Comments on FDA Concept Paper

1
Comments on FDA Concept Paper
Premarketing Risk Assessment

• Sidney N. Kahn, MD, PhD
• President
• Pharmacovigilance Risk Management, Inc.
• www.pvrm.com

2
General comments on risk management

• Benefit-risk optimization helps and requires all
  stakeholders
• FDA commendation for forward thinking and
  stakeholder involvement
• Caveats
• danger of disproportionate reaction to very small
  but visible risk
• adoption of untried and untested methods based on
  theoretical considerations of effectiveness and
  misperceptions of risk
• The ultimate goal is to ensure that efforts and
  costs are expended on effective processes that
  achieve a positive benefit/risk balance

3
Premarketing Risk Assessment - omission

• Increase emphasis on data quality
• cf. case follow-up in proposed rule on Safety
  Reporting Requirements (FR, March 14, 2003)
• refer to quality case components defined by
  CIOMS-V

4
Trial size acute use / serious conditions

• Impossible to generalize for unspecified safety
  analyses
• quantitative analysis of prespecified safety
  endpoints with known study population background
  rate
• dependent on patient population, trial design,
  type of comparator
• Recommendation
• continue current ad hoc safety data collection
  for most trials
• better qualitative data on individual important
  adverse events as or more important than
  quantitative information

5
Trial size chronic use

• Most AEs of concern idiosyncratic, rare
• If common, will be detected during development
  and usually lead to program termination
• Low-frequency AEs, even if anticipated, unlikely
  to be detected in any reasonably sized dataset,
  including large simple safety studies (rule of
  3)

6
Subgroup analysis

• Meaningful subgroup differences often difficult
  to demonstrate even for efficacy (single
  prespecified endpoint)
• Proposed analyses unlikely to have power to find
  meaningful differences
• Subgroup effects markedly different from overall
  study population likely to be detected using
  current methods
• Continue to conduct exploratory studies in
  relatively homogenous patient subgroups, e.g.
  elderly, renally impaired

7
Detection of unanticipated interactions

• Catch-22 anticipating the unanticipated!
• Suggestions
• identify existing drug usage patterns in target
  population
• compare PK/PD of investigational agent with
  common / known interacting products
• formal interaction studies of most
  likely/hazardous combinations
• concomitant treatment with less obviously
  interacting products for some subjects in later
  phase III studies
• compare incidences of AEs with clean study
  population
• identify trends for subsequent study and
  monitoring

8
Biomarkers - caveats

• False positive / false negative results
• Many mild abnormalities reflect only biological
  and/or analytical variability
• Mechanisms of common mild abnormalities not
  necessarily the same as those that lead to rare,
  severe outcomes

9
Biomarkers - suggestions

• Consider both absolute and relative (vs.
  individual baseline) values
• Using data from NDAs/BLAs, FDA could
• develop criteria for optimal biomarker
  sensitivity and specificity
• establish rates of different degrees of
  abnormality in treated and control populations
• correlate results with documented clinical
  outcomes in clinical trials and marketed use
• Pharmaceutical companies should allow use of
  proprietary submission data for this purpose.

10
Optimizing AE descriptions for signal detection -
I

• Great potential merit, requires refinement
• Current rigid adherence to verbatim term for
  codification
• investigator misclassification e.g. abnormal
  LFTs for jaundice, acute liver failure
  without encephalopathy, coagulopathy, or jaundice
• inconsistent clinical classification, e.g.
  similar findings reported variously as hepatitis,
  abnormal liver function, elevated
  aminotransferases and jaundice, etc.

11
Optimizing AE descriptions for signal detection -
II

• PvRM strongly supports FDA collaboration with
  sponsors to develop harmonized term definitions
• Ensuring consistency / minimizing subjectivity
  requires uniform approach across all sponsors and
  FDA offices and divisions cannot be done
  piecemeal / ad hoc
• Ideally, should be internationally agreed /
  acceptable to regulatory authorities in other
  jurisdictions
• Limited number of definitions developed
  (Reporting Adverse Drug Reactions Definitions
  of Terms and Criteria for Their Use CIOMS,
  Geneva, 1999).

12
Additional viewpoint CIOMS-VI (working draft)

• Events of critical importance consistently
  defined using standard criteria (e.g. acute liver
  failure), or in consultation with appropriate
  experts
• Describe definitions and requirements for use of
  particular terms in protocol and product safety
  plan
• Rarely, overrule investigators diagnosis for
  analysis if clearly erroneous document reason,
  maintain original verbatim for audit, show both
  terms in study AE tables
• If not reported, assign probable diagnosis when
  signs, symptoms, and/or treatment strongly
  suggest a defined syndrome (e.g. chest pain,
  elevated CK, acute thrombolytic treatment
  myocardial infarction)

13
Thank you