Issues and Case Studies in Clinical Trial Data Sharing ... · 5/17/2013 · Issues and Case Studies in Clinical Trial Data Sharing: Lessons and Solutions . Friday, May 17, 2013

Issues and Case Studies in Clinical Trial Data Sharing: Lessons and Solutions Friday, May 17, 2013 Wasserstein Hall, Milstein East

- 2 -

Welcome and Meeting Objectives

Mark Barnes, Barbara Bierer MRCT

- 3 -

Collaborating to Improve Multi Regional Clinical Trials

The MRCT Center’s Purpose is… To improve the design, conduct, and oversight of multi-regional clinical trials, especially trials sited in or involving the developing world; to simplify research through the use of best practices; and to foster respect for research participants, efficacy, safety and fairness in transnational, trans-cultural human subjects research.

Establish Best Practices

Develop Standards

Identify Opportunities for Improvement

Improve Transparency

Objectives

- 4 -

ACKNOWLEDGING OUR SPONSORS

MRCT Center Executive Committee Sponsors

MRCT Steering Committee Sponsors

- 5 -

Today’s Meeting Objectives

• Review the rationales for requirements to disclose participant-level clinical trials data

• Discuss implications of data disclosure requirements • Review evidence from recent experiences with participant-level

data disclosure

• Provide a new, multi-stakeholder perspective on potential solutions and criteria for access to participant-level data for public health and scientific research purposes

• Identify potential areas of collaboration on these issues among

stakeholder groups

- 6 -

Clinical Trial Data Sharing & Transparency Team

4 Subgroups formed: •Rationales for and Benefits of Data Sharing (Lead: Patricia Teden) •Responsible Use of Shared Data (Lead: Mark Barnes / Roshni Persaud) •Innovation and Incentives (Lead: Salvo Alesci / Jeff Francer) •Regulatory Implications (Lead: Jules Mitchel)

AJ Allen (Lilly), Mark Barnes (Ropes & Gray / MRCT), Barbara Bierer (Partners / MRCT), Melissa Binz (Novartis), Karen Craun (Sanofi), Amy Davis (PRIMR), David Dorsey (J & J), Jeffrey Francer (PhRMA), Kate Heffernan (KGH Advisors), Julie Kaberry (Harvard), Marcia Levenstein (Pfizer), Jennifer Miller (Harvard), Jules Mitchel (Target Health), Sandra Morris ( J&J), Pearl O’Rourke (Partners), Mercy Osakpawan (HSPH), David Peloquin (Ropes & Gray), Thomas Peppard (Gates), Patricia Teden (Teden Consulting), Fabio Thiers (Vis), Marc Wilenzick (MRCT), Roshni Persaud (MRCT)

Working Group Launched Feb 15, 2013 Co-chairs Salvatore Alesci (PhRMA), Michelle Mello (Harvard School Public Health)

- 7 -

Clinical Trial Data Sharing & Transparency Workgroup Output

• Establish common criteria for data-sharing models • Identify and describe reasonable data sharing models • Involve diverse stakeholders across pharma, biotech,

academia • Balance between:

assuring simplicity in operation, maintenance and access protecting participant privacy maintaining commercial incentives for medical product development meeting the public health and scientific requirements for data

transparency

- 8 -

Agenda Topics Timing Moderator / Presenter

Keynote Speakers 8:15-9:10 am Jeff Drazen AJ Allen

Introduction of Model for Data Sharing 9:10 - 9:20 am Michelle Mello

Session 1: Rationale for Increased Clinical Trial Data Sharing

9:20-10:45 am Michelle Mello

Break 10:45-10:55 am

Session 2: Safeguarding participant privacy, consent principles, and the integrity of data analyses

10:55 -12:20 pm Mark Barnes

Lunch served 12:20-12:40 pm

Lessons Learned from the implementation of FDAAA and the ClinicalTrials.gov Results Database (Working lunch)

12:40 -1:10 pm Deborah Zarin

Session III: Balancing companies’ intellectual property interests with public access to data

1:10-2:35 pm Justin McCarthy

Break 2:35 – 2:50 pm

Session IV: Assuming participant-level data is shared in the public domain, what are the ramifications?

2:50 – 4:45 pm Marc Wilenzick

Mark Barnes, Barbara Bierer,

- 9 -

Keynote Speaker –Jeffrey M. Drazen, M.D.

Data Sharing

Confirming

Testing a Formed Idea

Browsing

GEO (Gene Expression Omnibus)

GEO DataSet Browser

Data obtained by microarray must be submitted to a repository such as the Gene Expression Omnibus or ArrayExpress prior to submission. The raw and transformed data sets for each microarray experiment must be provided through the repository, and the Accession Number for each experiment or series must be provided in the Methods section. If the data are password-protected, the user name and password must be provided in the cover letter and the Methods section of the manuscript at the time of submission. A criterion of publication is full access to the relevant data sets through a publicly accessible repository.

FDA Mini-Sentinel Pilot Project http://www.mini-sentinel.org

Data Sharing

Confirming

Testing a Formed Idea

Browsing

Data Sharing

What data? For Whom? Under what rules?

Quality

Cost

All you can eat buffet

grade

Michelin 3 star grade

Almost free Very

expensive

RCT data

Routine Clinical data

Elizabeth Loder, BMJ & Harvard Medical School IOM Data Sharing Workshop

Elizabeth Loder, BMJ & Harvard Medical School

www.aamu.edu

- 33 -

Keynote Speaker –Albert J. “A.J.” Allen, M.D., Ph.D.

© 2012 Eli Lilly and Company

Perspectives on Patient-level Clinical Data Sharing from a Patient Advocate, Bioethicist and Industry Physician

Albert J. Allen, MD, PhD

Sr. Med. Fellow

Bioethics & Pediatric Capabilities

Eli Lilly and Company

FINANCIAL Disclosures

• Employee and shareholder, Eli Lilly and Company

• Member, SACHRP

Disclaimer

• The opinions expressed are those of the presenter and in no way reflect an official opinion of Eli Lilly.

• Nor are they the result of a systematic survey or review of the perspectives of individual patients, bioethicists or industry physicians.

My Background

• Education (1976-1995): • Univ. of Chicago, SB (chemistry) and SM

(biochemistry) • Univ. of Iowa, MD and PhD (pharmacology)

• Vice President of AMSA (1983-84) • Univ. of Iowa and NIMH, residencies and research

fellowship, psychiatry and child psychiatry • Univ. of Illinois at Chicago (UIC, 1995-2000)

• Institute for Juvenile Research, assistant professor, member of IRB before/after suspension of FWA in late 1990s

My Background (continued)

• Eli Lilly (2000-present) • Clinical research physician on global product

development team for an ADHD medication (2000-2003)

• Global medical director/sr. medical director on global product development team for an ADHD medication (2004-2011)

• Sr. medical fellow (2011-Present) • Chair of Bioethics Advisory Committee • Cochair, Pediatric Steering Committee

• Member or cochair of several safety advisory committees

• SACHRP (2011-present)

My Background (continued)

• Early, stage 1, multiple myeloma patient with renal insufficiency due to kappa light chain deposition disease (August, 2012-present) • Formal diagnosis: Feb. 11, 2013 • Chemotherapy started: Feb. 12, 2013

• Married to a retired journalist, aspiring women’s fiction writer who is a depression and breast cancer survivor

• As the only physician in the family, I am/have been the informal medical consultant to multiple relatives

“Call me Don Quixote…”

Are you sure? Don Quixote was…psychotic


A lot of people and groups say they are speaking in the interests of patients, but what

do patients really want?

And would they want the same thing if they were

intimately familiar with drug development and

bioethics from multiple perspectives?

What this patient wants, knowing everything I know about drug development

• I want to get well today, or as soon as possible. I want treatments with the best possible balance of benefits to risks.

• I want as many options for the future as possible, including new treatments that (hopefully) improve the benefit/risk balance.

• Respect me and my fellow patients as fellow humans. If you aren’t already, you will be one of us some day. If you want to study “my data,” I either want: • a say in the matter (consent), and/or • someone(s) independently looking out for my

interests

What this patient wants, knowing everything I know about drug development (continued)

• If someone is using “my data,” I expect the research to have high scientific integrity – I don’t want my data used for bad science – by anyone – that could hurt patients, including me. So I want: • someone(s) independently looking out for my

interests, making sure the science is sound

What this patient wants, knowing everything I know about drug development (continued)

• If someone says they are doing a secondary analysis using “my data” via a shared dataset “in the interests of patients,” then they should be accountable for that. • Audits, inspections, reviews, warning letters,

academic and funding sanctions, legal proceedings, Congressional action, etc. are all ways of creating accountability

What if? (All assume open, unregulated, publicly available access to patient-level data from industry clinical trials)

1. A self-funded, independent researcher analyzes the data for a recently approved vaccine for a pediatric disease with significant morbidity/mortality.

a. Researcher justifies the work as being done “in the interest of patients.”

b. The researcher announces at an international medical conference that the vaccine increases the frequency of a rare, potentially fatal event.

i. Results published in respected, peer-reviewed journal c. Use of the vaccine is limited from then on. d. Subsequent research raises many questions about the

methodology and scientific validity of the self-funded, independent analysis, and eventually the weight of evidence supports safety of the vaccine.

What if? (All assume open, unregulated, publicly available access to patient-level data from industry clinical trials) • Consider in light of the case:

• When researchers justify use of patients’ data to investigate scientific questions “in the interest of patients,” should they/do they owe accountability to those patients (and/or their representatives, surrogates)?

• If accountability is owed, how is this to be achieved in different research settings/scenarios, including the one outlined in the case ?

• What is the impact of publicity, litigation, education, etc. on medical practice and public health outcomes?

• What is the impact of publicity, litigation, education, etc. on future research/investment into treatments?

• What is the impact of diversion of resources from other research to examine questions raised by flawed analysis?

