Background Mechanism-based physiologically based pharmacokinetic (PBPK) models have been proven to be powerful tools in the generic drug development process and bioequivalence assessment.
Major advantages of PBPK modeling include the ability to integrate information from human physiology (system-related
parameters), drug physicochemical properties and interaction with enzymes and other protein structures, and drug product characteristics (e.g., the formulation type) (drug product-related parameters) to generate predictions regarding the absorption, disposition, metabolism and elimination of the drug in populations following administration of drug products via oral and non-oral routes.
In addition to integrating prior information on parameter distribution into these models, experimental data can be used to estimate unknown or low certainty parameters, leading to more reliable predictions based on a well characterized system.
Acknowledging the superiority of these predictive models, optimization routines have been incorporated into a variety of software programs currently used in the development of mechanistic PBPK models.
The optimization routines that currently perform better are the ones that are successful in identifying and characterizing inter- and intra-subject variability in the population.
These types of variability can be attributed to sources that include, but are not limited to, anthropometric characteristics (e.g., age, body weight/mass, and sex), experimental variability observed in in vitro dissolution/drug release studies, study design which accounts for inter-trial and inter-occasion variability and subject-by-formulation interactions.
Models such as the ones described above are valuable in the development of safe, effective and affordable generic drug products.
They allow the simulation of bioequivalence trials that compare the predicted in vivo performance of the brand name with the generic drug product one.
These trials would not be feasible for a variety of reasons; conducting bioequivalence studies in special populations such as pediatrics is challenging because of ethical concerns and extrapolating model outcomes from healthy volunteers to patients, the target population, is an attractive approach.
Additionally, these models allow the prediction of in vivo performance for formulations of interest based on the establishment of in vitro-in vivo correlations for drug products with similar drug delivery technologies.
However, a platform that supports and facilitates the development of the models described above and the design and simulation of virtual bioequivalence trials to inform decisions in the generic drug product development process is not available today.
Objective:
The purpose of this project is to develop and implement a virtual bioequivalence trial simulation platform that can be used to perform population-based statistical analysis in complex and computationally intensive physiologically based pharmacokinetic (PBPK) models developed to describe the absorption, distribution and elimination of active pharmaceutical ingredients formulated in complex and non-complex dosage forms administered via oral or non-oral routes.
The developed virtual bioequivalence trial simulation platform will be used by all stakeholders engaged in generic drug development including regulatory agencies, pharmaceutical industry and academia to generate predictions on in vivo drug product performance, to perform bioequivalence assessments between brand name and generic drug products and to inform regulatory decisions relating to generic drug development.
Detailed description:
The purpose of the current project is the development of a potentially open-source platform that would provide the capability to conduct virtual bioequivalence trial simulations.
Ideally, this user-friendly platform will provide the user with many user-defined capabilities.
Applicants are encouraged to discuss items that they consider integral parts of this platform in their proposals.
These platform components may include, but not be limited to:
1. Algorithm implementation for population pharmacokinetic analysis:
The current project describes the marriage between PBPK modeling and population-based statistical analysis.
The latter is necessary for parameter estimation when individual-level data is available.
A critical aspect of this coupling is the implementation of an algorithm that would allow the performance of population-based statistical analysis.
Although novelty is encouraged in regards to the algorithm, proposals could entertain approaches such as:
a.
Implementing the non-linear mixed effects theory, b.
Maximum log-likelihood algorithms (stiff, non-stiff Ordinary Differential Equation solving methods with linearization), c.
Exact maximum likelihood with Expectation-Maximization algorithms (no linearization), d.
Introducing a Bayesian population PBPK approach (Markov Chain Monte Carlo), e.
Nonparametric methods.
Proposals could include, but not be limited to, methods to evaluate the robustness of the proposed algorithm in terms of bias to initial estimates, precision, accuracy, convergence rate, and computation time, in full-body PBPK models of low, intermediate, and high complexity using simulated and real (rich and sparse) datasets.
2. PBPK model structure to meet the needs and requirements of developing a generic drug product that can be administered via an oral or non-oral route to healthy volunteers or special populations.
Features of the developed models may include:
a.
capability to administer a dosage form in the gut or through any other tissue/organ (ie.
skin, lung etc.) captured in the model.
b.
capability to mechanistically describe the administration of different dosage forms that could include, but not be limited to:
immediate release or extended release tablets or capsules, solutions or suspensions, aerosols, creams/ointments/emulsions/transdermal delivery systems.
c.
Capability to account for and mechanistically describe the potential impact of formulation critical quality attributes on in vivo drug product performance by incorporating them into the model.
These formulation characteristics could include, but not be limited to:
content uniformity, particle size and particle size distribution, viscosity and rheology characterization parameters and in vitro dissolution/drug release characterization, aerodynamic and adhesion properties.
d.
Capability to extrapolate model outcomes from one population to another by modifying physiology (system-related) parameters in the PBPK model.
Most of bioequivalence studies are conducted in healthy volunteers who are not the target population.
Therefore, the capability of simulating a virtual bioequivalence trial is advantageous in that it can provide an insight on the in vivo drug performance in patients to whom the drug product is intended to be administered and in special populations such as the elderly or pediatrics for whom bioequivalence studies are not always feasible to conduct.
e.
Capability to simulate different study designs that include, but are not limited to:
crossover, parallel, (fully) replicated study design, single and multiple (steady state) dose studies.
3. It desirable to be able, within the platform, to validate/qualify the previously developed PBPK models by utilizing appropriate datasets retrieved from independent and accredited literature sources, by utilizing in house data if available or by designing and conducting studies that would allow the generation of the necessary experimental datasets.
Appropriate datasets that capture the subpopulation, study design and dosage form characteristics that were incorporated in the models are expected to be utilized for model validation/qualification.
It is desirable that the utility and performance of the platform will be demonstrated following a well-defined research plan that will be outlined in the submitted research proposal.
Virtual bioequivalence studies can be simulated on the developed platform to determine whether generic drug products that have been shown to not be bioequivalent to their innovators (positive control) or whether generic drug products that have been shown not to be bioequivalent to their innovators (negative control) are deemed not bioequivalent or bioequivalent, respectively, based on model output.
Multiple scenarios should be simulated to demonstrate model flexibility, sensitivity and overall predictability within the developed.
Demonstrating platform performance and utility with the appropriated examples is desired.