Developing Chronic Disease Therapies Presentation to the FDA Science Board

1
Developing Chronic Disease TherapiesPresentation
to the FDA Science Board

• Robert M Califf MD
• Director Duke Clinical Research Institute
• Details in Health Affairs 2377-88 2004

2
Key Points

• Chronic diseases will be (are) the dominant
  health issue in our society
• Predicting whether a chronic disease therapy
  causes net benefit or harm is difficult
• Biomarkers and imaging are the way to go
• But mostly as a screen
• Cannot substitute for clinical outcomes in most
  circumstances
• Tremendous inefficiency in clinical trials

3
Life Expectancy
Years
Years
4
Global Burden of Disease Top 10
5
Age-Standardized Distribution of Disability
Manton KG et al. Proc Natl Acad Sci 2001986354-9
6
Total Number of Persons Age 65 or Older in the
United States 1900-2050
US Census Bureau, Decennial Census Data and
Population Projections
80
60
Millions of people
40
Age 65 or older
20
0
1900
1950
2000
2050

Year
projected
7
Medicare Growth 20022030

• Medicare enrollment
• 2002 40 million
• 2030 77 million
• N Engl J Med 200134492831
• Worker beneficiary ratio
• 2002 4.0 1
• 2030 2.3 1
• 2070 2.0 1
• www.whitehouse.gov

8
Annual Report to Congress, the Medicare Board of
Trustees

• The financial status of the fund has
  deteriorated significantly, with asset exhaustion
  projected to occur in 2019 under current law
  compared to 2026 in last year's report.
• Medicare will grow much faster than the economy
  as a whole, increasing from 2.6 percent of the
  gross domestic product last year to 3.7 percent
  in 2010, 7.7 percent in 2035
• Projected Medicare costs would exceed those for
  Social Security in 2024

-- March 23, 2004
9
Drivers of Healthcare Consumerism
10
(No Transcript)
11
The Cycle of Clinical TherapeuticsNew Model
Concept
Clinical Trials
Guidelines
Education and Feedback
Outcomes
Performance Indicators
Performance
12
Principles for Trials Designed to Define Balance
of Risk and Benefit of Therapy

• Treatment effects usually modest
• Qualitative interactions are uncommon
• Quantitative interactions common
• Unintended targets common
• Long-term vs. short-term effects may differ
• Combinations are unpredictable
• Class effect may not be valid
• Most treatments produce a mixture of benefits and
  risks
• Califf and DeMets Circ 2002

13
Principles of Trials Designed to Elucidate the
Balance of Risk and Benefit of Therapies

• Most beneficial therapies do not save money, but
  create incremental benefit for incremental cost
• Applying the results of clinical trials is
  beneficial
• Some areas of CV medicine are underserved and
  therefore lack evidence to guide practice
• Our current method of doing trials is
  unnecessarily expensive
• Participation is critical
• Califf and DeMets Circ 2002

14
Sample Size
Pts Randomized Chance of CommentsDeaths (Ris
k 10) Type II Error on Sample Size 0-50 500 0.9 Utterly inadequate 50-150 1000 0.7-0.9 P
robably inadequate 150-350 3000 0.3-0.7 Possibly
inadequate 350-650 6000 0.1-0.3 Probably
adequate 650 10000 Probability of failing to achieve p risk reduction 25
Yusuf Progr CV Dis 1985
9503CG01
15
CAST
Placebo (n 743)
Encainideor Flecainide(n 755)
Patients without Event ()
p 0.0004
Days after Randomization
Echt, N Engl J Med, 1991
16
Unintended Targets
17
Unintended Targets
18
Time
19
Time
Disease
True ClinicalOutcome
A
SurrogateEndpoint
Intervention
True ClinicalOutcome Flemming and Demets
Disease
B
SurrogateEndpoint
20
Time
Intervention
Disease
True ClinicalOutcome
C
SurrogateEndpoint
Intervention
True ClinicalOutcome Flemming and Demets
Disease
D
SurrogateEndpoint
21
Therapies Always Cause a Combination of

• Good Effects

• Bad Effects

Adapted from Furberg
22
Troponin and Risk Stratification in
ACSGUSTO-IIa and TIMI-IIIb
TnT -
p TnT
Mortality Rates TnT Pos 14 TnT Neg 5
p Troponin I (ng/ml)
Antman EM, et al. NEJM 1996
Ohman EM, et al. NEJM 1996
23
Combined All-cause Mortality and Morbidity
n5010
Event-free probability
100
95
90
13.3 risk reduction
85
80
75
Valsartan
Placebo
70
P0.009
65
0
0
3
6
9
12
15
18
21
24
27
Months
Cohn, et al. N Engl J Med. 2001.
24
Subgroup Results
Mortality and Morbidity
Subgroup pts Overall 100 Beta blocker
(yes) 35 no beta blocker 65 ACEI (yes) 93 no
ACEI 7
.87
.77
.97
RR of death P no ACEI ? 41 blocker ? 42 0.009



44.0 ?, P.0002

n366
1.25
1.0
0.75
0.50
Valsartan worse
Valsartan better
Cohn. N Engl J Med. 2001 FDA analysis/package
insert
25
Valsartan Approved for CHF(But with 3 subgroup
provisos)

• Indications and Usage
• Diovan is indicated for the treatment of heart
  failure (NYHA class II-IV) in patients who are
  intolerant of ACEIs. In a controlled clinical
  trial, Diovan significantly reduced
  hospitalizations for heart failure. There is no
  evidence that Diovan provides added benefits when
  it is used with an adequate dose of an ACEI.
• Concomitant use with an ACEI and a beta blocker
  is not recommended.

