Voriconazole NDAs 21266 and 21267 Empiric Antifungal Therapy of Febrile Neutropenic Patients Study 6

1
VoriconazoleNDAs 21-266 and 21-267Empiric
Antifungal Therapy of Febrile Neutropenic
PatientsStudy 603

• John H. Powers, M.D.
• Medical Officer
• Division of Special Pathogen and Immunologic Drug
  Products
• Center for Drug Evaluation and Research

2
Introduction

• Scientific and regulatory background in
  indication of Empiric Antifungal Therapy in
  Febrile Neutropenic Patients (ETFN)
• Discussion of primary composite endpoint and
  statistical definition of non-inferiority
• Discussion of selected issues with secondary
  endpoints
• breakthrough infections within 7 days of EOT
• survival at 7 days after end of therapy
• discontinuations due to lack of efficacy or
  toxicity
• defervescence prior to recovery from neutropenia
• baseline fungal infections
• Summary

3
Background fungal infections in neutropenic
hosts

• Neutropenia one of major risk factors for
  invasive fungal infections
• risk varies with depth and duration of
  neutropenia
• Candida and Aspergillus species most common
• Autopsy studies show 12 to 43 incidence of
  invasive fungal infections
• Difficulty in pre-mortem diagnosis with occult
  infections may occur in as many as 50
• High mortality of 48 to 80 in various studies

4
Backgroundbasis for treatment

• Idea of empiric therapy
• treat occult infections which would not be
  diagnosed by conventional means
• prevent infections in patients at risk
• Two randomized trials address question
• Pizzo et al. Am J Med 198272101-110
• EORTC trial Am J Med 198986668-672
• Empiric therapy has become standard of practice

5
Backgroundprior discussion of study design

• Two public meetings in 1994 and 1995 to discuss
  issues of study design in antifungal drugs
• non-inferiority study design recommended
• amphotericin B deoxycholate (AMB-D) suggested as
  comparator
• not FDA approved for this indication but approval
  was in 1956
• for approval need at least one study showing
  efficacy in another fungal indication plus one
  study in ETFN
• prove efficacy in documented infections

6
Backgroundimportant points from workshops

• Points from 1994 and 1995 workshops
• certainty of efficacy since resolution of fever
  rather than proven infection is determinant of
  sample size
• lower bound of 95 CI to determine
  non-inferiority suggested as -10
• composite endpoint recommended as trials with
  primary endpoint of BT infections may have
  prohibitively large sample size
• considered important to detect
• differences in proven fungal infections
• differences in mortality
• differences in fever within a 10 CI
• differences in safety

7
Background

• Dr. Alan Sugar at 4/14/97 Advisory Committee
• Empiric therapy is given to patients because
  some will actually need it. However, others will
  not, that is, they are treated unnecessarily. At
  issue is the diagnosis of invasive fungal
  infections and the difficulties associated with
  it.Also at issue is the degree of neutropenia.
  What places the patient at higher risk? Overall
  consensus is that patients with an ANC of lt100
  are at highest risk. The duration of neutropenia
  also contributes to risk.

8
Trial Design and Analysis

• Non-inferiority trial with non-inferiority margin
  of -10
• primary analysis population MITT
• Stratified by risk of fungal infections and
  prophylaxis
• Open label study design
• no oral systemically active form of L-AMB
• ethics of two IV infusions in seriously ill pts.
• Data Review Committee blinded assessments of
  fungal infections and outcomes

9
Backgroundprevious approvals in ETFN

• Two agents approved for ETFN using composite
  endpoint similar to Study 603
• Ambisome (liposomal amphotericin B, L-AMB)
• empirical therapy for presumed fungal infections
  in febrile, neutropenic patients
• Sporanox (itraconazole injection, oral solution)
• empiric therapy of febrile neutropenic patients
  with suspected fungal infections.
• overall response greater for itraconazole than
  AMB-D but more D/C due to lack of efficacy with
  itraconazole, more D/C due to toxicity with AMB-D

