The GARD™potency assay for potency-associated subclassification of chemical skin sensitizers – Rationale, method development and ring trial results of predictive performance and reproducibility

Poster presented at the 2021 World Congress on Alternatives and Animal Use in the Life Sciences

Robin Gradin, Angelica Johansson, Andy Forreryd, Amber Edwards, Veronika Hoepflinge, Florence Burleson, Helge Gehrke, Erwin Roggen, Henrik Johansson| SenzaGen AB, 22381 Lund, Sweden; Burleson Research Technologies, Morrisville, NC 27560, USA; Eurofins BioPharma Product Testing Munich Gmbh, 82152 Planegg, Germany; 43RsMC Aps, 2800 Kongens Lyngby, Denmark

Download a copy



  • GARDpotency is an assay for sub-categorization of strong sensitizers (CLP sub-category 1A), allowing for discrimination from weak sensitizers (CLP sub-category 1B) and non-sensitizers. The method is based on the GARD platform, combining human immune cells, a genomic biomarker readout and machine learning-assisted classifications.
  • Sequential combination of GARDskin and GARDpotency forms the GARD Defined Approach, for complete hazard and risk assessment of skin sensitizers into three categories (CLP 1A, CLP 1B, non-sensitizers).
  • A blinded ring trial, comprising 28 chemicals, demonstrated that GARDpotency is functional and reproducible, with an accumulated predictive accuracy of 91% across three laboratories. In the same dataset, the GARD Defined Approach classifies chemicals into three categories with 86% accuracy.


The advancement of non-animal approaches for hazard assessment of skin sensitizers have generated a variety of alternative assays with discriminatory properties comparable with those of accepted in vivo methods. However, hazard identification is rarely sufficient and information permitting the relative ranking of chemicals’ skin sensitization potency is desired. For example, the globally harmonized system of classification and labelling of chemicals (GHS/CLP) extends the binary hazard assessment with a qualitative subcategorization to distinguish between weak and strong skin sensitizers.

Though substantial efforts have been made towards developing alternative methods for potency assessment, none have gained regulatory acceptance, emphasizing that continued development of improved alternative assays remains a high priority.

The genomic allergen rapid detection (GARD) is an in vitro testing platform for assessment and characterization of chemical sensitizers, based on evaluation of transcriptional patterns of endpoint-specific genomic biomarker signatures in a human dendritic-like cell line following chemical exposure, in order to provide machine learning-assisted classifications of tested substances. The GARDskin assay was recently subjected to a formal validation procedure (OECD TGP 4.106) and reported a reproducibility between laboratories of 92%, as well as a predictive accuracy of 94%, for sensitization hazard assessment.

Here, we present the implementation of the related GARD application GARDpotency, for potency-associated subcategorization of chemical sensitizers. Following prediction model establishment, the functionality of the assay was validated in a blinded ring-trial, in accordance with OECD-guidance documents, by assessing predictive performance and reproducibility. It was found that the assay is functional and predictive, with an estimated cumulative accuracy of 88% across three laboratories and nine independent experiments. The within-laboratory reproducibility measures ranged between 63-89%, and the between-laboratory reproducibility was estimated to 61%. In conclusion, the in vitro GARDpotency assay constitute a standardized, functional assay, which could be a valuable tool for hazard characterization of skin sensitizer potency.