2019 Annual Meeting
The annual NCI Cohort Consortium meeting, sponsored by EGRP and the Division of Cancer Epidemiology and Genetics (DCEG), was held on November 18-20, 2019, at the NCI Shady Grove Campus in Rockville, MD. Project/Working Group meetings were also held during this time.
On this page...
Day 1: Monday, November 18, 2019
|11:00 a.m. – 6:00 p.m.||
Cohort Consortium Working Group Meetings
|11:00 a.m. – 12:30 p.m.||
Steering Committee Working Lunch
Day 2: Tuesday, November 19, 2019
|8:30 a.m. – 9:15 a.m.||
SESSION I: Opening SessionModerators
Dr. Kathy Helzlsouer, National Cancer Institute (NCI)
Dr. Celine Vachon, Mayo Clinic College of Medicine
|8:30 a.m. – 8:40 a.m.||
Welcome and Introductions
|8:40 a.m. – 8:50 a.m.||
Opening Remarks and National Institutes of Health (NIH) Updates
|8:50 a.m. – 9:05 a.m.||
Update on NCI DCEG Cohort
|9:05 a.m. – 9:15 a.m.||
|9:15 a.m. – 10:00 a.m.||
SESSION II: Strategic Planning for the Cohort Consortium: Updates
|9:15 a.m. – 9:35 a.m.||
Strategic Planning Updates and Summary of Results from 2018 Science Planning
|9:35 a.m. – 9:45 a.m.||Open Discussion|
|9:45 a.m. – 10:00 a.m.||Break|
|10: 00 a.m. – 11:30 a.m.||
SESSION III: Plenary Session - Return of Research ResultsModerator
Dr. Julie Palmer, Boston University School of Public Health
|10:00 a.m. – 10:30 a.m.||
Returning Genetic Research Results to Individual Participants: Who, What, When and How?
|10:30 a.m. – 11:00 a.m.||
Germline Genetics: An Institutional Experience With Research Results
|11:00 a.m. – 11:30 a.m.||Open Discussion|
|11:30 a.m. – 12:45 p.m.||Lunch|
|11:30 a.m. - 12:45 p.m.||
SESSION IV: Kick-off Session of the Cohort Consortium Associate Member CouncilModerator
Dr. Celine Vachon, Mayo Clinic College of Medicine
|11:45 a.m. – 11:50 a.m.||
Overview: Goals, Purpose, Charge
|11:50 a.m. – 12:10 p.m.||
Table Discussions with Senior PIs – Round 1
|12:10 p.m. – 12:30 p.m.||
Table Discussions with Senior PIs – Round 2
|12:30 p.m. – 12:45 p.m.||
|12:45 p.m. – 1:45 p.m.||
SESSION V: Poster and Networking Session
|1:45 p.m. – 2:00 p.m.||
|2:30 p.m. – 3:45 p.m.||
SESSION VI: Plenary Session: Methodological Issues in Cohort Studies Moderator
Dr. Roger Milne, Cancer Council Victoria, Australia
|2:00 p.m. – 2:30 p.m.||Integration of Geospatial Data into Existing Epidemiology Cohorts
Dr. Gianluca Severi, Inserm, France
|2:30 p.m. – 2:45 p.m.||Open Discussion|
|2:45 p.m. – 3:00 p.m.||Development of the Virtual Pooled Registry Cancer Linkage System
Ms. Castine Clerkin, North American Association of Central Cancer Registries, Inc.
|3:00 p.m. – 3:15 p.m.||Opportunities to Advance Subsequent Neoplasm Ascertainment in Survivors of Childhood Cancer
Dr. Gregory T. Armstrong, St. Jude Children’s Research Hospital
|3:15 p.m. – 3:30 p.m.||Cancer Ascertainment by U.S. Population-Based Cancer Registries, Self-Report and Death Certificates in the Nationwide U.S. Radiologic Technologists Cohort
Dr. Danping Liu, DCEG, NCI
|3:30 p.m. – 3:45 p.m.||Open Discussion|
|3:45 p.m. – 5:00 p.m.||
SESSION VII: Plenary Session – Grants: Best Practices and Lessons Learned in Supporting Consortial ProjectsModerators
Dr. Joanne Elena, NCI
Dr. Marc Gunter, International Agency for Research on Cancer (IARC), France
|3:45 p.m. – 4:00 p.m.||The Quest for the Holy Grail of Consistent Consortium Funding
Dr. Shelley Tworoger, Moffitt Cancer Center
|4:00 p.m. – 4:15 p.m.||Open Discussion|
|4:15 p.m. – 4:30 p.m.||The Experience of the Pooling Project on Alcohol and Cancer Consortium
Dr. Pietro Ferrari, IARC
|4:30 p.m. – 4:45 p.m.||Open Discussion|
|4:45 p.m. – 5:00 p.m.||
Wrap Up of Main Meeting
Dr. Celine Vachon, Mayo Clinic College of Medicine
|6:00 p.m.||Networking Social|
Day 3: Wednesday, November 20, 2019
|8:30 a.m. - 1:00 p.m.||
Cohort Consortium Working Group Meetings
Session I: Opening Session
Moderators: Drs. Kathy Helzlsouer and Celine Vachon
Welcome and Introductions
Drs. Kathy Helzlsouer and Celine Vachon
Dr. Kathy Helzlsouer welcomed participants and introduced the session moderator, Dr. Vachon, who is the current Chair of the Cohort Consortium Steering Committee. Dr. Helzlsouer also introduced the incoming 2020 Steering Committee Chair, Dr. Lynne Wilkens; Vice Chair and 2021 Chair-Elect, Dr. Roger Milne; and four new Steering Committee members joining in 2020. In addition, Dr. Helzlsouer recognized current Steering Committee members and the NCI organizers of this meeting. She announced that no election would be held for Steering Committee members in 2020. Dr. Gapstur also received an award for her work on strategic planning for the Cohort Consortium.
