Integrated approach to overcome a food effect in clinical studies: an example of how in vitro, in vivo and simulation tools can help in determining an appropriate strategy
M. Petrone (1), S. Beato (2), R. Gomeni (1), F. Gray (3)
(1) GlaxoSmithKline, Clinical Pharmacology Modelling and Simulation, Verona, Italy, (2) GlaxoSmithKline, Pharmaceutical Development, Verona, Italy, (3) GlaxoSmithKline, Clinical Pharmacology and Discovery Medicine, Harlow, UK
Objectives: The between-subject variability in the absorption parameters may be reduced in humans when a drug is given with a standard breakfast, especially when its solubility is pH-dependent [1,2,3]. However co-administration of a drug with food is often considered a burden for its further development. The objective of the present work is to suggest an integrated, data-driven formulative approach to overcome the liability to progress a drug to further clinical studies with the burden of this food effect.
Methods: All available pre-clinical and clinical data of a weak base new chemical entity (NCE) were reviewed to decide the best dosing strategy to meet the study objectives of high exposure and controlled variability. Chemico-physical information on solubility and its pH dependency, permeability and food-binding were considered. In addition, single dose data were obtained in fed and fasted state. Simulations of absorption profiles were conducted with the use of Gastroplus [4,5] and other pharmacokinetic simulation tools, assuming the administration of the compound with beverages stabilizing the pH of the gastric media at different values.
Results: Review of clinical data showed that co-administration with food reduced between-subject variability on both Cmax and AUC (0-24) with no significant impact on mean exposure; this was more apparent at high single doses. Food appeared to improve the PK profile playing the role of an absorption regulating factor: reducing absorption in fast absorbing subjects and improving absorption in the slow absorbing subjects with an overall reduced variability. The pH-dependent solubility of the NCE, the wide physiological range of pH found in the stomach in fasted and fed conditions, the observed food binding of the molecule, all may have contributed to the large PK variability seen when the molecule was dosed in the fasted state. As stomach pH ranges physiologically between 2 and 5, the solubility of the molecule was found to decrease by ten thousand-fold and even when administering at low doses, there was a significant risk of incomplete solubilisation. Different simulations accounting for all these factors showed that dosing the drug with beverages with appropriate acidity in the fasted state may reduce between-subject variability. The same PK profile and variability was then expected when dosing in the fed state compared to the fasted state with drinks such as coke or ribena with relatively low pH values, suggesting a formulation opportunity to overcome this food effect.
Conclusions: The integration of in-vitro and in-vivo data in a simulation tool like Gastroplus, provided a model-driven opportunity to integrate data and knowledge regarding an NCE. The benefit of this approach is to identify the optimal experimental strategy to progress compounds characterized by lower variability in fed condition. Understanding the source of variability guides most effective formulation design to address and mitigate the impact of physico-chemical properties on drug absorption.
 Physiological Pharmaceutics : Biological barriers to drug absorption (2001) N Washington, C Washington & C G Wilson pub Taylor & Francis (London) ISBN 0-748-40610-7 (hardback) & ISBN 0-748-40562-3 (paperback)
 Feldman M & Barnett C ,1991 Dig Dis Sci. 37: 866-869
 Chin T.; Loeb M.; Fong I. Antimicrobial agent and chemotherapy, Aug 1995, Vol. 39 N.8: 1671-1675
 Wei, H. ; European Journal of Pharmaceutics and Biopharmaceutics 2008 ISSN: 0939-6411 Vol. 69 Issue: 3, 1046
 Parrott, N.; Molecular Pharmaceutics 2008 ISSN: 1543-8384 Vol. 5 Issue: 5 , 760