2013 Cohort Consortium Annual Meeting

The annual Cohort Consortium meeting, sponsored by EGRP and the Division of Cancer Epidemiology and Genetics (DCEG), was held on November 18-19, 2013, at the NCI Shady Grove Campus in Rockville, Maryland. Project/Working Group meetings were also held during this time.


Agendas


Main Cohort Consortium Meeting Agenda

Tuesday, November 19, 2013
Time Event
11:30 a.m. - 12:00 p.m. Lunch
Note: All attendees are responsible for their own lunch and are encouraged to eat lunch in the conference room so the meeting can start promptly at noon.
12:00 p.m. - 1:00 p.m. Session I: Opening Session
Moderator: Anne Zeleniuch-Jacquotte, M.D., M.S.
12:00 p.m. - 12:10 p.m. Welcome and Introductions
Anne Zeleniuch-Jacquotte, M.D., M.S.
Professor, Department of Population Health
New York University School of Medicine
12:10 p.m.- 12:30 p.m. Diabetes and Cancer Initiative in the Cohort Consortium
Elio Riboli, M.D., HonFPH, FmedSci
Professor and Director, School of Public Health
Imperial College London, United Kingdom
12:30 p.m. - 12:45 p.m. Cohort Infrastructure Program Announcement
Daniela Seminara, Ph.D., M.P.H.
Senior Scientist and Consortia Scientific Coordinator
Epidemiology and Genomics Research Program
Division of Cancer Control and Population Sciences, National Cancer Institute
12:45 p.m. - 1:00 p.m. Sequencing Update
Stephen J. Chanock, M.D.
Director, Division of Cancer Epidemiology and Genetics
National Cancer Institute
1:00 p.m. - 2:00 p.m. Session II: Rapid Reports from Working Groups
Moderator: Julie E. Buring, Sc.D.
1:00 p.m. - 1:05 p.m. Julie E. Buring, Sc.D.
Professor of Medicine, Brigham and Women's Hospital / Harvard Medical School
Professor of Epidemiology, Harvard School of Public Health
1:05 p.m. - 1:15 p.m. BMI Mortality Pooling and Related Projects
Patricia Hartge, Sc.D.
Epidemiologist Emerita
Division of Cancer Epidemiology and Genetics, National Cancer Institute
1:15 p.m. - 1:25 p.m. Breast and Prostate Cancer Cohort Consortium (BPC3)
Peter Kraft, Ph.D.
Professor of Epidemiology and Biostatistics
Deputy Director, Program in Genetic Epidemiology and Statistical Genetics
Harvard School of Public Health
1:25 p.m. - 1:35 p.m. Human Papillomavirus (HPV) Infection and Risk of Head and Neck Cancer
Mattias Johansson, Ph.D.
Scientist, Genetic Epidemiology Group
International Agency for Research on Cancer, World Health Organization
1:35 p.m. - 1:45 p.m.

Ovarian Cancer Cohort Consortium (OC3)
Nicolas Wentzensen, M.D., Ph.D., M.S.
Senior Investigator, Division of Cancer Epidemiology and Genetics
National Cancer Institute

Liz Poole, Ph.D.
Instructor in Medicine, Channing Division of Network Medicine
Harvard Medical School and Brigham and Women's Hospital
1:45 p.m. - 1:55 p.m.

Premenopausal Breast Cancer Collaborative Group
Anthony Swerdlow, Ph.D.
Professor, The Institute of Cancer Research, United Kingdom

Hazel Nichols, Ph.D.
Research Fellow, Epidemiology Branch
National Institute of Environmental Health Sciences
2:00 p.m. - 2:15 p.m. Break
2:15 p.m. - 3:35 p.m. Session III: Transforming Epidemiology for 21st-Century Medicine and Public Health
Moderator: Muin J. Khoury, M.D., Ph.D.
2:15 p.m. - 2:30 p.m. Progress and Opportunities since Trends in 21st-Century Epidemiology Workshop
Muin J. Khoury, M.D., Ph.D.
Associate Director, Epidemiology and Genomics Research Program
Division of Cancer Control and Population Sciences, National Cancer Institute
2:30 p.m. - 2:45 p.m.

Model Open Access for Cohort Epidemiologic Data: Can We Learn from the World of Genomics?
Elizabeth M. Gillanders, Ph.D., M.P.H.
Chief, Host Susceptibility Factors Branch
Epidemiology and Genomics Research Program, Division of Cancer Control and Population Sciences, National Cancer Institute

Will talk about NIH-wide genomics data sharing and the recently announced Global Alliance and what it means for epidemiology cohorts in the context of sharing and pooling genomic and non-genomic data.
2:45 p.m. - 3:00 p.m.

Knowledge Integration Using Data Sciences Jennifer Couch, Ph.D.
Chief, Structural Biology & Molecular Applications Branch
Division of Cancer Biology, National Cancer Institute

Michelle Dunn, Ph.D.
Mathematical Statistician, Data Modeling Branch
Surveillance Research Program
Division of Cancer Control and Population Sciences, National Cancer Institute

NCI representatives on the NIH Big Data to Knowledge Initiative (BD2K) will talk about this NIH-wide initiative and how it could apply to epidemiology cohorts.
3:00 p.m. - 3:15 p.m.