Dr. Helzlsouer announced two Notices of Special Interest (NOSI) that recently were released focusing on geospatial approaches in population science and the use of biological information to understand the interaction between genes and environmental exposures. (NOSIs are similar to Program Announcements, but shorter). She recommended that participants listen to or attend the joint meeting of the Board of Scientific Advisors (BSA) and the National Cancer Advisory Board to be held December 2–3, 2019, to learn about upcoming funding opportunities. Participants also were encouraged to review and comment on the draft NIH Policy for Data Management and Sharing and Supplemental guidance (NOT-OD-20-013), which expires January 10, 2020. In addition, participants should subscribe to the EGRP newsletter to receive updates.
Steering Committee Activities
Dr. Celine Vachon
Dr. Vachon discussed several activities of the Steering Committee over the past year, including finalization of new bylaws and strategic initiatives. As Chair of the Steering Committee in 2019, Dr. Vachon focused on forming the Associate Member Council (AMC), which held its first networking event at this meeting. The purpose of the AMC is to engage and promote the professional development of all early-stage investigators in the Cohort Consortium. The AMC now has six members who have developed a charter. Dr. Vachon asked Consortium PIs to identify other early-stage investigators on their teams who should receive communications from the AMC.
The Steering Committee releases bimonthly newsletters to communicate with the membership, including information explicitly targeted to subgroups. Another communication tool, the Cohort Consortium Project and Publications webpage, will be enhanced over the next year. In addition, the Cancer Epidemiology Descriptive Cohort Database questionnaire is being updated and Dr. Vachon expects those updates to be implemented before the 2020 Cohort Consortium annual meeting.
The Working Group (WG) Subcommittee performs an ongoing review of WG projects and activities. Dr. Vachon emphasized the importance of WGs responding promptly to requests to provide updates on their projects. NCI and the Steering Committee need to report on the activities and progress of the WGs to justify the Cohort Consortium. In 2020, the Steering Committee will focus on identifying active WGs as opposed to those waiting for funding or those that no longer are active.
The Cohort Consortium now has 60 cohorts, including 4 new cohorts that recently joined. Based on the latest information received by the Steering Committee, at least 48 WGs or projects are active. Three new WGs were formed in 2019, including one focused on methodology.
Opening Remarks and National Institutes of Health (NIH) Updates
Dr. Robert Croyle and Dr. Stephen Chanock
Dr. Helzlsouer introduced Dr. Robert Croyle, Director of the Division of Cancer Control and Population Sciences (DCCPS) at NCI. She also introduced Dr. Stephen Chanock, Director of NCI’s Division of Cancer Epidemiology and Genetics (DCEG).
Dr. Croyle made the following announcements:
- NCI’s budgetary goal is to raise the R01 payline by 1% this fiscal year.
- Dr. Norman Sharpless returned as Director of NCI last week.
- Many Cancer Moonshot activities are underway. Over the past 2 years, the Moonshot has allowed DCCPS to engage in activities beyond the usual population science initiatives.
- DCCPS is focusing on population health guidelines and incentives for the Cancer Centers program.
- DCCPS has developed new informational materials that are available on the Division’s website.
- The NCI Center for Global Health will have a new director, Dr. Satish Gopal, from the University of North Carolina Lineberger Cancer Center. Many new global health priorities will accompany the change in Center for Global Health directors. New Center initiatives will be presented at the next BSA meeting.
Dr. Chanock highlighted the importance of supporting new investigators, which is the objective of the AMC, as well as acknowledging the Cohort Consortium in publications. He noted the report submitted to the BSA earlier this year focusing on the importance of cohorts.