Conclusion: What this patient wants

• Get well today, desirable benefit/risk balance • Options for future, desirable and hopefully

improved benefit/risk balance • Respect patients as persons – seek consent,

independent review to protect patient interests • Scientific integrity of the work is critical so that

research is most likely to benefit patients, not harm them • Review/approve proposed analysis in advance to

prevent problems • If you invoke the “interest of patients,” I want

you accountable to patients and/or their surrogates/representatives


Questions? [email protected]

Definition

• Accountable = responsible and answerable (Beauchamp, 2010) • Implies a conflict of interests exists, that there is a

responsibility to one or more of those interests for which a party is answerable in some way

- 51 -

Keynote Speakers – Q & A Jeffrey M. Drazen, M.D.

Albert J. “A.J.” Allen, M.D. Ph.D.

- 52 -

Michelle Mello, J.D., Ph.D.

- 53 -

Click to Edit Master Title Style

Potential Models for Data Sharing Michelle Mello, JD, PhD Professor of Law and Public Health Harvard School of Public Health

Issues and Case Studies in Clinical Trial Data Sharing: Lessons and Solutions

May 17, 2013

- 54 -

Core Principles: Models Should…

• Protect research participants

• Advance innovation and public health

• Balance risks with benefits of data sharing

• Treat all Data Generators equally

• Make data disclosure practicable by avoiding undue burdens on Data Generators and requesters

• Provide timely access to data

• Ensure adequate transparency

• Ensure accountability

- 55 -

Common Elements

• Requests and decisions posted on the web

• Requesters pre-commit to an analytical plan* • Requester’s identity and scientific plan are publicly

disclosed • Requester signs a data use agreement

• Decisions about data releases include both the Data Generator and other parties.

* May not apply to Open-Access Model

- 56 -

Two Gatekeeper Models

Learned Intermediary Model Data Generator Model Decision

-maker

Review Board that is independent of Data Generator

Data Generator

Criteria 1. Sound science: Is there a reasonable scientific hypothesis, sound analytical plan, and adequate plan to disseminate findings?

2. Benefit/risk balancing: Do the potential public health benefits of answering the proposed question(s) outweigh the probable adverse effects on the Data Generator (intellectual-property interests, competitive concerns, technical-support burden) and the potential risks to research participants?

3. Expertise: Does the research team have expertise sufficient to carry out the proposed analyses?

Process • Board reviews request, collects input from Data Generator, decides, and publicly documents rationale for decision

• Data Generator reviews request, decides, and publicly documents rationale for decision.

• Denials are appealable to independent Appellate Board, whose decision is final.

- 57 -

A Variation on the Gatekeeper Theme

“Black Box” / Database Query Model Decision-

maker

Independent review board or Data Generator

Criteria 1. Sound science

2. Benefit/risk balancing

3. Expertise Process • Requester submits a research query to the Data Holder

• Data Holder runs the query and returns results—not data

- 58 -

Removing the Gatekeepers

Open Access Model Decision-

maker

None.

Criteria Responsible-use attestation: All requests granted if Requester attests that data will not be used inappropriately (e.g., to re-identify research participants)

Process • Data Generator routinely posts data from trials when results are publicly reported or submitted to regulator, along with documentation to facilitate use of data

• Researchers can simply download the material

- 59 -

Session I: Rationale for Increased Clinical Trial Data Sharing

Moderator: Michelle Mello

- 60 -

Martha Brumfield, Ph.D.

Leveraging Data Sharing to Accelerate Biopharmaceuticals

Development

Martha A. Brumfield, Ph.D., CEO and President

Critical Path Institute (C-Path) 17 May 2013

Boston

61

Risk & Cost Sharing Model

Individual Companies or Research Institutions Will Not Solve Challenges to Future of Medicine Development Alone

No single entity has the answer

De-risking development and regulatory pathways is critical

Different Model for Partnering is Needed

Expertise from all sectors must be involved

International participation is necessary

Willingness to share critical information/data is required 62

C-Path: A Public-Private Partnership

Act as a trusted, neutral third party

Convene scientific consortia of industry, academia, and government for pre-competitive sharing of data/expertise

The best science

The broadest experience

Active consensus building

Shared risk and costs

Enable iterative EMA/FDA/PMDA participation in developing new methods to assess the safety and efficacy of medical products

Official regulatory recognition through “qualification” of Novel Methodologies and Drug Development Tools and acceptance of data standards

63

C-Path Consortia

Coalition Against Major Diseases UNDERSTANDING DISEASES OF THE BRAIN

Critical Path to TB Drug Regimens TESTING DRUG COMBINATIONS

Multiple Sclerosis Outcome Assessments Consortium

DRUG EFFECTIVENESS IN MS

Polycystic Kidney Disease Consortium NEW IMAGING BIOMARKERS

Patient-Reported Outcome Consortium DRUG EFFECTIVENESS

Electronic Patient-Reported Outcome Consortium DRUG EFFECTIVENESS

Predictive Safety Testing Consortium DRUG SAFETY

Seven global consortia collaborating with 1,000+ scientists and 41 companies

• Biomarkers • Clinical

Outcome Assessment Instruments

• Clinical Trial Simulation Tools

• Data Standards

C-Path Collaborators

65

Consortia Members

Partners

Define research goal

Identify data needed to accomplish goal

Apply data standards to enable integration

Pool data to create integrated database

Develop new drug development tools

C-Path Drug Development Strategy: Objective: Support the Work of the Consortium

66

After recent AD Phase III Failures…What’s Next?

67

Reasons for Phase III & Submission Failures: 2007-2010

Arrowsmith J. Nature Reviews Drug Discovery Feb 2011

Safety 21%

Efficacy 66%

7% 6%

• Sharing Knowledge

• Learning from Failures

• Public-Private Partnerships

68

that enable…

The Solution…

Collaborations

Coalition Against Major Diseases (CAMD)

69

The mission of CAMD is to advance innovative drug

development tools through a regulatory path that

accelerates therapies for neurodegenerative diseases.

Firsts: • Therapeutic Area clinical data standards published by CDISC (AD

and PD) • Unified CDISC database of Alzheimer’s disease clinical trial

information provided by multiple pharmaceutical companies • Clinical trial modeling and simulation tool advanced for a regulatory

decision • Neuroimaging biomarker for Alzheimer’s Disease qualified by a

regulatory agency (EMA)

Value of Data Sharing, Standards and Pooling

Nine member companies agreed to share data from 24 trials

The data were not in a common format

All data were remapped to the CDISC AD standard and pooled

A new in silico modeling tool was created through

the application of data standards and is under review by the FDA and EMA

Researchers utilizing database

Start Point

Result

70

24 studies, >6500 patients

What Was Learned? ADAS-Cog Variability

A B C D E F G

Item 1 Word Recall Word Recall Word Recall Word Recall Word Recall Word Recall Word Recall

Item 2 Commands Name Obj/fing. Name Obj/fing. Commands Name Obj/fing. Name Obj/fing. Name Obj/fing.

Item 3 Constr. Praxis Delayed recall Commands Constr. Praxis Commands Commands Commands

Item 4 Delayed recall Commands Constr. Praxis Delayed recall Delayed recall Constr. Praxis Constr. Praxis

Item 5

Naming Obj/fing. Constr. Praxis Idea Praxis Name Obj/fing. Constr. Praxis Idea. Praxis Idea. Praxis

Item 6 Idea. Praxis Idea Praxis Orientation Idea. Praxis Idea. Praxis Orientation Orientation

Item 7 Orientation Orientation Word Recog Orientation Orientation Word Recog Word Recog

Item 8 Word Recog. Word Recog. Remem. Instr. Word Recog Word Recog Remem. Instr. Spoken Lang Abil.

Item 9 Remem Instr. Remem Instr. Spoken Lang. Abil. Remem. Instr. Remem. Instr.

Spoken Lang. Abil. Comprehension

Item 10 Comprehension Spoken Lang. Abil.

Word Finding Dif.

Spoken Lang Abil.

Spoken Lang Abil. Word Finding Dif. Word Finding Dif.

Item 11

Word Finding Dif.

Word Finding Dif. Comprehension

Diff. Spont. Speech

Word Finding Dif. Comprehension Remem. Instr.

Item 12

Spoken Lang. Abil. Comprehension Concentration Comprehension Comprehension Concentration

Item 13 Number cancel. Concentration Concentration Concentration

C-Path’s Track Record: Data and Modeling & Simulation Tools

Integrated Data

Rogers et al., J Pharmacokinet Pharmacodyn. 2012 Oct;39(5):479-98

The C-Path Data Sharing Experience

Data Sharing/Use Agreement

Protects subjects and owners of data

Up front identification of objectives and governance

Rules for merging data

De-identify data to HIPAA “Safe Harbor” and EU DPD requirements

High value data in standard accepted by regulatory agencies

Rules for accessing data Obtain broadest possible data use agreement that meets regulatory

requirements for secondary use of data

Use access controls appropriate to research objectives

Governance Considerations

73


Legacy data conversion is resource intensive but worthwhile for specific projects

Assurance is needed that a specific dataset will be useful in achieving research/regulatory qualification objectives

New insights can be obtained from data converted to a common standard and aggregated to enable queries and analysis

Addition of standardized data from other sources (prospective, retrospective) becomes simplified and expands the power and utility of an integrated data resource

Key Insights Gained

74


Polycystic Kidney Disease Consortium

Aggregated data across three patient registries and two observational studies to support use of total kidney volume imaging as biomarker

Under review at FDA & EMA to support regulatory acceptance of new prognostic biomarker

Critical Path to TB Drug Regimens

CDC and company(ies) sharing patient level control arm data

Goal to develop TB disease model/simulation tool

MS Outcome Assessment Tool Consortium

Willingness expressed to share patient level clinical trial data (control and active)

Other Programs

75

Shared Learning Can Shorten the Timeline

76

Data Standardization and Sharing Biomarker Development and Qualification Outcome Assessment Measures Modeling and Simulation

Adapted from “A virtual space odyssey”, Cath O'Driscoll (2004) http://www.nature.com/horizon/chemicalspace/background/odyssey.html

Diverse Expertise

Open Dialogue

Collaboration

Efficiency

Resource Sharing

Data Sharing Close

Regulatory Interaction

Advancing Science

Challenges to Overcome

Challenges can be overcome with collaboration, diligence and focus

77

Patient privacy / informed consent

Review/approval for secondary use

Respect for confidentiality and IP

Data sharing / pooling mechanisms

Risk mitigation

Cultural resistance

Effective communication of why

and value proposition

Incentives

Funding sources

Data Sharing

78

Sharing Clinical Research Data Workshop Summary

• Released: March 29, 2013

• Meeting: October 4-5, 2012

• Chapters Include:

• Benefits of data sharing

• Barriers to data sharing

• Models of data sharing

• Standardization to enhance data sharing

• Changing the culture of research

Drug Information Association www.diahome.org

THANK YOU

79

C-Path gratefully acknowledges the support of

- 80 -

Robin Jenkins, MBA

The Project Data Sphere Initiative Robin Jenkins

May 6, 2013

- 82 -

LIFE

SCIENCES

CONSORTIUM

Project Data Sphere, LLC is an independent initiative of the CEO Roundtable on Cancer's Life Sciences Consortium

Video

83

- 84 -

Problem

● 7.6 million people per year die as a result of cancer1

● The decline of the cancer death rate has been slow compared to other disease areas such as heart disease

● Oncology R&D is generally not as productive as efforts in many other therapeutic areas

1Whose data set is it anyway? Sharing raw data from randomized trials. Trials, 2006. Vickers, A. doi:10.1186/1745-6215-7-15 http://www.trialsjournal.com/content/7/1/15. Last accessed January 5, 2013.