10
Overall Response Rate in Study 603

• Primary analysis was stratified overall response
  rate of composite (5 component) endpoint
• Lower bound of CI to select sample size and
  define statistical non-inferiority defined as
  -10 based on 50 overall response rate in MSG 32
  study with Ambisome vs. AMB-D
• Stratified response rates in Study 603
• voriconazole 23.7
• L-AMB 30.1
• difference -6.1 (-12.0, -0.1)
• FDA weighted 95 CI (-11.6, 0.1)
• Note lower bound of 95 CI falls below -10

11
Overall Response Rate in Study 603

• Statistical points about defining non-inferiority
• Lower bound of 95 CI below (more negative)
  pre-specified limit implies that drug may not be
  non-inferior to control drug
• Defining non-inferiority implies knowledge of
  efficacy of control drug over placebo/no
  treatment in superiority trial
• Lower bound of 95 CI used to define
  non-inferiority margin in non-inferiority trial
  cannot be greater than difference in control drug
  over placebo/no treatment

12
Overall Response Rate in Study 603

• Hypothetical example
• lower bound of 95 CI in a randomized, placebo
  (or no treatment) controlled trial shows an
  absolute benefit of control drug over no
  treatment of at least 7
• in subsequent non-inferiority trial, test drug
  with lower bound of 95 CI of -11 compared to
  control is potentially no better than placebo

13
Overall Response Rate in Study 603

• Clear that neutropenic patients develop invasive
  fungal infections
• At issue in selecting non-inferiority margin in
  ETFN
• Do antifungal drugs prevent breakthrough
  infections in neutropenic patients?
• If so, what is the magnitude of this benefit
  relative to no treatment?
• How does this impact the selection of a
  non-inferiority margin in clinical trials using a
  composite endpoint rather than BT infections as
  primary endpoint?

14
Previous Trials in ETFN

• Two previous trials in ETFN comparing AMB-D to no
  treatment
• not adequately powered to determine difference in
  breakthrough infections
• EORTC trial used resolution of fever as primary
  endpoint
• included mucosal as well as invasive disease in
  breakthrough infections
• did not include deaths and discontinuations as
  failures in analysis of BT infections

15
Previous Trials in ETFN

• True difference of AMB vs. no treatment in
  prevention of BT infections may be from 60
  better to 8 worse than no treatment
• How does this impact studies using the current
  composite endpoint?
• What is clinical relevance of non-inferiority
  margin of -10 in ETFN
• Can we extrapolate from studies 20 years ago to
  current rate of emergent fungal infections?

16
Secondary Endpoints and Subset Analyses

• When primary endpoint not met, can attempt to
  explain failure by looking at secondary and
  subset analyses
• secondary and subset analyses considered
  hypothesis generating in this situation
• Subset analyses of patients stratified according
  to risk of fungal infections and/or receipt of
  prophylaxis
• Secondary endpoints are 5 individual components
  of composite endpoint

17
Secondary Endpoints and Subset Analyses

• In attempt to adjust for multiple comparisons for
  the 5 secondary endpoints, 99 confidence
  intervals are presented
• Not primary endpoint of trial so study not
  adequately powered to determine true difference
  in secondary endpoints and subsets
• Secondary and subset analyses may have small
  numbers of patients in each group
• Look for consistency in outcomes of secondary
  endpoints as sensitivity analysis

18
Overall Response Rate in Study 603subset
analysis by risk of fungal infection

allogeneic BMTs and relapsed leukemia patients
19
Secondary Endpoints
all subsequent 99 CI based on these rates, so
that focus is on lower bound
20
Breakthrough Infections

• Fewer BT infections in voriconazole arm compared
  to L-AMB
• voriconazole 1.9 (8/415)
• L-AMB 5.0 (21/422)
• Difference of 3.1 in favor of voriconazole with
  99 CI (-0.37, 6.5)
• difference greatest in Aspergillus BT infections
  with 4 in voriconazole vs. 13 in L-AMB
• BT infections included proven and probable
  disease
• 6/8 patients proven disease in voriconazole arm
• 20/21 patients proven disease in L-AMB arm

21
Breakthrough Infectionssensitivity analysis

• Sensitivity analysis considering BT infections as
  primary endpoint may consider patients who die as
  failures
• voriconazole 9.2 (38/415)
• L-AMB 9.2 (39/422)
• difference of 0 with 95 CI (- 4.2, 4.2)