Dr. Chanock also emphasized the importance of initiatives that cut across disciplines, programs, and projects. DCEG has established trans-branch initiatives to stimulate conversations about common interests. Many research areas cut across disciplines and projects, such as data science and management, the relationship of early life and maternal factors to cancer across the life course, multiple primary cancers, translational epidemiology, and new descriptive epidemiology methods using geographic and other new tools.
DCEG has two new Moonshot projects focused on the human papilloma virus (HPV). The first is the One Versus Two Dose Non-Inferiority Trial in Costa Rica, which has enrolled a cohort of approximately 18,000 adolescent girls. The other project is using artificial intelligence to examine biomarkers of cervical cancer risk in women infected with HPV.
Finally, Dr. Chanock noted that DCEG is supporting a large prospective etiologic cohort—CONNECT—and he encouraged all participants to collaborate on this effort.
Update on NCI DCEG Cohort
Dr. Montserrat Garcia-Closas
Dr. Garcia-Closas delivered an update on the CONNECT project. CONNECT is creating a resource using new technologies and methods for the study of cancer etiology, from precursors to tumor progression and from cancer risk prediction through early detection to second cancers and survivorship. CONNECT is in the early stages; investigators are finalizing the study design and have not begun recruiting participants. The study will collect serial tissue specimens from various sources, including tumors. The study also will link to electronic health records (EHRs) and cancer registries to obtain complete case ascertainment. In addition, data will be collected through mobile and wearable technologies.
Several health care partners are participating in the study at nine locations across the United States: HealthPartners (Minneapolis, MN), Henry Ford Health System (Detroit, MI), Kaiser Permanente Colorado, Kaiser Permanente Georgia, Kaiser Permanente Hawaii, Kaiser Permanente Northwest (Portland, Oregon), Marshfield Clinic Health System (Marshfield, WI), Sanford Health (Sioux Falls, SD), and University of Chicago Medical Center. Kaiser Permanente Georgia recently joined the study to represent populations in the southern United States. The racial/ethnic distribution of the population across the participating sites is representative of the U.S. population. The study will be able to identify emerging exposures because of the relatively young population that will be recruited.
A CONNECT goal is to receive all health care data through a single network, but EHRs are not well integrated at this point. Regional networks used by partner organizations therefore will transmit data. A series of questionnaires also will be used to collect various types of health-related information. Data standards will be developed for CONNECT because of the planned use of multiple data linkages, including the virtual tissue repository (VTR). CONNECT will be designed for data sharing from the outset, following FAIR (Findable, Accessible, Interoperable, Reusable) guidelines. All CONNECT code will be collocated with data and shared through a cloud-based system. CONNECT will employ NIH Science and Technology Research Infrastructure for Discovery, Experimentation, and Sustainability (STRIDES) vetted cloud-managed services enabling a serverless epidemiological data commons that can be shared and used for future studies.
Participants noted that few of the 60 Cohort Consortium cohorts fully comply with FAIR guidelines. Participants recommended that the Consortium consider steps to achieve FAIR compliance across all cohorts. The Breast Cancer Risk Prediction Modeling project is examining a data structure that would adhere to FAIR principles. If this pilot project is successful, other Consortium cohorts could use its data structure as a model. Dr. Garcia-Closas would like to determine which cohorts already adhere to FAIR principles and which cohorts would be interested in applying a model to help them improve adherence to FAIR principles.
Participants asked what efforts DCEG is making to increase data sharing through the application of FAIR principles. DCEG is planning to develop an infrastructure that ultimately will ensure that all internal studies adhere to FAIR principles. Established cohorts already have made data available upon request, but accessing the data is not easy. A participant added that data from the Prostate, Lung, Colorectal, and Ovarian (PLCO) Cancer Screening Trial now are accessible in the cloud, so that downloading is not necessary.
A participant asked that cohort PIs ask their information technology (IT) staff to share the GitHUB link and protect time for IT staff to work together across cohorts. He also suggested that investigators continue to monitor the NIH Data Commons.
Data-sharing restrictions often impede adherence to FAIR principles, sometimes making it impossible for project staff to make data available in the cloud. The CONNECT feasibility study might highlight some strategies for overcoming these barriers to data sharing.
A participant asked that CONNECT investigators collaborate with similar cohorts to ensure that data can be harmonized by correlating or asking the same questions. She recommended creating an external advisory committee for CONNECT to develop shared questionnaires. That advisory committee would be composed of investigators from other cohorts with similar populations, research questions, and other characteristics. Dr. Garcia-Closas noted that CONNECT already has an external advisory committee, which includes investigators from other Cohort Consortium projects. The CONNECT questionnaire was developed with harmonization in mind, using many questions already used by existing cohorts, particularly for newer cohorts. Much of the work on the CONNECT questionnaires is expected to help improve alignment of variables across cohorts and support the creation of a Cohort Consortium data commons. Dr. Garcia-Closas is seeking ideas for ways to increase engagement of other cohort investigators and WGs in the CONNECT project.