Incorporating New Technologies (Mobile, Social Media, etc.) in the Setting of Multi-Decade Cohort Studies both for Collecting Data and Embedding Communications and Interventions
William Riley, Ph.D.
Chief, Science of Research and Technology Branch
Behavioral Research Program, Division of Cancer Control and Population Sciences, National Cancer Institute

Will talk about incorporating new technologies in the design of epidemiology cohorts and embedding communication, implementation, behavioral, and intervention research in cohorts.
3:15 p.m. - 3:30 p.m.

Using Epidemiology Cohorts to Accelerate Discovery and Validation of Biomarkers of Cancer
David F. Ransohoff, M.D.
Professor of Medicine and Clinical Professor of Epidemiology
University of North Carolina at Chapel Hill

Will summarize information from a collaborative workshop on using epidemiology cohorts to accelerate discovery, characterization and validation of cancer biomarkers.
3:30 - 3:35p.m.

Charge to the Breakout Groups
Muin J. Khoury, M.D., Ph.D.

How can we accelerate/foster: 1. more open access, 2. more knowledge integration using data science, 3. more rapid incorporation of new technology, 4. interventions and implementation efforts?
  • What are some opportunities?
  • What are the challenges?
  • Are there some initial pilot, validation, or methodologic efforts — or things that could be done in the short-term — that could be informative?
  • What type of long term effort/infrastructure would be needed to make this happen?
3:35 p.m. - 4:35 p.m. Session IV: Breakout Session (see rooms below)
 

Model Open Access for Cohort Epidemiologic Data: Can We Learn from the World of Genomics?
Room 6E 032/034
Chair: Julie E. Buring, Sc.D.
Professor of Medicine, Brigham and Women's Hospital / Harvard Medical School
Professor of Epidemiology, Harvard School of Public Health

Knowledge Integration Using Data Sciences
Room 2W 910
Chair: Anne Zeleniuch-Jacquotte, M.D., M.S.
Professor, Department of Population Health
New York University School of Medicine

Using Epidemiology Cohorts to Accelerate Discovery and Validation of Biomarkers of Cancer
Room 2E 908
Chair: William J. Blot, Ph.D.
Professor of Medicine, Vanderbilt University
Chief Executive Officer, International Epidemiology Institute

Incorporating New Technologies (Mobile, Social Media, etc.) in the Setting of Multi-Decade Cohort Studies both for Collecting Data and Embedding Communications and Interventions
Room 2W 912
Chair: Elio Riboli, M.D., HonFPH, FmedSci
Professor and Director, School of Public Health
Imperial College London, United Kingdom

4:35 p.m. - 5:15 p.m. Session V: Report Back from Breakout Groups (Room TE406)
Moderator: Muin J. Khoury, M.D., Ph.D.
 

(10 minutes each)

Julie E. Buring, Sc.D.
Anne Zeleniuch-Jacquotte, M.D., M.S.
William J. Blot, Ph.D.
Elio Riboli, M.D., HonFPH, FmedSci
5:15 p.m. - 5:45 p.m Session VI: Wrap up
Moderator: Anne Zeleniuch-Jacquotte, M.D., M.S.
5:15 p.m. - 5:25 p.m. NCI Cohort Consortium Governance Planning
Deborah M. Winn, Ph.D.
Deputy Director, Division of Cancer Control and Population Sciences
National Cancer Institute
5:25 p.m. - 5:45 p.m. Emerging Issues for Secretariat to Focus on Next Year
Open Microphone
5:45 p.m. Adjourn

Project Meeting Agenda

Monday, November 18, 2013
Time Meeting Name Room Number
9:00 a.m. - 11:00 a.m. Second Cancers Working Group 2W 912
10:00 a.m. - 12:00 p.m. Ovarian Cancer Cohort Consortium (OC3) 2E 908
1:00 p.m. - 3:00 p.m. PanScan (CLOSED) 2E 908
2:30 p.m. - 4:30 p.m. Diet Pooling Project (CLOSED) 2W 912
3:00 p.m. - 6:00 p.m. BMI and Cancer Incidence in the Cohort Consortium 2E 908
4:00 p.m. - 6:00 p.m. Lymphoid Malignancies Working Group 2W 910
4:30 p.m. - 6:30 p.m. Vitamin D Pooling Project of Breast and Colorectal Cancer (CLOSED) 2W 912
  2E 908 2W 910 2W 912
9:00 a.m.     Second Cancers Working Group
9:00 - 11:00 a.m.
9:30 a.m.    
10:00 a.m. Ovarian Cancer Cohort Consortium (OC3)
10:00 - 12:00 p.m.
 
10:30 a.m.  
11:00 a.m.    
11:30 a.m.    
12:00 p.m.      
12:30 p.m.      
1:00 p.m. PanScan
1:00 - 3:00 p.m.
[Closed - WG Members Only]
   
1:30 p.m.    
2:00 p.m.    
2:30 p.m.   Diet Pooling Project
2:30 - 4:30 p.m.
[Closed - WG Members Only]
3:00 p.m. BMI and Cancer Incidence
in the Cohort Consortium
3:00 - 6:00 p.m.
 
