Developing eStandards for Clinical Trials Data and Analyses

1
Developing e-Standards for Clinical Trials Data
and Analyses

• Steve Wilson
• Division of Biometrics II, CDER, FDA

22nd Spring Symposium New Jersey Chapter of the
American Statistical Association International
Harmonization and Electronic Submission Embassy
Suites, 121 Centennial Ave, Piscataway,
NJ Wednesday, June 6, 2001
2
Disclaimer

• Views expressed in this presentation are those of
  the speaker and not, necessarily, of the Food and
  Drug Administration

3
Acknowledgements

• John Clark, Center for Drug Evaluation and
  Research (CDER), FDA
• Michael Fauntleroy, Center for Biologics
  Evaluation and Research (CBER), FDA
• Randy Levin, Center for Drug Evaluation and
  Research (CDER), FDA

4
Outline

• Background / motivation
• Statistical review
• Electronic submission
• Developing guidance for e-standards
• Clinical data
• Documenting analyses
• CTOC and the eCTD
• Other data-related issues
• Too much data? Thinking parsimoniously
• The right data? Quality assurance and safety
• Non-clinical data? Stability and Carcinogenicity
• The Big Picture Electronic Submissions -- Data
  Repository -- Review Tools

5
Background / Motivation

• Statistical review
• Electronic submission
• Regulation
• Guidances
• MAPPs

6
Statistical Review
Background/Motivation

• Assess compliance with protocol / blinded
  analysis plans -- conduct of the study
• Check appropriateness of statistical models and
  conclusions
• Verify results reported in the NDA
• Answer review questions
• Modify models and assess robustness / sensitivity
  of the results

7
Statistical Review
Background/Motivation

• Modify data sets and reanalyze
• Examine the trial and data for potential bias
• Results by center
• Baseline predictors
• Important subgroups (sex, age, race, etc,)
• Assess impact of audits
• Due diligence

8
The Way We Were

• Statisticians requested data, program files and
  documentation (PROC Contents, annotated CRF,
  description of derived variables, etc.) at
  Pre-NDA meetings
• To assist review, sponsors submitted these
  electronic files to reviewers as desk copies --
  no formal archive
• Possibly a number of data and program requests
  during review cycle

9
Electronic Submission Regulation
Background/Motivation

• 21 CFR Part 11Electronic Records Electronic
  Signatures Final Rule
• Electronic Submissions Establishment of Public
  Docket Notice
• August 20, 1997
• www.fda.gov/ora/compliance_ref/part11/Default.htm

10
Regulation 21 CFR Part 11

• ...electronic records as equivalent to paper
  records...
• ...apply to all FDA program areas...
• ... intended to permit the widest possible use of
  electronic technology, compatible with FDA's...
  responsibility to promote and protect public
  health...

11
Regulation 21 CFR Part 11

• ...The use of electronic records as well as their
  submission to FDA is voluntary...
• ...docket No. 92S-0251 ... identify specifically
  what types of documents or parts of documents are
  acceptable for submission in electronic form
  without paper records...
• ...consult with the intended agency receiving
  unit for details

12
Electronic Submission NDA

• Electronic equivalents to paper
• Text and CRFs
• PDF (Adobes portable document format) organized
  in specified folders
• Navigate with bookmarks and hyperlinks
• CRTs -- the Data
• Version 5 SAS transport SAS Institute (SAS
  technical support TS-140.
• Business Case replace CRTs format...open format
  published by the -- save trees, already giving it
  to the statisticians

13
Electronic Submission GuidanceNDA -- Case
Report Tabulations

• ...provide a single transport file for each
  dataset. ...
• ...less than 25MB per file...
• ...data definition tables ...variable name, a
  description of the variable, the type of variable
  (e.g., number, character, date), and codes...
• derived variables...method of deriving the
  variable

14
Electronic Submission GuidanceNDA -- Case
Report Tabulations

• ...Variable names are limited to 8 characters...
• ...Descriptive name up to 32 characters...
• ...Further recommendations...for each specific
  submission type...
• ...discuss the content of the datasets with the
  review division prior to submission.

15
Appendix 2 Example Content of Specific Clinical
Datasets

• The following lists contain suggested data
  elements for the individual datasets.
• ... serve as a starting point for discussion
  between you and the review division on the
  content and organization of the datasets and,
  therefore, is not all inclusive.
• ... refining these data elements ...
• ...data needed for each indication varies
  specific information ... at the time of the
  pre-NDA meeting or earlier ...

