Exploring Experimental and Statistical Approaches to Control Oversensitivity of In Vitro Permeability to Excipient Effects

<b>Background/Objectives:</b> The static in vitro permeability assay based on cell monolayers has been widely used in the pharmaceutical industry and recognized by regulatory agencies as a surrogate method for BCS classification. However, the application of such an experiment to study the effects of formulations is limited by the oversensitivity to the excipient effect on drug permeability. In this article, we studied the effects of common excipients on the permeability of moderately and poorly absorbed model compounds across cell monolayers, using two approaches to control said oversensitivity. <b>Methods:</b> Drug permeability across MDCK-wt was assessed in the absence (control) or presence (treatment) of excipients, using minoxidil as a high-permeability marker. The effects of excipients were parameterized as a permeability ratio (PR = treatment/control) without or with normalization (nPR) by minoxidil permeability. Metrics were compared by either ANOVA (<i>p</i> < 0.01) or confidence intervals (CI90, as per bioequivalence metrics) to identify excipient effects. <b>Results:</b> Acyclovir and hydrochlorothiazide showed the highest and lowest number of interactions, respectively. The most impactful excipients were sodium lauryl sulfate, microcrystalline cellulose, and sodium starch glycolate. Unexpectedly, nPR increased the number of excipient effects across model drugs (19 vs. 21). Alternatively, the CI90 approach was more sensitive than ANOVA in identifying excipient effects (41 vs. 32). <b>Conclusions:</b> Minoxidil was not able to control the anticipated oversensitivity of cell-based permeability experiments. Meanwhile, ANOVA was overall able to reduce oversensitivity to excipient effects on drug permeability compared to CI90. Nonetheless, there might be a niche for CI90 analysis when comparing the performance of two formulations on the permeability of moderately and poorly absorbed drugs.