3:30 p.m.  
4:00 p.m. Lymphoid Malignancies
Working Group
4:00 - 6:00 p.m.
4:30 p.m. Vitamin D PP of Breast and Colorectal Cancer
4:30 - 6:30 p.m.
[Closed - WG Members Only]
5:00 p.m.
5:30 p.m.
6:00 p.m.    
6:30 p.m.      
Tuesday, November 19, 2013
Time Meeting Name Room Number
8:30 a.m. - 10:30 a.m. Liver Cancer Pooling Project and Biliary Tract Cancers Pooling Project 2W 914
8:30 a.m. - 10:30 a.m. Lung Cancer Cohort Consortium (LC3) 2W 030
9:00 a.m. - 11:00 a.m. Breast and Prostate Cancer Cohort Consortium (BPC3) (CLOSED) 6E 032/034
9:00 a.m. - 11:00 a.m. Pooled Analysis of Time Since Last Birth and Breast Cancer Subtype; and Pooled Analysis of Risk Factors for Premenopausal Breast Cancer 3E 030
10:00 a.m. - 11:00 a.m. African American BMI and Mortality Pooling Project (CLOSED) 5W 030
10:30 a.m. - 11:30 a.m. Kidney Cancer Working Group 2W 030
  2W 030 2W 914 6E 032/034 3E 030 5W 030
8:30 a.m. Lung Cancer Cohort Consortium (LC3)
8:30 - 10:30 a.m.
Liver Cancer Pooling Project and Biliary Tract Cancers Pooling Project
8:30 - 10:30 a.m.
     
9:00 a.m. Breast and Prostate Cancer Cohort Consortium (BPC3)
9:00 - 11:00 a.m.
[Closed – WG Members Only]
Pooled Analysis of Risk Factors for Premenopausal Breast Cancer; and Pooled Analysis of Time Since Last Birth and Breast Cancer Subtype
9:00 - 11:00 a.m.
 
9:30 a.m.  
10:00 a.m. African American BMI and Mortality Pooling Project
10:00 - 11:00 a.m.
[Closed – WG Members Only]
10:30 a.m. Kidney Cancer Working Group
10:30 - 11:30 a.m.
 
11:00 a.m.        

Return to Top

Meeting Summary

This summary reflects the open portion of the meeting conducted during the afternoon of November 19, 2013.

  • Session I: Opening Session
    • View Session I Details

      Introduction

      Anne Zeleniuch-Jacquotte, M.D., M.S.

      The Cohort Consortium is growing and now includes 50 cohorts. More than 130 participants attended this meeting. Since last year, three cohorts joined the Consortium: the Breast Cancer Surveillance Consortium (BCSC); Atherosclerosis Risk in Communities Study (ARIC); and CARTaGENE.


      Tumor Block Collection Survey

      Anne Zeleniuch-Jacquotte, M.D., M.S.

      The Cohort Consortium conducted a survey of investigators to determine how many were collecting tumor blocks. Of the 40 cohorts that completed the survey, 16 reported collecting tumor blocks. Five of these cohorts are in the pilot stage of collecting tumor blocks. Eleven breast, eight colorectal, seven ovarian, five lung, and five prostate cancer cohorts are collecting tumor blocks. Three rare cancer cohorts are collecting tumor blocks.

      Among the survey respondents, 22 cohorts indicated that they would consider collecting tumor blocks. Respondents noted that substantial effort would be required to obtain consent from study participants (i.e., reconsenting); locate and store specimens; satisfy Health Insurance Portability and Accountability Act (HIPAA) requirements; obtain Institutional Review Board (IRB) approval (at some locations); contact multiple hospitals with pathology laboratories/departments (this is particularly challenging for cohorts covering wide geographic areas); and/or develop collaborations with pathologists to assist with tumor specimen collection, supervise the creation of tissue microarrays (TMAs), and generate research ideas. Twenty out of the 22 cohorts indicated that they would require funding to support these activities.

      Despite this, the survey respondents expressed strong interest in participating in collaborative projects and working groups to establish best practices for tumor collection. Survey results will be posted on the secure Cohort Consortium portal.


      Diabetes and Cancer Initiative in the Cohort Consortium

      Elio Riboli, M.D., HonFPH, FmedSci

      The Diabetes and Cancer Initiative will examine the relationship between Type 2 diabetes and cancer risk and survival as well as the role of other risk factors (particularly body habitus) and cancer types in this relationship. Cohorts that are interested in participating in the Diabetes and Cancer Initiative were encouraged to contact Dr. Riboli (to date, 30 cohorts have agreed to participate in this Initiative).

      A large study of the relationship between diabetes and cancer may help to explain the many dichotomies that exist between pre- and post-menopausal breast cancer risk factors.

      Discussion

      One limitation of the Diabetes and Cancer Initiative is the primarily Caucasian study population. Dr. Riboli is discussing a possible collaboration with the Multiethnic Cohort Study to obtain a larger sample of non-Caucasian participants. Participants also recommended collaborating with the Asian Cohort Consortium. A collaboration with the Black Women's Health Study also has been considered because this study has a large number of diabetic participants. In addition, participants suggested linking the Diabetes and Cancer Initiative with the metabolomics project.