Participants noted that data on planned recruitment of Latino participants were missing from Dr. Garcia-Closas’s presentation. She agreed to share planned recruitment tables that separate Latino ethnicity from race when this information is received from all participating CONNECT centers. CONNECT might need to include additional locations to increase the diversity of the cohort. In response to a related question about engaging non-English speakers, Dr. Garcia-Closas noted that all questionnaires will be translated into Spanish. The questionnaires also might be translated into other languages, but this option has not been discussed yet.
Participants asked how CONNECT would take advantage of or contribute to existing data commons. The planned data infrastructure should facilitate pooling data from any other study with data located in the cloud. For example, CONNECT investigators are working with All of Us project investigators to combine data prospectively.
Some participants indicated that CONNECT would be a useful model for future cohorts. They noted that the CONNECT data infrastructure will not only facilitate data sharing, but also make it easier for individual participants to withdraw their data.
Session II: Strategic Planning for the Cohort Consortium - Updates
Strategic Planning Updates and Summary Results From 2018 Science Planning
Drs. Susan Gapstur and Nonye Harvey
Dr. Gapstur announced that the Cohort Consortium recently completed a 2-year long strategic planning process to set priorities, focus energy and resources, and strengthen operations. She now is seeking Cohort Consortium members to lead the implementation of different components of the strategic plan. She noted that the Steering Committee also would be willing to consider other strategic priorities if a member was willing to lead the implementation of that strategy. For example, at this meeting a participant agreed to lead a WG to develop research on contralateral breast cancer and survivorship. A brief strategic plan document stating the Cohort Consortium mission and identified goals and strategies was distributed to all participants.
Many Cohort Consortium members actively participated in the strategic planning process during the World Cafés held at annual Consortium meetings. During these World Café sessions, participants discussed ways to enhance Cohort Consortium operations and identified best practices for operationalizing scientific activities, as well as high-priority scientific directions and research areas. Five goals emerged from the 2017 World Café discussions: communication, career development, research facilitation, leveraging cohorts to fill scientific gaps, and responding to common methodologic challenges. 2017 World Café participants were most interested in fostering communication among investigators leading cohorts. High-priority research areas identified during 2017 World Café discussions included circulating biomarkers, data linkages, biospecimen and tissue biomarker collection and linkage to other data, rare cancers, and survivorship. 2018 World Café discussions focused on the science, and participants were asked specific questions about the research areas identified in 2017. Many of the ideas generated during the 2018 discussions overlapped and were consolidated into the two scientific goals focusing on leveraging cohorts and common methodologic challenges.
Dr. Gapstur emphasized the importance of a bidirectional approach to implementing the strategic plan, using data to inform next steps. For example, cohorts that are not ready to address certain priorities in the strategic plan, such as survivorship, can examine their surveys to ensure that questions related to survivorship are included. Dr. Gapstur added that many strategic areas overlap.
Dr. Harvey discussed Cohort Consortium past and current strategic planning initiatives for 2018 through 2021. She highlighted the extensive ongoing and planned efforts toward achieving the communication goal, such as usability testing of the Cohort Consortium website and improving tracking and reporting of WG activities. Dr. Harvey emphasized the importance of WGs and announced that a new WG progress report form had been pilot tested with a few WGs. She also noted that bylaws will be reviewed annually and encouraged participants to review the bylaws regularly and provide recommendations for changes or updates. To enhance member engagement, NCI will work with the Steering Committee to host webinars on such topics as data sharing. Participants should submit suggestions for webinar topics to Dr. Harvey or Ms. Camille Pottinger.
Strategic planning efforts focusing on the research facilitation goal will include—
- An exploration of options for leveraging cloud-based technology to support data sharing.
- Developing templates for standardized Data Use Agreements (DUAs). Standard DUAs have been drafted by some WGs and should be shared.
Dr. Harvey announced a Call to Action for science priorities and asked for volunteers/champions to address the scientific gaps identified in the strategic plan and common methodologic challenges.
Dr. Gapstur clarified that more than one WG likely will cover such broad topics as Survivorship or Rare Cancers research. Volunteers do not need to lead a broad topic area. They should lead WGs on topics that interest and are a priority for them. WGs might have the opportunity to produce publications in some topic areas. Dr. Harvey added that AMC members should be involved in developing any publications.
