How can computational predictions guide risk-based decision making in the presence of data gaps?
Risk-based chemical safety decision making requires an understanding of two fundamental components: hazard and exposure. In the absence of hazard characterization data, toxicologists turn to structure-based inference tools to approximate hazard. Threshold of Toxicological Concern is a level of chemical exposure that is presumed, based on studies performed using structurally similar compounds, to carry minimal risk of impact on human health. Compounds are grouped with similar actors using a decision tree originally conceived by Cramer et al. In-life data for compounds within these “Cramer classes” are used to benchmark the group, and additional safety factors are applied to ensure the resulting TTC for the class is sufficiently conservative. ScitoVation has systematically evaluated the Cramer decision tree for a corpus of 45,000 compounds, resulting in TTC values that can be compared to exposure estimates to formulate a risk estimate in the absence of chemical-specific test data. The results of this exercise are shared through a searchable, sortable online portal.