      Cohort Infrastructure Program Announcement

      Daniela Seminara, Ph.D., M.P.H.

      NCI is planning to reissue the Core Infrastructure and Methodological Research for Cancer Epidemiology Cohorts funding mechanism (PAR-11-167) , which was created to 1) assure continuity of funding for the cohort infrastructure, 2) provide targeted review, 3) support integration of innovative approaches, and 4) enhance transparency and data sharing.

      Since the PAR was implemented in 2012, cancer epidemiology cohorts supported by the Epidemiology and Genomics Research Program at NCI appear to be improving their cost effectiveness and are increasing number of grants and total spending. The average time period that these cohorts are funded is 12 years.

      The epidemiology cohorts include participants of all ages and races/ethnicities. Although disparities exist, EGRP is attempting to address disparities in representation of gender and racial/ethnic groups in these cohorts. There are efforts underway to increase enrollment of Hispanic participants in both national and international cohorts, and collaborations with the Center for Global Health to increase participation of cohorts in Spanish-speaking countries.

      The cancer epidemiology cohorts have become a resource for large genomic (and other -omics) studies and include large numbers of cancer survivors. The cohorts are used to examine a variety of treatments, including radiation. In addition, some cohorts enhance existing studies, such as the California Teachers' Study Biorepository.

      Dr. Seminara provided a brief overview of other initiatives that facilitate collaboration across cohorts, including the Data Management and Harmonization across Cancer Epidemiology Cohorts initiative. A workshop on data harmonization is planned for October 2014.


      Sequencing Update

      Stephen Chanock, M.D.

      This presentation outlined challenges and future directions of the Next Generation Sequencing in Cancer project. The upcoming conclusion of The Cancer Genome Atlas (TCGA) will present new opportunities to conduct the next generation of genomics research. Sequencing has uncovered many unexpected relationships between mutational rates and the nature of mutations for specific cancers. An important area for future investigation is the large "tail" of recurrent mutations, which may provide important epidemiologic information.

      Interest is increasing in somatic (as opposed to germline) genomics, but somatic mutations must be well-defined and transitioned into profiles that could be used as biomarkers or predictors of risk/outcome. Somatic genomics currently is focusing on the patterns of drivers/pathways that are evident in advanced cancers rather than mapping specific loci as with germline genomics. Somatic genomics can identify multiple targets for sequencing. Approximately 10,000 cases of a specific cancer subtype will be needed to begin to accurately identify the range of mutations. The Cohort Consortium could provide this number of cases with tissue specimens. The Center for Cancer Genomics is interested in developing large studies for further discovery of the tail of somatic mutations in specific cancers but with a strong focus on clinically annotated samples. In this regard, we also need to consider the value of epidemiological analyses linked to clinically annotated somatic characterization studies.

      There is a brisk and exciting discussion across NIH, specifically within NCI addressing the value of next generation sequencing in cohorts (e.g., NHGRI workshop in 2012 and Sequencing Center Projects). There is a concern about lack of precision and generalization and the proportion of cases versus controls. Investigators need to support sequencing in cohort studies and the development of methods for sequencing. The Sequencing Center Projects provides one opportunity to do this but requires well defined proposals. NCI also is developing big data for oncology genomics research through the Cancer Genomics Data Commons and Cloud Pilots which are about to be issued.

      Discussion

      Participants asked about the possibility of sequencing for tumor DNA to identify tumors at an early stage. This type of effort would require overcoming some of the technical challenges related to single cell sequencing and circulating tumor cells, DNA, and microRNA. NCI has supported a great deal of exploratory work in this area. Most of this work, however, has been difficult to replicate across centers and studies. Large-scale sequencing for DNA in early diagnostics and detection still is not ready for prime time.

      Participants also inquired about efforts to identify driver mutations as tumors evolve. The study of driver mutations examines the frequency at which a specific mutation occurs in a gene linked to a tumor type. Different driver mutations may be important as the cancer progresses. GWAS identified many potentially important mutations but these will require extensive time and effort to understand. The same is true for driver mutations. Pharmaceutical companies are especially interested in driver mutations to develop targeted therapies.

      Sequencing studies may provide an opportunity for developing cancer prevention strategies in the future. The more immediate opportunity, however, is to revisit susceptibility for specific tumor subtypes using new methods for subtyping and analysis. Members of the Cohort Consortium could examine these methods and technologies.

  • Session II: Rapid Reports from Working Groups
    • View Session II Details

      Body Mass Index (BMI) Mortality Pooling and Related Projects

      Patricia Hartge, Sc.D.

      This working group examines the relationship between BMI and physical activity (PA) and mortality and cancer incidence. Relevant study findings included:

      • A study that compared mortality in individuals with class 3 obesity with healthy-weight individuals found that class 3 obesity confers the same risk as current versus never smoking.
      • A pooled analysis of 20 prospective cohort studies of thyroid cancer found increased risk with increased BMI, waist circumference, and height but different risk relationships for different subtypes. It also found a positive relationship between prostate cancer and physical activity, which may be related to screening bias.
      • An examination of gall bladder cancer incidence found a strong increased risk with increased BMI, but the risk was attenuated by waist circumference.
      • A project that examined leisure time PA using pooled data from 12 cohorts. This analysis found that breast, colon, gall bladder, liver, head and neck cancer were inversely associated with PA.

