Parameterization of In Silico Oral Disposition Models: Focus on Pediatrics

Abstract

Owing to their biologically relevant design, physiologically-based pharmacokinetic (PBPK) models require quantitative knowledge of organism anatomy and physiology to facilitate appropriate parameterization. Within such models, an intrinsic relationship exists between the quality of input parameters and the confidence bestowed upon simulated outputs. Therefore, in order to instil confidence in PBPK model predictions of pediatric pharmacokinetics (PK), a fundamental understanding of age-specific changes in anatomy and physiology is required. However, due to a lack of consensus and general paucity of biological data denoting pediatric gastrointestinal (GI) physiology, parameterization of mechanistic oral disposition models in this population is quite challenging. The current dissertation expands our understanding of the ontogeny of key physiological aspects regulating oral drug disposition and serves to highlight differences between children and adults. In addition, the thesis describes essential processes involved in the development of pediatric PBPK models as well as demonstrates the use of such models as tools for identification of human physiological values – a utility that is of potential interest particularly for children, where several biological knowledge gaps persist. To illustrate the key processes involved in the rational development of pediatric PBPK models, a structured workflow was proposed and subsequently utilized to develop age-specific PK predictions for the benzodiazepine, lorazepam. Literature-based assessments of the age-dependency of small intestinal transit time (SITT), GI solubility, and α-1-acid glycoprotein (AAG) employed different methodologies. To discern the influence of age on SITT, random-effects meta-regression models were employed. Investigations assessing age-specific changes in GI fluid parameters (i.e. pepsin, bile acids, pH, osmolality, etc.) were collected from the literature and served to define the composition of a novel set of pediatric biorelevant media representative of the stomach and upper small intestine. Solubility assessments were conducted for seven BCS Class II compounds within the developed pediatric media and a set of reference adult media. Plasma AAG concentrations were assessed in both healthy subjects and those suspected of infection. The analysis evaluated use of linear, power, exponential, log-linear, and sigmoid Emax models to describe the ontogeny of AAG. Predictive performance of the most suitable ontogeny model was evaluated with regards to its ability to estimate pediatric fraction unbound in plasma (fu,p). Predictive performance was measured using average-fold error (AFE) and absolute average-fold error (AAFE) as measures of bias and precision, respectively. To demonstrate the use of PBPK modeling to facilitate predictions of human physiology, plasma concentration-time data depicting oral administration of acyclovir and chlorothiazide in adults were utilized to generate model-based estimates of small intestinal water volume (SIWV). Estimates were based on a framework that consisted of a whole-body PBPK model integrated with a compartmental absorption and transit (CAT) model. Use of the proposed workflow permitted for the development of age-specific PBPK models that provided relatively accurate estimates of lorazepam PK in children in comparison to a competing modeling technique (i.e. Population PK modeling). For SITT, age was not found to be a significant modulator. With regards to the age-dependency of GI solubility, for six of the seven BCS class II compounds investigated, solubility fell outside an 80-125% range from adult values in at least one of the developed pediatric media. The ontogeny of AAG was best approximated using a sigmoid Emax model in both healthy and infected subjects. For estimation of pediatric fu,p, the AAG ontogeny equation derived from this work (AFE 0.97; AAFE 1.24) provided a superior predictive performance in comparison to a previously proposed equation (AFE 0.74; AAFE 1.45). Model-based predictions of SIWV (~116 mL) closely approximated experimentally determined in vivo estimates, demonstrating the utility PBPK modeling as a rational method for investigating aspects of human physiology. The presented work serves to improve the parameterization of PBPK models tasked with simulating oral drug disposition in children; however, more research is still required to address additional knowledge gaps associated with pediatric GI physiology.