Enabling the Molecular Medicine Revolution in Cancer through Biomedical Informatics

1
Enabling the Molecular Medicine Revolution in
Cancer through Biomedical Informatics

• Ken Buetow, Ph.D.NCI Associate
  DirectorBioinformatics andInformation Technology

U.S. DEPARTMENT OF HEALTH AND HUMAN
SERVICES National Institutes of Health
2
(No Transcript)
3
Cancer is a Complex Adaptive System
base state
selection
selection
selection
mutation
malignantstate
mutation
mutation
4
Advanced Technologies are needed to characterize
disease
base state(s)
malignantstate(s)
Mutationstatus
Alleleloss
Constitutionalvariation
RNAexpression
Epigeneticvariation
5
caBIG Pilot Imperatives
Integrate the biological and clinical silos
Integrate IT infrastructure, software and data
Integrate institutions and people
Address the complexity of cancer
6
caBIG NCIs Answer to the Infrastructure
Challenge
7
Scientific Discovery Opportunities

• Identify the biomarkers that predict efficacy
  of a new cancer treatment.
• Correlate molecular profiles with clinical
  outcomes.
• Identify a new cancer subtype.
• Discover indicators that predict disease
  progression.

8
Molecular Medicine as a Complex Continuum
Molecular Medicine
Imaging
Clinical Research
Pathology
Molecular Biology
9
The People
10
The People
11
The People
12
The People
13
The Activities
Clinical TrialManagement
14
The Activities
Image Sharing Analysis
Cross-reference image archive to improve
detection and diagnosis.
15
The Activities
Tissue Banking
Collect, process, annotate, archive and
disseminate tissue samples from patients.
16
The Activities
Molecular Profiling
17
Cancer Center Landscape

• Integrated Systems
• Homegrown/Commercial
• Smooth navigation between applications
• Difficult to expand/extend
• Large IT staff
• 10Ms invested

• Heterogeneous Systems
• Complex mix of commercial and homegrown
  components (may be composed of dozens of
  components)
• No common interfaces
• Medium size IT staff
• 1Ms invested

• Informal/ no systems
• Use of productivity applications (e.g. Excel,
  Access)
• Complex manual processes
• Small or no IT staff
• 100Ks invested

18
caBIG Approach

• Modules that address specific needs
• Electronic/remote data capture
• Adverse Event Reporting
• Regulatory Reporting
• Hospital Information Systems Interfaces
• Trial lifecycle management
• Connect through defined Electronic interfaces
• Use of international data standards

19
Systems Interoperability Harmonization
One harmonized standard/model Biomedical
Research Integrated Domain Group (BRIDG) v1.0
(6/26/07)
20
Boundaries and Interfaces

• not on the internal details of how focus on
  boundaries, interfaces, how things fit together,
• once theyre built assume that will be diverse
  changing

21
Standards-based interoperability caCORE
biomedical objects

• Community driven
• Dynamic implementation
• Built to be upgraded as standards harden, and
  domains expand

common data elements
controlled vocabulary
22
Standards infrastructure and services

• Enterprise Vocabulary Services (EVS)
• Browsers
• APIs
• cancer Data Standards Repository (caDSR)
• CDEs
• Case Report Forms
• Object models
• ISO 11179 model
• caGrid
• Globus
• Mobius
• Introduce
• Grouper
• Dorian
• ActiveBPEL
• Developer Toolkits
• caCORE SDK

23
caGrid 1.0 Conceptual View
24
Grid of Grids
Bilateral Negotiations
NCRI ONIX
NHLBI CVRN
NCI caGrid
25
caBIG Product Suites

• Electronic Clinical Trials Management
  Applications
• Connecting through caBIG and its biomedical
  research applications
• Security and Data Sharing

26
CTMS Bundle

• The CTMS Bundle brings together a range of
  interoperable tools supporting the clinical
  trials enterprise.
• Functions include
• Patient Study Calendar (PSC)
• Participant Registry (C3PR)
• Adverse Event Reporting (caAERS)
• Clinical Source Data Integration (caXchange)
• Integration with Cancer Central Clinical Database
  (C3D), or with commercial clinical trials data
  collection tools at sites

CTMS Bundle
27
CTMS Bundle Participant Registry (C3PR)

• Tracks subject registrations to clinical trials
• Verifies registration criteria (study open,
  participant eligible, consent received)
• Stratifies subject into a stratum group,
  randomizes to the trial
• Tracks participants across sites (handles
  multi-site trials)
• Manages study personnel
• Reporting (federal/local requirements, supplies
  NCI Summary 3/4 data)

28
CTMS Bundle Clinical Source Data Integration
(caXchange)

• Enables automatic transfer of clinical data from
  point-of-care systems in medical centers, e.g.,
  clinical chemistry lab systems
• Accumulates results in a standards-based data
  warehouse with defined electronic interfaces
• Translation of multiple source data formats into
  standards-compliant data for use in clinical
  trials
• Incorporates Viewer enabling viewing and
  selection of data

29
Commercial clinical trials tools are an
importantpart of the caBIG CTMS Bundle concept

• Velos Comprehensive clinical trials system in
  use in the extramural Cancer Centers throughout
  the country.
• PercipEnz A comprehensive solution for managing
  all aspects of clinical research study setup
  and activation, scientific reviews, subject
  registration, compliance tracking, visit
  tracking, data collection, data and safety
  monitoring, financials management, data
  extraction, regulatory reporting, and outreach.
• Akaza Rsch Web-based, open source software
  platform for managing multi-site clinical
  research studies. It facilitates protocol
  configuration, design of case report forms,
  electronic data capture, retrieval, and
  management.

