In vitro-in vivo correlation (IVIVC) population modeling for the in silico bioequivalence of a long-acting release formulation of Progesterone.
E.M. Tosca (1), G. De Nicolao (1), M. Rocchetti (2), P.Magni (1), E. Pérez (4), C. Nieto (5), P. Bettica (3), J. Moscoso del Prado (4)
(1) Department of Electrical, Computer and Biomedical Engineering, University of Pavia, via Ferrata 5, Pavia, I-27100, Italy, (2) Consultant, Milano, Italy, (3) Italfarmaco S.p.A., Milan, Italy, (4) ITF Research Pharma SLU, San Rafael 3, 28108 Alcobendas, Spain, (5) Italfarmaco SA, San Rafael 3, 28108 Alcobendas, Spain
Objectives: Health authorities carefully evaluate any change in the batch manufacturing process of a drug before and after regulatory approval. In absence of an adequate in vitro-in vivo correlation (level A IVIVC) (1) an in vivo bioequivalence study is frequently required (2), increasing drug development costs and time to market. This work proposes a population modeling approach to establish a level A IVIVC between the in vitro release of two batches of Progesterone vaginal rings (PVRs), a dosage form designed for the continuous delivery in vivo, and the corresponding serum profiles observed during clinical studies. Estimates of the expected in vivo relative bioavailability of two tested batches can also be obtained from the model here proposed.
Methods: Experimental methods: In vitro data included time courses (24-408h) of the amount of released Progesterone for 2 batches of rings (reference batch A and test batch B) manufactured by Italfarmaco S.p.A.. Data at 4 different dose levels (125, 375, 750, 1500 mg) were available for batch A, while only 375 mg data were available for batch B. In vivo Progesterone serum level profiles were collected in clinical studies performed on batch A (54 subjects); for each subject, the total amount of Progesterone (P) released within the experimental period was also measured. Model-based approach: Development of: (i) a model describing the in vitro release profiles of P at each dose level (Pvitro Model); (ii) an in vivo release model accounting for the limited solubility of P in the finite volume of vaginal fluid (Pvivo Model); (iii) a global population IVIVC Progesterone ring (IVIVC P-ring) model including the Pvivo Model, able to predict, at each dose level, the in vivo serum P concentration profiles.
Results: For both batches and for all the doses, time profiles of the accumulated Pvitro release appeared well described by the sum of a biexponential model and an immediate release in the medium (a common behaviour for this type of dosage form). For each dose level, different rings (6-12) from the same batch were tested; however, a unique curve was estimated using a pooled data approach because the release profiles were so close to consider negligible their inter-ring variability. The Pvivo release Model follows the Pvitro one apart from the addition of a dose dependent inhibition function acting on the biexponential release. Then, the Pvivo release enters a two-compartment PK model with first-order absorption rates that yields the serum P concentration (IVIVC P-ring model). Fixing the Pvitro parameters to the values obtained from the in vitro data, in vivo model parameters were estimated in NONMEM performing a simultaneous fitting on all the dose levels of batch A. As recommended by the regulatory guideline (1), the predictive performance was evaluated internally assessing the absolute percent prediction errors (%PE) for AUC(0-t) of P serum concentrations. For AUC(0-408h), %PE was <2% for each dose, remaining always <7% at all sampling times. Model external predictability was assessed identifying the IVIVC P-ring model leaving out the 375 mg data that were considered as external dataset. Visual predictive check plots (500 simulated individuals) were used to assess the model predictive performance when compared with the external 375 mg dataset. For the average AUC(0-t), %PE was <10% for sampling times after 264h and <3% at the end of the experiment. In addition to the standard "level A IVIVC" procedure, the population approach here developed allows to performe simulation studies providing estimates of the relative bioavailability in vivo (F=AUCtest/AUCref) of any new batch tested in vitro in comparison to a reference one. For example, considering here the 375 mg dose, 500 serum P profiles were generated with the IVIVC P-ring model previously identified on batch A; then, other 500 serum P profiles were obtained for batch B using the same model parameters apart the in vitro parameters that were re-estimated from its in vitro release dataset. From the AUC values of the two populations, the in vivo relative bioavailability was directly obtained, F=0.913 with 90%CI = [0.878, 0.948].
Conclusions: In this work, a level A IVIVC model for PVRs was developed and its internal and external predictability was evaluated. In addition, using a population approach, the IVIVC P-ring model resulted a useful tool for the assessment of the in vivo bioequivalence from in vitro studies performed with a new batch.
 Food and Drug Administration. Guidance for industry. Extended release oral dosage forms: Development, evaluation, and application of in vitro/in vivo correlations. Rockville MD: US Department of Health and Human Services, Food and Drug Administration, Center for Drug Evaluation and Research; 1997
 Food and Drug Administration. Guidance for industry. SUPAC-MR: Modified Release Solid: Oral Dosage Forms: Scale-Up and Postapproval Changes: Chemistry, Manufacturing, and Controls; In Vitro Dissolution Testing and In Vivo Bioequivalence Documentation. Rockville MD: US Department of Health and Human Services, Food and Drug Administration, Center for Drug Evaluation and Research; 1997