Cancer Intervention and Surveillance Modeling Network: Reissuance RFA CA09025

1
Cancer Intervention and Surveillance Modeling
Network Reissuance RFACA-09-025

• Eric J. (Rocky) Feuer, Ph.D.
• Chief, Statistical Research and Applications
  Branch, Division of Cancer Control and Population
  Sciences
• Program Director, CISNET

2
Why Continue CISNET?

• I. Formidable and growing gap between the rapid
  pace of innovation in biomedicine and our ability
  to harness it to improve public health
• There is no capacity or infrastructure to meet
  the tsunami of basic research discoveries and
  move these discoveries rationally into clinical
  application.
• Kathy Hudson, director of Genetics and Public
  Policy Center, Johns Hopkins (Health Affairs,
  2008)
• II. Maturation of modeling in cancer sites beyond
  the top 4
• Cervical
• Esophagus

3
III. Encourage Modeling to Address Issues Across
the Full Cancer Control Spectrum
NEW DIAGNOSIS
TREATMENT
DYING FROM CANCER
HEALTHY
LIVING WITH
POPULATIONS
OF CANCER
OF CANCER
CANCER
Availability of Care Access to Care Quality of
Life Comorbidity State-of-the-Art Care Complian
ce
Availability of Care Access to Care Quality of
Life Comorbidity Hospice/ End-of-Life
Care Cancer Mortality
Cancer Incidence
Behavioral Risk Factors Screening
Examinations Environmental Exposures
Quality of Life Comorbidity Recurrence Cancer S
urvival Cancer Prevalence


4
Examples of Questions Addressed by CISNET
Landmark paper What are the contributions of
screening and adjuvant therapy on declines in
breast cancer mortality?
US Preventive Services Task ForceWhat are the
optimal starting and stopping ages, periodicity,
and combination of screening modalities to be
recommended for colorectal and breast cancer
screening?
National Coverage Determinations for colorectal
cancer screening tests What should CMS
reimburse for new more effective screening
technologies?
What are the number of lung cancer deaths averted
due to tobacco control efforts in the last half
century?
Do prostate cancers dedifferentiate (change
Gleasons score) during their screen-detectable
preclinical phase?
5
Schema for the Translation of Medical Research
Evidence-Based Practice Guidelines
Population Health Impact
Health Applications
Health Practice
T1
T2
T3
T4
Discovery
US Preventive Services Task Force

• CISNET models provide a platform for evaluating
  the downstream consequences of decisions and
  strategies that are made in earlier phases.

• 9 New Areas for Exploration

6
Current Schema for CISNET Modeling
7
I. Upstream Modeling
8
II. Multi-Scale Modeling
9
III. Incorporating Genomic and Family History
Risk Profiles
Integrative Cancer Biology Program Division of
Cancer Biology
10
IV. Optimizing Biomarker Development Strategies
CISNET models can provide a set of tools for
EDRN investigators to

• Project the likely impact of screening tests of
  given sensitivity on disease-specific deaths
• Investigate how early in the preclinical period
  the test needs to become sensitive in order to
  produce a target benefit in terms of lives saved
• Given specified test characteristics, project
  benefits and costs associated with different
  regimens of screening

11
V. Translation of Trial Results into Clinical
Guidelines and Public Health Policy

• European (ERSPC) and US (PLCO) prostate cancer
  screening trials differ with respect to
• Screening protocols, test positive criteria,
  compliance with biopsy recommendations,
  treatment patterns
• ERSPC (efficacy trial established protocol for
  follow-up of abnormal results) PLCO
  (effectiveness trial individual physicians
  determine follow-up)
• Probable co-existence of overdiagnosis for some
  and mortality benefits for others will further
  complicate guidelines.

12
Comparative Effectiveness Research

• a rigorous evaluation of the impact of
  different options that are available for treating
  a given medical condition for a particular set of
  patients. CBO, 2007
• Modeling can integrate evidence, extend available
  evidence from intermediate to long term outcomes,
  and balance trade-offs
• E.g. Radical prostatectomy vs. conservative
  management for prostate cancer (survival benefit
  vs. urinary and sexual dysfunction)

13
VII. Suggesting Optimal Routes to Reduce Health
Disparities

• Moving beyond standard racial/ethnic
  characterizations
• Education/Income
• Insurance Status
• Geographic Disparities
• Search for the largest leverage points to reduce
  disparities in mortality rates as a function of
• Risk Factors smoking rates, obesity, other risk
  factors
• Screening rates, follow-up to abnormal screening
• Treatment, quality of care

14
VIII. Evaluation of Diagnostic Tests

• A large number of diagnostic tests (conducted
  for symptoms or for known disease) are not
  supported by empirical studies showing that they
  affect patient outcomes
• Example Assistance to CMS in making coverage
  decisions about indications for using PET imaging
• Which cancers
• Diagnosis, staging, restaging and monitoring
  response to treatment

15
IX. Interactive Policy-Level Decision Tools

• Allow cancer control planners and policy makers
  to explore the impact of varying key parameters
  involved in their decision making.

