Data Sharing: A Trialists View Mahesh KB Parmar MRC Clinical Trials Unit London

1
Data Sharing A Trialists ViewMahesh KB
ParmarMRC Clinical Trials UnitLondon
2
Review

• Issues in data sharing for trials
• Our current approach
• Issues for the future

3
Issues

• Unlike most other scientific areas, trials have
  the potential to have an immediate and sometimes
  substantial impact on many different groups
• a) patients
• b) doctors
• c) health services as a whole
• d) consumer groups
• e) pharmaceutical companies
• f) financial traders of shares
• HERA - Herceptin in breast cancer
• ICON3 the use of paclitaxel in ovarian cancer
• a

4
Issues

• Results can be interpreted by different groups in
  different ways
• Considerable incentive for results to be
  misrepresented this can lead to widespread
  confusion, especially if original data are
  tortured until they submit
• truth will out in the long-term - of little
  practical use in the short to medium term

5
Example Icon3

• 2000 patient trial of paclitaxel plus
  carboplatin vs. non-paclitaxel chemotherapy in
  advanced ovarian cancer
• Reported no evidence of difference in PFS and OS
  between paclitaxel and no paclitaxel therapy (HR
  0.98, p 0.74 95 CI 0.87 to 1.10)
• 2 other studies (GOG111, OV10) came to the
  opposite conclusion, 1 supported ICON3 (GOG132)
  considerable heterogeneity in the results
• Although many tried to consider the clinical and
  scientific reasons for this heterogeneity
  number of individuals/groups tried to discredit
  the results of ICON3 and dismiss GOG132
• Led to considerable confusion for many years
• NICE ultimately ruled that both therapies should
  be discussed with the patient

6
ICON3 - Overall survival Control vs. TJ
Survival
paclitaxel plus carboplatin - -
- - - - - - Control
paclitaxel plus carboplatin
7
Example Icon3

• 2000 patient trial of paclitaxel plus
  carboplatin vs. non-paclitaxel chemotherapy in
  advanced ovarian cancer
• Reported no evidence of difference in PFS and OS
  between paclitaxel and no paclitaxel therapy (HR
  0.98, p 0.74 95 CI 0.87 to 1.10)
• 2 other studies (GOG111, OV10) came to the
  opposite conclusion, 1 supported ICON3 (GOG132)
  considerable heterogeneity in the results
• Although many tried to consider the clinical and
  scientific reasons for this heterogeneity
  number of individuals/groups tried to discredit
  the results of ICON3 and dismiss GOG132
• Led to considerable confusion for many years
• NICE ultimately ruled that both therapies should
  be discussed with the patient

8
Current approach

• Share data as widely as scientifically
  appropriate and practically possible
• Clear process for application, assessment and
  approval
• Aim to collaborate wherever possible and
  appropriate

9
Process

• Structured application received
• Sent to the Trial Management Group and CTU for
  consideration
• Sent to Trial Steering Committee (oversight body
  for the trial)

10
Structured application

• Background
• Rationale
• Objectives
• Summary of variables requested
• Timelines
• Statistical Methods
• Funding issues (inc. MRC CTU Costs)
• Ethical Approval
• Patient consent
• IPR issues
• Publication Policy

11
Conditions

• Ethics and Research Governance
• Confidentiality of the data, and use only for
  this project
• Any publication/presentation of results must be
  reviewed by CTU/TMG
• IPR held by MRC unless otherwise agreed

12
Conditions

• Ethics and Research Governance
• Confidentiality of the data, and use only for
  this project
• Any publication/presentation of results must be
  reviewed by CTU/TMG
• IPR held by MRC unless otherwise agreed

13
Data Sharing20 Cancer Approvals for 04 05

• Diseases and trials
• Colorectal (FOCUS 5 requests AXIS)
• Bladder (BS01, BS03, BS07, BA06)
• Testicular (TE04, TE19, TE20)
• Ovarian (ICON2, ICON3)
• Renal (RE01, 2 requests)
• Lung (BLT)
• Breast (AB01)
• Oesophageal (OE02)

14
Data Sharing20 Cancer Approvals for 04 05

• Types of project
• Molecular/Genetic Markers 6
• Socio-Economic Outcomes 5
• Meta-Analyses 4
• NICE 1
• Other (e.g prognostic factors) - 4

15
Data sent to CTU for MAs
 
16
Future

• Continue to share data as widely as
  scientifically appropriate and practically
  possible
• Resources are likely to become an issue, and a
  limit may have to be set

17
(No Transcript)