John DiBella, Vice President, Marketing & Sales, Simulations Plus, Viera Lukacova, Team Leader – Simulation Technologies and Dr Neelam Sayed, Application Scientist, Electrolab, give an insight about one of the first biowaivers granted by the US FDA for Class IV drugs through the use of mechanistic modelling
The use of in silico modelling to support drug discovery research has been around for over 25 years. Only recently, however, has mechanistic absorption modelling (MAM), coupled with physiologically-based pharmacokinetic (PBPK) modelling, gained widespread attention in the preclinical and clinical development stages. In 2016, the European Medicines Agency (EMA) and US Food and Drug Administration (US FDA) issued the first draft guidance documents focussed on the use of mechanistic modelling methods, where clear directives were given to sponsors planning to include these approaches in submissions. The use of this technology has been encouraged by regulatory agencies by waiving certain clinical studies (e.g., drug-drug interactions) based on modelling results.
Studies have shown how GastroPlus, the top-ranked MAM/PBPK modelling platform, has been used to justify biowaivers for BCS Class I, II, and III compounds.1,2,3,4
Simulations Plus, a leading provider of modelling and simulation (M&S) solutions for the pharmaceutical, biotech, and chemicals industries, worked closely with the sponsor to build models using GastroPlus software and prepared biowaiver reports that were ultimately approved.
Changes in a compound’s manufacturing process can affect the particle size distribution of the active pharmaceutical ingredient (API). In these cases, the US FDA requires proof that the API’s oral bioavailability (F per cent) and bioequivalence are not adversely impacted by these changes. In the case presented here, the granulation process was changed from high shear to fluid bed granulation, and an in-line milling step was added to the crystallisation process as part of the overall process improvement. This was introduced to reduce the fines in the crystallisation step and narrow the particle size distribution (particle engineered, PE). The API from the original crystallisation process (non-particle engineered, NPE) was used to manufacture several Phase 1 and 2 clinical supply lots. Examination of these different API lots showed that two early batches of the NPE API had a broader particle size distribution than those of the PE API lot. Evidence needed to be supplied to show that this change did not adversely impact either its F per cent or bio-equivalence.
Initially, the US FDA requested the sponsor to design an expensive relative bioavailability study for this BCS Class IV drug, delaying the approval of the product by many months. However, the sponsor was able to present a M &S plan for a biowaiver extension, utilising GastroPlus modelling with in vitro physicochemical API characterisation studies, that the US FDA reviewed and approved.
The modelling objectives
- Build the baseline mechanistic model for the API using clinical data from prior studies.
- Assess sensitivity of API particle size distribution on drug exposure.
- Compare predicted bio-equivalence of the tablets from the NPE and PE API lots.
All simulations were carried out with GastroPlus v7. The physicochemical and pharmacokinetic parameters values used as inputs into the model are defined in Table 1. The drug’s sponsor provided complete and/ or cumulative particle size distribution (PSD) data for all lots, along with pharmacokinetic data across three dose levels (50, 100, and 300 mg) in a patient population with four different NPE API lots.
The first step was to build the mechanistic model in GastroPlus and validate it against available clinical data. Figure 1 illustrates the validation of this model for 50, 100 and 300 mg doses of a particular NPE API lot. The graphs demonstrate that the same model does a reasonable job of predicting the observed plasma concentration-time data across three different doses of the NPE API lots.
Assessment of particle size changes
The PSA mode was used to establish particle size specifications, which is the allowed variability in mean particle size and standard deviation before any significant changes in predicted outcomes are observed. Since the drug is a Class IV compound with known dissolution limitations, simulations were run changing both the particle size and dose together to determine if an increase in dose would result in a higher or lower sensitivity to the change in particle size. Figure 2 illustrates the results of these simulations on absorption, which indicated that there would be very small changes, if any, in the fraction absorbed until the largest mean particle sizes of the NPE API lots (> 30 – 40 µm) were reached and the dose exceeded 100 mg.
Virtual bioequivalence trials
The Population Simulator in GastroPlus predicts likely population distributions of results across subjects. This feature is commonly used to simulate crossover studies and evaluate whether two formulations are likely to be bioequivalent. Here, the Population Simulator was used to carry out simulations for 10 different populations, each with 25 virtual subjects. A combination of default and observed variability from the analysis of individual subject data from the clinical studies was used to define inputs for the population runs. To account for intra-subject and inter-occasion variability, a random multiplicative error term with CV=15 per cent was added to each of the simulated profiles.
Simulated AUC and Cmax values for the tablets manufactured from the NPE and PE API lots were compared, with sample results seen in Table 2. These bioequivalence calculations, as defined by the FDA’s guidance on Bioequivalence Studies, were implemented to show the similarities between AUC and Cmax for the NPE API lots (up to 30 – 40 µm) versus the PE API lot.
The BCS system currently allows for biowaivers for rapidly dissolving immediate-release (IR) products of Class I drugs. Here, we show one of the first cases where mechanistic modelling with GastroPlus was used to extend biowaivers to IR products of Class IV compounds, which is a breakthrough in formulation development. By eliminating bioequivalence studies, where appropriate, significant reductions in both costs and time to market reformulated drug products can be realised while ensuring good drug product performance. Moreover, while the case study described here focuses on post-approval changes, the principle can be applied to generics when filing for ANDAs.
1. Application of gastrointestinal simulation for extensions for biowaivers of highly permeable compounds.
Tubic-Grozdanis M, Bolger MB, and Langguth P. (2008) AAPS J. 10(1): 213-26.
2. An investigation into the importance of “very rapid dissolution” criteria for drug bioequivalence demonstration using gastrointestinal simulation technology. Kovacevi I, Parojci J, Tubi-Grozdanis M, Langguth P. (2009) AAPS J. 11(2):381-4.
3. The biowaivers extension for BCS Class III drugs: the effect of dissolution rate on the bioequivalence of BCS Class III IR drugs predicted by computer simulation. Tsume Y, Amidon GL. (2010) Mol Pharm. 7(4):1235-43.
4. The role of predictive biopharmaceutical modeling and simulation in drug development and regulatory evaluation. Jiang W, Kim S, Zhang X, Lionberger RA, Davit BM, Conner DP, Yu LX. (2011) Int J Pharm. 418(2):151-60.