30
Biomedical Informatics Bundle

• The Biomedical Informatics Bundle brings together
  a range of caGrid-interfaced tools supporting
  biomedical informatics
• Functions include
• Tissue Banking (caTISSUE Suite)
• Gene Expression Database (caArray)
• Translational Medicine tools (caIntegrator)
• Biomedical Image Management (NCIA)
• Array analysis (geWorkbench)
• and the supporting caGrid infrastructure

Biomedical Informatics Bundle
31
Biomedical Informatics Bundle caTISSUE
Product Description caTissue Core is caBIG's
tissue bank repository tool for biospecimen
inventory, tracking, and basic annotation.
Version 1.1 of caTissue permits users to track
the collection, storage, quality assurance, and
distribution of specimens as well as the
derivation and aliquotting of new specimens from
an existing ones (e.g. for DNA analysis). It also
allows users to find and request specimens that
may then be used in molecular, correlative
studies.
Current Version Number Version 1.1 Release Date
of Current Version February 2007 caBIG
Compatibility Level SilverMaturity Assessment
Stable Release
32
Biomedical Informatics Bundle caARRAY
Product Description caArray is an open source
microarray data management system that allows
users to submit, annotate and download microarray
data. caArray was developed using the caBIG
compatibility guidelines, as well as the
Microarray Gene Expression Data (MGED) society
standards for microarray data. Compatibility with
these standards and guidelines will facilitate
data sharing and integration of diverse data
types including clinical, imaging, tissue and
functional genomics data. A number of analytical
tools that connect to caArray are already
available, including geWorkbench and GenePattern
that both provide a variety of data analysis,
visualization and annotation functions for
microarray and other data types.
Current Version Number Version 1.4 Release Date
of Current Version October 2006 caBIG
Compatibility Level SilverMaturity Assessment
Stable Release
33
Biomedical Informatics Bundle NCIA
Product Description The National Cancer Imaging
Archive (NCIA) is a searchable, national
repository integrating in vivo cancer images with
clinical and genomic data. NCIA provides the
cancer research community, industry, and academia
with public access to DICOM images, Image
markup, Annotations, and rich meta data.
Current Version Number Version 2.2 Release Date
of Current Version January 2007 caBIG
Compatibility Level SilverMaturity Assessment
Mature Product
34
Biomedical Informatics Bundle geWorkbench
Product Description geWorkbench provides an
innovative, open-source software platform for
genomic data integration, bringing together
analysis and visualization tools for gene
expression, sequences, pathways, and other
biomedical data. It gives scientists transparent
access to a number of external data sources and
algorithmic services, combining these with many
built-in tools for analysis and visualization (at
present more than 40 distinct analysis and
visualization modules are part of the platform).
Current Version Number Version 1.0.4 Release
Date of Current Version August 2006 caBIG
Compatibility Level In process Maturity
Assessment Stable Product
35
Biomedical Informatics Bundle caIntegrator
Product Description caIntegrator is a novel
translational informatics platform that allows
researchers and bioinformaticians to access and
analyze clinical and experimental data across
multiple clinical trials and studies. The
caIntegrator framework provides a mechanism for
integrating and aggregating biomedical research
data and provides access to a variety of data
types (e.g. Immunohistochemistry (IHC),
microarray-based gene expression, SNPs, clinical
trials data etc.) in a cohesive fashion.
Current Version Number Version Release Date of
Current Version caBIG Compatibility Level
SilverMaturity Assessment Stable Release
36
Data Sharing and Intellectual Capital

• The DSIC Bundle provides a critical range of
  processes, procedures, policies and template
  agreements that provide a framework for
  collaboration
• Bundle includes
• Master Guidance Document
• Flow document and questionnaire
• Decision tree
• Template agreements for MTA, IRB, etc.
• Security policies, procedures, and a framework
  for caGrid-wide authorization
• and a framework for participating in the DSIC
  process, refining the structure for data sharing
  throughout the program

DSIC Bundle
37
Scientific Discovery Realized

• Identify the biomarkers that predict efficacy
  of a new cancer treatment.
• Correlate molecular profiles with clinical
  outcomes.
• Identify a new cancer subtype.
• Discover indicators that predict disease
  progression.

caBIG Enables Molecular Medicine
38
NCI-based caBIG Support

• Application Support
• E-mail
• Phone support
• List Servers
• caBIG boot camps
• Developers
• Application Users
• Online Interactive Training
• Down-loadable User Materials
• Training Sessions at Scientific Meetings

