Modeling cortisol and cortisone binding to corticosteroid-binding globulin and albumin based on clinical protein binding data.

Abstract

Cortisol is a major endogenous glucocorticoid that regulates numerous physiological processes. In plasma, cortisol and its inactive metabolite cortisone bind to corticosteroid-binding globulin (CBG) and albumin, leaving only the unbound fraction available for receptor activation and metabolism. Changes in ligand or protein concentrations alter unbound fractions. Existing binding equations are difficult to extend to multiligand, multiprotein systems and do not readily capture competitive endogenous binding interactions. The goal of this study was to develop a plasma protein binding model that quantitatively describes binding species and predicts unbound concentrations across physiological states. Total and unbound cortisol and cortisone, CBG, and albumin were measured in plasma from healthy premenopausal women (n = 13) at baseline and after 7 days of 30 mg hydrocortisone treatment. Reversible 1:1 binding models were implemented in COPASI and MATLAB/Simulink, and dissociation constants (K_d) were estimated by fitting equilibrium binding models to observed unbound concentrations. A model describing simultaneous binding of cortisol and cortisone to CBG and albumin yielded in vivo K_d values for cortisol:CBG, cortisone:CBG, cortisol:albumin, and cortisone:albumin of 0.0130 μM, 0.169 μM, 172 μM, and 519 μM, respectively. Model predictions agreed with observed unbound cortisol and cortisone concentrations, and bootstrap resampling confirmed stable K_d estimates. This work provides a quantitative framework for predicting unbound cortisol and cortisone concentrations across physiological and disease states by accounting for changes in ligand and protein concentrations. This enables extrapolation to conditions such as pregnancy, adrenal insufficiency, and liver disease, informing interpretation of altered cortisol concentrations in these populations. SIGNIFICANCE STATEMENT: This work establishes a framework to predict in vivo cortisol and cortisone binding. The developed model was applied to predict unbound cortisol and cortisone concentrations in physiological and pathophysiological states and can be integrated into pharmacokinetic models. Our analysis demonstrates that cortisol and cortisone binding affinities estimated in the native plasma environment differ from those measured using purified proteins. These differences have important implications for predicting and analyzing unbound cortisol concentrations.