Colorectal Cancer Mortality Projections Website

• Ensure that the tools are understandable and
  relevant for target audiences.

16
Structure/Content of Applications

• Each application must be for a single cancer site
  (breast, prostate, colorectal, lung, cervical,
  esophageal)
• Multiple PI Applications Encouraged
• Coverage of the full cancer control spectrum
• A plan of reasonably comprehensive coverage of
  the cancer control issues amenable to modeling
  facing that cancer site
• Coverage of the 9 focus areas
• Coverage of every area and complete cancer
  control spectrum is not mandatory emphasis in
  different areas should be proportional to the
  importance of the area
• Include comorbidity (or life expectancy) if
  possible
• Anticipate emerging issues (release of major
  trial results)
• A plan of comparative modeling with mix of base
  cases, mini-base cases, and individual efforts
• Emphasis must be on model applications to
  important public health issues in the US context
• International issues can be included, but
  applications should explain how they can serve to
  inform U.S. policies in related issues

17
Modeling Groups

• Approximately 2-5 modeling groups per application
• Modeling groups work on a single or group of
  related models
• A single modeling group can include multiple
  institutions
• Applications may propose application, extension,
  refinement, and/or merging of existing models (de
  novo model development not supported)
• Applicants should carefully consider how to
  structure applications and how many groups to
  include to provide coverage of relevant areas,
  promote collaborative modeling, and also to
  facilitate communication, collaboration, and
  consensus
• Budget generally not to exceed 275K total
  costs in year 1 per modeling group (guideline not
  a rule) with 3 increases allowable in years 2-5
• Groups with limited role could have smaller
  budgets

18
Coordinating Centers

• One group (usually one of the modeling groups)
  also specified as a coordinating center
• Coordinating center institution is the submitting
  institution
• Duties of coordinating center
• Formulating, prioritizing, and coordinating work
  on base case and other questions (including
  outside requests with new funding opportunities)
  
• Negotiating common requests for outside data
  sources
• Consensus building and coordinating critical
  evaluation of disparate results
• Preparing inputs and collecting and processing
  common outputs for model comparisons
• Coordinating synthesis papers and group responses
  bringing together disparate information to inform
  policy makers
• Organizing conference calls and setting meeting
  agendas
• Voting member of Steering Committee and making
  sure that policies and procedures in their group
  are in compliance those of Steering Committee

19
Coordinating Centers

• Some coordination activities can be
  decentralized, e.g.
• Mini-base case with 2 modeling groups could be
  coordinated by one of the groups)
• Local arrangements for mid-year meetings
• Budget
• Generally not to exceed 125K in year 1 with 3
  increases after that (guideline not a rule)
• Discretionary Funds Years 2-5
• For centralized admin and coordination (up to 50
  of funds can be allocated for this purpose)
• Remainder is left open to
• Tap outside expertise
• Pay for access to data sources
• Provide funds for high priority efforts not
  originally anticipated
• Budget Year 2 up to 11 of Year 1 total budget
  of entire award with 3 increases after that

20
Budget Cap and Sample Budget

• 1.3M total costs in Year 1
• Sample Budget
• Year 1
• 3 modeling groups Budgets (250K, 275K, 300K
  with justification)
• Coordinating center _at_ 125K
• 950K total costs
• Year 2
• Basic Award (950K 1.03 978.5K)
• Discretionary funds (950K 0.11 104.5K)
• 1,083,000 total costs
• Year 3
• 1,083,000 1.03 1,115,490

21
Structure of Application Research Plan

• Maximum of 55 pages
• N1 Overall Objectives, Significance, Specific
  Aims
• 2-4 pages
• N2 Team Leadership and Coordinating Center
• 2-4 pages
• N3 Proposed Models and Previous Model
  Applications
• Overview and accomplishments of each model (if
  worked together joint accomplishements)
• Why models complement each other
• Summary table of cross models comparisions
• Summary table of model applications

22
Structure of Application

• N4 Proposed Model Extensions, Applications, and
  Comparative Modeling
• Summary table
• Row projects
• Columns items such as project title, areas of
  coverage, modeling groups included, timeline,
  coordinator
• Additional Sections (Not part of Research Plan)
• Resource Sharing Plan
• Up to date Model Profiles
• Active solicitation of model applications from
  outside groups
• Interactive decision making tools and/or
  executable versions of program
• Source Code in context of collaborations
• Leadership Plan