39
(No Transcript)
40
(No Transcript)
41
(No Transcript)
42
Built For the Community, By the Community

• Expanding the Network
• More caBIG compatible systems and tools
• Resources for adoption and support
• New partners from IT and biomedicine
• Extension to other health categories

2007
2006

• Delivering Software Tools and caGrid
• 40 software products delivered
• caGrid 1.0 launched December 18th
• 900 active participants at 80 institutions
• New partnerships private sector, regions,
  Federal agencies

• Establishing Connectivity
• Connectivity achieved between pilot nodes of
  caGrid
• Pre-existing software retrofitted for caBIG
  compatibility
• caBIG compatibility embedded into NCI Advanced
  Technology programs, Cancer Centers, and external
  product development activities

2005
2004

• Building Community
• caBIG pilot launched - February 2004
• Project plans developed and Working Groups
  established
• Standards conventions determined
• First generation software tools developed

2003

• NCI studies the IT challenges and develops
  strategic plan for a large-scale bioinformatics
  network

43
Current caBIG Governance
44
Ongoing Funding Developers, Adopters and
Participants
Developers
Adopters
Participants

• Developers
• Funding continues to support developing or
  modifying interoperable tools (e.g., software,
  infrastructure) that meet new community or
  scientific research needs

45
Ongoing Funding Developers, Adopters and
Participants
Developers
Adopters
Participants

• Adopters
• Funding continues to support the development and
  adoption of tools and applications for use in
  settings different from those in which they were
  developed

46
Ongoing Funding Developers, Adopters and
Participants
Developers
Adopters
Participants

• Participants
• Funding continues to support specific, targeted
  activities such as mentoring others in data
  model and tool development, software development,
  documentation or training activities
• Additional activities might include contributing
  to white papers in strategic, policy or
  technology areas, such as patient privacy or
  security architecture

47
(No Transcript)
48
Facilitating Next Generation Adoption
What
Who
Program Offices
Knowledge Centers
Service Providers

• Services open to all caBIG institutions
• Broad technical service support

• caBIG certified 3rd party support
• Partner with other groups for best customer
  service

Ongoing Tool Development, Adoption and
Participation
49
http//pid.nci.nih.gov
50
http//caintegrator.nci.nih.gov/cgems
51
(No Transcript)
52
(No Transcript)
53
http//caintegrator.nci.nih.gov/rembrandt
54
Glioma Molecular Diagnostic Initiative GMDI
The goal of the GMDI is to create a publicly
accessible web-based glioma data base, and
informatics platform consisting of in depth
pathologic, molecular and genetic data with
detailed clinical corollary data for hundreds of
individual brain tumors.
Data base should be invaluable for basic
scientists for aiding in tumor lineage
determination gene discovery new target
identification and validation Data base may be
invaluable for clinical investigators for a
prognostically more meaningful classification
system. toward individual patient-based
therapy selection .
55
GMDI Questions to Be Addressed

• Does there exist a biological basis for the
  current glioma classification schemas based on
  gene expression profiling? Toward a molecular
  taxonomy of gliomas.
• Do genetic/molecular determinants help to
  identify subgroups of gliomas within major
  standard histological groups that might have
  biological and/or clinical significance?
• Toward patient-specific tailored therapy
• Do genetic/molecular profiles predict patient
  survival?
• Do genetic/molecular profiles allow one to
  stratify patients into more homogeneous groups
  for more accurate assessment of treatment
  efficacy in clinical trials.
• Targeted therapeutics (i.e. EGFR, PDGF
  inhibitors)
• Can we identify genes and pathways involved in
  gliomagenesis that might serve as novel molecular
  targets?

56
Data Integration via Rembrandt
57
Workflows

• Quick search
• Show K-M plot based on EMP3 expression
• Show gene expression profile plots
• Advanced search
• Select good survival and poor survival patient
  groups
• Perform class comparison analysis using T-Test to
  identify genes that are differentially expressed
  in these groups
• Explore annotations to identify relevance to
  molecular changes in gliomas
• Survival analysis in a targeted group of patients
• Select patients with EGFR over-expression and
  PTEN deletion
• Display K-M survival chart for this group vs.
  rest of the patient population
• Plot copy number data from patient DNA samples
  against physical genomic location
• Display PCA chart for GBM and normal samples

58
Survival data on patients with EGFR amplification
and PTEN deletion
EGFR up PTEN del
Rest of the patients
59
Perform higher-order statistical analysis on
genomic and clinical datasets
60
Plot copy number data from patient DNA samples
against physical genomic location
61
Rembrandt facts

• Number of unique patient samples in the database
  - 500
• Gene expression data points - 20 million
• Copy number data points - 35 million
• Registered users - 250
• Average unique users per month - 150
• Average time spent per session - 45 minutes
• Longest visit as of June, 2006 283 minutes

62
Join the effort!!!

• More information
• caBIG.cancer.gov
• Join caBIG effort
• caBIG.nci.nih.gov