STUDY
DESIGN
Study design
considerations are important when conducting a study in pregnant women to
determine if the PK and/or PD are altered enough to require an adjustment from
the established dosage. Ideally, PK studies in pregnancy would be done
pre-pregnancy (for baseline comparison) and during all three trimesters,
especially for chronically administered drugs. Given the constraints of a study
design that enrolls women prior to pregnancy, an alternative can be to determine
PK/PD in the second and third trimesters, with the baseline assessment for
comparison to the pregnant state done in the postpartum period.
The Agency
recommends care be taken to select the most appropriate postpartum time for PK/PD
determination, if known. Cardiovascular and renal changes do not return to the
pre-pregnancy state until 3 months postpartum. Optimally, postpartum PK/PD
assessments for comparative purposes to PK/PD in pregnancy would be done when
the woman is neither pregnant nor lactating.
The PK and/or PD study
can also be nested within a larger clinical study on safety, efficacy, and
outcomes of interest (e.g., Prevost et al. performed a study on the PK of
nifedipine on a small subset of patients who were participating in a larger
clinical study to assess treatment for pregnancy-induced hypertension (Prevost
1992)).
A. Longitudinal Design
For drugs that are
administered chronically or given for several treatment cycles during pregnancy,
a longitudinal study design is most clinically meaningful. This allows for
intensive PK studies in pregnant women conducted serially so that each woman
serves as her own control, avoiding the common criticism that PK/PD studies in
pregnant women are flawed because of the comparison group employed (Reynolds
1991; Little 1999). Such a study would focus on comparing each pregnant woman
enrolled at one trimester of pregnancy to the same patient at a different
trimester as well as during the postpartum period. We recommend that the
rationale for which trimesters are chosen be stated clearly in the study
protocol. This longitudinal design minimizes interindividual variability across
gestational ages; however, intraindividual variability would be taken into
account when determining the sample size. It is important that the analytical
plan take into consideration the repeat measures characteristics of a
longitudinal design.
Because physiologic
changes are continuous throughout pregnancy, and abrupt changes do not
necessarily coincide with each trimester shift, the Agency recommends that
investigators consider narrowing the time of sampling from trimester to a
window of time during each trimester. For example, 4-week windows
can be selected for second trimester (e.g., 24-28 weeks) and third trimester
assessments (e.g., 34-38 weeks).
The Agency recommends
that each woman serve as her own control and have PK/PD determinations performed
at different trimesters and in the postpartum period. For certain drugs that
are given acutely (e.g., single dose or short course of therapy) it can be
difficult to implement a longitudinal design using the same subjects throughout
and after pregnancy. For example, in certain circumstances drug therapy may no
longer be medically essential in the postpartum period. In these situations, a
multi-arm study can be designed to compare different pregnant subjects at
different trimesters and in the postpartum period.
B. Population PK Design
A population PK approach with nonlinear mixed
effects modeling techniques can be used as an alternate way to enroll pregnant
women in PK studies and minimize the number of blood draws and PD assessments.
The population PK approach can assess the impact on the PK of a drug on various
covariates, such as maternal characteristics (e.g., age, gravity, parity, race,
weeks or trimester of gestation), concomitant medications, and underlying
medical conditions. For example, a measure of pregnancy status such as weeks
gestation can be one of the covariates, making it possible to model the
relationship between gestational age of pregnancy and PK parameters such as the
apparent clearance of the drug (CL/F).
In principle, a population PK study design and
analysis might detect PK differences large enough to warrant dosage adjustment
if the study has enough pregnant and nonpregnant women enrolled with sufficient
representation of second and third trimesters (with a continuum of gestational
ages from 13 to 40 weeks). Typically, each patient is only sparsely sampled to
obtain plasma drug concentration data and/or PD data. Due to the intrinsic
characteristics of a population PK study, the controls for this study design can
differ from other study designs and can potentially include matched healthy
nonpregnant female volunteers. To ensure the ability to determine the
inter-occasion variability and prevent a parallel group trial design, a cohort
of study subjects would have data collected from all trimesters and the
postpartum period. Considering the number of subjects in the study and the key
objective of the study, efforts can be made to reduce the number of influential
covariates such as concomitant medication.
Some investigators have proposed conducting a
population PK study as a preliminary step and to subsequently conduct a standard
intensive PK/PD study if the population PK study suggests changes between the
pregnant and nonpregnant women (Stika 2000). For further information about the
population PK approach, see the Guidance for Industry
|
