From genome-wide arrays to tailor-made biomarker readout – Progress towards routine analysis of skin sensitizing chemicals with GARD

TLDR

Allergic contact dermatitis is a major public health concern, and EU regulations together with advances in skin‑sensitization mechanisms have spurred the development of non‑animal in‑vitro tests such as GARD, though existing protocols are not yet suitable for standardized screening. This work aims to refine GARD so it can transition from biomarker discovery to a reliable, cost‑effective routine assay for identifying chemical allergens. The authors migrated GARD’s gene‑expression readout to the NanoString nCounter platform, optimized normalization for serial sample processing, and introduced a novel batch‑correction strategy. Testing 29 compounds yielded 94 % sensitivity, 83 % specificity, and 90 % accuracy, demonstrating that the updated workflow maintains predictive performance while enabling standardized screening and integration into testing strategies.

Abstract

Allergic contact dermatitis (ACD) initiated by chemical sensitizers is an important public health concern. To prevent ACD, it is important to identify chemical allergens to limit the use of such compounds in various products. EU legislations, as well as increased mechanistic knowledge of skin sensitization have promoted development of non-animal based approaches for hazard classification of chemicals. GARD is an in vitro testing strategy based on measurements of a genomic biomarker signature. However, current GARD protocols are optimized for identification of predictive biomarker signatures, and not suitable for standardized screening. This study describes improvements to GARD to progress from biomarker discovery into a reliable and cost-effective assay for routine testing. Gene expression measurements were transferred to NanoString nCounter platform, normalization strategy was adjusted to fit serial arrival of testing substances, and a novel strategy to correct batch variations was presented. When challenging GARD with 29 compounds, sensitivity, specificity and accuracy could be estimated to 94%, 83% and 90%, respectively. In conclusion, we present a GARD workflow with improved sample capacity, retained predictive performance, and in a format adapted to standardized screening. We propose that GARD is ready to be considered as part of an integrated testing strategy for skin sensitization.