Creating A Framework For Success Clinical Strategies for FDA Approval

1
Creating A Framework For SuccessClinical
Strategies for FDA Approval

• Kent R Thoelke
• Senior Vice-President, Therapeutic Expertise
• Scientific Medical Affairs
• PRA International

2
Results of poorly planned, designed and
implemented clinical strategies.

• Neopharm Phase III fails to meet efficacy
• AstraZeneca Stroke Drug Fails Phase III
• NXY-059 shows NO efficacy
• Telik Cancer Drug Failure
• Stock Plummets 70
• 3 phase III studies (1200 pts) fail to meet
  endpoints
• GPC Biotech drops 61 on Failed Phase III
• Drugs Failure Rocks Sonus, Stock falls 84
• Phase III showed Tocosol to be less effective and
  more toxic than standard taxol in 800 breast
  cancer patients
• Hepatitis C Drug Viramidine Fails Pivotal Phase
  III Trial Again
• Both VISER 1 and VISER 2, two Phase III trials
  failed
• Threshold Cancer Drug Fails To Meet Study Goal

3
Everyone Anticipated Success

• And yet these and many more failed
• The cost of these failures affects
• Investors
• Employees
• Patients
• Poorly designed or unsupported Phase III trials
  waste a critical resourcePatients

4
A Critical Resource is Wasted

• Oncology Trials
• 2006, 679 trials in all phases for lung, breast
  and prostate would need 238,000 patients
  corresponds to ½ of all cancer incidence
• However, current participation rates in trials
  are roughly 3-5 NOT 50
• The goal of clinical research is to bring new
  therapies to patients
• Yet poorly designed Phase III trials cost patient
  lives

5
The Failures

• lt12 of compounds entered into human testing make
  it to market
• Dmitrienko, Chuang-Stein DAgostino, 2007
• Ph III failure rate reported as high as 50
• Chuang-Stein, 2004
• Ph III failure has greatest impact
• accumulated resources spent
• Financial and Patient Resources
• indefensible marketing application

6
Where is the attrition?
Pritchard, Risk in CNS Drug Discovery
7
Success Rate Depends on Target
8
Why Drugs Fail
9
Analysis of Phase III Failures

• Gordian et, al (4/06) analyzed 73/212 failed
  Phase III compounds in CNS, Infectious Disease,
  CV, Endocrinology, Oncology and Respiratory
• Analysis looked at 3 major areas
• Efficacy
• Safety
• Differentiation (against relative comparators)

10
Analysis of Phase III Failures

• 50 of cases failed due to lack of efficacy
• Trials could not demonstrate they were more
  effective than placebo
• 30 of cases failed due to safety
• 20 failed because the new drug could not be
  prove safer or more effective than currently
  marketed drugs

11
Analysis of Phase III Failures

• Even after Phase II, 50 failed due to poor
  efficacy
• In some instances differences in endpoints can
  account for the unexpected failure
• In heart failure Phase II looks at hemodynamic
  EP, Phase III has hard EPs like mortality and
  cardiac hospitalization
• In ACS EP identical in IIIII, but Phase II 30
  days follow-up, Phase III (6-)12 months
• In Oncology many trials will use response rate or
  progression free survival data in Phase II and
  Overall Survival endpoints in Phase III

12
Why is Efficacy the Primary Fail Point?

• Two primary issues
• Endpoint Definitions
• Trials with more objective (diagnostic endpoints)
  more successful (Survival, Tumor Progression,
  Imaging, Viral Load)
• Soft endpoints such as QOL, subjective endpoints
• Mechanistic Novelty
• Drugs that used novel mechanisms of action failed
  2x more often in Phase III as those with known
  MOA
• Drugs with both a novel MOA and less objective
  endpoints failed 70 of the time
• Drugs with a validated MOA and objective
  endpoints failed 25 of the time