      Participants specifically recommended examining the relationship between prostate cancer and PA in European cohorts (which may be less subject to screening bias), and whether the interaction between gender and PA is significant in predicting non-Hodgkin lymphoma (NHL) risk.


      Breast and Prostate Cancer Cohort Consortium (BPC3)

      Peter Kraft, Ph.D.

      BPC3 activities during the previous year included:

      • Conducting a genome-wide study of estrogen receptor negative (ER─) breast cancer and advanced prostate cancer, which resulted in the identification of several new loci for both types of cancer.
      • Contributing to the breast and prostate subprojects within the Discovery, Biology, and Risk of Inherited Variants in Breast Cancer (DRIVE) and Elucidating Loci Involved in Prostate Cancer Susceptibility (ELLIPSE) projects.
      • Characterizing single nucleotide polymorphisms (SNPs) emerging from other studies.
      • Examining gene-environment interactions and characterizing the joint effects of genes and the environment.

      Proposing an exome-wide association study to examine rare coding variation (GWAS examines common variation). This study was not funded, but BPC3 was able to genotype most of the proposed cases and controls for the OncoArray study.

      Future BPC3 activities likely will focus on smaller, R01-supported projects involving intensive analyses of existing data and focused assays.


      Human Papillomavirus (HPV) Infection and Risk of Head and Neck Cancer

      Mattias Johansson, Ph.D.

      This project aims to investigate whether antibody biomarkers against HPV infection can be used as pre-diagnostic biomarkers for head and neck cancer and anal cancer, two cancers that are on the rise in the U.S. This initiative is jointly led by Investigators from the NCI and the International Agency for Research on Cancer (IARC). Preliminary data from European and U.S. cohorts indicate that between 35% and ~50% of oropharyngeal cancers are seropositive for HPV16 E6, whilst being extremely rare among controls that remain cancer free (<0.5%). These initial data indicate that the HPV16 E6 antibody biomarker may be a highly informative and specific biomarker for predicting future oropharyngeal cancers, the head and neck cancer sub-site predominantly linked to HPV infection.


      Ovarian Cancer Cohort Consortium (OC3)

      Nicolas Wentzensen, M.D., Ph.D., M.S. and Liz Poole, Ph.D.

      The OC3 study will examine 1) risk factors for different histologic subtypes of ovarian cancer, and 2) surrogates for different etiologies such as tumor dominance (e.g., fallopian tube versus ovary) and fatality. A study goal is to refine the ovarian cancer risk model by examining different risk factor associations with subtypes and validating results in the cohorts. OC3 investigators are interested in conducting biomarker studies with the many prospective blood samples already collected. These investigators plan to seek additional funding for this undertaking.

      Several new projects are planned for OC3. The Consortium was able to include a subset of participant DNA in OncoArray for genotyping of new cases that will facilitate refinement of risk models. OC3 also is collaborating with the Ovarian Cancer Association Consortium (OCAC) to develop a base risk model. OC3 will provide data to estimate baseline bilateral salpingo-oophorectomy (BSO) rates. OCAC/OC3 data recently were used to compare pooled and meta-analysis methods for estimating ovarian cancer subtype risk with oral contraceptive use, parity, and tubal ligation. This analysis found no differences between pooled and meta-analysis estimates.


      Premenopausal Breast Cancer Collaborative Group

      Anthony Swerdlow, Ph.D. and Hazel Nichols, Ph.D.

      The Premenopausal Breast Cancer Collaborative Group plans to examine 1) risk factors for pre- and perimenopausal breast cancer, and 2) the etiology of premenopausal breast cancers by subtype and potential effect modifiers.

      The Collaborative Group is in the process of recruiting cohorts. As of the meeting, >15 cohorts in the NCI Cohort Consortium have indicated that they have the requisite data for the Premenopausal Breast Cancer study (e.g., reproductive history, physical activity, height and weight information) and wish to join.

  • Session III: Transforming Epidemiology for 21st-Century Medicine and Public Health
    • View Session III Details

      Progress and Opportunities since Trends in 21st Century Epidemiology Workshop

      Muin J. Khoury, M.D., Ph.D.

      In response to the recommendations from the last Cohort Consortium meeting in December 2012, NCI formed a working group to determine which recommendations can be implemented within the Consortium and which ones will require collaboration outside of NCI and the Consortium. NCI efforts to address the recommendations included:

      • Creating a Cancer Epidemiology Matters Blog.
      • Launching the Cancer Genomics and Epidemiology Navigator (CGEN), a tool for linking NCI-funded projects with project publications and thus allowing NCI to identify research gaps.
      • Developing an Epidemiologic Science Evaluation Metrics working group (PQRST). The National Heart, Lung, and Blood Institute (NHLBI) launched a similar metrics initiative and is currently collaborating with NCI.

      NCI is seeking ideas to increase the efficiency and effectiveness of cohorts. In the future, the Institute will focus on funding survivorship cohorts, mapping cohorts, and maximizing resources that promote collaboration across cohorts and data harmonization.