84

http://www.trialsjournal.com/content/7/1/15

- 85 -

IP

● Unique challenges in healthcare

● Multiple very valid attempts

● Attitude is “don’t share unless I can prove no harm occurs”

Historical Barriers

Privacy Security Resources Realizing Benefits

85

OVERVIEW AND VISION

86

- 87 -

CEO Roundtable on Cancer

● Project Data Sphere, LLC, is a wholly owned non-profit subsidiary of the CEO Roundtable on Cancer and arose as part of the CEO Roundtable on Cancer’s mission

● “Life Sciences Consortium” is a task force of the CEO Roundtable on Cancer

● Address issues in cancer research

● Accomplish together what no single company might consider alone

87

ADDRESSING THE BARRIERS AND ENABLING SUCCESS

88

- 89 -

● Four key obstacles:

Why Hasn’t this Happened?

Contracts Privacy Security Resources

89

- 90 -

1. Privacy

Privacy

●De-identifying patient level data for research purposes is possible

●An example of one company who has done it –

● Worked with a statistical expert using one way that the US regulatory framework acknowledges – Expert Determination Model

● Approx. 40 hrs of programming time, specific to the sample data set

● US regulatory environment provides other alternatives

●HHS website provides more information on de-identification

o (http://www.hhs.gov/ocr/privacy/hipaa/understanding/coveredentities/De-

identification/guidance.html)

10

http://www.hhs.gov/ocr/privacy/hipaa/understanding/coveredentities/De-identification/guidance.html



- 91 -

2: Security

● Hardened SAS hosting environment

● Firewall ● All access to data behind SAS firewall

● “Secure Socket Layer” (SSL) Protocol transmissions ● All transmissions encrypted

● Content virus scanning ● All documents scanned before made public

● Enrollment ● Role based permissions

● Password policy – to reduce chances of unauthorized entry

● Application acceptance required for access

Security

91

- 92 -

Contracts ● Uniform legal agreements to support comparator arm data contributions ● Data Provider

● Data User

92

● The data provider retains ownership and all existing IP while contributing meaningful datasets

● Limited restrictions of the use of the data ● Users may not seek patent protection for research procedures or research designs

that result from their research using the database ● Users may not assert against a data provider any patent right that results from the

user's use of the provider’s data. ● No other restrictions or limitations on users seeking to patent inventions that

result from their research using the database

● Publication acknowledgment but not manuscript review

3: Contracts

- 93 -

Resources

4: Resources

93

● Minimal resources required ● Especially relative to trial cost and benefits of sharing

● Leverage internal IT, legal, and biostats infrastructure

● Only requirements ● “Champion” of data sharing to facilitate effort within organization

● Data preparation

● Legal review

● Minimal upload time

● Available PDS Support ● Toolkit

- 94 -

The Mission

94

The universal platform to responsibly share and analyze oncology clinical trial datasets to revolutionize cancer research

Solution

Gather Allow contributed data to be leveraged by researchers across the world

Analyze Compile data with either Project Data Sphere tools or your own analytics

Disseminate Share observations and learnings within the scientific community

Collate Integrate and standardize data

95

The Vision: To Revolutionize Cancer Research

--- ----------.-...-..-

.. _ _ , , , ---

Peer- Reviewed

Publications

Project Data Sphere

. I

-1"

- - I 'I

'I I ,

- -" ' '

---------- :--\------- \ I

Government ......:a.... --- .::: . ...........

96

- 97 -

Our Objective: Changing the Curve

● Project Data Sphere initiative’s objective is to alter the trajectory of the cancer death rate

Objective

97

- 98 -

Home Page

● The primer:

● ~10 phase III datasets/CRFs /protocols/data descriptors

● The Goal

● 60+ datasets, by priority disease area, by key LSC members end 2013

18

- 99 -

Data Analytics Tools

● Project Data Sphere LLC, is exploring the possibility of including analytical tools from 3rd parties as part of phase II of the initiative

19

- 100 -

Realized Benefits

Reduce cost and duplication of effort

Establish research standards Prevent selective data

reporting Design more efficient trials

Maximize the impact of patient participation in studies

Study the impact of risk factors across studies

More valid population estimates for epidemiological work

29

- 101 -

Value

Access pre- competitive drug discovery data from oncology clinical trials through a single software platform

101

Collaborate on cancer research and optimize trial designs that improve transparency, patient selection, and data analysis; to establish real-world, epidemiology-based data standards in oncology

Patients who have participated in these trials have the opportunity to see their efforts reach beyond a single study to reach more patients with cancer

- 102 -

Load data into system

Project Data Sphere Initiative Timeline

2012 / Early 2013 February March April May

Solicit input from a broad spectrum of stakeholders in the fight against cancer to enable clinical data sharing

June

Public Launch

• Gather input

• System build

• User/beta testing

• Biostats & legal work

• Patient advocacy meetings

Build community, data, and support network

Multiple datasets available

102

- 103 -

Contact Information

103

John Dornan Executive Director, CEO Roundtable on Cancer

[email protected] 919-531-0966

Charles Hugh-Jones

Member, Life Sciences Consortium, CEO Roundtable on Cancer

[email protected]

mailto:[email protected]




- 104 -

LIFE

SCIENCES

CONSORTIUM

Project Data Sphere, LLC is an independent initiative of the CEO Roundtable on Cancer's Life Sciences Consortium

- 105 -

Patricia Teden, MBA

- 106 -


Rationales and Benefits of Data Sharing Patricia Teden, Teden Consulting LLC


May 17, 2013

- 107 -

Key Questions

• What are the specific rationales for data sharing? What public health benefits arise from being able to share and access data?

• What kinds of data need to be shared to reach those goals, and in what format?

• What conditions must be present to ensure that data sharing adequately achieves the identified goals?

- 108 -

Group 1 Members

AJ Allen Eli Lilly & SACHRP Karen Craun Sanofi Oncology Amy Davis Public Responsibility in Medicine and

Research (PRIM&R) Mercy Imahiyerobo Harvard School of Public Health Michelle Mello Harvard School of Public Health Jennifer Miller Safra Center, Harvard University David Peloquin Ropes & Gray Tom Peppard Global Health Program, Gates Foundation Pat Teden Teden Consulting Fabio Thiers ViS Research Institute

- 109 -

Data Sharing Offers Many Potential Benefits

Rationales and Benefits 1 Appropriately improve public confidence in study results

2 Deter the inflation of benefits and minimization of risks in reported study results

3 Meet ethical imperative to minimize risk for study participants

4 Meet ethical imperative to study participants that their participation advances science

5 Improve safety surveillance

6 Speed new discoveries

7 Facilitate secondary analysis to verify results, regulatory decisions, public policy

8 Enable patients and advocacy groups to access data for their specific medical issue

9 Facilitate secondary analyses to explore new scientific questions

10 Achieve operational efficiencies in trial conduct

11 Improve strategic decision-making regarding research portfolio

- 110 -

Typologies of Stakeholders and Data

Stakeholders 1. Public – general

2. Patients and providers

3. Research participants

4. Scientific community

5. Regulators

6. IRB/ECs and investigators

7. Trial sponsors

Data Formats

1. Summary results (e.g., clinicaltrials.gov)

2. Participant-level datasets

3. Participant-level datasets, analysis programs, SAPs (“All datasets”)

Documents

1. Clinical Study Report (CSR)

2. CSR synopsis

3. Lay summaries

4. Scientific publications

- 111 -

Stakeholders Benefited:

Public – general

Patients and providers

Research participants

Scientific community

Regulators

IRB/ECs and investigators

Trial sponsors

Stakeholders Benefitted

Rationales and Benefits 1 Improve public confidence in study results




5 Improve safety surveillance

6 Speed new discoveries




10 Achieve operational efficiencies in trial conduct


All Stakeholders Stand to Benefit

- 112 -

Rationales and Benefits Material That Should Be Shared 1 Improve public confidence in study results Data summaries, participant-level datasets

Documents: Lay summaries


Data summaries, participant-level datasets, all datasets Documents: CSRs


Data summaries Documents: scientific publications


Data summaries, participant-level datasets Documents: scientific publications

5 Improve safety surveillance Data summaries, participant-level datasets, all datasets Documents: CSRs

6 Speed new discoveries Data summaries, participant-level datasets, all datasets Documents: CSRs


Data summaries, participant-level datasets, all datasets


Data summaries, participant-level datasets, all datasets Documents: CSRs, lay summaries, scientific publications


Data summaries, participant-level datasets, all datasets Documents: CSRs, scientific publications

10 Achieve operational efficiencies in trial conduct Data summaries Documents: CSRs


Data summaries Documents: CSRs

- 113 -

Key Points of Consensus

1. Data sharing rules should apply to equally to all study sponsors and data generators

2. Something beyond a purely voluntary regime is desirable to create a level playing field

3. There should be global standard formats for clinical-trial data and documents

4. The rationales and benefits presuppose that initial and re-analyses of shared data will reflect sound science

– Data sharing system should have mechanisms for promoting responsible use of data

– Accountability standards should be similar for the initial sponsor or data generator, and a researcher conducting a re-analysis

5. Data sharing system must be practical

- 114 -

Points of Consensus: Participant-level Datasets

6. Many of the rationales/benefits require participant-level datasets – Facilitate secondary analysis to verify results, regulatory decisions, public policy – Improve safety surveillance – Speed new discoveries

7. Important mechanisms for a data sharing system: – Ensure adequate scientific expertise among the analytical team – Provide technical support sufficient to permit users to understand the data

8. Some benefits are difficult to achieve in a sponsor-controlled model

9. Timing of availability for both summary and participant-level data should be 1 year after primary study completion. – Assuming an adjudicated process to obtain participant-level datasets, evaluation

of the purpose for the participant-level datasets could be different (‘tighter’) prior to product approval.