13
Steps to Success

• Pick a good drug candidate
• Is the mechanism well defined
• Does mechanism support the target
• Are there valid pre-clinical models
• Sufficient pre-clinical models to support
  clinical target
• In case of biologicals this is tricky as e.g.
  many MABs do not work in anything but primates
  and then in primates the PoC models are not
  available
• Sufficient ADME/Tox data
• Dont ignore showstoppers, Risk/Benefit
• Do animal tox early avoid post patient animal
  tox
• Dosing/Formulation is it a commercial drug?
• Oral drug for mucositis? CIV infusion x 5 days
  for outpatient population

14
Steps to Success, continued

• Choose a development strategy that is supported
  by preclinical/tox
• Pick the right patient population
• Evaluate the use of markers to enrich the
  population Her2/Herceptin
• Be realistic in assessing patient populations,
  competition, treatment paradigms
• Pick appropriate endpoints and realistic
  expectations for a win
• Meet with the FDA (EMEA) early and often
• Pre-IND, CMC, Phase I, End of Phase II, SPA
• Use experts with experience gain from others
  prior pain/failures

15
Product Development Path

• Challenging, inefficient and costly
• FDA Challenge and Opportunity on the Critical
  Path to New Medicinal Products (March 2004)
• 3 major scientific/technical dimensions in
  critical path
• Safety assessment
• Effectiveness
• Product scale-up (industrialization)
• Do not underestimate CMC issues Phase II CTM vs
  III
• right standards and better toolkits needed

16
FAIL FAST

• Compounds should be killed in Phase I/II or as
  early as possible in Phase III
• Identify and admit early flaws
• Safety (Dont ignore or think issues will go
  away)
• Efficacy (Anecdotal early evidence)
• Drugability
• Clinical
• Commercial (Routes of Administration, Costs, etc)
• Temptation to keep drugs alive is high
• Financing decisions, capital raising, emotional
  investment

17
Tough Decisions Sooner

• Phase I and II trials should weed out the
  ineffective compounds
• However evidence suggests companies are not using
  Phase II data to guide decisions appropriately
• Wishful thinking is not an effective decision
  making tool
• Lost objectivity
• Wall street

18
Reducing Attrition at Late StageStatistical
Considerations

• Phase I and II are not merely a means to an end
• Use Phase I and Phase II data to ensure a robust
  Phase III program
• Get More information in Ph I and II
• Adaptive designs
• e.g. Continual Reassessment Methods in Ph I
  oncology
• e.g. Adaptive Dose Response in Ph II
• Multiple Phase II programs with larger numbers of
  pts
• Adaptive or Group Sequential Methods in Stage III
• Use interim/futility analyses to terminate early
  if little chance of success
• Futility analysis cost little in the way of power
  and patient numbers but can save greatly in and
  patients

19
Adaptive Designs

• Allow modifications to some aspects after
  initiation
• e.g.
• sample size re-estimation
• early stopping for efficacy or futility
• as in classical group sequential designs
• response adaptive randomization
• dropping inferior treatment groups

20
Dont Sacrifice Approval for Speed

• Take the time to do Phase I, II and IIa correctly
• Too many drugs go into Phase III without the
  right dose
• Too many drugs pick the patient population based
  on blockbuster revenue targets, pick the target
  based on pre-clinical and mechanism, confirm with
  Phase I
• 1-2 responses in Phase I does not support a Phase
  III trial
• Small, unpowered Phase II trials do not predict
  Phase III behavior strong rationale for a
  futility interim analysis.
• Alternate is to do larger, randomized phase II

21
The Go/No-Go Step in Phase I/II

• The more robust the decision making process in
  which drug candidates to take forward.the higher
  the likelihood of Phase III success!
• Strong data in pre-clinical, Phase I (PK) and
  Phase II (PK/PD, Dose-Range, Efficacy) will
  support a successful discussion with FDA
• Dont skimp on the PK/PD data to help determine
  Phase III dose

22
Increasing Likelihood of Success

• All decisions should be firmly based on the
  science!
• Work with FDA early and often to determine which
  paths/endpoints will support approval based on
  pre-clinical, Phase I and Phase II data
• Accurately define PK/PD modeling
• Find the right dose in Phase II
• Use drug development experts to design trials not
  academicians
• Select the right patient population and endpoints
• Not the largest or highest revenue model
  population