IV-05 Robert Leary Automatic framework for bioequivalence studies from In Vitro test to In Vivo study design
Kairui Feng(1), Robert H. Leary(1), Michael Dunlavey(1), Amin Rostami-Hodjegan(1,2)
1. Certara, USA, 2. University of Manchester, UK
Objectives: The traditional deconvolution-based method of in vitro-in vivo correlation (IVIVC) for bioequivalence (BE) studies pre-dates modern population approaches to PK modeling. For example, it does not incorporate inter-individual (IIV) and inter-occasion variability (IOV), is limited to linear one and two compartment models, and does not accommodate study protocols based on clinical trial simulation such as parallel, crossover and replicate designs. It is not flexible enough and sufficiently free from bias to meet all the requirements in BE studies such as single dose, multiple doses (steady-state) and food effects. Hence the IVIVC method has not been widely used for human BE studies and Abbreviated New Drug Applications (ANDAs) based on IVIVC are rarely successful . This project is intended to create an automatic framework that addresses the deconvolution-based limitations and provide a user-friendly automatic methodology to help BE studies from in vitro test identification to in vivo study design and thus facilitate improved bioequivalence strategies for more timely approval of ADNAs.
Methods: The automatic framework can be described in different stages including before BE studies, after BE studies, and after pilot studies. The relationship between these stages includes the following.
A direct IVIVC model incorporating a pharmacokinetic PK model (D-IVIVC-PK) is created , including steps:
Selection of a dissolution function from a candidate set such as Hill, cumulative Weibull, cumulative double Weibull, Hiquchi, Makoid-Banakar, Kopcha, or user-provided custom function, to best fit (e.g, according to AIC) the in vitro dissolution data.
Differentiation of the in vitro function for use as a forcing function in a differential equation-based IVIVC and PK model.
Collection of all relevant in vivo data such as single dose (IV & oral), multiple dose, food effects, including covariates and formulations (categorical covariates as well as IOV) for use in an NLME model, with any possible variabilities such as IIV, IOV, or sequence effect, period effect, formulation effect or carryover effect.
Nonlinear mixed effect modeling (NLME) for model estimation to fit the in vivo plasma/blood data.
Model diagnostics and model evaluation reporting, such as VPC, OFV etc.
This model is used to support clinical trial simulation (CTS) and the required bioequivalence ANOVA test, including:
Use of the fitted D-IVIVC-PK model for simulation and study design including parallel, crossover and replicates design
Computation of PK parameters such as Cmax, Tmax, AUC as well as other relevant secondary parameters and information such as sequence, subject, period, formation
A variety of possible bioequivalence studies are created accommodating features such as single dose, multiple dose and food effects.
AVOVA test is provided for average, population and individual bioequivalence test
In vitro test experiment support includes:
Automatic comparison from a list of in vitro dissolution tests on the reference formulation (e.g. different pH, volumes, media type, rotation speed, USP etc.) from the above D-IVIVC-PK models
Ranking the list of in vitro dissolution methods and using the best in vitro dissolution test for the clinical trial simulation and design
Both reference and test formulation are used for the above CTS and BE test to determine the likely success of test formulation
Ideally, IVIVC can serve as a surrogate for human BE studies includes:
Collection of all the test formulations (such as slow, median/reference and fast release) in vitro dissolution data and in vivo human data together to build D-IVIVC-PK model as above
Validation of data for each formulation can be performed via standard model diagnostics
A new test formulation is used for the above CTS and BE ANOVA test by comparing with the reference formulation
Results: By selecting a few options and a few clicks from a user-friendly screen, a user can follow the automatic framework to make the best decision for formulation strategies as well as bioequivalence strategies.
Conclusions: The built automatic framework for BE studies can save resources and get more timely approval from ADNAs. This automatic framework also helps generic pharma companies eliminate common mistakes and provides a step-by-step work-flow easy for regulatory ADNA assessment.
References:  Kaur et al. Application of In Vitro-In Vivo Correlations in Generic Drug Development; AAPS; 2015; Vol. 17 No 4  Buchwald, P. Direct, differential-equation-based in-vitro-in-vivo correlation (IVIVC) method; JPP; 2003; 55:495-504