- 115 -

Reactions to the Four Models

• Learned Intermediary: Appealing in principle. Who would run and pay for it?

• Data Generator Review: Insufficient to garner public trust and ensure consistency across all sponsors / data generators.

• Database Query: Flawed on the grounds of transparency and practicability.

• Open Access: Maximal transparency, but at the cost of ensuring scientific integrity in how data are used.

- 116 -

Comparison of Data Summaries, Clinical Study Reports and Participant-Level Datasets

Data Summaries

CSRs Participant-level Datasets

Format Data with visual on webpage

Document Data

Common standards cross-study?

Yes, within portal ICH template Limited standardization; considerable documentation needed for context

Scope of data included

Limited Wide Fully complete

Effort to create Input still viewed by industry and academia as an ‘extra’ effort

Industry – standard document . Extra effort needed to review/redaction some info? Other sponsors?

All sponsors have some electronic data. Effort to support and manage the documentation. Infrastructure to support the ‘model’

Risk to patient confidentiality

Very low Low Varying opinions from low to high

- 117 -

How Well Do Existing Initiatives Measure Up?

Initiative Summary Impression

GSK and Roche Policies for Sharing Results

“Independent’ committees to adjudicate requests

Not in place long enough to evaluate effectiveness

BMJ policy Authors must make data available upon ‘reasonable request’

One group’s contribution to the issue.

Mini-Sentinel Active safety-surveillance system whose access is limited to ‘partners’

Group had no information about effectiveness

Life Sciences Consortium Collaboration to speed new cancer discoveries

Adjudicated access. Effectiveness?

YODA Adjudicated researchers to reanalyze data; ‘Independent’ committees advise

Many aspects of the models being considered. Deliveries to be made in near future.

OSTP Initiative Research funded with public funds will have results shared with the public

Not specific to clinical research. Waiting to hear FDA and NIH plans.

- 118 -

Discussion Panel Q & A

AJ Allen, Lilly John Orloff, Novartis Martha Brumfield, CPI James Ware, NEJM, HSPH Patricia Teden, Teden Consulting Sally Okun, PatientsLikeMe Robin Jenkins, Sanofi Moderator: Michelle Mello

• What are the specific rationales for data sharing? What public health benefits arise from being able to share and access data?

• What kinds of data need to be shared to reach those goals, and in what format?

• What conditions must be present to ensure that data sharing adequately achieves the identified goals?

- 119 -

Session II: Safeguarding Patient Privacy, Consent Principles, and the Integrity of Data

Analyses

Moderator: Mark Barnes

- 120 -

Case Study, GSK: Jessica Scott, M.D., J.D.

Access to Anonymised Patient Level Data from GSK Clinical Trials

GSK Publicly Discloses Clinical Research Information

Protocol summary posted

Study Start

Study Completion

Result summary posted

Manuscript submitted

Full protocol and clinical study report* posted on the GSK Clinical Study Register

All human subject research studies that evaluate investigational or approved medicinal products – (phase I-IV, meta-analyses, observational studies)

8-12/18 months

18-24 months

Time of publication

* CSR posted after approval or termination of the medicine

Result Summaries and Publications Have Limitations

Publicly disclosed results:

Do not include the primary data from each research participant

Summarise data from the study population with statistics to compare treatment groups

Primary Efficacy Results: Total Population

Emetic Episodes Day 1 To Day 5 Dose 1 Dose 2 Dose 3

Treatment Response, n (%)

Complete (0 Episodes) 7 (19) 8 (22) 10 (31)

Major (1-2 Episodes) 10 (28) 14 (39) 10 (31)

Minor (3-5 Episodes) 0 1 (3) 0

Failure (>5 Episodes/Rescued) 19 (53) 13 (36) 12 (38)

p-value (stratified for centre)

Dose 2 vs Dose 1 0.848

Dose 2 vs Dose 3 0.467

Benefits of Greater Access to Patient Level Data

Enables the identification of trends and associations that may provide greater insight or help develop hypotheses and theories for further research

Enables the review of results from clinical trials to validate the results

Strengthens trust in clinical research through enhanced openness and transparency

Helps ensure the data provided by research participants are used to maximum effect in the creation of knowledge and understanding

Main Issues

Protecting the privacy and confidentiality of research participants

Ensuring the data are used for valid scientific investigation

Practicalities of anonymising data and providing data in ways that enable external researchers to understand and navigate the information

A Solution

Research sponsors

Independent Data Custodian

Anonymised patient level data provided after completion of the project and publication

Researchers

Undertakes scientific review of proposals

Manages privacy

Agreements to protect privacy, and publish the results

Submits scientific proposals and analysis plans Reviews

expertise and management of any conflicts of interest

The aim of GSK’s initiative is to help realise a broader solution with an Independent Data Custodian

How can we Realise this Solution?

Request site (https://clinicalstudydata.gsk.com)

Request site (https://clinicalstudydata.gsk.com)

Researchers can submit research proposals and request data from clinical studies we have listed on the site

Studies are listed after the medicine studied has been approved by regulators or terminated from development and the study has been accepted for publication

We have included global studies conducted since 2007; approximately 200 studies are listed

Over the next two years we will add global studies going back to the

formation of GSK

We will also include all studies (including local studies) we start in or after 2013

We estimate that over 100 studies will be added in September 2013

Available Clinical Studies will Build

Researchers Submit Research Proposals Using an On-line Form with these Items

Section A: Research Plan

Title Lay Summary

Study Design Studies Selected and Study Populations Primary and Secondary

Endpoints Statistical Analysis

Plan Publication Plan

Section B: Information about the Research Team

Name Post or Position

Employer, Company, Research Institution or

Affiliation

Education, Professional Qualifications and

Memberships

Section C: Funding of the Proposed Research

Section D: Potential Conflicts of Interest (CoI) and management of CoI

Section E: Other Information

Research Proposals will be Reviewed by an Independent Panel

The scientific rationale and relevance of the proposed research to medical science or patient care

The ability of the proposed research plan (design, methods and analysis) to meet the scientific objectives. This is a high-level review

The qualifications and experience of the research team to conduct the proposed research (a statistician with a degree in statistics or a related discipline should be part of the research team)

Real or potential conflicts of interest that may impact the planning, conduct or interpretation of the research and proposals to manage these conflicts of interest

The panel will consider:

The publication plan for the research

Researchers can Ask us About the Availability of Data from GSK Studies not Listed Before they Submit a Research Proposal

GSK Review of Enquiries

Have the studies been published or accepted for publication?

Are the studies of terminated or authorised medicines (in approved indications)?

Do we have the data? (e.g. many observational studies use data from third party databases)

Do we have the legal authority to provide the data? (e.g. the medicine may have been out-licensed to another company)

Are there any practical constraints? (e.g. there may be issues related to the size of genetic databases)

Can we effectively anonymise the data? (e.g. for studies of rare diseases we will consider this on a case by case basis)

What resources are required to retrieve and anonymise the data and documents?

Our Intent is to Provide Access to Data Where we Can

Where we are able to provide access, we will do so if the Independent Review Panel approves the submitted research proposal and we receive a signed Data Sharing Agreement

A Data Sharing Agreement will be put in place

The Data Sharing Agreement includes requirements to:

Only use the data for the agreed purpose

Not to attempt to establish the identities of research participants

Inform regulatory authorities and GSK of any safety concerns as soon as they are identified

Provide GSK with a copy of the manuscript after submission to a peer reviewed journal

Post and seek publication of the research

Allow us to use any invention that comes out of the research and negotiate any further rights in good faith

Access to Data is Provided in a Secure Password Protected Internet Site

Controls in place to prevent data being downloaded or transferred

Analytical software provided (e.g. “R” and SAS)

Data can be combined and analyses downloaded

This helps to protect the privacy and confidentiality of research participants and helps ensure the data are used for the agreed research purpose

Data and Documents Provided for Each Study

Raw dataset

Analysis-ready dataset

Protocols with any amendments

Annotated case report form

Reporting and analysis plan

Dataset specifications

Clinical study report (appendices which include patient level data are not included)

Helpline support will be available to help researchers understand and navigate the data

Summary

We have taken a first step and established a system where researchers can request access to anonymised patient level data from our studies

We recognise that there may be different ways to provide greater access to patient level data for further research and that our approach is likely to evolve as we gain experience and receive feedback

This is a first step towards the ultimate aim of having a broader system in which data from multiple companies and organisations are made available for research through an independent data custodian

- 140 -

Dr. Yaniv Erlich

- 141 -


How anonymous is genetic research data?

Dr. Yaniv Erlich, Whitehead Institute


May 17, 2013

Yaniv Erlich Yaniv Erlich 5/17/13

Penetration tests…

Intercom

Fingerprint reader

IT department of a major bank


Co-segregation between Y-chr and surnames

www.ysearch.org: Y

Y

Smith Smith

Smith

Y

Smith

Intro. Risk assessment

The Venter case Anonymous datasets Summary

Erlich


Exploiting genetic genealogy databases Intro. Risk assessment


An anecdote?


Can we recover the identity of anonymous sequencing datasets using public resources?