16
Appendix 2 Example Content of Specific Clinical
Datasets

• ... programs that you used in your statistical
  analysis ... final analysis for principal
  efficacy and safety data...placed in the
  appropriate subfolder of the crt folder.
• ...The programs should contain sufficient detail
  to allow the reviewer to follow the logical flow
  of the program...

17
Appendix 2 Example Content of Specific Clinical
Datasets

• Demographics
• Age
• Sex
• Race
• Weight
• Height
• Country
• (Consult the review division )
• Inclusion criteria
• Vary by protocol consult ...
• Exclusion criteria
• Vary by protocol consult ...

• Concomitant medications
• Drug name
• Drug start date
• Drug stop date
• Drug started before study (yes/no)
• Drug type
• Dose
• Reason for medication
• (Consult the review division...)
• ...

18
Appendix 2 Example Content of Specific Clinical
Datasets

• Medical history
• Disposition
• Drug exposure
• Efficacy results
• Human pharmacology and bioavailability /
  bioequivalence data

• Microbiology data
• Physical examination
• Adverse events
• Vital signs
• ECG
• Labs

19
Electronic Submission CBER

• REVISED Guidance for Industry Providing
  Regulatory Submissions to the Center for
  Biologics Evaluation and Research (CBER) in
  Electronic Format - Biologics Marketing
  Applications Biologics License Application
  (BLA), Product License Application (PLA) /
  Establishment License Application (ELA) and New
  Drug Application (NDA) 11/12/99

www.fda.gov/cber/guidelines.htm
20
CDER MAPP 7600.6

• Requesting and Accepting Non-Archivable
  Electronic Records for New Drug Applications
• cannot be accepted in lieu of the archivable
  electronic record as outlined in the guidance.
• SAS transport file to the EDR.

21
Mid-Course Review

• Regulation 21 CFR 11
• Guidances
• General considerations
• NDA
• CBER
• MAPP

www.fda.gov/cder/regulatory/ersr
22
Regulation Vs. Guidance
23
The Way We Are

• Electronic submissions becoming routine for some
• Still dealing with paper for a number of others
• Statistical reviewers caught in the middle --
  CRTs and analysis files
• Confusion on both sides

24
Developing Guidance for e-Standards

• Leveraging / observing
• Standardizing clinical data
• Documenting analyses

25
Leveraging /Observing

• Leveraging is the creation of relationships
  and/or formal agreements with others outside the
  FDA that will ultimately enhance FDA's ability to
  meet its public health mission.
• CRADA -- Cooperative Research And Development
  Agreement ...appropriate only with collaborators
  who will make significant intellectual
  contributions.
• Observing -- we can look, but we cant touch

26
Standardizing Clinical Data
Developing Guidance for e-Standards

• Recognized need / advantages
• CDISC -- Clinical Data Interchange Standards
  Consortium
• Agency working group -- guidance
• Safety data and patient profiles
• Cautionary note

27
CDISC

• CDISC (Clinical Data Interchange Standards
  Consortium) -- an open, multidisciplinary,
  non-profit organization committed to the
  development of industry standards to support the
  electronic acquisition, exchange, submission and
  archiving of clinical trials data and metadata
  for medical and biopharmaceutical product
  development.
• www.cdisc.org

28
CDISC Goals

• Nearly seamless exchange of data within a
  company, between collaborating companies, and
  with regulatory agencies across protocols,
  companies and compounds
• Effortless archiving of data and metadata for
  future review or regulatory audit
• Integration of data from a wide variety of
  applications and systems
• Facilitated reviews of regulatory submissions
• Improvements in data quality cleaner data
  faster

29
The CDISC Approach to Submission Standards

• Follow the lead of the FDA Submission Guidelines
• Consider Regulatory Reviewer(s) as primary
  customer(s)
• Define basic safety metadata standards to guide
  dataset organization -- not rigid structures
• Aim for 80 of domains and 80 of variables
• Use representative examples rather than hard
  rules
• Allow flexibility for science and sponsor
  differences
• Start with 12 safety domains then develop a
  library for therapeutic areas over time
• Post standards openly and encourage ongoing input
  by all.

