A direct oral anticoagulant rivaroxaban fails to prevent stroke and systemic embolism in one-to-several percent of patients with nonvalvular atrial fibrillation (NVAF), but the reasons are unknown. The study used semi-mechanistic in vitro-in vivo prediction (IVIVP) modeling to explore the reasons for ineffective thrombosis prevention in NVAF patients. Steady-state drug concentrations in plasma were measured at 0 h (Ctrough), 3 h (C3h), and 12 h post-dosing in thirty-four patients treated with 20 mg rivaroxaban daily. The clinical data were compared against "virtual twins" generated with a novel IVIVP model that combined drug dissolution modeling, mechanistic description of gastric drug transit, and population pharmacokinetics defining the variability of drug disposition. The nonresponders had significantly lower C3h and Ctrough than the responders (p < 0.001) and the covariates included in the population pharmacokinetic submodel did not fully explain this difference. Simulations involving varied gastrointestinal parameters in the "virtual twins" revealed that lower small intestinal effective permeability (Peff), rather than a slower stomach emptying rate, could explain low rivaroxaban exposure in the nonresponders. IVIVP modeling was effectively used for exploring pharmacotherapy failure. Low Peff, found as a major determinant of ineffective rivaroxaban treatment, encourages further research to find (pato)physiological factors influencing suboptimal absorption.
Keywords: Atrial fibrillation, biopredictive dissolution; Fed state; In-vitro-in-vivo modeling; Pharmacotherapy efficiency; Population pharmacokinetics.
Copyright © 2023. Published by Elsevier B.V.