Skin Sensitization Application based on an Integrated Testing Strategy

Allergic contact dermatitis (aka., skin sensitization) accounts for 20% of all contact dermatitis cases (1) and hence has an estimated annual cost up to $200 million. It is also a public health problem, responsible for more than seven million outpatient visits annually. Currently, there are more than 3,700 substances that are identified as contact allergens.

At Douglas Connect (DC), we continue to support innovation through the adoption of state-of-the-art scientific approaches to help risk assessors and regulators perform a better job. At the SOT 2017 conference, we launched DC ITS SkinSens, a solution for assessment of chemicals' safety for skin sensitization. The solution is based on the integrated testing strategy (ITS-3) developed by Joanna Jaworska et al (2). It combines data from three validated in vitro assays (DPRA, KeratinoSens and h-CLAT) as well as in silico predictions for bioavailability as an alternative to performing LLNA animal experiments. The tool allows scientists to evaluate the skin sensitization hazard of their chemicals utilizing the expert-knowledge embedded in the skin sensitization adverse outcome pathway (AOP) as published by OECD.

The tool provides a quantitative assessment of the skin sensitization potential using a Bayesian network approach applied against AOP key events. The network has an overall accuracy of 80% matching the accuracy of the published network (2). The network also provides confidence in prediction (using Bayesian factors). The accuracy of the high confidence predictions is 98% while moderate confidence predictions have an accuracy of 71%.

The prediction of chemical potency for sensitization is first calculated in the form of a probability distribution over four potency classes (non-sensitizer, weak, moderate and strong). The probability distribution is then transformed into Bayes factors to remove prediction bias from the training set distribution and to calculate uncertainty in a quantitative manner therefore judging the confidence in the prediction. ITS-3 is based on a database of 207 chemicals for which the complete set of in vivo and in vitro data are available.

Additionally, DC ITS SkinSens can direct consequent testing by value of information calculations, i.e. it suggests the experiments to be conducted in order to achieve maximum information gain reducing uncertainty in prediction.

Continuing our philosophy of supporting open science, the development prioritized the use of cheminformatics tools that are open (such as RDKit) or that offer friendly academic license schemes (e.g, ChemAxon calculators, OCHEM web services).

DC ITS SkinSens can be integrated into existing workflows for industrial partners, academia and regulators using Application Programming Interfaces (see here). It can also be easily deployed behind the corporate firewall due to its containerized architecture.

DC ITS SkinSens builds on the Bayesian network for skin sensitization potency assessment to provide a web application built on open and accessible components for supporting decision making by a risk assessor. It offers a potency hypothesis with mechanistic interpretation and quantitative weight of evidence. By assessing the maximum potential knowledge gain through mutual information, it also offers an adaptive testing strategy for chemicals.

Acknowledgments: We would like to thank Dr. Jaworska for her support to the project and the cheminformatics community for providing open tools that made such development possible.

REFERENCES

(1) Coman, G.; Zinsmeister, C.; Norris, P. Occupational Contact Dermatitis. Allergy, Asthma, Clin. Immunol. 2008, 4, 59–65.

(2) Jaworska, J. S.; Natsch, A.; Ryan, C.; Strickland, J.; Ashikaga, T.; Miyazawa, M. Bayesian Integrated Testing Strategy (ITS) for Skin Sensitization Potency Assessment: A Decision Support System for Quantitative Weight of Evidence and Adaptive Testing Strategy. Arch. Toxicol. 2015, 89, 2355–2383.