RESUMEN
We performed an ab initio next-generation risk assessment (NGRA) for a fragrance ingredient, benzyl salicylate (BSal), to demonstrate how cosmetic ingredients can be evaluated for systemic toxicity endpoints based on non-animal approaches. New approach methodologies (NAMs) used to predict the internal exposure included skin absorption assays, hepatocyte metabolism, and physiologically based pharmacokinetic (PBPK) modeling, and potential toxicodynamic effects were assessed using pharmacology profiling, ToxProfiler cell stress assay, transcriptomics in HepG2 and MCF-7 cells, ReproTracker developmental and reproductive toxicology (DART) assays, and cytotoxicity assays in human kidney cells. The outcome of the NGRA was compared to that of the traditional risk assessment approach based on animal data. The identification of the toxicologically critical entity was a critical step that directed the workflow and the selection of chemicals for PBPK modeling and testing in bioassays. The traditional risk assessment and NGRA identified salicylic acid (SA) as the "toxdriver." A deterministic PBPK model for a single-day application of 1.54 g face cream containing 0.5% BSal estimated the Cmax for BSal (1 nM) to be much lower than that of its major in vitro metabolite, SA (93.2 nM). Therefore, SA was tested using toxicodynamics bioassays. The lowest points of departure (PoDs) were obtained from the toxicogenomics assays. The interpretation of these results by two companies and methods were similar (SA only results in significant gene deregulation in HepG2 cells), but PoD differed (213 µM and 10.6 µM). A probabilistic PBPK model for repeated applications of the face cream estimated the highest Cmax of SA to be 630 nM. The resulting margins of internal exposure (MoIE) using the PoDs were 338 and 16, which were more conservative than those derived from external exposure and in vivo PoDs (margin of safety values were 9,705). In conclusion, both traditional and ab initio NGRA approaches concluded that the daily application of BSal in a cosmetic leave-on face cream at 0.5% is safe for humans. The processing and interpretation of toxicogenomics data can lead to different PoDs, which can subsequently affect the calculation of the MoIE. This case study supports the use of NAMs in a tiered NGRA ab initio approach.
RESUMEN
This study aimed to develop valid in vitro models for preclinical evaluation of proliferative and anti-inflammatory effects of human allogeneic serum eye drops for dry eye disease (DED) treatment. A DED wound healing model was developed by analyzing the influence of coating and serum concentrations on human corneal epithelial (HCE-T) wound closure. Further, intralaboratory variance, freeze-thaw cycle effects, donor variability and stability assays were conducted. Interleukin-1ß (IL-1ß) and tumor necrosis factor α (TNFα) were used to induce the gene expression of matrix metalloproteinase 9 (MMP9), cyclooxygenase 2 (COX2), transforming growth factor-ß (TGFß) and IL-1ß. MMP9 induction was optimized using a design-of-experiments (DoE) approach and applied to examine serum under static and dynamic conditions. MMP9 protein expression was analyzed by ELISA. The DED wound healing model detected proliferative effects of serum down to 1% with a small intralaboratory variance. Serum stability was shown over six months, donor variance could be detected, and freeze-thaw cycle effects did not affect wound closure. Serum decreased MMP9 expression on the gene and protein levels. The induction method was successfully optimized using DoE modeling and transferred to a dynamic setting mimicking tear film fluidics. The DED wound healing and inflammatory DED model present useful in vitro models for the preclinical evaluation of allogeneic serum eye drops without the use of animal experiments.