The main idea – a systematic study




Databases of interest

www.smgf.org www.ysearch.org Publicly available and free 135,000 surname-YSTR records…




Empirical test: what is the probability to recover a surname?

Y-STR of a real person Querying Ysearch and SMGF

Calculating surname confidence score

Inferring surname

Intro.

Risk assessment


Comparing the predicted surname to the true one

x900

Expectation for US Caucasian males from middle and upper class: 12% Successful recoveries


Distribution of inferred surnames

• Given a success, what is the expected distribution of surnames?



Most of the inferred surnames are relatively rare


Age+state+surname points on ≤12 males in more than half of the cases

Age+state+surname

Only age+state

Simulation: can we identify an individual?

Age State

0.0%

0.5%

1.0%

1.5%

2.0%

0 10 20 30 40 50 60 70 80 90

Surname

Adams



100,000 rounds


Putting it all together: the Venter case

Yaniv Erlich



www.ysearch.org:

lobSTR DYS458: 17 repeats

Try it yourself: bit.ly/craig_venter_haplotype_updated

• We got a surname from whole genome sequencing data


Finding Craig Venter

Searching for: 1. Venter 2. California 3. Born in 1946 4. Male In USsearch.com and PeopleFinders.com



Two matches, including:


Can we identify anonymous personal genomes?




Recovering the identifies of CEU individuals



8 Surname predictions with Utah ancestry

10 CEU genomes

Found an obituary that has the exact description of the pedigree

Probability of a random match < 5x10-9

Winfield Utah

Yaniv Erlich Yaniv Erlich 5/17/13 Yaniv Erlich

Beginner’s luck?



In total: 5 successful surname recoveries Breaching the privacy of close to 50 CEU samples.

Successful surname recovery (targeted individual)

Patrilineal line from source to target Person tested by genetic genealogy service (source)

p<5x10-9 p<10-5 p<5x10-6 p<5x10-6 p<10-5


Summary



Our approach: - No experimental work involved. - The identifying information propagates via deep genealogical ties. - The attack completely relies on public resources.

Testing close to 1000 Y-STR haplotypes, demonstrating complete identification of Venter and close to 50 CEU individuals.

Yaniv Erlich Yaniv Erlich

Melissa Gymrek (HST – Harvard/MIT) Amy McGuire (Baylor) David Golan (Tel-Aviv University) Eran Halperin (Tel-Aviv University)

Melissa Gymrek

Acknowledgements

1/18/13

Open Access (with FREE registration)

- 157 -

Mark Barnes, J.D., LL.M

- 158 -


Responsible Use of Data

Mark Barnes, MRCT and Ropes & Gray LLP

Issues and Case Studies in Clinical Trial Data Sharing: Lessons and Solutions May 17, 2013

- 159 -

Key Questions

• What are the risks of clinical trial data sharing in regard to privacy protection, and how can they be balanced against the potential social benefits of data sharing?

• Would privacy concerns related to clinical trial data sharing deter

prospective research participants from participating in clinical studies? • Does de-identification of data solve the problem of risks to participant

privacy and confidentiality? • Should participants’ consent serve as a precondition for sharing of

clinical trials data? • What should be the consequence (e.g., liability) if/when privacy is

compromised as a result of increased clinical trial data sharing?

- 160 -

Group 2 Members

Mark Barnes MRCT, Harvard Law School, Ropes & Gray LLP

Melissa Binz Novartis Pharmaceuticals

Jeffrey Francer PhRMA

Kate Gallin Hefferman KGH Advisors

Michael Hughes Harvard School of Public Health

Mercy Imahiyerobo Harvard School of Public Health

Julie Kaberry Harvard School of Public Health

Marcia Levenstein Pfizer

Pearl O’Rourke Partners HealthCare

David Peloquin Ropes & Gray LLP

Roshni Persaud MRCT

- 161 -

I. What are the risks of clinical trial data sharing in regard to privacy protection, and how can they be balanced against the potential social benefits of data sharing?

- 162 -

• Overall, the risks associated with data sharing appear to be limited, depending on process of sharing and form in which data are shared.

• We must weigh the “social benefits” of data

sharing against the possible negative impacts of data sharing, including the possibility of misuse and poor-quality secondary and tertiary data analyses.

• Risks are mostly related to the amount and detail

of data released at the research participant/subject level.

Overview of Privacy Concerns

- 163 -

Risks associated with data sharing

Risks to the privacy of participants and clinical study personnel

• Research Participants • Fear that employers or insurance companies might access data

and re-identify • Persons involved in studies for sensitive conditions may fear being

identified due to possible stigma/discrimination • Genetic information is of special concern, as many studies have an

“add on” component in which genetic information is derived and stored

• Patients/participants might identify themselves in shared data sets

• Clinical Study Personnel • May fear retribution or stigma if their participation becomes known

to others

- 164 -

Risks associated with data sharing

Risks to discrete and insular minorities • Heightened risk because privacy can be undermined even if individual

identities cannot be ascertained

• Example: Havasupai case • Tribe members provided DNA samples to university

researchers investigating diabetes. • Tribe members later learned that samples provided were

being used for other matters, including theories of the tribe’s origin that contradicted its traditional beliefs.

• Risk suggests that researchers should disclose the full extent to

which collected data will be used.

- 165 -

II. Would privacy concerns related to clinical trial data sharing deter prospective research participants from participating in clinical trials?

- 166 -

Effect of privacy concerns on clinical trials participation

Privacy issue has two primary aspects:

i. Risk of re-identification ii. Adequacy of informed consent

No substantial evidence indicates that prospective participants are greatly disincentivized from clinical trial participation out of a fear of privacy violations. Nevertheless: • Some participants have heightened sensitivity to privacy concerns • Certain types of trials pose increased risk of privacy violations because of

uniqueness of data, e.g., pediatric and orphan drug trials or multi-year trials • IRBs may be reluctant to approve clinical trials involving participant level data

sharing

- 167 -

III. Does de-identification of data solve the problem of risks to participant privacy and confidentiality?

- 168 -

Is de-identification a solution for patient privacy concerns?

Consensus of Subgroup

• De-identification is not consistently defined; EMA definition is more vague, less detailed and thus possibly quite different than the HIPAA definition

• De-identification is a moving target due to improving

technology

• Genetic information is becoming increasingly identifiable, which may make the HIPAA de-identification standards obsolete

• Degree of de-identification is inversely related to data

usefulness: the more identifiers removed, the less

useful the data become to subsequent researchers

- 169 -

IV. Should participant consent serve as a precondition for sharing of clinical trials data?

- 170 -

Participant consent should be a precondition for sharing of clinical trials data

Note: Consent/authorization to release information is distinct from consent to participate in a clinical trial

• Data sharing must be included as a specific part of the informed consent process

• Participant consent to data sharing should be a precondition to

participation in a clinical trial, if the EMA or other agencies will require sponsors to make participant-level data publicly available

• Participants should be informed about how their data will be used

and by whom, regardless of whether data are made freely available on the internet or are subject to a gatekeeping process

- 171 -

V. What should be the consequence (e.g., liability) if/when privacy is compromised as a result of increased clinical trial data sharing?

- 172 -

Suggested sanctions for compromise of privacy

The range of potential consequences for privacy violations depends on a variety of factors.

First preference Recommend that legislative and/or regulatory measures be created that impose civil or criminal liability on data recipients who engage in data “misuse,” such as re-identifying participants or sharing the clinical data set with additional users without authority to do so. Second preference Enforce data-sharing provisions through agreements between a learned intermediary and the data user.

• Data requesters sign a pledge that they will not “misuse” information

• Breaking of the pledge can lead to liability for breach of contract

- 173 -

Reactions to four models

• Learned Intermediary: Intermediary can assess the risk of re-identification for each data set disclosed, deny or rigorously control access in situations of heightened risk, and tailor a data use agreement to the appropriate risk level

• Data Generator Review: From a privacy perspective, this is nearly identical to the learned intermediary model

• Database Query: Safest from a privacy perspective because the data requester never gains access to the data set; may be difficult to operationalize because the system must respond to a heterogeneous group of data queries

• Open Access: Most dangerous from the perspective of privacy; need to make sure appropriate penalties are in place to discourage misuse of data

- 174 -

How Well Do Existing Initiatives Measure Up?

Initiative Summary Impression GSK and Roche Policies for Sharing Results

“Independent’ committees to adjudicate requests

Privacy impact depends on decisional principles used and on strength of data use agreements required

BMJ policy Authors must make data available upon ‘reasonable request’

Effects on privacy are unclear at this point

Mini-Sentinel Active safety-surveillance system whose access is limited to ‘partners’

Strong privacy protections

Life Sciences Consortium Collaboration to speed new cancer discoveries

Group was unable to assess privacy concerns

YODA Adjudicated researchers to reanalyze data; ‘Independent’ committees advise

Privacy controls are fairly rigorous due to extensive de-identification; however, very long and costly process

OSTP Initiative Research funded with public funds will have results shared with the public

Unclear how strong privacy protections will be

- 175 -


Kristen Henderson, Quintiles Yaniv Erlich, Whitehead Jessica Scott, GSK Mark Lim, Faster Cures Claudia Emerson, Sandra Rotman Centre

- 176 -

Key Questions

• What are the risks of clinical trial data sharing in regard to privacy protection, and how can they be balanced against the potential social benefits of data sharing?

• Would privacy concerns related to clinical trial data sharing deter

prospective research participants from participating in clinical studies? • Does de-identification of data solve the problem of risks to participant

privacy and confidentiality? • Should participants’ consent serve as a precondition for sharing of

clinical trials data? • What should be the consequence (e.g., liability) if/when privacy is

compromised as a result of increased clinical trial data sharing?

- 177 -

Lunch Served

- 178 -

Deborah A. Zarin, M.D.

The Limits of Summary Data Reporting:

Lessons from ClinicalTrials.gov

Deborah A. Zarin, M.D.

