A systematic framework to integrate preclinical data within PBPK models: from global sensitivity analysis to middle-out approaches

Abstract

Whole-body physiologically-based pharmacokinetic (PBPK) models have many applications in academic and pharmaceutical research and drug development. However, optimization of parameters in such complex models by fitting models to observed data is a challenging and time-consuming process. The models are often unidentifiable/over-parameterized given the large number of parameters and availability of data which are mostly limited to plasma observations. It is common practice to fix some parameters and estimate others. However, the decision on which parameters to fix and which to estimate is subjective and therefore the final model and parameters may vary significantly between different investigators. This was a concern highlighted by regulatory agencies in their guidance documents for PBPK modelling. Hence, the overall aim of this thesis was to develop a systematic approach for integrating preclinical data within PBPK models to address this issue. The first part of this thesis was focused on identifying key drug-dependent and physiological parameters that influence predictions of tissue-to-unbound plasma partition coefficients (Kpus) and thus drug distribution in PBPK models. The impact of these parameters was evaluated on the Kpus predicted by the Rodgers and Rowland model using sensitivity and uncertainty analyses. For most drug classes, LogP and fraction unbound in plasma (fup) were generally the most influential parameters for Kpu predictions. Uncertainty in tissue composition parameters especially acidic phospholipid concentrations and extracellular protein tissue:plasma ratios, could have a large impact on Kpu predictions for all classes. For parameter estimation involving PBPK models and dimensionality reduction purposes, less influential parameters might be assigned fixed values depending on the parameter space, while influential parameters could be subject to parameter estimation. Secondly, several model reduction approaches were investigated to simplify PBPK model structure or dimensionality and thus facilitate the estimation process during PBPK model development. Tissues were clustered according to physiological information reducing the number of unknown parameters without changing the overall PBPK model structure. The investigated mechanistic models in conjunction with preclinical in vivo data were able to provide suitable estimates of Kpus using the nonlinear mixed effect method. To that end, diazepam was used as a case example. This analysis provided a basic framework for PBPK model development and estimation of distribution parameters and subsequent applications of PBPK modelling, especially translation of drug distribution from animals to humans. Subsequently, the use of the investigated mechanistic models for interspecies extrapolation was evaluated. The models that could best fit data in rats and monkeys were applied for translation of drug distribution to humans. The performance of these best models was assessed for three compounds (diazepam, midazolam and basmisanil) and compared to the whole-body PBPK model with Kpu predictions from the Rodgers and Rowland model. Using the approach of simplified PBPK models with common scalars and the best models, PK profiles could be well described in preclinical species and plasma profiles were successfully predicted in human for diazepam and midazolam. This proof of concept was shown for lipophilic weakly basic compounds. For an exhaustive evaluation, the work and models proposed herein may be extended to different drug classes and more compounds. The PBPK modelling framework presented in this work for drug distribution and prediction of human PK could also be applied to translation within species e.g., from an adult to a paediatric population.

Date of Award	31 Dec 2021
Original language	English
Awarding Institution	The University of Manchester
Supervisor	Leon Aarons (Supervisor), Kayode Ogungbenro (Supervisor) & Adam Darwich (Supervisor)