Session 6
Beyond data sharing - towards data transparency, management, mining and application to predictive safety assessment
Programme of the Session
- S06-01
The IMI eTOX initiative – data mining, read-across and predictive models for target evaluation and early drug candidate assessment
Thomas Steger-Hartmann1, Francois Pognan2, Ferran Sanz3
1Investigational Toxicology, Bayer AG, Berlin, Germany; 2PreClinical Safety, Novartis Institute for Biomedical Research, Basel, Switzerland; 3Research Programme on Biomedical Informatics (GRIB), Universitat Pompeu Fabra, Barcelona, Spain - S06-02
Preclinical ontologies - a key feature for toxicity data exchange and mining
Philip Drew, Francois Pognan, Philippe Marc, Carlo Ravagli
PreClinical Safety, Novartis Institute for Biomedical Research, Basel, Switzerland - S06-03
Toxicity databases of chemical substances in Japan to improve in silico approaches for regulatory safety assessment
Takashi Yamada1, Makoto Hayashi2, Masamitsu Honma1
1National Institute of Health Sciences, Tokyo, Japan; 2Makoto Intenational Consulting, Ebina, Japan - S06-04
Linking in vivo toxicity data to ToxCast/Tox21 in vitro assay data
Richard Judson
National Center for Computational Toxicology, US EPA, RTP NC, United States - S06-05
Progress in in silico toxicity model development - lessons learnt analysing complex toxicity data
Manuel Pastor, Ferran Sanz
Department of Experimental and Health Sciences, University Pompeu Fabra, Barcelona, Spain
Session Abstract
The pharmaceutical industry has accumulated a wealth of data residing in private company archives under various shapes and formats. To lower safety-related attrition rate the IMI eTOX project (“electronic toxicity”; www.e-tox.net) established processes for pre-competitive safety data sharing and subsequent exploitation. Enhanced data availability and sharing among companies allows for chemical and biological read-across of early drug candidates. In addition, the shared preclinical data is used for mining to build predictive tools which leads to a more efficient drug development process and also contributes to the reduction of animal use (3R) and eventually also change the regulatory risk assessment.
The eTOX project, a public-private partnership between 13 US and European pharmaceutical companies, 11 academic institutions and 6 small-to-medium size enterprises (SMEs) was launched under the umbrella of the European Innovative Medicines Initiative (IMI) to leverage existing data in drug safety assessment. The project developed a consolidated standardized toxicity database consisting of preclinical data from both marketed drugs and failed drug candidates. Prerequisite for enabling multi-parametric searches (structures + toxicity findings for a broad variety of toxicity endpoints) was the unprecedented establishment of controlled vocabularies and ontologies, which are aligned with FDA SEND (Standard for Exchange of Nonclinical Data). The data were then used to build an in silico systems for read-across and predictions for organ and in vivo repeated-dose toxicity. To date, a new compound without safety data can be mapped to more than 8000 systemic toxicity reports and 2500 structures in order to derive hypotheses on the expected safety of the drug candidate.
In parallel to the data sharing initiative for pharmaceutical data similar initiatives were launched for data from the chemical industry (the Japanese Ames NIHS Database and Hazard Evaluation Support System Integrated Platform [HESS]). Combining the efforts of the different initiatives will broaden the chemical space accessible to searches, but also reveal commonalities of underlying toxicities, which will help to validate Adverse Outcome Pathways (AOPs).
The wealth of the collected data can be further leveraged by combining it with consortia, which have an in vitro mechanistic approach, such as the Tox21/ToxCast initiative. eTOX has provided access to its database to the National Center for Competency Testing at EPA in order to assess the degree of overlap between the eTOX data and ToxCast and to explore correlations between ToxCast and in vivo preclinical data
The symposium will present the current achievements of the three initiatives as well as the results of hypothesis-driven data mining and derived computational models with a particular focus on the data exchange and interactions between the mentioned international consortia. In addition, an outlook will be provided on future approaches to sharing and analysing "big data" in the field of preclinical toxicity.
The eTOX project, a public-private partnership between 13 US and European pharmaceutical companies, 11 academic institutions and 6 small-to-medium size enterprises (SMEs) was launched under the umbrella of the European Innovative Medicines Initiative (IMI) to leverage existing data in drug safety assessment. The project developed a consolidated standardized toxicity database consisting of preclinical data from both marketed drugs and failed drug candidates. Prerequisite for enabling multi-parametric searches (structures + toxicity findings for a broad variety of toxicity endpoints) was the unprecedented establishment of controlled vocabularies and ontologies, which are aligned with FDA SEND (Standard for Exchange of Nonclinical Data). The data were then used to build an in silico systems for read-across and predictions for organ and in vivo repeated-dose toxicity. To date, a new compound without safety data can be mapped to more than 8000 systemic toxicity reports and 2500 structures in order to derive hypotheses on the expected safety of the drug candidate.
In parallel to the data sharing initiative for pharmaceutical data similar initiatives were launched for data from the chemical industry (the Japanese Ames NIHS Database and Hazard Evaluation Support System Integrated Platform [HESS]). Combining the efforts of the different initiatives will broaden the chemical space accessible to searches, but also reveal commonalities of underlying toxicities, which will help to validate Adverse Outcome Pathways (AOPs).
The wealth of the collected data can be further leveraged by combining it with consortia, which have an in vitro mechanistic approach, such as the Tox21/ToxCast initiative. eTOX has provided access to its database to the National Center for Competency Testing at EPA in order to assess the degree of overlap between the eTOX data and ToxCast and to explore correlations between ToxCast and in vivo preclinical data
The symposium will present the current achievements of the three initiatives as well as the results of hypothesis-driven data mining and derived computational models with a particular focus on the data exchange and interactions between the mentioned international consortia. In addition, an outlook will be provided on future approaches to sharing and analysing "big data" in the field of preclinical toxicity.
