QSARs Mix: Development of high accuracy QSARs for mixtures to bridge the data gaps in skin and eye irritation

As a part of the toxicological information required by regulators, the skin or eye irritation potential needs to be determined for all chemical substances being manufactured in quantities of over ten tonnes per year. Improving the predictive ability of the computational models for gathering this information would bolster the scientific merit of these in silico approaches and progress them towards direct use in regulatory studies, where in future they could replace animal testing. 

Unlike existing tools that may be used to predict skin and eye irritation endpoints, the team proposes an in-silico method which will consider toxicological and physicochemical properties simultaneously to generate an accurate, quantifiable prediction. The proposed tool will function by determining specific structural alerts and then relativizing their toxic potential by accounting for tissue exposure using appropriate physicochemical properties such as solubility, volatility, surface tension etc. The toxicity of a mixture can be determined following a phase equilibrium thermodynamic approach similar to that already defined and validated for aquatic mixture toxicity for MOA1 compounds.

This project will utilise the available expertise at KREATiS (in-silico modelling of High Accuracy QSARs) and CEHTRA (consultants in toxicology and environment). The model will then be evaluated for its reliability using their expertise in statistics for applicability domain assessment and the model validated following the five OECD principles.  

Full details about this CRACK IT Challenge can be found on the CRACK IT website.

KREATiS, in collaboration with Challenge Sponsors Shell, has developed iSafeRabbit, a set of high accuracy QSARs for predicting the skin and eye irritation and corrosivity potential of petrochemical substances and mixtures, to replace in vivo animal studies.

The iSafeRabbit skin/eye models have been validated for 12 chemical families and their predictions have a true positive/negative rate of 95% using validated in vivo experimental data as reference. The models also satisfy the five recommended OECD principles for QSAR models making them fit for regulatory purposes.

To maximise its application and wide-scale use as an in silico alternative to in vivo experimentation, KREATiS is looking to extend the scope of iSafeRabbit and validate the model to include new chemical groups. Through CRACK IT Solutions, KREATiS is seeking in vitro and in vivo irritation data from various industrial domains, in particular cosmetics and household products, but also data on active ingredients from pharmaceuticals, pesticides and biocides.

To find out more about iSafeRabbit visit the KREATiS website or contact the KREATiS team.  
 

Figure 1. iSafeRabbit skin and eye irritation QSARs.​

Posters

Sahigara F, Charmeau-Genevois C, Bicherel P, Perea M, Sarang S, Eadsforth C, Austin A, Thomas P. (2016). A High Accuracy QSAR based on rabbit data to predict the human skin irritation potential of individual constituents and mixtures.