22
Breakthrough Infections subset analysis by risk
of infection and prophylaxis
23
Survival at 7 days after EOT

• More deaths in voriconazole arm vs. L-AMB
• voriconazole 8.0 (33/415)
• L-AMB 5.9 (25/422)
• difference of 2.1 in favor of L-AMB with 99 CI
  (-6.9, 2.7)
• difficulty of attribution of death in seriously
  ill patients
• attribution of death not blinded or reviewed by
  DRC

24
Discontinuations

• More discontinuations due to lack of efficacy or
  toxicity in voriconazole arm
• voriconazole 9.9 (41/415)
• L-AMB 6.6 (28/422)
• difference of 3.3 in favor of L-AMB with 99 CI
  (-8.4, 1.8)
• Composite endpoint combines discontinuations due
  to lack of efficacy with those due to toxicity
• combines efficacy and safety endpoint
• can obscure important differences in outcomes
• Look at various reasons for discontinuations

25
Discontinuationslack of efficacy and toxicity

• More discontinuations due to lack of efficacy in
  voriconazole arm compared to L-AMB
• voriconazole 5.3 (22/415)
• L-AMB 1.2 (5/422)
• difference 4.1 in favor of L-AMB
• Fewer discontinuations due to toxicity in
  voriconazole arm compared to L-AMB
• voriconazole 4.6 (19/415)
• L-AMB 5.5 (23/422)
• difference 0.9 in favor of voriconazole
• open label trial and potential for bias

26
Discontinuationslack of efficacy

• More discontinuations with persistent fever as
  reason for lack of efficacy in voriconazole arm
• voriconazole n 14 out of 22
• L-AMB n 2 out of 5
• Failure to become afebrile in neutropenic
  patients may indicate presence of occult fungal
  infections

27
Defervescence Prior to RFN

• Fewer patients in voriconazole arm meet
  definition of defervescence prior to recovery
  from neutropenia
• voriconazole 32.5 (135/415)
• L-AMB 36.5 (154/422)
• difference 4.0 in favor of L-AMB with 99 CI
  (-12.7, 4.7)
• results highly dependent on definition of
  defervescence
• previous trials required defervescence at any
  time prior to RFN
• Study 603 required 48 hours without fever prior
  to RFN

28
Defervescence Prior to RFNrates of fever
resolution prior to recovery from neutropenia in
ETFN trials
29
Defervescence Prior to RFN

• Sensitivity analysis for overall response
  changing definition of defervescence in Study 603
• defervescence within 24 hours prior to RFN
• voriconazole 35.9
• L-AMB 40.8
• difference -4.9 with 95 CI (-11.7, 1.9)
• defervescence any time prior to RFN
• voriconazole 50.1
• L-AMB 56.2
• difference -6.1 with 95 CI (-13.1, 0.9)
• results of sensitivity analyses are supportive of
  primary analysis

30
Baseline Infections
31
Baseline Infections

• Small numbers of patients with BL infections
  expected given exclusion criteria
• Difficult to compare efficacy in treatment of BL
  infections with small numbers
• Two of 6 voriconazole successes developed
  invasive Candida infections 2-4 months after EOT

32
Summary

• Drugs in ETFN should have proven efficacy vs.
  documented Candida and Aspergillus infections
• Aspergillosis global study
• Candida esophagitis data and ongoing candidemia
  study
• Since therapy is empiric, patients in ETFN may
  receive treatment without having fungal
  infections
• safety profile of drug

33
Summary

• Voriconazole fails to meet definition of
  statistical non-inferiority in Study 603
• Subset analysis of overall composite endpoint
  showed numerical advantage of voriconazole in
  high risk patients but advantage of L-AMB in
  moderate risk patients
• Secondary analyses of BT infections showed
  numerical advantage of voriconazole especially in
  prevention of Aspergillus infections
• sensitivity analysis including deaths as failures
  showed no difference between drugs
• Secondary analysis of other 4 components of
  composite endpoint in favor of L-AMB

34
Considerations for the Committee

• Clinical relevance of non-inferiority margin of
  -10 in ETFN trials with composite endpoint
• Secondary analyses in study which did not meet
  primary endpoint
• Need to consider safety of drug in empiric
  therapy where patients may receive drug who do
  not have infections