The GARDskin Dose-Response assay for determination of a point-of-departure (PoD) for Next Generation Risk Assessment (NGRA) of skin sensitizers: A case study using isocyclocitral

Presented at Eurotox 2023

The GARDskin Dose-Response assay for determination of a point-of-departure (PoD) for Next Generation Risk Assessment (NGRA) of skin sensitizers: A case study using isocyclocitral

Peter Nählstedt2, Shashi Donthamsetty1 , Andy Forreryd2, Paul Sterchele1, Xiao Huang1, Robin Gradin2, Henrik Johansson2, Ulrika Mattsson2, Isabelle Lee3, Anne Marie Api3, Gregory Ladics1

1 International Flavors & Fragrances USA ,2 SenzaGen AB Sweden, 3Research Institute for Fragrance Materials USA

Download a copy



  • The continous readout from the assay is reproducible and the assay predicts LLNA EC3 and human NESIL values with high correlation to reference benchmark data (geometric mean fold-misprediction factors of 3.8 and 2.5 respectively)
  • The assay provides a nice tool for the fragrance industry to predict the NESIL value which can be used for conducting the quantitative risk assessment for generating the IFRA standard.


New Approach Methods (NAMs) for the assessment of skin sensitizers have been adopted as OECD Test Guidelines (TGs), supporting hazard- and GHS potency classifications. However, more granular potency information, preferably on a continuous scale, is needed to derive a point-of-departure (PoD) for Next Generation Risk Assessment (NGRA) of new chemical entities, which still represents a missing element in the application of NAMs for sensitization assessments.

The GARDskin assay (OECD TG 442E) provides a novel and mechanistically different method to monitor the Key Events (KE) in the Adverse Outcome Pathway (AOP) for skin sensitization and is the first harmonized test guideline based on genomics and machine leaning. A modified version of the validated protocol incorporating dose-response measurements has recently been described which uses linear models for the prediction of LLNA EC3/Human No Expected Sensitization Induction Levels (NESIL) values.

The aim of the following study, which represents a cross-sector collaboration involving industry, assay developer, and a non-profit research institute, were to perform a pre-validation exercise to evaluate the precision and reproducibility of the continous potency predictions from the GARDskin dose-response assay, and to demonstrate how the derived continous potency predictions can be implemented into available NGRA-framework to determine safe use levels in consumer products.

Predictive performance was estimated in a blinded study by evaluating a total of 17 fragrance materials, and reproducibility of the continous predictions was assessed by evaluating 11 of the materials in three replicate experiments. Results illustrate that predicted LLNA EC3/human NESIL values from the GARDskin Dose-response assay correlate well with reference data (geometric mean fold-error: 3.8 and 2.5, respectively), and that the continuous potency predictions are reproducible between experiments (geometric mean fold-change: 3.1). A case study using isocyclocitral was used to illustrate how the assay can be implemented into an NGRA-framework, which is an exposure driven risk assessment methodology. The predicted NESIL value from GARDskin Dose-response was used within a weight-of-evidence  approach to derive a PoD for use in QRA. Sensitization assessment factors were applied to the PoD to determine acceptable exposure levels at which no skin sensitization induction would be expected for different product types based on exposure.

In conclusion, the results reported from this study demonstrate that the predicted potency values from the GARDskin Dose-Response assay are reproducible between experiments and show good concordance with the published values. The case study illustrates a proof of concept and establish the assay as a relevant source of information to derive NESIL values for conducting QRA evaluations for fragrance materials without any new animal data.