Director, ClinicalTrials.gov

May 2013

179

Types of Clinical Trial Data

• Participant Level Data

– Uncoded data

– Abstracted

– Coded

– Computerized

– Edited/cleaned

– Analyzable

• Summary Data

– Analyzed/summary

180

Summary Data

• Decision makers (other than FDA) rely on summary data

– Clinical decision making

– Policy decision making (e.g., payors)

• Characteristics of Summary Data

– Convenient

– Assume they are accurate reflection of underlying participant level data—(assume little room for subjectivity)

181

Three Key Problems with the Published Literature

• Not all trials are published

• Publications do not always include all prespecified outcome measures

• Unacknowledged changes are made to the trial protocol that would affect the interpretation of the findings

– e.g., changes to the prespecified outcome measures

182

ClinicalTrials.gov

• Registry (est. 2000)

– At trial inception

– Contains key protocol details

– >130,000 trials

• Results Database (est. 2008)

– After trial completion

– Summary results

– >7000 trials

183

The Results Database

• FDAAA enacted in September 2007

• Results Database launched in Sept 08

• Design based on statutory language and informed by CONSORT and other relevant standards

• Requires reporting of “minimum data set” that was specified in the trial protocol

• Tabular format for data with minimal narrative

• EMA is developing a DB based on our model

184

4 Scientific Modules

• Participant Flow

• Baseline Characteristics

• Outcome Measures

• Adverse Events

185 185

What We Have Learned?

• Registration – Inconsistent adherence to protocols – Continued evidence of selective publication – Evidence of selective reporting of outcomes

• Results Database – Lack of clarity about who is in charge of the science – Data analysis practices are not always rigorous – Subjects (and data) are commonly left out of analyses

186

Examples of Incoherent Entries

• 823.32 mean hours sleep/day

• “time to survival”

• 36 eyeballs in study of 14 people

• “mean time to seizure” = 18 people

• “first occurrence of all cause mortality (adjudicated)”

187 187

Examples of Changed Outcome Measures

• Quality of life scale is replaced by a depression scale

• One month data are replaced by 3 month data

• “# people with an event” is replaced by “time to event”

• “all cause mortality” is replaced by “time to relapse”

188

Structural Changes to Studies

• Arms come and go

• Participants come and go

• Participant Flow and Baseline Characteristics Tables describe different population than the Outcomes Tables

• Data providers cannot explain the “denominators”

189 189

Drug Placebo Drug (All) Placebo (All, Pre-CO)

Placebo (CO, Post-DB)

Total # participants affected/at risk 153/297 (51.52%)

164/302 (54.3%)

191/297 (64.31%)

185/302 (61.26%)

26/47 (55.32%)

Blood and lymphatic system disorders Neutrophils # participants affected/at risk 1/297

(0.34%) 0/302 (0%)

1/297 (0.34%)

0/302 (0%)

0/47 (0%)

Blood and lymphatic system disorders Hemoglobin # participants affected/at risk 8/297

(2.69%) 6/302

(1.99%) 11/297 (3.7%)

7/302 (2.32%)

0/47 (0%)

Serious Adverse Events

?

?

190

Initial Assumptions about ClinicalTrials.gov Data Requirements

• Required data are generated routinely after a clinical trial

– Required reporting based on the protocol for each trial

– Required data would be necessary to understand the results of the trial

– Required data would be necessary to write a journal article

• Burden of reporting would be mainly due to data entry and time requirements

191

Our Initial Assumptions Were Wrong

• Protocol may be vague, or may not be followed

• Summary Data NOT always readily available, even for trials that had been published

– For many trials, nobody could explain the structure or analysis

• There is not an objective, easy to describe route from initial participant level data to the summary data—Many people and many judgments are involved

192

193

Results: “In the nonblack cohort [n=938], a sustained virologic response was achieved: • in 125 of the 311 patients (40%) in group 1, • in 211 of the 316 patients (67%) in group 2 (P<0.001), and • in 213 of the 311 patients (68%) in group 3 (P<0.001)”

Summary Data: Journal vs. ClinicalTrials.gov

• 110 matched “pairs” of ClinicalTrials.gov results entries and publications

• 82% had at least one important discrepancy, e.g. – 24% in data for primary outcome measure

• Numerator • Denominator

– 30% in Serious Adverse Event data

194

Not a Straight Line from Uncoded Data to Summary Data

Uncoded

Data Type

Abstracted Coded Computerized Edited/cleaned

Analyzable Analyzed/Summary

Leve

l of

info

rmat

ion

Max

Min

Individual Participant-Level Data Aggregated Data

195

Documents that may help to explain the journey

• Protocol and Amendments

• Investigator Brochure

• Statistical Analysis Plan (SAP)

• Informed Consent Form(s)

• DSMB Reports

• Clinical Study Reports

• AE Reports

• Other ??

• . 196

Key Facts

• 1,200 New Trial Registrations per Month in ClinicalTrials.gov

• 2,900 New Trial Publications per Month in MEDLINE

– 35 per Month in Annals, BMJ, JAMA, Lancet, NEJM

– 5,300 Total Journals in MEDLINE

• 190 New Results Entries per Month in ClinicalTrials.gov

• 8 Original New Drug (NDA and BLA) Application Approvals per Month in Drugs@FDA (Jan-Jun 2012)

197

In Sum

• Decision makers will always need summary data

• The “journey” from initially collected participant-level data to summary data is not completely objective or reliable

• Structured curated data help to mitigate against acts of commission and acts of omission

• Participant-level data might allow for – Audit/accountability function

– Subgroup and other analyses not possible with summary data

– Pooling of data leading to potential new discoveries

• Non-systematic data release could also generate a new kind of “disclosure bias”

198

Session III: Balancing Companies’ Intellectual Property Interests with Public Access to Data

Moderator: Justin McCarthy, Pfizer

- 200 -

Richard Kingham

Recent EU Transparency Initiatives: Legal and Policy Implications

Richard Kingham Covington & Burling LLP Washington and London

May 17, 2013

Topics

• Public access to nonclinical and clinical data in marketing authorization files

• Disclosure of information relating to clinical trials

• Legal and policy implications

Access to Information in MA Files

Legal Framework

• Regulation (EC) No. 1049/2001

– Governs public access to documents held by EU institutions, including European Medicines Agency

– Article 4 includes exception where disclosure would undermine protection of “commercial interests of a natural or legal person, including intellectual property”

– Unless there is an overriding public interest in disclosure

Legal Framework

• TRIPS article 39(3) – – Members, when requiring, as condition of approving

the marketing of pharmaceutical … products which utilize new chemical entities, the submission of undisclosed test or other data, the origination of which involves a considerable effort, shall protect such data against unfair commercial use. In addition, Members shall protect such data against disclosure, except where necessary to protect the public, or unless steps are taken to ensure that the data are protected against unfair commercial use.

Traditional EMA Interpretation as to Nonclinical and Clinical Data in MAs

• Case report forms and patient listings not normally submitted, so not available to disclose

• Full study reports (with tabulations) treated as confidential commercial information and not disclosed

• Assessment reports disclosed, with redactions

• Policy consistent with approach in other developed countries

EU Ombudsman Recommendations

• Official of European Parliament

• Issues nonbinding recommendations concerning “maladministration” by EU entities

• November 2010 recommendation in case brought by Danish academic researchers seeking data in MAs for Xenical and Orlistat – Suggests “private” disclosure, which EMA rejects

– Holds that studies are not confidential information

– Fails to address TRIPS

– EMA agrees to revise disclosure policy

March 2012 Guidance

• Issued by EMA and heads of national medicines agencies (HMA)

• “Information encompassing non-clinical and clinical development of the medicinal products … is not per se confidential”

• In general, the entire content of modules 4 and 5 (nonclinical and clinical study information) can be released, including “case report forms and individual patient listings, when submitted”

Litigation in EU General Court

• Two MA holders seek annulment of decisions to release full clinical study reports

• In both cases, president of court has granted preliminary measures (in effect, a stay pending review)

Developments Outside Litigation

• EMA plans to continue releasing clinical study reports not directly affected by court order

• Possibility of more streamlined procedures

• Possible requirement for submission of individual patient listings or case report forms

• Data suggest that many requests have been submitted by competitors or law firms

– A minority have been submitted by researchers

Disclosure of Information Relating to Clinical Trials

October 6, 2012 Commission Guidance

• Legal basis includes general provisions of Pediatric Medicines Regulation and Centralized Approval Regulation

• Requires sponsors of clinical trials to submit summary reports to EudraCT (current EU clinical trials data base)

• Results of pediatric trials submitted within 6 months, other trials within 1 year

• Non-pediatric phase 1 studies will not be released • All others will be made public within 15 days of posting • No exception for studies completed prior to marketing

authorization • In process of implementation

Proposed Clinical Trials Regulation

• Issued July 2012 • Now before European Parliament on first reading • All clinical documents would be submitted to a single

“EU portal” and incorporated in an EU data base – Would include full clinical trials applications (CTAs),

correspondence, inspection reports, etc. – Plus summary reports of all clinical studies conducted

under the regulation, submitted within 1 year of completion

• EU data base will be publicly accessible unless exemptions apply (including confidential commercial information)

Legal and Policy Issues

• Two main incentives for private investment in drug development

– Patents

– Regulatory exclusivity periods

• Data exclusivity (e.g., 5- and 3-year periods under Hatch-Waxman; 8+2+1 period in EU)

• Market exclusivity (e.g., orphan exclusivity)

• Both are potentially affected by the EU initiatives

Patents

• Public disclosure of clinical trial results prior to submission of applications could undermine patentability

• Regulatory requirements for early disclosure of clinical trial results could force submission of applications before full information is available to support them

Regulatory Exclusivity Periods

• Data exclusivity provisions differ from one jurisdiction to another

• But most were designed with the expectation that full safety and effectiveness data would remain confidential

• In some cases, persons in possession of full safety and effectiveness data submitted by competitors might use them, along with bridging studies, to circumvent DE periods, in whole or in part – Initial submissions – Supplemental submissions (e.g., for new indications)

• Clinical study reports (without CRFs and patient listings) might be accepted in some jurisdictions, in full or partial fulfillment of data requirements

Other Implications

• Even if full reports are not used directly to support competitive applications, they can contain valuable commercial information

– Research approaches that were tried and failed

– Endpoints, subanalyses and other key elements of study design that would be useful to competitors

– Future development plans (second-generation products, new indications, etc.)

