Optimization of the Transwell® System for Assessing the Dissolution Behavior of Orally Inhaled Drug Products through In Vitro and In Silico Approaches

Abstract

The aim of this study was to further evaluate and optimize the Transwell® system for assessing the dissolution behavior of orally inhaled drug products (OIDPs), using fluticasone propionate as a model drug. Sample preparation involved the collection of a relevant inhalable dose fraction through an anatomical mouth/throat model, resulting in a more uniform presentation of drug particles during the subsequent dissolution test. The method differed from previously published procedures by (1) using a 0.4 µm polycarbonate (PC) membrane, (2) stirring the receptor compartment, and (3) placing the drug-containing side of the filter paper face downwards, towards the PC membrane. A model developed in silico, paired with the results of in vitro studies, suggested that a dissolution medium providing a solubility of about 5 µg/mL would be a good starting point for the method’s development, resulting in mean transfer times that were about 10 times longer than those of a solution. Furthermore, the model suggested that larger donor/receptor and sampling volumes (3, 3.3 and 2 mL, respectively) will significantly reduce the so-called “mass effect”. The outcomes of this study shed further light on the impact of experimental conditions on the complex interplay of dissolution and diffusion within a volume-limited system, under non-sink conditions.

Characterization of Membrane-Type Dissolution Profiles of Clinically Available Orally Inhaled Products Using a Weibull Fit and a Mechanistic Model

“…It could therefore be argued that the relevant spatial length scale exceeds the particle size and that a constant stagnant-layer thickness therefore would be appropriate. 29 Consistent with this, Amini et al 17 note that the stagnant layer, as a result of the mentioned factors, will be different from the one used by May et al 16 Third, we have considered the characteristic dissolution time to be a free parameter that absorbs effects resulting, for example, from the presence of micelles (see Section 3.7 below), incomplete wetting, hydrodynamics, and so forth. In the work by Amini et al, 17 a correction factor F was used to the same effect.…”

“… 29 Consistent with this, Amini et al 17 note that the stagnant layer, as a result of the mentioned factors, will be different from the one used by May et al 16 Third, we have considered the characteristic dissolution time to be a free parameter that absorbs effects resulting, for example, from the presence of micelles (see Section 3.7 below), incomplete wetting, hydrodynamics, and so forth. In the work by Amini et al, 17 a correction factor F was used to the same effect. The predictions by the two modeling approaches will nevertheless be similar (at least as long as sink conditions prevail in the acceptor compartment) because they are based on the same assumptions.…”

“… 28 As a result, our account of polydispersity, presented in Section 2.10.2 , and the transition from sink to nonsink conditions, elaborated upon in Section 2.10.3 , are both novel. Second, we have assumed a time-independent stagnant-layer thickness, contrary to Amini et al 17 (as well as May et al 16 and Hintz and Johnson 18 ) who considered a thickness that decreased with the particle size. The main motivation for using constant thickness is that the particles are lying on, and likely partly embedded in, the filter onto which they were deposited.…”

“…In a recently published study by Amini et al, such a mechanistic model was indeed formulated in terms of binned particle-size data, where each bin corresponds to an impactor stage. 17 This model utilizes and extends a methodology developed by Hintz and Johnson, 18 who represented the underlying continuous particle-size distribution by a discrete distribution with 16 particles sizes and also considered permeation across a membrane, in a similar manner as in a Transwell dissolution setup. The Hintz and Johnson model is commonly used to describe dissolution in physiological based pharmacokinetic modeling and simulation.…”

“…However, as observed by May et al, polydispersity can have a significant effect on the overall dissolution (and permeation) profiles, implying that this feature should be included in mechanistic modeling. In a recently published study by Amini et al, such a mechanistic model was indeed formulated in terms of binned particle-size data, where each bin corresponds to an impactor stage . This model utilizes and extends a methodology developed by Hintz and Johnson, who represented the underlying continuous particle-size distribution by a discrete distribution with 16 particles sizes and also considered permeation across a membrane, in a similar manner as in a Transwell dissolution setup.…”