Food Effect in Pediatric Populations: Current Practice, Challenges, and Future Potential for Use of Physiologically Based Biopharmaceutics Modeling

Abstract

In drug development, the effects of food on oral pharmacokinetics are usually assessed by performing a clinical study in adults where a single dose of the drug is given in a crossover design, and pharmacokinetic parameters derived after dosing in the fasted state are statistically compared to those obtained after a high-calorie meal.¹ When it comes to children, ethical concerns limit the conduct of such studies and current regulatory guidance recommends that new pediatric formulations should be assessed for their food effect in adults to guide dosing in children.¹ However, the validity of this practice can be questioned. For instance, food effects might differ between children and adults because many of the determining physiological factors, such as stomach volume, gastrointestinal pH, gastric emptying time, intestinal bile salt concentrations, and liver blood flow are age-dependent.^{2, 3} Furthermore, meal types and feeding patterns in children are quite different from those in adults, and the high-fat and high-calorie meal used in adult food effect studies can be inappropriate to project effects in young children. For example, a study was performed where adults were dosed with pediatric formulations of paracetamol and ibuprofen in fasted and fed states in a crossover design.⁴ In one arm, the fed state was represented by a 990 kcal standard adult meal whereas an infant 520 kcal formula meal was used in a second arm. Quite distinct fed-state pharmacokinetic profiles were seen for these different meal types. Although the extent of absorption was comparable, the pediatric meal caused slower absorption than the standard adult meal showing that, even for BCS1 drugs, the impact of the meal type should be considered and a pediatric meal may result in different absorption. Further doubts on the validity of the direct transfer of food effects between adults and children were raised by clinical food effects collected for a set of antibiotic suspensions.⁴ Only one out of seven drugs shows a food effect that is qualitatively similar in adults and children (Table 1).

Additional evidence that the current approach for the prediction of pediatric food effects is not optimal was provided in a recent report from Tunehag and colleagues at the FDA.⁵ They analyzed pediatric drug development studies submitted from 2012 to 2022. In that 10-year period, 102 drug products were approved for use in children <6, and 43 drug labels give dosing recommendations regarding food directly transferred from adult findings. Fourteen products are recommended to be taken without food in infants aged less than 2, which is problematic considering that children of this age feed more frequently than adults, typically every 2–3 h, and tend to remain in a semi-fed state. On the other hand, for the drug products that were recommended to be taken with food, labeling often does not specify the food-type.

In the search for better methods to project food effects and guide appropriate dosing in children, physiologically based biopharmaceutics modeling (PBBM) appears to be one of the most promising. PBBM is that part of physiologically based pharmacokinetic (PBPK) modeling that focuses on the integration of physiological knowledge with drug properties and formulation characteristics to predict absorption. PBBM applications currently make up ∼20% of the overall publications in PBPK,⁶ with key applications including formulation selection, pH-dependent drug–drug interaction risk assessment, and food effect projection. PBBM is also increasingly being used to support drug product quality questions.⁷ PBBM predictions of food effect in adults that integrate food-related changes in physiology with data from in vitro biorelevant solubility methods date back more than 20 years.⁸ Since then, multiple publications have reported successful applications; however, predictions can also fail for reasons that are not always clear, and so the regulatory impact has been limited by a lack of best practices, which reduces confidence in predictions. This was recognized by the International Consortium on Innovation and Quality in the Biopharmaceutical Industry (IQ), which took steps to advance this area by verifying PBBM predictions of adult food effect. They followed a standardized workflow for a set of 30 compounds with diverse properties and diverse food effects. Overall, this showed that the food effect could be predicted within a twofold margin for about 80% of the drugs.⁹ In the successful cases, the mechanism of the food effect was linked to well-understood mechanisms such as the changes in gastrointestinal solubility with food as measured in biorelevant media, or to well-known physiological changes such as delayed gastric emptying. The remaining 20% of unsuccessful predictions tended to be associated with more complex mechanisms where in vitro tools are not sufficiently advanced or where modeling platforms cannot include the relevant mechanisms.

The use of PBPK modeling to predict doses in children has been a growth area because it offers the potential to overcome ethical challenges in recruitment and allows consideration of age dependencies in physiology. However, most pediatric-PBPK models have used simple absorption rather than mechanistic absorption models accounting for pediatric physiology. The application of pediatric PPBM for food effect prediction in children is now timely, and recently, a strategy for translation of food effects from adults to children was proposed,¹⁰ including a workflow for stepwise verification of models in adults, leveraging data from clinical studies with the pediatric formulation.

Using the previously described adult clinical pharmacokinetic data for pediatric formulations of paracetamol and ibuprofen, Statelova and colleagues applied PBBM to understand the mechanisms driving the pediatric food effect. They predicted absorption and pharmacokinetics in children using knowledge of age dependencies in physiology included in a PBBM framework. Importantly, they first modeled absorption mechanisms for the pediatric suspensions based on data from the study in adults and, for the fed state, scaled the gastric emptying kinetics to children based on the caloric content of the adult and infant meals. In this way, they could project fasted and fed state pharmacokinetics in children while accounting for formulation specifics, age-dependent physiology, meal type, and meal size.^{10, 11} By comparing their PBBM simulations to clinical data collected in infants, they confirmed that understanding is gained by modeling the pediatric suspensions in adults. The model predictions based on clinical pharmacokinetic data obtained with an infant formula meal were more successful than those based on data obtained with an adult high-calorie meal.

A pediatric PBBM included in a submission to the FDA's Office of Clinical Pharmacology was described for entrectinib.¹² The submitted model was able to capture oral pharmacokinetics in subjects older than 4 by accounting for the lumenal solubilization changes due to changes in gastrointestinal bile salt concentrations with food. Uncertainties in physiological development limited the model reliability in children <4 years of age.

Research into the possibilities and limitations of pediatric PBBM was performed with an amorphous solid dispersion formulation of tacrolimus, which was chosen as a model drug because of the availability of rich pharmacokinetic datasets in children.¹³ As input data, customized in vitro dissolution testing was performed with conditions relevant to pediatric physiologies of different ages. The modeling used two commercial PBBM platforms, Simcyp and Gastroplus. Uncertainties were handled by exploring the sensitivity of the simulated absorption to pediatric-relevant in vivo luminal fluid volumes and bile salt concentrations. It was concluded that the enhanced in vivo solubility achieved via amorphous solid dispersion results in complete absorption in children across a range of ages. However, the precise prediction of oral pharmacokinetics in children was limited by uncertainty in the first-pass metabolism, which is high for tacrolimus.

The projection of food effect in children by direct transfer from studies in adults is not optimal. Although clinical data on pediatric food effects are very limited, several cases have shown distinct behavior in children compared to adults and so better methods are needed. These methods need to account for age-dependent changes in physiology and differences in formulation, food type, and feeding patterns. Physiologically based biopharmaceutics modeling provides a framework integrating knowledge on age-dependent physiology and food effect mechanisms and has already shown its value for the projection of food effects in adults and—in a very few examples—in children. There is room for improvement in PBBM when it comes to the mechanistic description of drug release from enabling formulations such as amorphous solid dispersions or lipid-based formulations.¹⁴ Furthermore, important gaps in our knowledge of the ontogeny of the pediatric GI tract need to be filled (Table 2)¹⁵ and improved in vitro dissolution tests for children developed and validated.¹⁴ However, pediatric PBBM should play a key role in advancing the prediction of pediatric doses by integrating growing knowledge on physiology development with biorelevant in vitro and clinical data.

The author declares no potential conflicts of interest with respect to the research, authorship, and publication of this article.

The author received no financial support for the research, authorship, and publication of this article.