– Other

Conclusion

• Policies on public disclosure of safety and effectiveness data should be developed in a manner that takes account of the need to preserve incentives for innovation, as well as the legitimate needs of researchers and others for information relating to the drug development and approval process

Richard Kingham [email protected]

- 220 -

Jeff Francer

- 221 -


Maintaining Incentives to Invest in Research Jeffrey K. Francer, J.D., M.P.P. Pharmaceutical Research and Manufacturers of America


- 222 -

Key Questions

•What intellectual property rights, proprietary interests, and competitive concerns do companies have that may be adversely affected by data sharing [by either voluntary or mandated clinical data disclosure policies]? •Would the impingement on these interests that could accompany data sharing likely affect public and private investments in R&D and, ultimately, innovation? •How should these concerns be balanced against the potential benefits of data sharing? •What strategies might effectively address companies’ legitimate concerns while maximizing the public benefit of data sharing? •Is imposing a “learned intermediary” between those who seek access to data and the data sources a possible approach to ease competitive concerns while still allowing reasonable access for independent researchers?

- 223 -

Salvatore Alesci PhRMA Melissa Binz Novartis Jeffrey Francer PhRMA Rebecca Li MRCT Justin McCarthy Pfizer Sandra Morris Johnson & Johnson David Peloquin Rope & Gray Roshni Persaud MRCT Fabio Thiers ViS Research

Group 3 Members

- 224 -

Data Sharing Policy Objectives

• Biopharmaceutical companies seek responsible data sharing arrangements that respect— • Research participant privacy • Integrity of the regulatory system • Incentives to invest in biomedical research

• Question Posed: What intellectual property rights, proprietary interests, and competitive concerns do companies have that may be adversely affected by data sharing [by either voluntary or mandated clinical data disclosure policies]? • Patentability issues • Regulatory approval and data exclusivity issues • Economic “free rider” issues

- 225 -

Investment in Research and Development

• Developing a new medicine takes an average of 10-15 years. • According to a 2007 study, it costs an average of $1.2 billion to develop one new drug.

Source: J.A. DiMasi and H.G. Grabowski. "The Cost of Biopharmaceutical R&D: Is Biotech Different?" Managerial and Decision Economics 2007; 28: 469–479.

- 226 -

Patentability Issues

• Compelled early disclosure could negatively affect the ability to patent discoveries. • Earlier disclosure could require inventors to file certain patent

applications earlier, and possibly in less complete condition, resulting in increased difficulty in prosecution of applications and uncertain patent protection in many countries.

• Earlier filing also could reduce the effective patent term for a marketed pharmaceutical invention.

• Undermining intellectual property protection will affect a companies’ ability to recoup investments in research and development.

- 227 -

Regulatory Approval, Data Exclusivity, and “Free-riding”

• Mandated disclosure of certain regulatory files could have secondary consequences that will affect public health:

• Could allow competitors to obtain regulatory approval in other countries without having invested in the research (“free-rider problem), thus affecting incentives for investment in biomedical research

• Many data exclusivity provisions protect only data that have not been disclosed to the general public

• Additional negative effect on incentives to invest in biomedical research

- 228 -

One Solution: Data Sharing with Appropriate Protections

• What is the goal of clinical trial data sharing?

• Help practitioners and patients through more complete understanding of available medicines and medical devices

• Inform scientific discourse • Advance medical technology

• Can goals be balanced against risks?

• Research participant confidentiality (e.g., providing line listings) • Harming incentives to invest in biomedical research

- 229 -

Potential strategies to address concerns while maximizing public benefit of data sharing

• How might privacy concerns be addressed?

• De-identification (but may not completely remove risk that a research participant can still be identified)

• How might concerns regarding investment incentives be addressed?

• Restricting access of data to “qualified” individuals / institutions (not competitors) • Restricting the use of data through a contractual arrangement

• Specify acceptable uses of data for research purposes • Specifying requirements for keeping data set confidential / not transferable • Prohibiting the filing of patent applications on inventions made from data set • Potential penalties for misuse

• Delay disclosure of data • “Learned intermediary”?

- 230 -


Moderator: Justin McCarthy

Susan Forda, Lilly Ben Roin, HLS Richard Kingham, Covington Jeffrey Francer, PhRMA Ira Shoulson, Georgetown Aaron Kesselheim, HMS Sandra Morris, J&J

• Question Posed: What intellectual property rights, proprietary interests, and competitive concerns do companies have that may be adversely affected by data sharing [by either voluntary or mandated clinical data disclosure policies]?

•Patentability issues •Regulatory approval and data exclusivity issues •Economic “free rider” issues

- 231 -

Afternoon Break

- 232 -

Session IV: Assuming Research Participant Data IS Shared in the Public Domain, What are

the Ramifications?

Moderator: Marc Wilenzick, MRCT

- 233 -

Sabine Haubenreisser, PhD

An agency of the European Union

EMA update on Clinical Trial Data transparency

MRCT Center at Harvard

Cambridge, 17 May 2013

Presented by: Sabine Haubenreisser European Medicines Agency Liaison Official at the U.S. FDA

A change of minds and hearts

“Clinical trial data is not commercial confidential information”

EMA position

235

Open clinical trial data for all – why?

Build trust and confidence in the system

Ethical responsibility to the patients enrolled in clinical trials

Public health benefit: independent (re)analysis of data broadens

knowledge base

Scientific progress: sharing of complex data can open new

horizons

236

237

What happened ?

• Growing pressure from academic groups and patient/consumer

advocates to release clinical trial data, mounting distrust in

industry and regulators

• Highly publicised case: ‘EMA refuses to make available data on

Tamiflu’ European Ombudsman: ‘public health interest

overrides commercial confidentiality’

• Change of Agency policy (2010): Clinical trial data will be

made available upon request (‘re-active’) and we will explore

how to make available pro-actively.

• Joint EMA-Heads of Medicines Agencies (HMA) document on

what is considered ‘commercially confidential information’

238

Where are we today?

• The Agency initiated a broad discussion on pro-active

publication of clinical trial data data

• Workshop on access to clinical trial data and transparency on

22 November 2012 to help define how to provide access

• For now, most, but not all, stakeholders are very positive

about the Agency’s initiative

2011: Journal of the

American Medical Association

2012: Journal of the

American Medical Association

2

3

9

240

Who is unhappy?

(Parts of) the drug industry; their arguments:

• Availability of detailed clinical trial datasets will encourage

flawed ‘secondary analyses’ which will give rise to

unfounded public health scares.

• Such data could be used by competitor companies to

produce copycat drugs in less regulated markets.

“Not all our products are protected by patents and we are relying on data

protection to keep our data secret. If you put the whole file on the internet

someone could send it to other countries such as Australia, Canada or

countries in the rest of the world where regulation is not as robust”

• Industry will lobby against full transparency.

Advisory groups – balanced membership

EMA (as Coordinator)

Patient groups

Pharmaceutical industry, consultants, law firm

Research institutes, NGO’s

Healthcare professionals

Academia

Regulators

Media

241

Developing EMA’s policy on proactive

publication of clinical trial data:

Steps taken so far

Nominations for membership in December 2012

Sessions for each group from January-March 2013

Final advice from each advisory group published 30 April 2013

242

Protecting patient confidentiality

How can the Agency ensure through its policy that patient and

other personal information will be adequately protected, that

patients cannot be retroactively identified when clinical trial data

are released, and that applicable legislation, standards, and rules

regarding personal data protection will be respected?

243

Clinical trial data formats

How can the Agency ensure through its policy that clinical trial

data can be shared, in the interests of public health, in a clear

and understandable format that enables appropriate analyses and

a swift implementation without undue burden to stakeholders?

244

Rules of engagement

Are there rules or conditions that should be in place before an

external stakeholder can download clinical trial data (e.g. formal

acceptance of the need to respect personal data rules, uploading

of analysis plans)?

245

Good analysis practice

Are there good analysis practice guidelines that the Agency could

ask external requestors of data to consider or be aware of, and

that the Agency can apply when confronted with additional

analyses from external parties?

246

Legal aspects

Are there any legal aspects other than personal data protection

that need to be addressed when drafting the Agency’s policy?

Are there exceptional circumstances under which data can be

claimed to be commercially confidential?

247

Outcome of advisory group discussions

Stakeholders expressed divergent views and suggested different

options on particular aspects

Different viewpoints are reflected in the final advice from each

group

EMA will make the final decision which option to choose

248

EMA draft policy – considerations (I)

Enable public scrutiny and secondary analysis of clinical trials

• verify regulatory authority’s position and to challenge it

• take regulatory decision-making closer to citizens and

promote better use of medicines

• independent replication of clinical trial data is a legitimate

scientific and societal goal

249

EMA draft policy – considerations (II)

Access to clinical trial data in an analysable format benefits

public health

• more efficient drug development, learning from past

successes and failures

• scientific community develop new knowledge

250

EMA draft policy – considerations (III)

Ensure that personal information on patients is adequately

protected

• need guarded approach to sharing of patient-level data

• learn from sharing of patient-level data whilst preventing

patient identification

251

EMA draft policy – considerations (IV)

Respect boundaries of informed consent

• patients participate to support development of particular

medicine

• any other use (e.g. for commercial purposes) oversteps the

boundaries and should not be enabled

252

EMA draft policy – considerations (V)

Protect public health and regulatory decisions against claims of

inappropriate secondary data analyses

• address conflicts of interest and set quality standard

Ensure transparency both ways

• same standard of transparency applies to secondary analyses

253

EMA draft policy – considerations (VI)

Ensure transparency in the best interest of public health without

impinging on intellectual property rights

• Guard against unintended consequences on intellectual

property rights that might disincentivise future investment in

R&D

Balance between access to clinical trial data and rules for

engagement

254

255

Developing EMA’s policy on proactive

publication of clinical trial data:

next steps

• The Agency will go on as planned

• EMA draft policy to be published by 30 June 2013 for public

consultation

• End of public consultation phase: 30 September 2013

• Publication of final EMA policy (including comments received):

30 November 2013

• Coming into force: 1 January 2014

256

The Agency’s current position

(anticipated outcome?)

