CSS 2012 WG6 WhitePaper
- 1 Overview
- 2 Needs and Challenges: Identification and Prioritization
- 3 Frameworks for Collaboration on Prioritized Issues
- 4 Needs and Issues: SEND 3.0 Implementation
- 5 Summary and Next Steps
- 6 Disclaimer
White Paper development
On March 19-20, 2012, FDA and PhUSE co-hosted a Computational Science Symposium that brought stakeholders from all over industry and government to work collaboratively to solve common needs and challenges. The main objective of the meeting was to create a space for collaboration where experts can get together to identify issues, discuss possible solutions and ultimately devise and execute solutions. Several workgroups were created based on Agency's priorities and interests and energy from stakeholders identified during previous meetings. One of the workgroups was dedicated to improving nonclinical assessments and regulatory science by identifying key needs and challenges in the field and then establishing an innovative framework for addressing them in a collaborative manner among regulators and industry.
During the hands-on working session, a diverse group of experts from pharmaceutical industry, government, contract research organizations, and software and technology providers identified, discussed, and prioritized needs and challenges for nonclinical data users, determined the overlap among stakeholders, and then developed the framework(s) for how to develop and pilot solutions that address the top priority needs during the next year.
From the prioritization effort, the group created a framework for moving 5 projects forward to support nonclinical informatics efforts. These include
Endpoint Predictivity This group will identify and pilot 3 use cases to examine the feasibility and usefulness of relating nonclinical (nonhuman) to human clinical findings to assess predictive value. Historical Controls This group will focus on determining the feasibility and practical uses of standardizing historical control data Nonclinical Data Standardization Roadmap This group will canvas current data standardization efforts and needs for the future, develop a roadmap for future standardization efforts, and identify/pilot innovative approaches to standardization Data Integration Starting with pharm class and later expanding to other areas, this group will focus on determining key integration opportunities across nonclinical data silos Interorganizational SEND - This group will attack the issues of the transfer and handling of SEND datasets between organizations, including analysis of chain of custody and GLP considerations. Primary focus will be on CRO-Sponsor exchange for subchronic toxicity and carcinogenicity studies.
The expectation is that this working group will continue to meet and work throughout the next year on the identified projects. The workgroup will refine and further develop the concepts and frameworks and address how this collaboration will occur.
- FDA Perspective on the Initiative
There have been stunning advances in technology to provide effective solutions to data heavy projects in translational science, genomics, and business analytics to name a few. These advances have begotten fields and businesses that CDER can leverage to meet reviewer and regulatory science needs. However, currently, the vast majority of nonclinical data submitted to the Agency is in paper or ‘electronic paper’ (pdf) format. Recently, CDER has adopted the SDTM/SEND standard as the preferred, supported standard for relevant studies; currently SEND provides for a SDTM representation of animal general toxicology and carcinogenicity studies. The submission of study data formatted as SEND datasets will greatly increase the usability of the study data. However, for a multitude of other data and study types, data continues to be submitted in static and ‘locked in’ paper or pdf formats, continuing inefficiencies and missed opportunities. Additionally, the accessibility of the data is only the first step, an enabling step. The full benefit of standardized data can only be actualized when combined with systems, tools, and procedures that allow the user/reviewer to easily find answers to the scientific and regulatory questions that face them.
The FDA has a responsibility protect and enhance the public health. Within the context of drug application review, this means to perform effective reviews, efficient reviews, move regulatory science forward, & have informative communication with stakeholders.
To support this vision, computational science should be harnessed. Study data, metadata, and resource information must be rapidly accessible, searchable, stored and structured with context in a way that allows for correlations/comparisons with other data types and those data types must be accessed with the appropriate tools and techniques to answer relevant questions.
In practice this will entail different solutions for different data types – and for some may entail no changes at all. The solutions developed should be based upon questions related to providing an end state described above. To move forward towards this vision, CDER & CBER recognize that other entities share some of the same challenges. A partnership among stakeholders with similar needs and vision is better positioned to take advantage of innovation and a greater diversity of possible solutions. We envision collaborations that break down needs and challenges into the smallest bits, progress quickly, fail and pivot quickly, using small pilots to develop solutions that can then be broadly vetted.
- Industry Perspective of the Initiative
“Anything that helps us standardize for SEND will help us for many other things” ….. a sanofi-aventis R&D Senior Director
SEND, a robust example of a nonclinical data standardization tool, provides opportunities for industry sponsors to internally realize the benefits of aggregation, advanced query and data mining across a single study, a drug project or a portfolio of nonclinical studies. For more than a decade, as the density and volume of data collected to support the discovery and development of new drugs has increased exponentially, pharmaceutical industry scientists and laboratory operational staff have faced larger and larger data management challenges. Aggregation and curation of data collected from various sources, in various forms are heavy tasks in scientific environments without standards and associated tools to make the job of transforming data into knowledge and fast, efficient, highly repeatable process.
For more than a decade, many industries, including pharmaceutical R&D have had ambitions to warehouse and mine their data into knowledge, but run into the same high barriers: curating the existing data into a form that allows intelligent aggregation and establishing the standards to collect future data in multi-usable, integrate-able formats. Observation in Harvard Business Review article, circa 2002: “Senior managers quickly realized that global data definitions, which would facilitate information sharing among business units, would be difficult to implement.”
