Fraunhofer Institute for Toxicology and Experimental MedicineDevelopment of a QSAR-based prediction tool
The overall aim of the project “RespiraTox” is to develop a QSAR model and user-friendly tool that reliably predicts human respiratory irritancy of single compounds and mixtures.
RespiraTox
Binary classification of compounds - Irritation in the respiratory tract.
For many substances, inhalation is the most relevant route of occupational exposure, but it is also of relevance to the consumer, i.e. after spray application of household products and cosmetics. Inhalation of substances can lead to local effects such as respiratory irritation, acute and chronic inflammation. To date, there is no specific guideline test to identify respiratory irritants.
The project is subdivided into three main parts:
- Gathering of relevant data on respiratory irritation in a project database
- Development of workflows that model repiratory irritation
- Development of a user-friendly tool
We are currently gathering a sound dataset comprising reliable data on respiratory irritation, e.g.,
by considering Alarie assays, effects reported in acute and repeated inhalation studies (e.g. from the Fraunhofer database RepDose) as well as classification and labeling (C&L) data, e.g., from the ECHA inventory. Based on these data, we will group compounds as tissue and/or sensory respiratory irritants.
The QSAR model will be developed by evaluating different machine learning techniques. It will fulfill the five OECD principles for QSAR validation to demonstrate statistical and mechanistic reliability of the model. This model will then be integrated into a user-friendly tool that provides the QSAR prediction, visualizes the associated uncertainty and shows closest neighbors with experimental data. Besides the prediction for a single compound, mixture toxicity will also be addressed by using case studies.
Dr. Sylvia Escher and her team from Fraunhofer ITEM are contributing a profound knowledge in toxicological data, while Dr. Andreas Karwath and his team from the University of Birmingham are responsible for the modeling and machine learning part.
Use of the in-silico model in the regulatory context
Assessment of whether a chemical will cause respiratory irritation in humans is often based on observations in rodent acute (single) and repeated-dose inhalation toxicity studies. However, there are no specific test protocols in place to determine the irritancy potential of respiratory toxicants or allergens. In the absence of specific or well defined guidelines, respiratory irritation results are extrapolated from acute inhalation toxicity studies (the Organisation for Economic Cooperation and Development (OECD) test guidelines (TG) 403 and 436) performed in rats (OECD, 2009a, OECD 2009b). This involves the modification of protocols to include endpoints for respiratory irritation and requires additional dose groups.
Sensory irritation is tested in the Alarie test, which measures the RD50 value, a concentration of a given compound that causes a 50% decrease of the respiratory rate in rodents (Alarie Y, 1966). Extrapolating the rodent respiratory hazard data to human respiratory irritation, however, is difficult.
Under the REACH (Registration, Evaluation, Authorisation and Restriction of Chemicals) regulations, the registrant may be able to demonstrate that a substance poses no respiratory risk, if exposure via the inhalation route is not to be expected. For most substances, however, exposure via the inhalation route is likely to play a role, and if the substance is a skin or eye irritant, it may be difficult to justify a waiver for acute inhalation studies. Without robust models for respiratory irritation, it is possible that chemicals may pass through the R&D pipeline and reach the marketplace with the potential liability of being respiratory irritants.
Visualized outline for the “RespiraTox” project.