Bio: Dr Sylvia E. Escher is a chemist and a human toxicologist by training. She joined the Fraunhofer Institute of Toxicology and Experimental Medicine (ITEM) in Hanover, Germany, in 2006. ITEM is the EUROPEAN leading public non-profit research institute in inhalation toxicology. She currently leads the division “Safety Assessment and Toxicology” and the department “In Silico Toxicology”. Her research interests include the integration of alternative methods into regulatory risk and hazard assessments, starting with in silico approaches and progressing to in vitro methods such as omics. Her team also develops toxicological databases such as RepDose® or VICT3R, which are used to improve the TTC concept or to develop virtual control groups, respectively. ITEM also develops PBKiT, a bottom-up PBK model which predicts the kinetic properties of e.g. airborne compounds using in vitro ADME assays.

Abstract: In silico tools are frequently used by experts in data-poor situations, e.g. to predict toxicological properties. This presentation illustrates the development of two in silico models, i) a QSAR model for respiratory tract irritation and ii) virtual control groups. The talk outlines the process of data curation, the usefulness of ontologies, and how data/tools will be made available to the public. The Respiratox project1 developed a QSAR method to predict human respiratory irritants. The curated project database comprised 1997 organic substances, being classified as irritating (1553) or non-irritating (444) based on data from e.g. inhalation studies with acute and repeated exposure. Gradient-Boosted-Decision-Trees provided the best discrimination between both compound classes. The QSAR model is available online and provides a prediction together with a list of structurally similar analogues to facilitate expert review. Virtual-control groups2,3 shall replace study-specific control groups and can thus reduce 25% of animals in animal studies. To achieve this, large databases with control group animals from dog/rodent studies were collected. Laboratory data and histopathological findings were standardized using SEND terminology. The ongoing work aims to identify the necessary parameters to define a virtual-control group. Both approaches reduce animal testing and emphasize the importance of data sharing.

References:

1. https://doi.org/10.1016/j.yrtph.2021.105089

2. VICT3R - Developing and implementing Virtual Control Groups to reduce animal use in Toxicology Research

3. Virtual Second Species | Innovation Platform