Presentation on theme: "Cathy Barrows, Ph.D. GlaxoSmithKline Co-Leader of CDISC Pilot Project CDISC ADaM Team An Update on the CDISC SDTM/ADaM Pilot Project FDA/Industry Statistics."— Presentation transcript:
Cathy Barrows, Ph.D. GlaxoSmithKline Co-Leader of CDISC Pilot Project CDISC ADaM Team An Update on the CDISC SDTM/ADaM Pilot Project FDA/Industry Statistics Workshop - 29 September 2006
3 CDISC Pilot Project was to… Demonstrate that data submitted to the FDA using the CDISC Standard will meet the needs and expectations of both medical and statistical FDA reviewers. Produce a worked example implementation of the available CDISC standards.
4 How? By conducting a case study legacy data (real clinical trial data, warts and all) CDISC SDTM domains and ADaM datasets and associated metadata submission of case study package to FDA for mock review And in the process, identify any issues to be resolved in SDTM and ADaM models
5 FOCUS: the package not the process Choices/decisions guided by timeline realities of a team of volunteers from multiple companies quick, efficient, effective - not necessarily the most preferred option are not making recommendations re process!
6 Pilot Submission Deliverables Submission package Includes SDTM datasets, analysis datasets, all relevant metadata, analysis results, abbreviated report Review package tied together using metadata in DEFINE.XML Summary report of the pilot submission project issues encountered, strengths and weaknesses incorporate what we learned from the FDA feedback Both to be made available to the public on the CDISC website
7 Criteria for success of the Pilot Project FDA statistical and medical reviewers will evaluate the submitted datasets (SDTM and AdaM), metadata and documentation Usable with their tools? Reproducibility of analyses, derivations? Navigable? Contents – what and where are OK? etc.
8 And the verdict? We DID produce an electronic submission that met the reviewers needs The overall tone of the reviewers feedback was very positive noted easier learning curve The FDA review was very thorough and they provided constructive criticism Issues encountered both by team and in FDA feedback are already beginning to be addressed by CDISC teams
9 Now to details…
10 Who? 15 core team members from industry FDA involvement Unprecedented level of involvement Provided co-leadership employees involved 12 consistently in contact with team includes medical and statistical reviewers Interactions: regular team teleconferences Feb. face-to-face meeting to define the project (expectations/requirements) Pre-submission encounter Feedback from review
11 FDA representatives - expectations, requirements, wish list Key messages: Consistency, accuracy, completeness are extremely important - follow the specifications! Define file crucial, but needs to be accurate Clear mapping between the plans for analysis, the tabulation data, the analysis data, and the analyses performed SDTM and Analysis datasets should be available for both medical and statistical reviewers
12 Presubmission Encounter Important opportunity to communicate about those crucial data-type issues that we run out of time for at usual meetings Discussed data to be submitted - structures, variables FDA was able to make specific requests, for example: Hys Law analysis dataset (liver hepatotoxicity)
13 Legacy data used in the pilot submission Real clinical trial data, provided by Eli Lilly Data de-identified, documents redacted Indication: Alzheimers Randomized, double-blind, placebo-controlled, parallel-group study Three treatment arms: low dose, high dose, placebo Approximately 300 patients, multiple centers Representative set of endpoints and analyses included in package
14 CDISCPILOT01 M1 (Administrative) Cover Letter PDF M5 (Clinical Study Reports) Study Report Study Report PDF Datasets Analysis DEFINE XML Analysis Datasets XPT Tabulation Annotated CRF PDF DEFINE XML SDTM Datasets XPT Reviewers Guide PDF CDISC Pilot Submission Package Content PDF TOCs and eCTD folder structure
15 Legacy documents received Decisions regarding data analysis Write SAP Map blank CRF to SDTM (aCRF) Create SDTM data metadata Create analysis data metadata Create SDTM datasets (little derived data) Create analysis datasets Receive legacy data Create 0-obs analysis datasets Coding of events data & con.med. data Write study report Create 0-obs SDTM datasets Finalize SDTM datasets Generate analyses Derived data to SDTM Create analysis results metadata Note that create includes QC steps. Write reviewers guide Write cover letter Create DEFINE Create XPT files Building the CDISC Pilot Submission Package
16 Presentation of the Define FDA expectation: Pilot package in Define.xml The Define file integrates tabulation dataset (SDTM) metadata analysis dataset (ADaM) metadata analysis results (ADaM) metadata New implementation! Exciting!
17 Illustration of the Pilot Submission Package
18 Top of Define file Table of Contents Ideally would also have included a link to reviewers guide
19 Define file: List of Analysis Datasets
20 Define file: Metadata for ADSL
21 Define file: Code List
22 Define file: Computation Method
23 Define file: List of Tabulation Datasets
24 Define file: List of Analysis Results
25 Define file: Analysis Results Metadata
26 Define file: Linked to Study Report
27 Inclusion of Programs? Elected to not include entire program Included program code in metadata for repeated measures analysis This WAS used in the review
28 Next Steps: Wrap up tasks for this iteration Some revisions to current package Implement some of the FDA feedback Fix a few things that are errors or oversights Incorporate some things we wish we had done Complete the project report Publish the package and the project report
29 Future iterations might: Fully and completely work this example (rather than a subset) More fully develop the metadata Include datasets in the DEFINE file, instead of separately XPT files Use studies from other therapeutic areas Go beyond a single study Test submitting different sets of sponsor data (e.g., NDA, safety update) Address how to send updates (e.g. additional derived variables) effectively Test newer versions of models, e.g.: ADaM PK data Pre-clinical data ODM Test other analysis strategies
30 Our Advice - Wisdom is scar tissue in disguise Or, as one FDA Review Team member said: In order to get a standard we have to suffer Its worth it!