Session III: Plenary Session - Return of Research Results
Moderator: Dr. Julie Palmer
Many cohorts now have shared samples and are collaborating to perform whole-genome or targeted DNA sequencing. These activities are generating data on potentially clinically actionable pathogenic/likely pathogenic gene mutations including some autosomal dominant with high penetrance. Increasingly, cohorts are having to make decisions about sharing research-based results of genomic data on mutations with participants and their families. Dr. Julie Palmer introduced Dr. Angela Bradbury, an expert in medical ethics and cancer genetics and Dr. Les Robison, the founding PI of the Childhood Cancer Survivor Study and co-PI of the St. Jude’s Lifetime Cohort Study.
Returning Genetic Research Results to Individual Participants: Who, What, When, and How?
Dr. Angela Bradbury
Dr. Bradbury discussed several considerations in determining whether and how to return genetic testing results to study participants. Researchers studying genetic mutations now are being confronted with information about clinically available information about genetic mutations. Doctors are using this type of information to evaluate risk and implement measures to prevent advanced cancers in affected patients and their family members. Researchers therefore feel an ethical obligation to return information about certain genetic mutations to study participants. Researchers traditionally have not shared genetic test results with study participants because of the separation between research and clinical care, potential confusion or disinterest regarding the results, and cost considerations. Many researchers now favor sharing genetic test results with study participants to benefit them and respect their autonomy, create a more reciprocal relationship between investigator and study subject, as well as because clinical care often is linked to cancer research. Some consensus has been reached on best practices, including the responsibility to return clinically actionable results that have been confirmed in a Clinical Laboratory Improvement Amendments (CLIA)-certified laboratory and the need to explain the study policies regarding the return of individual research results as part of the informed consent process. The American College of Medical Genetics has published a policy statement that makes recommendations for reporting incidental findings in clinical exome and genome sequencing and that lists 56 genes (23 of which are cancer susceptibility genes) that should be reported to participants as “incidental findings” with clinical sequencing.
The 2013 report of the Presidential Commission for the Study of Bioethical Issues provides guidelines for returning incidental research results to individual participants, as does a 2014 article published by Dr. Gail Jarvik and colleagues in the American Journal of Human Genetics. In 2018, the National Academies of Sciences, Engineering, Medicine (NASEM) also published guidelines that recommended returning actionable results and provided guidance for returning results not confirmed in a CLIA laboratory.
Dr. Bradbury discussed ways to apply existing guidelines to situations that might be encountered by Cohort Consortium investigators, using the RESPECT Studies and the Penn BioBank as examples. Questions considered by the investigators for these studies included the following: What results do you have? What kinds of results are you interested in returning to participants? Which results are you returning, and what is your rationale for returning them? Who is your cohort, and what did the original consent state (in the case of retrospective decisions to return results)? Who will need to approve sharing of results, especially when it involves a change in approach? Problems include changing Institutional Review Board (IRB) members, multiple IRBs, and institution decisions that can override IRB decisions. Early engagement of all stakeholders is important before deciding on an approach. Even when all stakeholders are engaged in developing well-justified procedures for returning results, these procedures should be modifiable and include plans for responding to potential problems.
Dr. Bradbury recommended that cohorts share their experiences in returning genetic test results. Lessons learned from the RESPECT study and Penn Biobank should inform other cohort studies. For example, the RESPECT study found that in-person and genetic counseling was no more effective than Web-based counseling, which presents a cost-saving opportunity. Scalable models for sharing genetic test results are needed.
Germline Genetics: An Institutional Experience With Research Results
Dr. Les Robison
Dr. Robison discussed his institutional experience of determining how to share germline genetic information with pediatric cancer survivors participating in the St. Jude Lifetime Cohort (SJLIFE). This study recruited patients at St. Jude Children’s Research Hospital, so separation between clinical care and research was less defined than most cohort studies. The study collects biospecimens at baseline and every visit, and thus consent forms and tissue banking protocols evolved over time with the changing practices for genomic research. With the success of the Pediatric Cancer Genome Project, a collaboration between St. Jude and Washington University, the SJLIFE investigators obtained approval to perform whole-genome and exome sequencing of the first 3,000 patients in the cohort.
The study is described in an article in the Journal of Clinical Oncology (Wang et al, 2018). The SJLIFE genomics project examined pathogenicity for 156 cancer predisposition genes and found that 11.5% of the cohort had a pathogenic or likely pathogenic mutation in a known cancer predisposition gene, with 5.8% having a mutation in one of 60 genes with autosomal dominant inheritance and high to moderate penetrance. The study found that patients with a pathogenic mutation had a significantly increased risk for second cancers, leading investigators to recommend the referral of all pediatric cancer survivors to genetic counseling.
