Critical Issues for Bipolar Disorder

1
Psychopharmocological Drugs Advisory Committee

• Critical Issues for Bipolar Disorder

Gary S. Sachs, M.D. Harvard Medical
School Massachusetts General Hospital
October 25, 2005
2
Treatments for Chronic Mental IllnessPrinciples

• Approval standards should reflect the interest of
  patients
• Research design should be informed by clinical
  epidemiology
• The best design methodology is that which
  optimizes validity and feasibility

3
Bipolar Disorder Untreated vs TreatedStandardize
d Mortality Ratios
29.2
Adapted from Angst, 2000
Zurich Cohort, n406 1959-1997
plt 0.001 plt 0.05
Untreated
Treated
6.4
2.2
1.6
2.0
2.2
1.4
1.7
2.0
1.6
1.3
1.3
1.3
0.6
All Causes
Neoplasm
Accidents
Suicide
Cerebro- vascular
Cardio- vascular
Other
4
Should approval for an acute indication include a
requirement for demonstration of long-term
efficacy?

• The Public Health interest is not served by
  establishing this requirement
• A policy intended to protect patients against
  acutely efficacious treatments without proven
  long-term benefit would benefit few, while
  depriving many of potentially helpful treatments

5
Is there a need to protect patients from
treatments only proven to have short-term
efficacy?

• Patients remain on ineffective medications
  briefly
• Effective medications are frequently discontinued
  over relatively short time periods
• Most patients using medications long-term are
  those who responded acutely and either perceive
  continued benefit or have suffered recurrence
  when attempting to taper.

Few get long-term treatment in the real world
Lack of efficacy
Unable to tolerate
Continues long-term treatment
Based on Altshuler et al. AJP. 2003
6
Lithium Use and Discontinuation in a Health
Maintenance Organization
Adapted from Johnson, Am J Psychi 1996
Lithium under used - only 8 used Li for 90 of
days enrolled - median continuous use 76 days
0
500
1000
2000
1500
Days
Length of first continuous period of Lithium use
(n1594)
7
Principles of Current Clinical Practice

• Bipolar disorder is a life long condition in
  which acute episodes are separated by period of
  variable remission
• The episodic nature of bipolar disorder provides
  the organizing principle of treatment
• Episodes are the target of acute and prophylactic
  treatment
• Multiphase Treatment model
• Acute, Continuation, Maintenance/Discontinuation
• Maintenance phase treatment reflects prior acute
  phase experience

8
Multiphase Treatment Strategy
Recovering
Recovered
Euthymic
Natural Course
Treated Course
Depressed
9
Basic Treatment paradigm is iterative

• Critical
• decision point

Evidence A-F
Individual factors
Intervention
10
Each treatment trial is evaluated for

• Benefit Adverse Effect Cost

Iterative Process Individual Response Directs
Treatment
Benefit
adverse effects expense
11
Maintenance Therapies in Bipolar Disorder Study
Design
Weekly Acute treatment
Preventative tx bi-weekly x 12 weeks then
monthly x 2 years
Patient and family attend psychoeducational
workshop
IPSRT protocol pharmacotherapy
IPSRT protocol pharmacotherapy
Recovery
ICM protocol pharmacotherapy
Acute episode
IPSRT protocol pharmacotherapy
ICM protocol pharmacotherapy
Recovery
IPSRT protocol pharmacotherapy

• 4 wks average
• HRSD lt7
• Bech-Rafaelsen lt7

12

p .03
Frank et al., Journal of Abnormal Psychology,
in press
13
How well do our current treatments work?

• Acute Bipolar depression
• No single agent has FDA approval for treatment of
  BP depression
• STEP-BD and Altshuler/StanleyFBN data indicates
  need for more efficacious treatments.
• Acute mania (8 approved agents)
• The data from 3-4 week double blind trials
  indicates that on average participants receiving
  active treatment finish the study with symptom
  severity scale scores above the severity score
  required for study entry.
• Prophylaxis (4 approved agents)
• None consistently efficacious for all relevant
  outcome variables
• None shown effective for subjects remaining well
  longer than 6 months

14
Acute Bipolar DepressionPlacebo controlled
therapy Trialswith adequate sample size
combination
monotherapy
Lamotrigine Olanzapine Quetiapine
Olanzapine Fluox.

