Purpose: In vitro dissolution tests are pharmacopoeially used as a quality control tools to justify batch-to-batch reproducibility of pharmaceutical solid dosage forms and to assist with determination of bioequivalence (BE). Several apparatuses are used for evaluating the in vivo dissolution behavior of drug products in the gastrointestinal tract.
In pharmaceutical development of Drug Product X, Dissolution Method A was successfully developed in order to detect changes of critical material attributes: drug substance particle size and uncoated tablet density. During the pharmaceutical development, since the formulation composition of Drug Product X was changed for commercial purpose after pivotal clinical study was launched, a BE study was conducted using different formulations: Formulation A (original formulation) and Formulation B (changed formulation). The in vitro dissolution profiles of two modelled formulations (Formulation A vs. Formulation B) were similar. Unfortunately, the two modelled formulations failed to show bioequivalence, leading to a non-bioequivalence (non-BE) outcome. Therefore, Formulation C (revised formulation) was further developed. In this regard, bioequivalence was successfully shown for Formulation A and Formation C. Therefore, it was important to develop an in vitro dissolution method that accurately predicts the in vivo performance of drug products.
In the past, several studies suggested that physiologically based pharmacokinetic (PBPK) modeling can be applied to predict the PK behavior in humans using clinical data. An in vitro-in silico-in vivo approach was proposed as a mean to screen product performance and target specific formulations for BE/non-BE assessment.
The objective of the present study was to demonstrate that in silico PBPK model developed using actual pharmaceutical development data was useful to develop an in vitro dissolution method showing clinically relevant behavior.
Dissolution studies: In vitro drug release was determined using the USP XXVIIII rotating paddle method (Type NTR-6100A, Toyama Sangyo Co., Ltd., Osaka, Japan): Dissolution Method A (900 ml, buffer A, 50 rpm, 37°C, n = 12) and Dissolution Method B (900 ml, buffer B, 25 rpm, 37°C, n = 12).
Pharmacokinetic studies: Two clinical studies (healthy volunteers) were conducted to verify BE of Drug Product X: Clinical Study I (Formulation A vs. Formulation C) and Clinical Study II (Formulation B vs. Formulation C). Human plasma concentration-time profiles after orally administered with Drug Product X from the two clinical studies were used to develop and verify in silico PBPK model.
In silico PBPK modeling: Commercially available software, GastroPlusTM, version 8.5 (Simulations Plus Inc., Lancaster, CA, USA) was used for the modeling.
Evaluation of in silico model predictability: Absolute percent prediction error (%PE) was evaluated according to the reference of “Guidance for Industry, Extended Release Oral Dosage Forms: Development, Evaluation, and Application of In Vitro/In Vivo Correlations”. In silico PBPK model validation criteria were predefined as not more than 10% of %PE.
In general, the in silico PBPK model was constructed based on the physicochemical and physiological properties of drug substance contained in Drug Product X. Figure 1 shows the observed and simulated plasma concentration-time profiles in clinical study I. The results indicated that the PBPK model successfully demonstrated the plasma concentration-time profile, since the %PE of Cmax and AUC (9% and 1%) met the predefined validation criteria. In order to develop a clinically relevant dissolution method, various dissolution media and paddle rotational speeds were examined using Formulations B and Formulation C. As a result, Dissolution Method B (f2 value of 38, Figure 2b) showed stronger discriminatory power compared to Dissolution Method A (f2 value of 55, Figure 2a). Based on these results, Dissolution Method B was verified to be clinically relevant by evaluating the in silico PBPK model predictability incorporated dissolution data for PK behaviors of Formulations B and Formulation C. Figure 3-a) and Figure 3-b) show the simulated and observed plasma concentration-time profiles of formulations B and C in clinical study II. The in silico PBPK model incorporated the dissolution profiles of the two modelled formulations was successfully verified because all %PEs of Cmax and AUC met the criteria (1% and 10% for BE batch, 0% and 7% for non-BE batch). By using the PBPK model incorporated dissolution data, Dissolution Method B was demonstrated to have clinically relevance. These results demonstrated that in silico PBPK modeling was able to predict a developed dissolution method showing clinically relevant behavior.
The usefulness of in silico PBPK model for developing a clinically relevant dissolution method was evaluated using actual pharmaceutical development data.
Hiroshi Nakagawa– Daiichi Sankyo Co., Ltd.
Tsuyoshi Mikkaichi– Daiichi Sankyo Co., Ltd.
Takuya Miyano– Daiichi Sankyo Co., Ltd.
Kazuko Maeda– Daiichi Sankyo Co., Ltd.
Tomoyuki Watanabe– Daiichi Sankyo Co., Ltd.
Shuichi Ando– Daiichi Sankyo Co., Ltd.