overview on etox – a project in the framework of …axlr8.eu/workshops/steger-hartmann.pdfencoding...

Overview on eTox – A Project in the Framework of Innovative Medicines Initiative (IMI)Dr. Thomas Steger-Hartmann, Bayer HealthCare

• Overview eTox • 23. May 2011

Changes in Technology – From “Forecast“ to Prediction


Why should there be in vivo toxicity prediction?

Benefit of early in silico prediction:

� Improved selection/exclusion of candidate compounds, lowering attrition in later phases

� Qualification of impurities or metabolites whithout additional in vivo studies (3Rs)

� Safety assessment of chemicals in the context of REACH without additional in vivo studies (3Rs)

� Development of a more targeted in-vivo testing strategy

� Better predict human toxicities and/or safe starting doses


In vivo Toxicity Prediction – Gaps

� Data from public domain sources on toxicity is often biased towards

toxic effects (negative tox data is usually not published)

� The data quality of tox reports in the public domain can hardly be

assessed and is often questionable

� The chemical space of published tox data is dominated by industrial

or household chemicals (pharmaceuticals are underrepresented)

� Prediction is mostly directed to pure a chemistry approach -

integration of pharmacodynamic/toxicodynamic and DMPK data

lacking


How Can We Get Better?

The tremendous wealth of high

quality toxicology data in the

archives of the pharmaceutical

companies is not yet leveraged!

High Quality

Pharmacology

Broad chemical spaceICH conform

Different species

Multipleendpoints

Buried in toxicology archives

High QualityTox Data

Repository


What is IMI - Innovative Medicines Initiative

Who? European Federation of Pharmaceutical Industries and Associations

(EFPIA) + European Community founded a Public-Private Partnership for pre-

competitive collaboration: Innovative Medicines Initiative (IMI). It represents a

Joint Technology Initiative within the EU’s 7th framework

Objective? To adress the key “bottlenecks” in the biomedical R&D process in

Europe, i.e. to accelerate discovery and develop more effective innovative

medicines with fewer side-effects.

How much? Budget of 2 billion € over 5 years (framework program 7)

2 Billion EURO

960 Mio. Euro 960 Mio. Euro


Translational

MedicineClinical

development

Pharmaco

vigilance

Predictive

pharmacology

Predictive

toxicology

Identification

of biomarkersPatient

recruitment

Validation of

biomarkers

Benefit/Risk

assessment

with regulatory

authorities

Research in 2 key areas

Underpinning themes

Drug development process

Areas of bottlenecks

Discovery

Research

Preclinical

development

Efficacy

Safety

Knowledge Management

Education & Training

eTox project

The four IMI pillars


eTox: Project Plan and Deliverables

� Data sharing: Exploit legacy preclinical reports from the

pharmaceutical industry to link chemical features to pathology findings.

� Establishment of a toxicological database with high quality in vivo and

in vitro data. This repository will form the basis of prediction model

development.

� Construction of integrated prediction models for target organ toxicity.

� Validation of new prediction models: the validation experience will be

shared between companies and with regulators.


Participating EFPIA Partners

Novartis Pharma AG (Francois Pognan)

AstraZeneca AB

Bayer HealthCare

Boehringer Ingelheim International GmbH

Laboratorios del Dr Esteve, S.A.

GlaxoSmithKline Research and Development LTD

Janssen Pharmaceutica NV

H. Lundbeck A/S

Pfizer Limited

F. HOFFMANN-LA ROCHE AG

Servier

Sanofi-Aventis

UCB Pharma SA


Contributions of EFPIA Partners

� Provision of high quality data (mostly GLP) from sy stemic toxicity studies (different species), DMPK studies, safety pharmacology studies and pharmacophore investigatio ns (ligand binding)

� Data extraction from legacy reports (done in-house or with external help)

� Expertise in the field of QSAR, pharmacophore modelling and ontologies

� Contacts to regulatory authorities (including interfacing reg. databases)

� Profound toxicological expertise


An Estimate for the Treasure

Additional envisaged data points / endpoints:

• Safety Pharmacology (in vitro including MDS or CEREP screen)

• Safety Pharmacology (Core battery)

• Toxicokinetic data (exposure!)

