PulmoSim: A Physiologically-Based Mathematical Model Software Package to Predict Lung Retention and Inhaled Pharmacokinetics of Therapeutic Candidates
Balaji M. Agoram, Massimiliano Germani*, Maurizio Rocchetti*, Francesca Del Bene*, Rhys Jones & Piet H van der Graaf,
Pfizer Global Research & Development, Sandwich CT13 9NJ, UK and *Accelera, Nerviano (Milan), Italy
Objectives: Retention within the lung tissue is widely believed to enhance the duration of action of inhaled therapeutics. However, the physicochemical factors that influence lung retention are not fully understood; hence rational design of potentially lung-retained inhaled molecules faces substantial hurdles.
The main aim of this work is to develop an in silico approach for designing lung retention. The specific objectives are: (1) integrate in vivo processes such as dissolution, retention, absorption, and systemic elimination of inhaled compounds quantitatively within a physiologically-based mathematical model with physicochemical determinants, (2) evaluate the model by comparing predictions of retention with experimental rat data for seven inhaled compounds and (3) create a user-friendly software package implementation of the model.
Methods: The model is based on a system of 15 ordinary differential equations describing drug dissolution, absorption across lung epithelium, lung tissue binding, systemic distribution and elimination processes. In vitro permeability and tissue binding data were generated for 7 small molecule compounds of widely different physicochemical characteristics and target mechanisms using Caco-2 systems and rat lung tissue. Plasma concentration data were also generated for these compounds in rats after intravenous (IV) and intratracheal (IT) solution dosing. Stochastic model simulations, accounting for uncertainty in model parameters such as permeability and tissue distribution, were performed to predict PK of IT compared to IV dosing. Retention of compound in the lung was assessed based on a visual examination of predicted lung and plasma IT PK profile compared to the IV profile. The model was implemented using MATLAB programming language.
Results: The model predictions were classified as either high or low confidence based on the width of predicted confidence intervals. Among high confidence predictions (5/7 candidates), all predictions of lung retention and lack thereof were accurate. Among low confidence predictions (2/7 candidates), 1 prediction of retention was accurate.
Conclusions: A mathematical model representing physiological processes during inhaled absorption has been developed and accurately predicts lung tissue retention in rats for the range of molecules tested. The model has the potential to substantially impact inhaled drug discovery and hence "inhalation by design" paradigm. The model has been implemented in a user-friendly software package and can run as a standalone application without additional software.