      Model Open Access for Cohort Epidemiologic Data: Can We Learn from the World of Genomics?

      Elizabeth Gillanders, Ph.D., M.P.H.

      NIH data-sharing policies and processes for improving access to epidemiologic data were discussed. Changes to the NIH policy have been proposed based on the GWAS data-sharing policy and the Federal Public Access Plan. These proposed changes were released for public comment, which closed on November 20, 2013.

      The number of studies contributing data to the Database of Genotypes and Phenotypes (dbGaP) is increasing daily, including extensive genotyping data. Data access requests to NCI also are increasing rapidly and the time taken to process data access requests has decreased. In addition, since 2008, the number of publications generated by dbGaP users has risen markedly.

      Because data sharing is particularly important in an era of reduced resources, Cohort Consortium members were encouraged to expand sharing of epidemiologic data or at least make available a minimal set of epidemiologic data. Sharing measures and several criteria for shared measures were also suggested.


      Knowledge Integration Using Data Sciences

      Jennifer Couch, Ph.D.

      The NIH Big Data to Knowledge initiative (BD2K) will focus not only on genomic data but on other -omics data, as well as imaging, exposure, and clinical data from electronic health records. NIH has supported the creation of big data sets but has been slower to develop tools and methods for analysis and storage of big data. The NIH Data and Informatics Working Group generated a report citing several challenges and opportunities relevant to big data such as organizing, managing, processing, locating, accessing, and shipping those data. In response to this report, NIH created BD2K.

      BD2K is currently exploring:

      • Ways to tag data so that they can be located online.
      • Developing a data index.
      • How best to support community-driven data standardization efforts.
      • Methods and software for analyzing, integrating, and visualizing data as well as compressing and storing data. The capabilities of major cloud providers and supercomputing facilities for data management and storage.
      • Opportunities to enhance training in the use of biomedical big data.
      • Establishing Centers of Excellence for Biomedical Big Data that will focus on one or more key areas of data science.

      Cohort Consortium members can participate in BD2K efforts by 1) responding to relevant Requests for Information (RFIs), 2) participating in videocast workshops on topics such as clinical data or cloud storage, 3) taking advantage of relevant funding opportunities (RFAs and PARs) that support big data research, and 4) using social media and other methods to disseminate big data information.

      Other NIH initiatives relevant to big data include the Division of Cancer Biology (DCB) Integrative Cancer Biology Program, the Interagency Modeling Analysis Group, the U01 on Bridging the Gap between Cancer Mechanism and Population Science, and the Citizen Science in Participatory Research initiative.


      Incorporating New Technologies (mobile, social media, etc.) in the Setting of the Multi-Decade Cohort Studies both for Collecting Data and Embedding Communications and Interventions

      William Riley, Ph.D.

      New methods, measures, and technologies to support behavioral science were discussed. The intensive, longitudinal monitoring that is facilitated by these methods, measures, and technologies offers many advantages for cohort studies such as high sensitivity in natural experiments and automated intervention delivery.

      Relevant technological advances include:

      • Item Response Theory (IRT) and computer adaptive testing (CAT), which allows item banking to more precisely and efficiently estimate the underlying trait. IRT also improves co-calibration across items, which is especially relevant to data pooling.
      • Ecological Momentary Assessment (EMA) randomly samples experiences throughout the day using personal digital assistants (PDAs) such as smartphones.
      • Passive Sensor Technologies (e.g., accelerometers, chips to monitor ingestion of medication) collect near-continuous data on many behavioral indices without requiring an active response from the participant.

      Technologies that have not been used extensively but show promise for behavioral research include autosense technologies for psychophysiology measures; chemical sensors to measure environmental exposures; and a portable device to analyze cells for biomarkers, DNA, bacteria, viruses, and drugs. Another promising technology is biosensors that may be implanted subcutaneously or in a tattoo. Important new modalities for voluntary data collection include citizen science, crowdsourcing, and opt-in Internet panels. These modalities collect data from individuals who, for the most part, are willing to share their data.


      Using Epidemiology Cohorts to Accelerate Discovery and Validation of Biomarkers of Cancer

      David Ransohoff, M.D.

      Leveraging epidemiology cohorts can help accelerate discovery and validation of biomarkers of cancer for screening purposes. A workshop on this issue was held in August 2013 to discuss why blood tests for cancer screening frequently are inaccurate and steps to address this problem.

      Dr. Ransohoff presented examples of problems with a proteomics assessment for ovarian cancer screening. This test appeared to have very high sensitivity and specificity in a comparison of individuals with and without ovarian cancer. The results of this study, however, were biased by the fact that the measurements for the cancer patients and those without cancer were performed on different days, and the signal generated by the mass spectrometer machine used to perform the measurements varied on different days and that variance was hard-wired into the specimens' results. Other promising tests have been developed in the past that turned out to be inaccurate due to biased comparisons resulting from samples taken at different clinics or other important comparison group differences, e.g., comparing older men (the cancer group) to younger women (the non-cancer group).

