observed data well .
[35]
Dong et al. developed a dosing strategy using individual patient PK profile to reduce the time to reach
[54]
MTD and maximize the effect of HU. Using D-optimal design, they identified that only three plasma
samples at sampling times of 15-20 min, 50-60 min, and 3 h post-dosing are needed to estimate HU
exposure. They described the relationship between the average PK parameter and continuous covariates
shown in the following equations by normalized power models or linear models :
[54]
(10)