Still, at many, many meetings and symposiums, Industry participants speak about, and listen to, ambitions such this quote from an Emerging Safety Science Workshop Summary, circa 2008…. “These technologies have the potential to identify safety issues much earlier …, reducing the number of expensive clinical trials, leading to more promising research avenues, and decreasing the exposure of human subjects and patients to products with safety problems.”
Now with standardizations such as SEND, a plethora of data mining and analysis tools at the fingertips of Industry scientists and consortiums all around trying to utilize pre-competitive data for R&D benefits, the technical resources are available for addressing real scientific nonclinical data management and use challenges. It is time to strongly engage in using the tools, even if there is still some debris of the barriers here and there.
Needs and Challenges: Identification and Prioritization
- Collecting perceived needs and challenges
- Pre-meeting materials; request for feedback.
- Received feedback from x stakeholders (e.g. x industry, x CROs, and FDA)
- As reps from all stakeholders, core group prioritized the overall list from 37 comments to 11 priority areas that fell in 4 major categories (Categories: Data Sharing, Predictivity, Standardization, Data Integration).
- SEND Implementation Needs and Challenges were separated for discussion during the SEND track
- Pre-meeting materials; request for feedback.
Resulting 11 issues, by category:
- Data Sharing
- Sharing of data across industry (CSS6-0006) for potential data mining
- Historical Control (CSS6-0005)-Leveraging existing historical control information
- Lessons from Failed Compounds or Existing Drugs with New Safety Warnings (CSS6-0015)
- Endpoint Predictivity (CSS6-0033)-Relate nonclinical findings or patterns to clinical adverse events/outcomes
- Cardiovascular Data to Predict Phase 1 Outcomes (CSS6-0002)
- QSAR Databases (CSS6-0032)
- Roadmap for Nonclinical Study Data Standardization (CSS6-0030)
- Standardizing ‘omics Data, Biomarker, Biosignatures (CSS6-0034)
- Data Integration
- Integrating Pharm Classes with Study Data (CSS6-0037)
- Interconnectivity between Nonclinical Data Silos (CSS6-0017)
- Integrating Lot and Impurity Data with Study Data (CSS6-0035)
- Phase 1: Breakouts and Voting
- Break out into 4 groups to discuss and prioritize the 11 key needs
- Each group had 5 votes – allocate them against each need
- Tally group votes and report back to the overall group
- New ideas go in “parking lot” for further discussion
- Participate, Challenge, Debate; Report back in 20 minutes;
- Phase 2: Discussion/consensus agreement
- Votes tallied for the 11 needs
- Discussion over top 4 based on highest overall votes
Frameworks for Collaboration on Prioritized Issues
Process for Creating
- Groups formed for each of the 4 issues selected
- Participants chose the topic of interest and joined the groups
- Participants were asked to complete a template (describe)
- Each group identified quick win projects that could be completed in 1 year
- Participants were advised that
- There may be multiple streams under each project to ‘bite’ off small pieces.
- For many, the first task for the solution may simply be defining the problem (e.g. ‘omics) issue.
- Define accomplishments for next ~3-9mo.
Proposed framework and identified projects
Roadmap for Nonclinical Data Standardization
- Develop strategy to prioritize & maximize standardization efforts
- 1st deliverables:
- Develop priority list and define scope
- Create catalog of current standardization efforts
- Define and develop pilot for investigating nonclinical to clinical prediction
- 1st deliverable: Develop scientific use cases
- Determine value and feasibility
- 1st deliverable: List of prioritized perceived benefits of historical control data types and recommendations for next steps
Interconnectivity between Nonclinical Data Silos
- Breaking down data ‘silos’ to integrate nonclinical data
- 1st deliverables:
- Define “interconnectivity” – need to work with WG3.
- Develop list of Pharm Class attributes for starting point for piloting
- White Paper. Describe the identification and prioritization of needs and challenges – the collection, process, and outcomes.
- Call for pilot participation
- Broaden the participation in the Projects to ensure necessary stakeholders are represented.
- Published through wiki the draft charters developed during the meeting in 4 weeks.
- Communication strategy (2 weeks) with identified project co-leads.
Needs and Issues: SEND 3.0 Implementation
- In Scope - SEND-specific needs and challenges related to:
- Obtaining and Managing SEND creation capabilities
- Creating SEND datasets and Quality Control
- Creating submissions with SEND data
- Exchanging SEND datasets within industry
- Out of Scope -Creation of new SEND domain definitions
- Participants were asked to prioritize items for resolution
- For each issue with high priority, participants defined a project, identifying:
- Scope of problem to be solved
- Participant roles needed (the skills and organization types that are needed to solve the issue)
- Core team (leaders and committed members for each role)
- Participants were advised to divide high-value needs that will take a long time to complete into smaller achievable issues for prioritization
- The objective was to launch the first 4 groups at FDA PhUSE meeting and by end of April, launch Forum and Wiki
- Teams formed for SEND implementation issues around 4 primary areas of focus
Summary and Next Steps
FDA comments are an informal communication and represent the individual's best judgment. These comments do not bind or obligate FDA. The contents of this wiki are from the individual contributors and do not necessarily reflect the view and/or policies of the Food and Drug Administration, the employers of the individuals involved or any of their staff.