1. Computational Toxicology and Virtual Development in Drug Design
Dale E. Johnson, Pharm.D., Ph.D. Chief Scientific Officer
ddplatform LLC

2. The “Problem” in pharmaceutical R&D
~ $700 MM and over 10 years to develop novel drug
Approximately 75% of overall R&D cost attributed to failures
The “Solution” for R&D
Identify/eliminate problematic drugs early
Design desirable properties into drugs

3. Drug Discovery: the hunting process where is toxicology today?
Target Selection
Lead Identification
Lead Optimization
Identification of potential targets
Screen development
Lead explosion/ optimization
Target verification
High-throughput screening
Potency in disease
Target selection
Secondary assays/ mechanism of action
Pharmacokinetics
Hits to leads
Early toxicology
From: Rosamond and Allsop, Science 287, 1973 (2000)

4. Early toxicology at the Lead Optimization Step: still a high failure rate – high cost to R&D
ADME, PK, TOX Lead optimization
Primary & secondary efficacy screening
Secondary in vitro screening
In vivo and
mechanistic
screens
Lead
selection
Chemical Libraries
Chemical Libraries
Development
Candidate
65%
Drop Out
IND enabling studies
Phase I, II

5. The toxicology solution
Incorporate predictive toxicology concept throughout discovery & development
Design reduced toxicity into chemical libraries
Create expert systems to accelerate and increase success rate
Expert systems must be multi-disciplinary for real impact

6. Major needs in Predictive Toxicology: Recent industry surveys
Predictive software with updated databases
Improved data mining capabilities
Enhanced in vitro mechanistic screens
Ready access to human hepatocytes and other cells
Relevant application of new technologies ie. toxicogenomics

7. Major needs in Predictive Toxicology: Recent industry surveys
Predictive software with updated databases
Improved data mining capabilities
Enhanced in vitro mechanistic screens
Ready access to human hepatocytes and other cells
Relevant application of new technologies ie. toxicogenomics

8. Missing elements in the toolbox
Quality data from controlled sources
Newly created database(s) using “pharmaceutical” chemical space
Multi-disciplinary chem-tox Information / decision tools
Data mining via “med chem building blocks”
Flexibility to incorporate all data from internal and external sources
Web-based, platform independent

9. LeadScopeTM Technology
Structural analysis based on familiar structural features
Powerful graphical representations and dynamic querying
Refine structure alerts to reflect new assay results
Statistically test structural hypotheses

10. RTECS database & liver toxicity
~7000 compounds with liver toxicity codes
Expert conversion to grades (risk)
Ordinal ranks using severity of findings, dose, regimen, species
Create 1o liver tox – chemical space
Data mining with ToxScopeTM: correlations between chemical structure and liver toxicity

11. Feature Hierarchy
Graphic Panel
Filter Panel
Information Windows

12. Portion of the Heterocycles hierarchy showing 3 levels of the pyridine subhierarchy
Selected subset of compounds containing a pyridine substructure with an acyclic alkenyl group in the 2-position
Subset contains 2 compounds

13. Each structure feature in the hierarchy is defined as a substructure search query
Structural definition
atom and bond restrictions

14. Compounds containing a pyridine, 2-(alkenyl, acyc) substructure

15. Uncovering bias in chemical space within data sets
Detect + and – coverage within a desired chemical space
Understand decision errors that can be introduced with biased space

18. Structural alerts Can rapidly find structural alerts
Can view new libraries in relation to structural alerts
Can evaluate impact of alert on optimization scheme

20. RTECS grade 5 only

21. ToxScopeTM Components
LeadScopeTM Enterprise Technology
Several public or commercial databases
New databases using “pharmaceutical" chemical space
New specific organ toxicity database
Structural alerts
Continual updates on target organs

22. Conclusion
“… an in silico revolution is emerging that will alter the conduct of early drug development in the future.”
“Preclinical safety must transition from an experimental-based process into a knowledge-based, predictive process, where experimentation is used primarily to confirm existing knowledge”

23. Acknowledgements Grushenka Wolfgang, Co-author
Julie Roberts Kevin Cross
Bill Snyder Michael Crump
Chris Freeman Jeff Miller
Don Swartz Michael Murray
Ilya Utkin Mark Balbes
Wayne Johnson Zhicheng Li
Allen Richon Yan Wang
Paul Blower Limin Yu
Glenn Myatt Sighle Brackman
Emily Johnson Lisa Balbes