• DMPK data

• Further species (Cynomolgus)

• Gene expression data

• …


Participating Public Partners

Fundació IMIM (Ferran Sanz, Carlos Diaz, Eva Molero)

Fundación Centro Nacional de Investigaciones Oncológicas Carlos III

European Molecular Biology Laboratory

Liverpool John Moores University

Technical University of Denmark

Universität Wien

Vrije Universiteit Amsterdam (VUA)

Inte:Ligand GmbH

Lhasa Limited

Molecular Networks GmbH

Chemotargets SL

Lead Molecular Design S.L.


Contributions of Public Partners

� Expertise in bioinformatics

� Expertise in database development and hosting (honest broker

concept)

� Expertise in the development of QSAR models and Expert

systems

� Expertise in DMPK modelling (CYP, transporters)

� Expertise in validation of predictive models


SoftwareDevelop-ment &validation FIMIM

CNIO

EMBL

LJMU

DTU

UVIE

VUA

IL

LL

MN

CT

LMD

In silico

pharmacology

& toxicologyDatabases

Bioinformatics(including ontologies & text mining)

DMPK-relatedtoxicology

Molecularapproaches

FIMIM

CNIO

EMBL

LJMU

DTU

UNIVIE

VUA

IL

LL

MN

CT

Transporters

LMD

In silico

pharmacology

& toxicology

Expertise of the Public Partners


eTOX – The Workflow

QSAR and Expert Systems for in silico Toxicity Prediction

2. Data sharing andcreation of data base

3. Building the models

N

4b. Validation with new structures

4a. Testing alerts

5. Refinement (new structuresor limited early toxicity studies)

New Report

N

OH

OH

N

O

N

OH

Chemical structure & descriptors

Bibliography

Public databases

Pharma reports

1. Data collection


� Classification of data: confidential (majority), non-confidential

(structure and EPAR in the public domain)

� Data transfer to “honest broker” bound by legal framework

� Encoding of sensitive structural information by honest broker

(“descriptors”)

� Provision of IT infrastructure to prevent loss and illegal down-load

of sensitive data

Overcoming the hurdles – safe data sharing and IP protection


� Joint database will be managed by a “neutral trusted institution”

(=honest broker).

� Bilateral non-disclosure agreements signed between each pharma

company and the trusted institution.

� The chemical/biological records of single compounds classified as

sensitive will never be disclosed to any pharma company or academic

partners different from the owner.

� The use of the joint database by the developers of models and expert

systems will be done by accessing the joint DB in the trusted

institution without capability of downloading it or accessing to single

records.

Component 1: The Honest Broker


Databasecontains toxicologicaldata and descriptors (encodedstructural information)

Novartis Bayer GSK AZ J&J Roche ….

Novartis Bayer GSK AZ J&J Roche …

Ow

ned

only

by E

FP

IAco

mpa

ny

Fre

e ac

ces

by a

ll pa

rtic

ipan

ts

Established and curated

by honest broker(O

wnder of database source codes)

.…

Databases contain pharm-tox data and structures - background

Publicpartners

Component 2: Selected Accessibility

Derive modelsand build tools

Physical Access BarrierEncoding

Non-Encoded

Database containsfull information –

foreground with data

EFPIApartners


� Start of project: January 2011

� Duration: 5 years

� Budget:EFPIA: 7.9 Mio. €IMI-JU funding: 4.7 Mio. €Total costs: 13.9 Mio. €

� Data base schema developed

� Ontology development started

� CROs for data extraction identified

� Public data sources identified

eTox Project Status (05/11)


Project achievements

� Creation of a complex framework of legal statutes and IT-technical provisions to overcome the hurdles of sharing proprietary data of EFPIA companies.

� Development of a first version of a toxicity ontology for seamless data gathering, integration and exploitation.

� Design and successful testing of strategies for the masking of sensitive structural information of compounds.

� Design and setup of the first version of the eTOX central database.

� Compilation and assessment of public data sources.

� Agreement on the (modular) architecture of the eTOX predictive system.

� Analysis and benchmarking of current models for toxicity prediction, and definition of quality criteria for method selection and development.

� Development of an innovative multi-scale modelling strategy for the prediction of cardiotoxicity (J. Chem. Inf. Model. 2011; 51:483-92)


� Interfaces to –omics databases (e.g. Innomed, FDA

liver tox etc.): Identification of mechanisms and

safety biomarkers

� Interface to Clinical and Pharmacovigilance

databases (the ultimate goal is to predict proband’s

or patient’s safety)

eTox and Beyond – The Vision


More information at…

…www.e -tox.net

Thank you for your attention!

overview on etox – a project in the framework of …axlr8.eu/workshops/steger-hartmann.pdfencoding...

Documents