Utilizing MechDermA model for early feasibility analysis in topical drug discovery

Objectives: One key challenge in preclinical topical drug selection is predicting effect of drug and formulation properties on skin permeation. Simple models relating molecular weight (MW), logP etc. to skin permeation (eg. flux) have been characterized using small datasets and simple formulations [ 1, 2, 3, 4]. However, the dermal drug development field is moving into novel formulation regimes, and these models are limited when extrapolating to these spaces. One example is the uncharacterized non-linear interaction between formulation and physchem properties on skin flux. Therefore, there is an unmet need to (a) apply mechanistic models that capture influence of complex formulations, skin physiology and compound physchem properties on skin flux, and (b) validate these workflows for practical lead identification in the preclinical phase. Here, we answer these questions using a mechanistic skin PBPK framework. Methods: The existing physiology within the SimCYP MPML MechDermA model, version 2024 was used for modeling [ 5]. For database selection, we utilized the ex vivo human skin permeation (IVPT) which is commonly generated in the preclinical phase. The first database utilized standardized protocol which made it amenable for initial model calibration and contained finite dose IVPT capturing realistic clinical trial situations [ 6]. The second database was a more expanded database for physchem properties [ 7]. After this comprehensive calibration of the model, it was used to predict the IVPT skin permeation of Pfizer’s early phase screening compounds with different types of formulations like emulgel, ethanol:propylene glycol and oil blends. Results: Parametric sensitivity analysis (local and global) indicated that diffusivity and partitioning into the stratum corneum controlled skin permeation post compound entry into the skin. Within the formulation, the most sensitive parameters were the thermodynamic solubility and particle distribution for complex multiphasic formulations. Interestingly, the viscosity was 10-fold less sensitive than the other parameters. The early evaporation rate controlled the Cmax and tmax (including lag phase) of skin PK, with later phases being less sensitive to the end point of the evaporation profile. For high logP (>4) compounds, additional parameter of importance was binding within skin compartment (depot effect). In addition, it was necessary to include time dependent changes in SC partitioning coefficients in the presence of excipients. In the validation phase, the model was able to accurately predict the flux of Pfizer discovery compounds and identify the best physchem properties for each formulation type. Interestingly, the optimal properties were different depending on the formulation type, which provides a better way to tailor dermal drugs for different applications. Conclusions: The new version of the model can be used to identify better compounds for each type of formulation especially in cases where only certain types of formulations are commercially acceptable for certain indications eg. acne. This provides a rational way to do early feasibility analysis of a chemical series and formulation and make cost saving go/ no-go decisions.