• Positive

Li Paroxetine Li Imipramine
Negative or Failed
Imipramine
power to detect a difference gt 0.8
15
Placebo-Controlled Bipolar Depression
Studies with Adequate Samples
Tohen et al.
Calabrese et al
Calabrese et al.
MADRS Change From Baseline
plt0.05 vs. PBO
OFC-olanzapine fluoxetine combination. Montgomery
Asberg Depression Rating Scale. Calabrese et al.
J Clin Psychiatry. 1999. Tohen et al. APA. 2002.
Calabrese et al APA 2004
16
A rough metric of clinical effectiveness CE
Response Rate x Completion Rate
Study
Response rate
NNT
Completion rate
CE
26.7 36.2 20.6
16.4 6.4
Calabrese 1999
OlZ 39.0 OFC 56.1 Placebo
30.4
Tohen 2003
48.4 64.0 38.5
18.9 35.9 11.7
13.9 4.1
Calabrese 2005
QTP 300 58.2 QTP 600 57.6 Placebo 36.1
66.9 54.4 59.1
38.9 31.3 21.3
5.7 10.0
17
Real World Pharmaco-epidemologyNew treatment
starts at MGH Bipolar Clinic
N 466 with at least 4 visits in year
Median tx duration (in days)
Recovered/Recovering
Majority use lt 6 months
18
Clinical intent to treat Bipolar Depression
Results from Stanley Foundation Bipolar Network
Acute Phase
Maintenance Phase
N1,078
1.0
Stanley Bipolar Network
0.8
549 received AD
0.6
0.4
186 AD gt60 days
84
0.2
Medication continuation group Medication
discontinuation group
15 remitted
0.0
0
8
16
24
32
40
48
Altshuler et al. AJP. 2003
Number of Weeks Until Relapse
19
What is the benefit of standard antidepressants?
Enriched sample of 84 Remitters 15 of total
25 relapse lt 4 months on or off antidepressant
41 relapse 12 months on antidepressant
71 relapse lt 4 months off antidepressant
Remaining Well gt 1 year
15
11
9
4
Discontinued using Antidepressant lt6 months
Ever
4 months
12m -AD
12 months using Antidepressant
Adapted from Altshuler et al. AJP. 2003.
20

STEP-BD Naturalistic data also indicates great
need for better acute treatments
Acute Phase
Maintenance Phase
15
Days well after reaching Recovered Status(8
Wks Well)
2000 BP Subjects
377 Intent to treat Acute depression
10
Median 96.5
Percent
Majority have remission lt 6 months
Outcome at 90 Days recovered
5
With AD 21.5 No Ad 27.2
0
0
40
80
120
160
200
240
280
320
360
Days Well
21
Psychiatric patients need more acute treatments

• The proportion of patients achieving an acute
  benefit is modest
• There is a compelling need for more acutely
  efficacious medications
• Raising the bar to require long-term efficacy is
  not yet practical
• Limiting patient choice to only new treatments
  with proven long-term efficacy will leave
  potential responders untreated or seeking options
  unavailable here

22
Less than 10 will benefit

• The proportion meeting criteria for response or
  remission in most pivotal trial overstates the
  benefit of acute treatment
• The proportion depressed meeting DSM-IV criteria
  for recovered is modest (lt20)
• Among those achieving recovery, gt 50 experience
  a recurrence within 3 months
• The vast majority will be negatively impacted by
  virtue of not having access to acutely beneficial
  treatments

23
Is the Public Health interest well served by
requiring demonstration of long term efficacy at
time of registration?

• No compelling benefit
• The proportion of patients achieving an acute
  benefit is modest
• There is no surplus of acutely efficacious
  medications
• Patients usually stop ineffective medications
  quickly
• Even effective medications are discontinued over
  relatively short time periods
• New obstacles to approval of treatments with
  acute efficacy will harm patients

24
Psychopharmocological Drugs Advisory Committee

• Critical Design Issues in Conducting Long-Term
  Trials in Bipolar Disorder

Gary S. Sachs, M.D. Harvard Medical
School Massachusetts General Hospital
October 25, 2005
25
Relapse Study Design Stabilization
PeriodDesign Should Reflect Validity
Feasibility

• DSM-IV defines recovery from a mood episode as
  8 weeks well
• New proposal calls for 6 months for all chronic
  conditions
• EU guidelines are condition dependent 2 6
  months
• Data suggests one duration does not fit all
  conditions
• The design should reflect clinical epidemiology
  of each condition
• Successful studies use enriched designs
• Study designed to enrich with atypical patients
  can be misleading