Dr. Robison highlighted the initial plan to inform participants with a pathogenic/likely pathogenic mutation of their research results. However, before the plan was initiated, an institution-wide policy was put in place not to return any results but, instead, offered genetic testing to every patient and genetic counseling at no cost to the patient. Few are declining testing, but investigators still are awaiting response from most participants. Few participants responded to mail or email offers of testing, but 66% expressed interest in testing when contacted by telephone and 80% expressed interest when approached during their follow-up care visits.
Participants asked how investigators dealt with family members whose cancer risk also might be predicted by the participants’ genetic test results. Dr. Bradbury wanted to share information with relevant family members but had to defer to the participant’s decision. She recommended educating patients and their family members about when and how to obtain genetic testing. Regarding deceased patients, IRBs determine whether and how genetic test results will be shared with next of kin. The St. Jude’s study only examined living people and only cancer-related mutations. Dr. Robison added that hospital social workers can help obtain coverage for genetic testing of family members. Results demonstrating a pathogenic mutation in a relative provides evidence of medical necessity to insurers.
Investigators have not yet explored the reasons for 10% of St. Jude patients declining genetic testing when approached by their physicians. One possible reason is that the patients were able to make informed decisions following counseling. The investigators also are examining anxiety experienced by patients pre- and post-testing. Participants questioned whether the offer to test other family members could lead some patients to decline genetic testing.
Participants discussed the chain of custody of DNA samples and how investigators can ensure that the samples belong to the correct individuals after many years. The RESPECT investigators had a relatively high level of confidence that the samples matched the correct participants, but they were prepared for discordant results. Dr. Bradbury recommended that cohort investigators who are concerned about chain of custody confirm results at a CLIA laboratory first. Her paper discusses the confirmation testing process. She noted that all participants with high penetrance mutations obtained testing, but not all participants with lower penetrance mutations obtained confirmation. Education about genetic testing always is an option when not all patients will receive their results.
St. Jude has an institutional biobanking protocol, from collection through CLIA laboratory confirmation testing. Once samples are transported to an outside laboratory for sequencing; however, chain of custody cannot be guaranteed. Investigators checked concordance between their original samples and those sent out for whole-genome sequencing and found a small proportion of discordant samples. Cohort investigators need to plan for this type of problem.
Participants asked about returning results to people with fully negative findings. The RESPECT study was not able to confirm all “no findings” results. Investigators wanted to share results with all participants, but one IRB did not allow this. Participants with “no findings” still received education about genetic testing because these results were not confirmed. Investigators plan to confirm “no findings” results for 50 actionable mutations.
In response to a question, Dr. Robison indicated that he did not know whether participants who did not respond had read the letter or email, but all will be invited back for testing unless they decline. Everyone who had a likely pathogenic mutation was called by a counselor so that they would have the option to receive confirmation testing in a timely manner.
Participants suggested including several caveats in the informed consent documents when genetic testing will be performed. When results are returned to participants, they should be instructed to discuss them with their physicians. Currently, investigators are moving toward a lighter approach to returning results of genetic testing conducted for research purposes. They have found that many participants overreact (e.g., prophylactic surgery based on results that indicate a low penetrance mutation) and others ignore the results. Approaches are needed that address all potential harms. Participants added that many people will not obtain confirmatory testing on their own. They might go to their doctors, who sometimes make poor decisions regarding genetic testing and might not understand the difference between research and clinical testing.
A study is underway at Kaiser Permanente to examine the best time to engage a genetic counselor. Many people are now bringing results from 23andMe, Ancestry, and other genetic testing companies to their physicians, who do not know how to respond.
Participants asked about the possibility of returning results indicating somatic mutations. No standard exists for counseling about somatic mutation test results. One study found that some doctors are sharing somatic mutation results with patients and others are not. Tumor repositories provide an important opportunity for testing for somatic mutations. The COMET study compared e-health counseling and usual care for patients with identified somatic mutations. The results of this study might be presented at the next eeting of the American Society of Clinical Oncology (ASCO).
Session IV: Kick-Off Session of the Cohort Consortium Associate Member Council
Moderator: Dr. Celine Vachon
Overview: Goals, Purpose, Charge
Dr. Kimberly Bertrand
The kick-off networking and brainstorming session for the AMC was conducted for early-career investigators. Dr. Bertrand provided an overview at the beginning of the session. She noted that the AMC now has a charter and a mission statement. The AMC’s mission is to foster career development and help new investigators identify research and collaboration opportunities within the Cohort Consortium.
Attendees at the AMC Kick-off Meeting engaged in small-group discussions about what they want from the AMC and ideas for potential initiatives over the next year. They rotated tables, with each table assigned specific discussion questions. Although each table focused on specific questions, participants were allowed to go off topic.
