Defining level A IVIVC dissolution specifications based on individual in vitro dissolution profiles
I González-García1, A García-Arieta2,a, M Merino-Sanjuan1,3, V Mangas-Sanjuan1,3 *, M Bermejo4
1 Pharmacy and Pharmaceutical Technology Area. University of Valencia, Spain; 2División de Farmacología y Evaluación Clínica, Departamento de Medicamentos de Uso Humano, Agencia Española de Medicamentos y Productos Sanitarios. aThis manuscript represents the personal opinion of the authors and does not necessarily represent the views or policy of the Spanish Agency for Medicines and Health Care Products; 3Institute of Molecular Recognition and Technological Development (IDM), Polytechnic University of Valencia and University of Valencia, Spain; 4 Department of Engineering, Pharmacy and Pharmaceutical Technology Area. Miguel Hernandez University, Spain.
Objectives: The purpose of this work is to compare the classical approach (the use of mean data) with a new methodology in which we have used individual data in order to assess the probability of declaring bioequivalence for a new batch based on an IVIVC. Furthermore, we have evaluated the impact of these two different methodologies on the establishment of dissolution specifications.
Methods: A slow, medium and fast dissolving drug formulations were used to develop the IVIVC. Dissolution data sets were generated for 12 units (e.g. tablets) based on a first-order dissolution model and forced to show a similarity factor (f2) below 50 between the medium and fast/slow formulation. A level A IVIVC using differential equations was established using these three drug formulations, where the link between in vitro and in vivo performance of the drug products was related between in vitro and in vivo dissolution rate coefficients (kd). Plasma profiles were generated using a one compartment model with first order dissolution, absorption, and elimination kinetics. Twelve individual units were considered for each formulation or batch. Batch suitability was assessed using six additional batches (12 units each). For each batch, simulation (n=1,000) of a dissolution assay with 12 units was generated through Monte Carlo simulation approach. The percentage of BE batches was computed for each approach. BE of a new batch was concluded when the Cmax ratio between reference and new batch formulations was within ±20%. Dissolution specifications were established as follows: (i) classical approach, in vitro dissolution limits of each formulation were computed using the batch whose ratio was the closest to ±20%; (ii) individual approach, the STSF and FTFF whose ratio was exactly ±20%. The simulations were performed in NONMEM 7.3. Graphical and statistical analysis were performed using R software and RStudio®.
Results: According to the results from the classical approach, the Cmax ratio from the six batches fulfill the ±20% range under linear level A IVIVC. Similar results were observed for Batches 2-6 when non-linear level A IVIVC was developed, but only 78.6% of the simulations with Batch 1 achieved a Cmax ratio within the ±20% difference. However, when the individual approach was applied under linear level A IVIVC, a significant amount of simulations with Batches 1 and 2 were out of ±20% limits: 53.3 and 58.1%, respectively. Greater differences between classical and individual approaches were observed for the non-linear relationship (scenarios 4-6), where the suitable number of batches of Batch 1 and 2 diminished to 0.3 and 15.5%, respectively. Additionally, 23.1% of the simulations with Batch 3 resulted in a Cmax ratio greater than ±20% compared to the reference formulation. The dissolution performance of Batch 3 was more similar to the reference formulation than Batches 1 and 2, but differences were not detected when the classical approach was applied. The batches that were closest to ±20% difference on Cmax (Batches 1 and 2) were used to establish the dissolution limit specifications (Table 4). The classical approach provides narrower specification limits because it is established based on the mean in vitro dissolution profile that is closest to ±20%, whereas the individual approach provides the dissolution specification limits that exactly achieved ±20% difference on Cmax between reference and new batch.
Conclusions: An individual approach has been proposed to establish the dissolution specifications using a level A IVIVC, ensuring BE of all units within the new batch developed. This methodology takes into consideration the in vitro and in vivo variability observed, providing the dissolution specification limits that ensure in vivo ratios exactly to 80-125. Thus, the widening of dissolution specification is a consequence of using individual data, but ensures the BE of all tablets, which is not always achieved using the classical approach.
 Rossenu, S., Gaynor, C., Vermeulen, A., Cleton, A., Dunne, A., 2008. A nonlinear mixed effects IVIVC model for multi-release drug delivery systems. Journal of pharmacokinetics and pharmacodynamics 35, 423-441.
 Bauer, R., 2011. NONMEM users guide: introduction to NONMEM 7.2.0, in: Solutions, I.D. (Ed.), Elicott City.