30
CDISC Metadata Description

• Specified in Guidelines
• Domain Dataset Name (e.g., DEMO)
• Description (Demographics)
• Location (crt/datasets/1234/demo.xpt)
• Metadata model proposes adding Structure
• Defines the key structure and unit of analysis
  for a row or observation
• Useful when multiple datasets are needed for the
  same clinical domain
• Differentiates crt datasets from redundant
  analysis datasets.

31
CDISC Metadata Example Dataset Redundancy

• Is the lab value normal? (1 rec/pat/visit/lab
  test)
• Did the lab value change over time? (1
  rec/pat/visit)

32
CDISC Submission Dataset Definition
33
CDISC and HL7

• The Associate Charter Agreement signed by HL7 and
  CDISC calls for the creation of a Clinical Trials
  Special Interest Group (CTSIG) within HL7 that
  will convene jointly with representatives from
  the existing CDISC Working Teams.

34
Documenting Analyses Analysis Dataset Models
(AdaM)
Developing Guidance for e-Standards

• DRAFT Guidelines for the Creation of Analysis
  Files and Associated Documentation for Submission
  to the FDA
• PURPOSE provide guidelines for the creation of
  files and associated documentation that are
  submitted to the FDA statistical reviewer in
  support of the primary and important secondary
  study objectives

35
eIND, eCTD and CTOC

• eIND -- electronic Investigational New Drug
• CBER Pilot
• eCTD -- electronic Common Technical Document
• CTOC -- Cumulative Table of Contents
• XML
• Pilots

36
Other Data-Related Issues

• Too much data?
• Oncology clinical trials
• The right data?
• Quality Assurance
• Safety
• Non-clinical data?
• Carcinogenicity
• Stability

37
Too Much Data? Thinking Parsimoniously A
Proposal for Oncology Trial Data

• Draft Guidance
• Cancer Drug and Biological Products Clinical
  Data in Marketing Applications
• Contact Grant Williams, CDER
• www.fda.gov/cder/guidance/3983dft.htm

38
Too Much Data? Thinking Parsimoniously A
Proposal for Oncology Trial Data

• Investigator to sponsor Why all these data?
• Answer FDA might want it.
• Sponsor to FDA How much data do you need?
• Answer Good question, weve never been asked.

Grant Williams, May 2001
39
Too Much Data? Oncology Proposal

• The Agency recognizes that the collection,
  quality control, and entry of data in a database
  is an expensive and time-consuming processIn
  fact, many of these data may not be called for in
  a marketing application for therapyWe therefore
  encourage discussion of specific data
  requirements at end-of-phase-2 meetings to
  minimize unnecessary data collection
• Draft Guidance for Industry Cancer Drug and
  Biological Products -- Clinical Data in Marketing
  Applications

40
Too Much Data? Benefits of data reduction

• Decrease cost
• Increase numbers of patients in trials
• Improve quality of important data
• Decrease audit citations
• Grant Williams, May 2001

41
The Right Data

• Quality Assurance
• Safety

42
The Right Data?Metrics for Data Monitoring and
Data Management -- A Proposal

• Extremely difficult during review to assess the
  impact of data monitoring and data management on
  the reported results of the trial.
• May lead to inefficiencies in the review process
• Government trial experiences -- time was spent
  thinking and worrying about this issue

43
The Right Data? Describing Data Monitoring and
Management

• Sample Confidence Codes
• I Empty field -- imputed value
• E Empty field -- filled-in after check with
  source documents / investigator
• C Failed edit check -- confirmed as actual
  value
• R Failed edit check -- value was replaced
  when checked against source
  documents/investigator
• P Failed edit check -- suspicious and not
  able to confirm as correct / retained
• S Failed edit check -- not able to confirm /
  imputed

44
The Right Data? Describing Data Monitoring and
Management
45
Non-Clinical Data?

• Carcinogenicity
• Stability

46
Carcinogenicity

• NDA Electronic Submission Guidance Appendix 1
  Example Nonclinical Pharmacology AndToxicology
  Datasets And Data Elements
• Guidance for Industry Statistical Aspects of the
  Design, Analysis, and Interpretation of Chronic
  Rodent Carcinogenicity Studies of Pharmaceuticals
• www.fda.gov/cder/guidance/815dft.pdf

47
Electronic Submissions -- Data Repository --
Review Tools
From Randy Levin, CDER, FDA
48
THANK YOU

• wilsons_at_cder.fda.gov
• 301 827-5583