Statistical Considerations on NDA 21-200

1
Statistical Considerationson NDA 21-200

• Sonia Castillo, Ph.D.
• Division of Biometrics 2
• June 26, 2000

2
Outline

• Background
• Efficacy Analyses by Gender
• Abdominal Pain as an Efficacy Endpoint
• Laxative Use
• Pooling of Studies
• Summary

3
Background on Clinical Studies
4
Background on Clinical Studies

• Study 351 was completed and analyzed first
• Protocol specified efficacy results were not
  significant
• Definition of a responder was modified
• Post-hoc analysis of data gave significant
  results
• Led to protocol amendments for studies 301 and
  307
• There are two studies in which the analysis
  presented is prospective and one in which it is
  post-hoc

5
Amendments to Protocols

• Original protocol for all three studies
• Subject Global Assessment of abdominal
  discomfort/pain was the primary efficacy variable
• Enrollment of 591 intent-to-treat patients

6
Subject Global Assessment of Abdominal
Discomfort/Pain
Patients placed a vertical mark on the line in
response to the following question "How much of
a problem was your abdominal discomfort/pain
over the last week?
7
Definition of Responder forSGA of Abdominal
Discomfort/Pain

• A responder was a patient who had a ? 20 mm and
  ? 40 reduction in the mean VAS at study
  endpoint (the last 4 weeks on treatment)
  compared to the baseline value

8
Amendments to Protocols

• An amendment was submitted prior to start of all
  studies
• Subject Global Assessment of relief was added as
  a second primary efficacy variable
• Enrollment of 693 intent-to-treat patients

9
Subject Global Assessment of Relief
Patients responded on a weekly basis to the
following question "Please consider how you
felt this past week in regard to your IBS, in
particular your overall well-being, and symptoms
of abdominal discomfort, pain and altered bowel
habit. Compared to the way you usually felt
before entering the study, how would you rate
your relief of symptoms during the past week?"
Completely relieved Considerably
relieved Somewhat relieved Unchanged Worse
10
Definition of Responder for SGA of Relief

• A responder was a patient who had Complete or
  considerable relief at least 50 of the time at
  study endpoint (the last 4 weeks on treatment)
• Definition of responder took into account the
  following
• Number of days with laxative use during treatment
  period is ?5 and no laxative use during the last
  28 days of treatment
• Duration of exposure to study medication is ?28
  days
• At least one post-baseline SGA of relief

11
Amendments to Protocols

• After post-hoc analysis of study 351, another
  amendment was submitted before breaking the blind
  in studies 301 and 307
• Definition of responder for SGA of relief was
  modified Complete or considerable relief at
  least 50 of the time at study endpoint
  OR Complete or considerable or somewhat relief
  100 of the time at study endpoint
• SGA of relief became the only primary efficacy
  variable
• SGA of abdominal discomfort/pain was changed from
  a primary to a secondary efficacy variable

12
Subject Global Assessment of Relief
13
Subject Global Assessment of Abdominal
Discomfort/Pain
14
Subject Global Assessment of Relief by Gender
15
Gender Conclusions

• Efficacy results are mixed in female patients
• Efficacy is not demonstrated in male patient
• The lack of efficacy in male patients may be due
  to the small sample size
• The results for female and male patients may
  indicate a difference in the pathophysiology of
  C-IBS between genders

16
Importance of Abdominal Pain inEvaluating
Efficacy

• Changing SGA of abdominal discomfort/pain from a
  primary to a secondary efficacy variable is of
  clinical concern
• Pain is an important clinical component of IBS
• Treatment difference for SGA of abdominal
  discomfort/pain was not significant in any of the
  3 studies

17
Not Accounting for All Laxative Use Affects the
Efficacy Assessment

• Laxative and bulking agent use affects bowel
  habit and abdominal discomfort
• Laxative and bulking agent use need to be
  accounted for in the definition of a responder
  for all analyses of SGA of relief and SGA of
  abdominal discomfort/pain
• Data on the amount of bulking agent taken was not
  collected

18
Reasons Why Pooling is not Appropriate to
Demonstrate Efficacy

• Pooled analysis not pre-specified in the protocol
• Design of study 307 is different from the other
  studies
• Different endpoint interpretations
• Study 307 tests a fixed dose group and a
  titration regimen group
• Studies 301 and 351 test two fixed dose groups
• 4-12 mg titration group is not the same as 12 mg
  group
• Up-titration was based on original definition of
  responder

19
Reasons Why Pooling is not Appropriate to
Demonstrate Efficacy (cont.)

• Pooling the 3 studies into one large study is
  inadequate
• Gives a larger study population and a smaller
  p-value
• Each study is adequately sized to show efficacy
  individually Intent-to-treat sample size at
  study end was larger than planned 27 larger in
  study 301 (881 vs 693) 20 larger in study 307
  (835 vs 693)
• We are left with one study and no replication of
  results

20
Reasons Why Pooling is not Appropriate to
Demonstrate Efficacy (cont.)

• Pooling 1 negative study with 2 positive studies
  is problematic
• Analysis is data driven and biased ?
  uninterpretable p-value
• Negative results of study 307 appear acceptable
  in the light of a positive overview

21
Summary

• Efficacy was demonstrated in female patients in
  study 301 and supported in a post-hoc analysis of
  study 351 but not replicated in study 307
• Efficacy was not demonstrated for either dose in
  study 307. Why?
• Efficacy has not been adequately assessed in male
  patients
• The effect of treatment on abdominal pain has not
  been adequately assessed and demonstrated
• The effect of laxative and bulking agent use
  remains to be determined
• Pooled study results are not appropriate to
  demonstrate efficacy