26

• Subgroup analysis a machine for generating false
  negatives

-- Richard Peto
ISIS-2 ASA Placebo RR P Gemini or
Libra 11.1 10.2 1.09 NS Others 9.0 12.1 0.72 0.00001
Lancet. 1988
27
Small Trials (low number of deaths) are Unreliable
28
Mortality 1 year-GUSTO 4
22,4
25
18,3
20
15,7
12,8
15

9,3
6,9
7
7,9
10
NT-proBNP, quartiles ng/L
4,1
3,8
4,1
5
1869
3,5
2,7
669- 1869
2,7
1,2
Troponin T, quartiles ug/l
237- 669
0
1,6
0.47

0.12 - 0.47
0.01 - 0.12

29
Mortality 1 year-GUSTO 4
23,4
25
17,2
20
16,7
15
15

9
8,1
10
NT-proBNP, quartiles ng/L
6,4
7,1
4,8
4,9
5
3,6
1869
2,8
2,7
669- 1869
2,7
CRP, quartiles mg/L
0,5
237- 669
0
1,6
9.62

3.96- 9.62
1.84- 3.96

30
Mortality 1 year-GUSTO 4
30
25,7
25
16,4
20
12,3
12,2

15
8,2
8
NT-proBNP, quartiles ng/L
8,2
6,9
10
5,2
6,8
3,1
1869
5
2,1
2,9
669- 1869
1,8
1,6
237- 669
0
Creatinine clearance, ml/min
0,3


51-66
66-84
84
31
Mortality 1 year
17,8
18
16
14,1
13,8
14
11,6
10,2
12
8,8
8,9
10

7,7
7,7
8
5,1
Troponin T, kvartiler ug/l
6,1
5,6
6
5,8
5,1
4
3,2
0.47
2
0.12 - 0.47
1,7
0
0.01 - 0.12
9.62

3.96- 9.62
CRP, quartiles mg/L
1.84- 3.96

32
Information systems can now draw meaningful
statistical inferences pertinent to each
individual from massive data sets that include
genomic data, imaging results, and biomarker
analyses along with traditional clinical
variables. Such evidence, made available to
clinicians working at the point of care, can
direct the most appropriate preventative and
therapeutic actions.
RS Williams, Science, April 2003
33
Predicting Clinical Benefit and Toxicity

• Biomarkers and imagingthe pathway
• If we dont develop effective approaches to
  interoperability of these data we will suffer
• Multivariable nature of biology increases needed
  sample sizes by log orders
• No single entity (corporation or institution) has
  enough cases to validate complex biological
  predictive models
• Allowing the market to sort it out is playing
  Russian roulette with the public health

34
Background

• Estimated 26.4 billion spent annually for RD,
  4.1 billion on Cardiovascular trials
• Hypothesis design of the operational plan
  impacts overall costs, and operational plans are
  put together in a risk averse manner
• Objective assess the importance of trials
  functions to total costs and estimate cost
  savings through changes in trial design and
  management

35
Background - contd

• Methods - two clinical trials case studies were
  prepared
• participants completed a simple 16 question
  survey based on budget drivers, not the science
• applied a ballpark pricing model
• Think-tank meeting including government
  agencies, academicians, and industry

36
(No Transcript)
37
(No Transcript)
38
(No Transcript)
39
Waste in Clinical Trials

• Imagine building bridges without engineering
  schools
• Lack of clinical research as a discipline
• Lack of standards
• Lack of empirical analysis of value of changes

40
Changes In Lipoproteins
Percent Change fromBaseline to Year 1
LDL
HDL
Triglycerides
41
HERS/HERS II and WHI Clinical Outcomes
42
WHI HRT and Quality of Life
Hayes, N Engl J Med 2003
43
Hormone Therapy and the Coagulation System

• Estrogen Progestin
• Factor XII Increases ?
• Factor VIIa Increases Decreases/Neutral
• Fibrinogen Decreases Decreases
• F1.2 Increases Increases
• Fibrinopeptide A Increases ?
• Protein C Variable Variable
• Protein S Variable Variable
• Antithrombin III Decreases Decreases

44
WHI Results Interpretation

• How large are the treatment effects?
• In 10,000 women treated for one year
• CHD 7 more
• Stroke 8 more
• PE/DVT 8 more
• Breast CA 8 more
• Colorectal CA 6 fewer
• Hip fracture 5 fewer
• Cumulative differences will increase with time
• 100 excess events over 5 years
• 60,000 excess events in 6 million users

45
Therapies Always Cause a Combination of

• Good Effects

• Bad Effects

Adapted from Furberg
46
Key Points

• Chronic diseases will be (are) the dominant
  health issue in our society
• Predicting whether a chronic disease therapy
  causes net benefit or harm is difficult
• Biomarkers and imaging are the way to go
• But mostly as a screen
• Cannot substitute for clinical outcomes in most
  circumstances
• Tremendous inefficiency in clinical trials