1. Changing the trajectory of drug R&D
Robert Plenge, MD, PhD
American College of Rheumatology
November 15, 2016

2. Disclaimer I am a full-time employee at Merck

3. Our two fundamental challenges
Cost to develop an asset has increased by 1/3rd since 2010
Average peak sales per asset has halved since 2010

4. Where things go wrong & what that costs
$$$
About 1 out of 10 programs make it to launch
Paul et al NRDD 2010

5. Where things go wrong & what that costs
But critical phase is choice of target and early development
$$$
About 1 out of 10 programs make it to launch
Paul et al NRDD 2010

6. We relied on preclinical models to pick targets and estimate efficacy in heterogeneous human populations
It was…
Discovery
(new targets)
Optimization
(pre-clinical)
Early Development

7. Today, humans are the model organism of choice for new targets and precision medicine
Discovery
(new targets)
Optimization
(pre-clinical)
Early Development

8. Human genetics is a powerful tool to establish “causal human biology” at the very beginning
Science Translational Medicine, July 2016

9. First, an example from cardiovascular disease

10. Many genes influence cholesterol levels and risk of heart disease
There are examples of human genetics leading to new drug targets (PCSK9)
Many genes influence cholesterol levels and risk of heart disease
Atherosclerotic Plaque
Blood Flow
PCSK9 mutations associated with high and low LDL cholesterol levels (and heart disease risk)
…and design studies to find drugs that fix the underlying molecular defects – for example, blocking PCSK9 lowers LDL (or “bad”) cholesterol in the blood.
Lysosome
LDLR
PCSK9
LDL-C
mAb
Recycling

11. Now, examples from osteoporosis and rheumatoid arthritis

12. A quick primer on genetics of osteoporosis and related traits
>100 common variants associated with osteoporosis
Additional genes mutated in rare forms of bone mass loss / accrual
Experimental studies determine function, including gain- vs loss-of-function of risk allele
While many genes implicated, only a few have led to novel therapies
David Karasik et al NRR 2016

13. osteoblast Teriparatide recombinant PTH Romosozumab approved
anti-sclerostin
phase III
osteoblast
Denosumab
anti-RANKL
approved
Estrada et al NG 2012

14. A very similar story in RA
>100 common variants associated with rheumatoid arthritis
Additional genes mutated in rare forms of immunodeficiency or autoinflammation
Experimental studies determine function, including gain- vs loss-of-function of risk allele
While many genes implicated, only a few overlap with approved therapies or those in early development
Okada et al Nature 2013

15. PADI4
Small molecule inhibitor
early development
abatacept
CTLA4-Ig
approved
tocilizumab
anti-IL6R
approved
TYK2
Small molecule inhibitor
early development
Smolen et al Lancet 2016

16. Most are retrospective examples…what is a prospective approach to new targets?

17. Pick a human phenotype for drug efficacy
High
Low
GOF
LOF
Gene function
Plenge et al NRDD 2013

18. Pick a human phenotype for drug efficacy
High
Low
GOF
LOF
Gene function
Nelson et al NG 2015

19. Pick a human phenotype for drug efficacy
High
Low
GOF
LOF
Gene function
Okada et al Nature 2013

20. Pick a human phenotype for drug efficacy
High
Low
GOF
LOF X X
Identify a series of alleles with range of effect sizes in humans (but of unknown function) X X X X X
Gene function

21. Pick a human phenotype for drug efficacy
High
Low
GOF
LOF X X
Assess biological function of alleles to estimate “efficacy” response curve X X X X X
Gene function

22. New target for drug screen! Pick a human phenotype for drug efficacy
High
Low
GOF
LOF
Toxicity X X
Assess biological function of alleles to estimate “efficacy” response curve X
Assess pleiotropy as proxy for ADEs X X X
This provides evidence for the therapeutic window at the time of target ID & validation. X
Gene function

23. RANK-RANKL and denosumab Pick a human phenotype for drug efficacy
Rare variants & osteopetrosis
Efficacy
Common variants & BMD, fracture risk
Osteoporosis
High bone
density
Low bone
GOF
LOF X X
Toxicity X
Assess pleiotropy as proxy for ADEs X X
Rare RANK variants & Paget’s disease (no known GoF mutations in RANKL) X
This provides evidence for the therapeutic window at the time of target ID & validation. X
Gene function

24. RANK-RANKL and denosumab Pick a human phenotype for drug efficacy
Rare variants & osteopetrosis
Efficacy
Common variants & BMD, fracture risk
Osteoporosis
High bone
density
Low bone
GOF
LOF X X
Toxicity
No “obvious” pleiotropic effects that could be ADEs X X X
Rare RANK variants & Paget’s disease (no known GoF mutations in RANKL) X
This provides evidence for the therapeutic window at the time of target ID & validation. X
Gene function

25. Allele that protects from RA is associated with loss-of-function (LoF)

26. Same LoF allele has no obvious increased risk of infection

28. Therapeutic hypothesis:
Inhibiting TYK2 will protect from RA without risk of infection

29. Tokarski et al JBC 2015

30. But (and there is always a but…)

31. Limitations of the approach
Not all successful drugs will have genetic support
Little direct genetic evidence for TNF as a target
Other approaches to causal human biology
Even those targets with genetic support may fail in clinical development
Cathepsin K (CTSK) mutations cause pycnodysostosis
Odanacatib failed in Phase III due to safety

32. Introducing novel therapies is an important component of our future health care system…
…but we need to do more to deliver affordable medicines that matter

33. Questions?
@rplenge