Session VI: Plenary Session - Methodological Issues in Cohort Studies
Moderator: Dr. Roger Milne
Integration of Geospatial Data Into Existing Epidemiology Cohorts
Dr. Gianluca Severi
Dr. Gianluca Severi discussed the integration of geospatial data into existing epidemiology cohorts to estimate environmental exposures. He noted several examples of the use of geospatial data in public health. For example, mobile telephones have been used to track places where vaccinations occurred, and geographic information systems (GIS) have been used to generate interactive maps of health events.
Dr. Severi focused his discussion on the European Prospective Investigation into Cancer (EPIC)-France cohorts, which cross generations. The goal of this study is to recruit three generations, and two generations (E3N and E4N) have been recruited to date. Geospatial data were used to map trends in participant characteristics, such as average number of children or age of menarche by department of residence. The study considers changes in residence over time and validates all cases.
The study found that urban residence increased breast cancer risk independent of other risk factors. This finding led to the XENAIR nested case-control study of breast cancer exposure to air pollutants. This study involves geocoding and examines both residential and professional addresses over time. Investigators are attempting to link geocoded addresses with environmental and health data collected through various sources. Investigators are building an inventory of sources of various forms of pollution (e.g., pesticides, radon, noise, light) and examining participant proximity to these sources of pollution. The built environment and access to health care also are being examined, and wind direction is being factored into the analysis to estimate exposures over time. The technical characteristics and activity of various pollution sources were used to estimate emissions. Special software is being developed for data analysis. Investigators identified large variations in the cohort in terms of exposure, but they did not find a significant relationship between breast cancer risk and exposure to dioxin or cadmium. Other exposures still are being studied. High levels of uncertainty, however, make significant results difficult to achieve. Dr. Severi noted that the applicability of GIS-based methods might be limited for some exposures in historical cohorts. Estimates for pre-1990 cohorts are particularly difficult to obtain.
Participants expressed interest in Dr. Severi’s examination of noise pollution. Sleep disturbance has been linked to obesity, and noise pollution resulting in disturbed sleep might explain some of the obesity epidemic in the United States. The impact of noise pollution might vary depending on how well a person’s residence was soundproofed. Listening to high-volume sound using headphones also constitutes noise pollution. France has a lower prevalence of obesity, and prevalence is even lower in the EPIC cohort, which is composed of a more health-conscious population. Nevertheless, Dr. Severi is studying obesity, as well as sleep quality and quantity in cohort participants. His study is beginning to examine the effects of noise pollution, but no results are available yet. Individual exposure to noise will be measured and can take into account the use of headphones and other sources of noise that the individual controls.
In response to a question about the easiest and most difficult aspects of incorporating geospatial data into the EPIC cohort, Dr. Severi noted that the collection and integration of geospatial data is extremely time consuming, particularly the data cleaning process, which involves some automation but also fairly extensive manual checking of data. The first round of geocoding is fairly imprecise, so repeated questionnaires are used to check and correct geocoded information.
Participants asked about correction for minimum error. Dr. Severi explained that his study lacks the data needed for true validation due to limitations of the measures, many of which were developed by the emitting source organizations. These measures tend to underestimate emissions.Dr. Severi clarified that his study is collecting data on where participants live and work, as well as on their transport to and from work, but only residential data have been geocoded and analyzed at this point. Cohort investigators are writing grants to obtain funding for geocoding and analysis of workplace and commute data. A large proportion of the cohort is teachers, with schools as workplaces.
Development of the Virtual Pooled Registry (VPR) Cancer Linkage
Ms. Castine Clerkin of the North American Association of Central Cancer Registries (NAACCR) discussed the Virtual Pooled Registry (VPR). The VPR was developed to facilitate linkages with cancer registry data by eliminating the need for researchers to approach each registry individually for data and creating a central IRB to streamline the review process, which can involve multiple local and state IRBs. Currently, the approval process for linking to cancer registry data can take as long as a year. A templated IRB registry application (TIRA) also has been developed based on existing registry applications for data linkage. The TIRA alone can be used to apply for data linkage at 29 registries (4 registries use both the TIRA and state forms, and 4 use state application forms only) of the 38 VPR pilot test registries. The VPR also will track applications and linkages and provide auto-notifications of deadlines, so that researchers do not need to create their own tracking systems.
Ms. Clerkin explained the planned VPR linkage process, which is being implemented in two phases. In the first phase, the online system is being tested at 38 cancer registries with six federally funded studies. Phase I facilitates a standardized linkage across participating registries and provides the researcher with aggregate match counts within weeks of submitting their study data file. In addition, more than 95% of cases were accurately identified by the linkage software without the need for manual review of the potential matches. Phase II is beginning and involves a streamlined application process using the TIRA for registry/IRB review and approval to release individual-level data to the researchers. The status of these reviews and key dates are captured in the VPR tracking system to assist the researcher in managing the study. In the future, the Central IRB will also be made available to support the review process.
