Viktoria Stachanow (1,2), Johanna Melin (1,2,3), Robin Michelet (1), Oliver Blankenstein (4), Uta Neumann (4), Wilhelm Huisinga (5), Richard Ross (6), Martin Whitaker (6), Charlotte Kloft (1)
(1) Freie Universitaet Berlin, Germany, (2) Graduate Research Training Program PharMetrX, Germany, (3) Quantitative Clinical Pharmacology, AstraZeneca, Gothenburg, Sweden, (4) Charité-Universitaetsmedizin, Berlin, Germany, (5) Institute of Mathematics, Universitaet Potsdam, Germany, (6) The University of Sheffield, UK
Objectives: Congenital adrenal hyperplasia (CAH) is a rare form of adrenal insufficiency leading to low or no biosynthesis of cortisol. Patients require a lifelong cortisol replacement therapy from birth. For paediatric patients hydrocortisone (HC, synthetic cortisol) is the recommended glucocorticoid [1]. HC has complex pharmacokinetics (PK) with saturable binding to corticosteroid-binding globulin (CBG), non-saturable binding to albumin and erythrocytes [2] and a dose-dependent oral bioavailability due to saturable absorption [3]. Dried blood spot (DBS) sampling is a sampling technique which is suitable for paediatric patients as it allows to collect full blood samples of small volumes (10–20 µL) for multiple times [4] and is therefore commonly used when monitoring the treatment of paediatric CAH patients. The objective of this analysis was to extend an established paediatric HC PK-model [5] with DBS data in order to improve the data density and to further understand and characterise the binding processes of cortisol, including binding to erythrocytes. The extended model is intended to be used to conduct simulations comparing different dosing regimens and thereby pave the way towards the optimisation of the cortisol replacement therapy in children.
Methods: A semi-mechanistic HC PK model that was based on adult plasma data from a phase I study for a novel HC formulation [6] has previously been reduced to a paediatric model using sparse paediatric plasma data from a phase III study in 24 patients with adrenal insufficiency [7]. In this phase III study cortisol plasma concentrations and DBS concentrations of cortisol and 15 other steroids, including the CAH biomarker 17α-hydroxyprogesterone (17-OHP), were collected. Additional DBS concentrations of cortisol and 14 further steroids were obtained from a follow-up study. The relation between plasma and DBS samples was characterised by a graphical evaluation using R (3.4.4). Finally, nonlinear mixed-effects modelling was applied using NONMEM 7.4 in order to extend the existing binding model, that included binding to CBG and albumin, with binding to erythrocytes and thereby informing the established HC model with DBS data.
Results: Plasma concentrations of cortisol were considerably higher than the corresponding DBS concentrations which were sampled at the same time points. The plasma/DBS cortisol concentration ratios ranged between 2 and 8 with similar inter- and intraindividual variability. The relation between the cortisol DBS concentrations and cortisol plasma concentrations showed nonlinear behaviour with plasma/DBS cortisol concentration ratios that decreased with increasing concentrations. The one-compartment model including plasma and DBS data incorporated nonlinear binding of cortisol to CBG, (equilibrium dissociation constant Kd= 9.71 nmol/L) as well as linear binding to albumin and to erythrocytes, described by linear non-specific binding parameters (NSalb=6.48 and NSery=1.93, respectively). Goodness-of-fit plots showed that the plasma data were adequately described by the model while DBS concentrations were partly underpredicted at lower concentrations.
Conclusions: Observing that plasma concentrations are higher than DBS concentrations was expected since these differences in the concentrations of full blood and plasma samples are due to the removing of the cellular component, reflected in the haematocrit (Hct). However, we also observed very high plasma/DBS cortisol concentration ratios as well as high inter- and intraindividual variabilities. The nonlinear relation between DBS and plasma concentrations mirrors the nonlinear binding kinetics of cortisol to CBG, which potentially affects the binding of HC to erythrocytes. The next step in the model development is to optimise the model and its predictive performance for DBS data and to identify covariates such as individual Hct values and CBG concentrations for describing the substantial differences and variabilities between plasma and DBS concentrations that were identified in the graphical analysis.
References:
[1] Merke, D. P. et al.: Lancet. 2005, 365: 2125-2136.
[2] Lentjes, E. G. W. M. et al.: J Clin Endocrinol Metab. 1999, 84: 682-687.
[3] Toothaker, R. D. et al.: J Pharmacokinet Biopharm. 1982, 10: 147-156.
[4] Wilhelm, A.J. et al: Clin Pharmacokinet. 2014, 53: 961-973.
[5] Melin, J.: Dissertation 2017.
[6] Melin, J. et al.: Clin Pharmacokinet. 2018, 57: 515-527.
[7] Neumann, U. et al.: Clin Endocrinol. 2018, 88: 21-29.
Reference: PAGE 28 (2019) Abstr 9097 [www.page-meeting.org/?abstract=9097]
Poster: Drug/Disease Modelling - Paediatrics