26
Stabilization Period EU GuidelinesFlexibility
Depending on Condition
27
STEP BD Data Time to Relapse and Roughening
after Recovery
( 8 weeks stable remission)
1.00
Steady gradual slope
0.75
Full Episode
0.50
Survival Distribution Function
0.25
Roughening
0.00
0
50
100
150
200
250
300
350
400
Time to Event/Censoring
28
Placebo vs Olanzapine in Combination with
Lithium or Valproate
Randomized after symptomatic remission of mania
(YMRS ?12) and depression Ham-D ? 8
Remainingin Remission
Immediate steep slope
(Days)
.
29
Placebo vs Olanzapine in Combination with
Lithium or Valproate
Time to recurrence of mania after symptomatic
remission of mania (YMRS ?12)
100
80
60
Probability of Remainingin Remission ()
P0.005
40
Immediate steep slope
20
0
0
100
200
300
400
500
Time to Recurrence (Days)
Tohen et al. Poster presented at XXIII Congress
of CINP 2002 Montreal, Quebec.
30
2nd Lamotrigine Maintenance Study
Randomized subjects well on monotherapy 1 week
1
PBO (n119)
LTG v. PBO, p 0.029 Li v. PBO, p 0.029 LTG v.
Li, p 0.915
LTG200/400 (n165)
0.9
Li (n120)
0.8
0.7
0.6
0.5
Survival Estimate
0.4
0.3
Immediate steep slope
0.2
0.1
0
0
10
20
30
40
50
60
70
Week
31
Relapse Following Open, Prospective Non-Random
Antidepressant Discontinuation vs. Continuation
Enriched Sample of 84 Remitters ? 6 weeks CGI-S ?
2
1.0
53 BPI 24 BPII
0.8
Steady gradual slope
0.6
0.4
0.2
Medication continuation group Medication
discontinuation group
0.0
0
8
16
24
32
40
48
Relapse CGI-S ? 4 Altshuler et al. AJP. 2003
Number of Weeks Until Relapse
32
Aripiprazole Maintenance of Stability in Bipolar
Mania
Required 6 Weeks Well for Randomization
1.0
Aripiprazole Placebo
0.9
0.8
0.7
Proportion of Patients Without Relapse
Steady gradual slope
0.6
0.5
Log-rank P value 0.020
0.4
0
14
25
42
55
70
84
95
112
125
140
154
168
182
195
210
Days
Relative risk 0.523 (0.30, 0.913). Data on
file, Otsuka America Pharmaceutical, Inc.
33
Successful studies use enriched designs
Proportion Randomized
Acute Phase Responders

Total Entering Study
Lithium
Enrichment
Most
Least
34
Inherent Maintenance Design Problem

• Conservative Assumptions Great Obstacles to
  Feasibility

• Eligibility Criteria for Randomization
  Responder
• True Responder /All meeting response criteria
  50
• Assumes Response for Active 50 Placebo 25
• Enriched Sample is a composite of
• 50 True Responders
• 50 Pseudo Responders

In the Double blind Phase Pseudo Responder
Relapse Rates Placebo Active
35
Inherent Maintenance Design Problem

• Study Powered for relapse rate
• Placebo 60 vs Active 40

Enriched Sample N200 50 True
Responders 50 Pseudo Responders
Active Cell n100
Placebo Cell n100 50 True Responders 50
Pseudo Responders
50 True Responders
50 Pseudo Responders
Subjects relapsing
60
30
10
Positive result requires active agent sustains
effectiveness in at least 80
36
Inherent Maintenance Design Problem

• Longer stabilization phase
• Study Powered for relapse rate
• Placebo 40 vs Active 20

Enriched Sample N200 50 True
Responders 50 Pseudo Responders
Active Cell n100
Placebo Cell n100 50 True Responders 50
Pseudo Responders
50 True Responders
50 Pseudo Responders
Subjects relapsing
40
20
0
Positive result requires active agent sustains
effectiveness in 100
37

produces a sample unrepresentative of patients
seeking treatment
Use of a 6 month stabilization
15
Days well after reaching Recovered Status(8
Wks Well)
Summary Statistics Number of Obs 281 N
Missing 19 N 262 Median 96.5 Mean 116.57 Std
Deviation 83.21 Mode 63 Maximum 365 Minimum 3.5
10
Percent
Majority have remission lt 6 months
5
0
0
40
80
120
160
200
240
280
320
360
Days Well
38
Limiting randomization to subjects 6 months
will alter study results
Antidepressant Discontinuation vs.
Continuation Conditional on 6 months well
Narrows the separation
Medication continuation group Medication
discontinuation group
Number of Weeks Until Relapse
39
Consequence of 6 Month Stabilization
Conditional on remaining well for 6 months
Original curves after 6 months
Frank et al., Journal of Abnormal Psychology
40
Trial Design Summary of Issues

• Validity
• One size does not fit all conditions
• Data shows what got you well keeps you well
• 8 wks well is adequate to achieve stabilization
• 6 month stabilization produces sample
  unrepresentative of patients seeking treatment
• 6 month stabilization will make it difficult to
  detect a true difference between a NME and placebo

41
Trial Design Summary of Issues

• Feasibility
• 6 month stabilization will require a larger
  sample
• Larger samples will increase enrollment phase
  8-24 month heightening rater drift and sample
  variability
• An increase in enrollment times will delay or
  prevent patient access to innovative new
  treatments
• This proposal is not in the best interest of
  patients