Pharmacokinetic Modeling of Hydrocortisone by Including Protein Binding to Corticosteroid-Binding Globulin

Abstract

Background: Patients with adrenal insufficiency are treated with oral hydrocortisone (HC) to compensate for the loss of endogenous cortisol production. Intrinsic imperfections of cortisol replacement strategies in mimicking normal cortisol secretion are the underlying cause of the increased morbidity and mortality of patients suffering from secondary adrenal insufficiency (SAI). To improve oral hydrocortisone substitution therapy, a better understanding of its pharmacokinetics (PK) is necessary. The previous PK model did not include protein binding. It is known that protein binding can impact hydrocortisone pharmacokinetics. The aim of this study is to describe HC pharmacokinetics including the protein-binding state using Edsim++ (Mediware, Prague) pharmacokinetic modeling software, paving the way for an in-silico tool suitable for drug delivery design. Methods: A total of 46 patients with SAI participated in a randomized double-blind crossover study Patients randomly received a low dose of HC (0.2–0.3 mg/kg body weight/day) for 10 weeks, followed by a high dose (0.4–0.6 mg/kg body weight/day) for another 10 weeks, or vice versa. Plasma samples were obtained and analyzed for free and total hydrocortisone. Single compartment population pharmacokinetic analysis was performed using an extended Werumeus-Buning model built in Edsim++. This model includes a mathematical approach for estimating free cortisol by Nguyen et al., taking the protein binding of HC to albumin and hydrocortisone-binding globulin (CBG, transcortin) into consideration, as well as different states of CBG which affect binding kinetics to HC. The goodness of fit for observed versus predicted values was calculated. Results and conclusions: Nguyen’s formula for free cortisol estimation was successfully implemented in a pharmacokinetic model. The model shows high Spearman’s correlation for observed versus predicted hydrocortisone concentrations. Significantly higher correlations (Spearman’s r, 0.901 vs. 0.836) between total and free hydrocortisone AUC24 (area-under the curve over 24 h) are found when comparing new and old models. This new model was used to simulate the plasma concentration–time behavior of a more suitable hydrocortisone formulation.