Be as transparent as possible,

cognisant of (some limited) risks – but …

“the benefits of transparency outweigh the risks”

Sabine Haubenreisser, MSc, PhD

European Medicines Agency

Liaison Official at the FDA

Building 31, Office 3434, Silver Spring

+ 301 796 8748

[email protected]

[email protected]

257






- 258 -

Elizabeth Loder, M.D., M.P.H.

Anticipated Impact of BMJ’s Data Sharing Policy

Elizabeth Loder, MD, MPH Clinical Epidemiology Editor, BMJ Associate Professor of Neurology Harvard Medical School

Issues and Case Studies in Clinical Trial Data Sharing MRCT Center at Harvard Cambridge, MA May 17, 2013

What I aim to cover

What is the BMJ’s data sharing policy?

Why do we have this policy?

What is the anticipated impact?

What is the BMJ Policy?

• Applies to drug and device trials submitted from January 2013

– Regardless of when conducted

– Regardless of funding or sponsorship

• Does not currently apply to trials of diagnostic tools or surgical operations or other interventions that are not drugs or devices

What is the BMJ Policy?

• Drug and device trials will be considered for publication only if:

– Authors commit to making relevant anonymized patient level data available on reasonable request

– This commitment must be detailed in the article’s data sharing statement which is published at the end of every research paper

Sample Data Sharing Statement

“Data sharing: patient level data (and/or) full dataset (and/or) technical appendix (and/or) statistical code are available at (/doi) (with open access/with these restrictions) (from the corresponding author at). “

Sample Data Sharing Statement (continued)

“Participants gave informed consent for data sharing (or…consent was not obtained but the presented data are anonymized and risk of identification is low…or consent was not obtained but the potential benefits of sharing these data outweigh the potential harms because…)”

BMJ Data Sharing Policy

If there are no such further data available, please use this wording: “data sharing: no additional data available.” This option is not available for trials of drugs or devices.

Required for trials, encouraged for all

• All authors of research articles are encouraged to link articles to raw data

• We recommend researchers should seek informed consent for sharing at recruitment, even if there are no current plans to share data

What is a reasonable request?

• “We will expect requesters to submit a protocol for their re-analysis to the authors and to commit to making their results public.”

• “We will encourage those requesting data to send a rapid response to bmj.com describing what they are looking for. If the request is refused we will ask the authors of the paper to explain why.”

What is meant by relevant?

• “All anonymized data on individual patients on which the analysis, results, and conclusions reported in the paper are based.”

Inclusiveness

Ris

k O

R R

ewar

d

Everyone

Any researchers

Other approved researchers

Original researchers

Sponsor

What does “sharing” mean?

Why the new policy?

Godlee F, Groves T. The new BMJ policy on sharing data from drug and device trials. BMJ 2012;345;e7888 doi: 10.1136/bmj.e7888.

“It is no longer possible to pretend that a report of a clinical trial in a

medical journal is enough to allow full

independent scrutiny of the

results.”

Why the new policy?


“Journals have a responsibility

to use what power they

have to push for greater

transparency.”

Why the new policy?


“If research if to help doctors and patients make the best clinical decisions, it must be

reliable and reproducible, but

these are qualities that current peer review processes cannot assure.”

Why the new policy?

• It’s not that new: “Just one step up” from our previous policy

• Since 2009 authors encouraged to share data and required to say whether they will or not

• One step closer to a data-sharing culture

“This may require the global organisation of a suitable shared database for all raw data from human trials...Concealment of data should be regarded as the serious ethical breach that it is, and clinical researchers who fail to disclose data should be subject to disciplinary action by professional organisations. This may achieve quicker results than legislation in individual countries, although this is also desirable.”

Many advocates

Organization for Economic Co-operation

and Development (OECD)

World Health Organization

National Institutes of

Health

Bill and Melinda Gates Foundation

Hewlett Foundation

US Congress

UK Medical Research Council

Wellcome Trust

The Cochrane

Collaboration

Funders’ statement on data sharing 2011: I

“We believe that making research data sets available to

investigators beyond the original research team in a timely and

responsible manner, subject to appropriate safeguards, will

generate three key benefits:

• faster progress in improving health

• better value for money

• higher quality science“

Major funders’ Jan 2011 joint statement on sharing research data to improve public health (http://www.wellcome.ac.uk/About-us/Policy/Spotlight-issues/Data-sharing/Public-health-and-epidemiology/WTDV030690.htm)

http://www.wellcome.ac.uk/About-us/Policy/Spotlight-issues/Data-sharing/Public-health-and-epidemiology/WTDV030690.htm













Public health benefits: funders’ position

Signatories:

Wellcome Trust, Hewlett Foundation, NIH,

MRC, CIHR, Gates Foundation, INSERM, DFG,

AHRQ among other major funders

Scientists should communicate the data they collect and the models they create, to allow free and open access, and in ways that are intelligible, assessable and usable for other specialists

Six key areas for action: 1. Scientists need to be more open among themselves and with public and media 2. Greater recognition of the value of data gathering, analysis and communication 3. Common standards for sharing information, to make it widely usable 4. Publishing data in reusable form to support findings must be mandatory 5. More experts in managing and supporting use of digital data 6. New software tools need to be developed to analyse the growing amount of data

http://royalsociety.org/policy/projects/science-public-enterprise/report/

Anticipated Impact

• Not the end of the story

• The policy has clear limitations

– Sympathetic to arguments about lack of funding, worries about consent and privacy

– But some arguments seem spurious

A step in the right direction

• “Our first substantial target is to achieve proper independent scrutiny of trials of all drugs and devices in current use. Journals and their contributors will now have to ensure that we are as rigorous in overseeing and critiquing this new breed of re-analyses as we have tried to be of the originals.”


Anticipated Impact

Practical and Scientific

Detection and deterrence of selective or inaccurate reporting of research

Reliable access to valid information about previously performed trials and avoidance of duplication

Accelerated research and enhanced collaborations Restoration of trust in the clinical research enterprise

Ethical and Moral Meet obligations to research participants Minimize known risks and potential harm from

unnecessary exposure to previously tested interventions Recognition of medical research as a public good

Very similar to the arguments for trial registration! Krleza-Jeric K et al. Principles for international registration of protocol information and results from human trials of health related interventions: Ottawa Statement. BMJ 2005;330:956-958.

The goal: a data-sharing culture

Thanks !

twitter@eloder

[email protected]

[email protected]



Dr. Jules T. Mitchel

- 287 -


Regulatory Implications of Data Sharing

Jules Mitchel, Target Health Inc.


- 288 -

Key Questions

• What are the implications of public sharing of clinical trial data for regulatory processes?

• Do the potential benefits of data sharing for regulatory processes outweigh the risks (e.g., second-guessing regulatory agencies, premature or incorrect conclusions on risk/benefit profile of medicines)?

• Can a move toward increased public data sharing jeopardize ongoing efforts toward improved regulatory harmonization?

- 289 -

Group 4 Members

Barbara Bierer MRCT Brigham and Women’s Hospital

David Dorsey Janssen Research & Development

Rebecca Li MRCT

Jules Mitchel Target Health

Sandra Morris Johnson & Johnson

David Peloquin Ropes & Gray

Roshni Persaud MRCT

Marc Wilenzick MRCT

- 290 -

I. What are the implications of public sharing of clinical trial data for regulatory processes?

- 291 -

Assuming that data is publicly available, what happens from a regulatory standpoint?

1. As the results of analyses become publicly available, will there be any issues that the FDA or other regulatory bodies need to address?

2.Will there be obligations imparted onto the FDA, or other regulatory bodies as a result of any secondary analyses?

3.For example, should the information be sent to a FDA Advisory Committee?

4.What are the implications for drug or device labeling?

5.What are the regulatory processes that need to be followed by sponsors?

- 292 -

Assuming that data is publicly available, what happens from a regulatory standpoint?

6. Will there be a regulatory mechanism for researchers conducting secondary analyses to provide their respective findings to regulators?

7. Since drug companies and medical device manufacturers have certain reporting obligations (i.e. adverse events or patient safety issues) to regulatory agencies, what will be the minimum reporting requirements to sponsoring companies for unaffiliated researchers conducting secondary analyses?

8. Journals may become inundated with publications from those outside the company performing sub-studies or post hoc analyses and this may lead to second guessing of labeling, etc.

- 293 -

II. Do the potential benefits of data sharing for regulatory processes outweigh the risks (e.g. second-guessing regulatory agencies, premature or incorrect conclusions on risk/benefit profile of medicines)?

- 294 -

Unintended consequences of transparency:

1. There could be serious implications if there is selective disclosure about unapproved uses of a drug or medical device that are positive but do not include the negative results.

2. Proponents and opponents of a specific drug may swiftly move to conduct subset analyses of their competitor’s product, and this may inevitably lead to challenges with respect to regulatory determinations.

3. Having personal data available, researchers may be able to conduct investigations involving targeted medical treatments which could define responder rates for subgroups within an indication. What happens then?

- 295 -

III. Can a move toward increased public data sharing affect ongoing efforts toward improved regulatory harmonization?

- 296 -

1.Due to cultural, political and regulatory differences among regions of the world, there may be conflicts between regions on the use of shared patient level data.

2.Awareness and sensitivity to these issues must be taken into consideration when initiating global clinical programs.

- 297 -


Robert O’Neill, FDA Sabine Haubenreisser, EMA Toshi Tominaga, PMDA Agnes Klein, Health Canada Deborah Zarin, NIH Elizabeth Loder, BMJ Jules Mitchel, Target Health Inc. Evgeny Rogoff, Russia Moderator: Marc Wilenzick

• What are the implications of public sharing of clinical trial data for regulatory processes? • Do the potential benefits of data sharing for regulatory processes outweigh the risks (e.g., second-guessing regulatory agencies, premature or incorrect conclusions on risk/benefit profile of medicines)? • Can a move toward increased public data sharing jeopardize ongoing efforts toward improved regulatory harmonization?

- 298 -

Wrap Up and Closing Remarks

Barbara Bierer, M.D MRCT

Mark Barnes, J.D. MRCT

Rebecca Li, PhD MRCT

Holly Lynch, J.D. Petrie-Flom Center