      An unbiased comparison, like one that might be obtained from cultivated cohort, can be used to assess validity of some cancer screening tests. A cultivated cohort design is basically a nested case-control study within a larger cohort. It is possible that biospecimens can be collected with no clear study question in mind and then the later study is designed on top of the biospecimen sample collection created within the initial cohort. Cohorts within the NCI Consortium provide opportunities to create appropriate biospecimen collections and accompanying data for studies for discovery and validation of cancer screening tests. The majority of these cohorts collect serial biospecimens prior to cancer diagnosis that could be used for discovery and validation studies. Two questions need to be answered:

      1. What effort would be required from the Consortium cohorts to utilize existing specimen collections and, in the future, to cultivate new collections?
      2. Is this effort worthwhile and feasible?

      Recommendations generated by the August meeting included:

      • Developing an NCI inventory of biospecimens; this project is currently underway.
      • Using a tool that facilitates collaboration and sharing. Templates for Material Transfer Agreements and Data Use Agreements exist at the NIH level and can be used for this purpose.
      • Developing models for cost recovery and methodology pilot studies to examine, for example, methods for use of discarded blood (as is done now in some HMOs) or improved collection and storage of new biospecimens.
  • Session IV: Breakout Sessions
    • Topics discussed in the breakout sessions were reported during Session V
  • Session V: Reports from Breakout Sessions
    • View Session V Details

      Model Open Access for Cohort Epidemiologic Data: Can We Learn from the World of Genomics?

      Julie Buring, Sc.D.

      The group concluded that sharing data is different from open access to data. The group noted two options for providing increased access to Cohort Consortium data: 1) make data available only to investigators within the Cohort Consortium, or 2) make data available to outside investigators. Group participants agreed that access to Consortium data should be expanded, but wanted Consortium members to participate in the decision process related to making these data more available. The group considered several different approaches for making Consortium data more available, but each approach had several challenges that would need to be carefully evaluated. The group therefore would like more time to discuss the issue of Consortium data release before it is mandated.

      Data release may be particularly problematic for cohorts that are bound by informed consent. Re-contacting large cohorts that have existed for several years to obtain consent for release of their data frequently leads to loss of participants. Strategies are needed for efficiently completing the necessary procedures and paperwork for the release of data from the different cohorts. For example, a small amount of money for a part time programmer would be critical to a successful, efficient release of Consortium data. Any efforts to expand access to Consortium data should begin with information that is most easily released.

      The group recommended a special Consortium meeting (virtual or in-person) to discuss the data release issue. The meeting product would be recommendations to NCI/National Institutes of Health regarding data release policies for different situations. The group also suggested publishing these recommendations. Many breakout group members likely would be willing to participate in the proposed meeting because of the huge potential impact on their work. All members of the Cohort Consortium should be invited to the special meeting on data access issues. The group did not specifically discuss establishing a Consortium committee or working group to address these issues.


      Knowledge Integration Using Data Sciences

      Anne Zeleniuch-Jacquotte, M.D., M.S.

      A discussion of knowledge integration technology is timely because of the new -omics data (e.g., metabolomics and proteomics) becoming available. Several potential sources of funding exist to support the development of methods and software needed to integrate and visualize the data. Funds also are available to develop infrastructure through a U01 supplement. In addition, a recently announced R21 for research on gene-environment interactions could be used to support data integration efforts.

      Legal and ethical issues related to knowledge sharing and integration vary by country. An ongoing challenge for investigators is protecting the privacy of study participants while collecting data that might identify those individuals. Secure data storage is critical. Data harmonization is an important priority at this point in time. The Westat data harmonization project could lead to the development of a common data model, which would benefit knowledge integration efforts.

      The group proposed developing an inventory of the various types of data sets collected by cohorts beyond genetic data (e.g., questionnaire, tumor registry, or Medicare data). The group also discussed other types of data that could be usefully integrated with Consortium data (e.g., data on the built environment and pollution). A knowledge integration effort would benefit from the involvement of statisticians, bioinformatics experts, and systems biologists.

      Participants noted that data integration is a moving target. They suggested conducting pilot projects to examine integration of various data sources from multiple levels (micro- to macro- levels). This idea would align with the recommendation for multi-level analyses across cohorts.


      Using Epidemiology Cohorts to Accelerate Discovery and Validation of Biomarkers of Cancer

      William J. Blot, Ph.D.

      This group discussed the collection and use of biospecimens. Serial collection of biospecimens is ideal, but many cohorts have collected biospecimens at only one point in time. The group suggested surveying the cohorts to obtain more detailed information about which cohorts have repeated collections of biospecimens and the timing between specimen collections (i.e., within X years of diagnosis). While many cohorts have collected blood or buccal cells for DNA extraction, members generally agreed that tumor tissue needs to be collected to examine both germline and somatic variations.

      The group distinguished between markers for early detection and markers of cancer risk. Even if cohorts collected biospecimens at a single point in time, nested case-control studies could examine the timing between sample collection and cancer diagnosis. Biomarkers associated with cancer diagnosis and collected five, ten, or more years before diagnosis could be considered good indicators of cancer risk, while those associated with cancer diagnosis and collected within a few years of the cancer onset could be considered potential markers for early detection.

