Results
Fifty-nine participants were included from three studies: 10 pregnant
women with preterm preeclampsia from the PIE trial and 49 non-pregnant
individuals (55% female), 30 from Helgadóttir et al . and 19 from
Hunfeld et al .13–15 Median (range) age was 30
(21 – 43) years for pregnant patients and 24 (18 – 46) years for the
non-pregnant individuals while weight was 99 (56 – 126) kg for pregnant
patients and 74 (54 – 107) kg for the non-pregnant individuals.
Genotype information was available for the study by Hunfeld et
al ., with 1 poor metabolizer and 18 extensive metabolizers. Participant
characteristics are summarised in Table 1. A total of 1064
concentrations were obtained, of which 68 (6%) were BLQ. Ten (1%)
samples from PIE were identified as outliers through goodness-of-fit and
individual plots and were removed from the analysis.
Esomeprazole pharmacokinetics was best characterized by a
two-compartment disposition model (ΔOFV=-225, p < 0.0001
compared with one-compartment) with first-order elimination and
absorption with transit compartments. Allometric scaling with body
weight improved the model fit (ΔOFV=-14.4). Final parameter estimates
and precision are presented in Table 2. For a typical 70-kg individual,
clearance in extensive metabolisers was more than 3 times faster than in
slow metabolisers, 24.3 (20.5 – 29.3) versus 7.87 (6.04 – 9.75) L/h.
In pregnant women, clearance was 42.2% (14.9% − 61.6%) slower
compared to non-pregnant after single dose
(∆OFV=-8.57, p < 0.005).
Bioavailability was not found to be significantly different in pregnant
women. In non-pregnant individuals, clearance was 54.9% (48.2% –
63.5%) slower in extensive metabolizers on day five compared to day one
(∆OFV=-142, p << 0.0001), and this effect could not
be estimated for poor metabolizers. Bioavailability was 33.0% (10.0% -
52.0%) higher on day five (∆OFV=-14.4, p < 0.0001), and this
effect could not be isolated by genotype.
We found a fast absorption with large between-occasion variability of
485% (211% – 916%). Considering differences in formulation and
mealtimes between the three studies (Table 1), study (dataset) was added
as a covariate on absorption parameters. Absorption was slowest for
individuals in the study by Helgadóttir et al ., with a mean
transit time (MTT) of 1.74 h, compared to those in PIE and the study by
Hunfeld et al . (0.491 h and 0.988 h, respectively). The visual
predictive check (Figure 1) shows acceptable agreement between the model
and the data.