      Group members discussed the possibility of the Cohort Consortium working together to share the burden of biospecimen collection and accelerate biomarker discovery efforts. The burden on single cohorts would be less and the involvement of multiple cohorts would encourage replication of findings.

      Group participants noted the complexity of statistical analyses in -omics research and methods that are being developed to analyze the huge data sets generated by this type of research. The Consortium might support -omics research by establishing an online portal where researchers could post questions or results of analyses that others could comment on. The Consortium also could assist with standardizing methods: for example, the bar coding of all biospecimens.

      The Consortium could help determine whether collection of archived tumor tissue should be routine for all/most cohorts or be done on a case-by-case basis, and whether to prospectively collect tumor tissue specimens. The group did not achieve a clear consensus regarding tumor tissue collection across the cohorts. Group members noted that the ideal approach is to collect biospecimens (including tumor tissue) for a specific research question, but generally agreed, however, that routine collection and storage of biospecimens is important to ensure that blood and tissue samples are not lost. On the other hand, requesting funds for routine (non-specific) biospecimen collection may be a handicap when the cohort investigators apply for U01 renewal because they cannot request funding for assays, only biospecimen collection, and reviewers tend to want a strong rationale for collecting biospecimens. Without a strong rationale for biospecimen collection, the application may receive a low score.

      Participants noted that, if the Cohort Consortium is to encourage routine collection of tumor tissue specimens, it should provide guidelines for collection and processing of biospecimens to ensure that the highest proportion of biospecimens can be analyzed using the latest technologies. Participants recommended examining BCAC Consortium standards for collection of tissue specimens. Many clinical studies also are engaged in both prospective and routine collection of biospecimens and likely can provide models for preparing biospecimens. These models would be useful for interpreting clinical information. To help ensure that biospecimen studies are quickly translated into clinical and etiologic outcomes, biospecimen collection efforts endorsed by the Cohort Consortium should align with similar efforts already underway across NCI.

      The group recommended sharing lessons learned by different cohorts regarding the collection, processing. and use of tumor blocks. The group proposed a meeting to discuss best practices for collecting tumor blocks and possibly involving societies of pathologists. Best practices and lessons learned could be published in a paper that would encourage collaboration with pathologists.


      Incorporating New Technologies (mobile, social media, etc.) in the Setting of Multi-Decade Cohort Studies both for Collecting Data and Embedding Communications and Interventions

      Elio Riboli, M.D., HonFPH, FmedSci

      Technologies and methods for collecting, using, and disseminating data generated by cohort studies generally have not kept pace with major advances in this area over the past 15 years. The group discussed new technologies and standards that would benefit big data collection efforts, and identified three key areas where improved technologies are needed: 1) objective measurement of behavioral factors (diet, smoking, physical activity, alcohol use); 2) measurement of low-level environmental exposures (in water or air or via occupational activities), ideally using portable devices; 3) measurement of physiological and clinical factors (e.g., metabolic, functional, fitness). Studies generally show that new technologies for data collection stimulate participation.

      An important area for investment will be bringing together new technologies for data collection. For example, one study is pilot testing real-time input of data from multiple clinical measures (e.g., ECG, spirometry) into the same server. Issues relevant to the use of new technologies for the collection, storage, and integration of big data include: 1) scalability – methods that can capture data for very large numbers of subjects; 2) cost – opportunities to fund large-scale data collection at reduced cost; 3) privacy – problems with social media and other technologies that change rapidly; and 4) feasibility and practicality of using new technologies.

      Group members considered possible future directions for new technologies (e.g., mobile phones, in-home sensors/monitors), particularly for combined collection of different types of information (e.g., food intake, blood pressure, pulse rate, movement). For example, a worldwide program for aging individuals is focused on technologies for monitoring health in elderly people. Another important area for future development is the measurement of exposures that vary over time.

      Group members discussed whether the Consortium should invest in new technologies only for new cohorts or consider investing in new technologies for use in existing cohorts. The feasibility of using new technologies in existing cohorts may need to be examined. Many existing cohorts are comprised of older individuals who may not respond well to new technologies.

  • Session VI: Wrap-Up
    • View Session VI Details

      NCI Cohort Consortium Governance Planning

      Debbie M. Winn, Ph.D.

      Because the Consortium has grown substantially, the Steering Committee will be discussing governance issues for the Cohort Consortium over the next few months. The Consortium is governed by a Secretariat that includes five extramural cohort investigators and NCI program staff (intramural and extramural). All members of the Secretariat have agreed to revisit the Consortium governance model to improve productivity, flexibility, efficiency, etc. The objectives of this effort will be to 1) better clarify roles of extramural and NCI staff members, 2) increase time that the Secretariat spends on strategic planning rather than day-to-day activities; 3) improve transparency of Secretariat activities and increase involvement of Consortium members; and 4) improve efficiency in facilitating research. The overall goals of this effort are to improve Consortium processes, better position the Cohort Consortium to obtain resources, improve understanding of future directions, and improve support of Cohort investigators. The Secretariat welcomes input from Cohort Consortium members and will keep members apprised of the process to improve governance.

Return to Top

Return to Top

Contact

If you have questions about the meeting, contact Nonye Harvey, M.P.H. at the National Cancer Institute.

Return to Top