Ms. Clerkin recognized that current registry Data Use Agreements (DUAs) do not include language that allows cohort studies to share or pool linked data, even when it is de-identified. When data are shared, a separate application, approval, and sometimes DUA is required. NAACCR, NCI, and registry staff have met to discuss these barriers and develop solutions. Proposed solutions include a cloud-based system to facilitate data sharing, linkage, and analysis without a download. Data privacy and security could be improved by collapsing some potentially identifying categories and not providing the state identifier. The cloud-based solution will be tested with PLCO Screening Trial data. A DUA also is being developed to facilitate secondary sharing of de-identified data using the Federal Demonstration Project DUA and registry DUAs as models.
NAACCR and NCI are forming a VPR Data Sharing WG to obtain researcher and registry input. Educational webinars also will be conducted to assist VPR users and increase awareness about NCI data-sharing policies and the Common Rule requirements. NAACCR and NCI also are considering developing a standard data-sharing approach with standard definitions of de-identified data, and standards for approving, tracking the use of, and protecting secondary data. Ms. Clerkin asked participants to make their data-sharing plans available to inform these efforts.
Opportunities to Advance Subsequent Neoplasm Ascertainment in Survivors of Childhood Cancer
Dr. Greg T. Armstrong
Dr. Greg Armstrong discussed the use of the VPR to perform rapid case ascertainment for the Childhood Cancer Survivor Study. The study is collecting detailed data on cumulative dose chemotherapy and radiation treatment, as well as a wide range of health outcomes. The study is attempting to identify and confirm subsequent neoplasms in childhood cancer survivors through periodic self-report and review of pathology reports. This process is challenging because of the need to work with community practices after receiving permission from patients to access their medical information. Some facilities do not want to release medical records or cannot locate them.
Dr. Armstrong expects that the VPR will facilitate the identification and ascertainment of second neoplasms and noted that the pilot study generated a large number of childhood cancer survivors with subsequent neoplasms (3,899). Although Dr. Armstrong expects that manual review still will be important for identifying missing data needed for linkage, the VPR will provide the opportunity to evaluate all childhood cancer survivors, as opposed to only study participants; reduce reliance on self-reports; and reduce the resources needed for linkage and ascertainment.
Cancer Ascertainment by U.S. Population-Based Cancer Registries, Self-Report, and Death Certificates in the Nationwide U.S. Radiologic Technologists Cohort
Dr. Danping Liu
Dr. Danping Liu discussed case ascertainment efforts for the U.S. Radiologic Technologists (USRT) Cohort that used the VPR to access data at 43 cancer registries across the United States. He compared registry linkage to self-report data and found that approximately 46% of VPR-identified cases agreed with self-reports. The dropout/non-response rate, however, was nearly 28%. Only approximately 9% of living patient cases demonstrated misreporting. The remaining cases were deceased. Older age was associated with increased odds of disagreement due to dropout. Misreporting, however, did not appear to be related to age. Dr. Liu noted that the VPR missed a relatively high proportion of non-small cell lung cancers in the USRT. Some of these cases might have been residing outside the coverage area, but many in the coverage area still were missed. Dr. Liu and colleagues will investigate the reason for these cases being missed by the VPR. One possible reason is change of residence, because residence information is updated at registries only every 10 years.
Dr. Liu discussed the advantages and disadvantages of self-report and VPR for case ascertainment. Coverage might be incomplete in the VPR because of nonparticipating registries, but the VPR still improves the completeness and accuracy of cancer ascertainment. The VPR also provides detailed information on cancer stage, histology, and other medical factors that cannot be collected through patient self-report. Self-reports and death certificates also under-ascertain cancer incidence in longitudinal cohort studies. Overall, cancer registry data are important for improving case identification and characterization in cohort studies. Dr. Liu is considering ways to combine registry and self-report data and is developing a model of reporting error for types of data.
Participants agreed that the VPR will be useful for national studies. A participant noted, however, that his study found that self-report identified approximately 5% more cases than the registries, but this proportion could depend on the software used. Participants agreed that self-report still is important, and tools might be needed to combine data from registries and self-reports. VPR efforts need to consider sharing of registry data outside of the United States.
Participants asked if the VPR would cover all 51 cancer registries when it becomes available next year. Ms. Clerkin explained that differences between the registries make it unlikely that all will participate initially. Some registries have more data-sharing restrictions or insufficient funding required to participate in the VPR. These registries will need to be approached directly.
Participants also expressed concern about collapsed 5-year age and diagnosis date categories. These categories would potentially be collapsed to address registry concerns, but NAACCR would prefer to make single-year date and age information available in the VPR.
Participants emphasized the importance of sharing the DUAs used by different groups. The Cohort Consortium Steering Committee has requested DUA templates from cohorts in the past and the response rate has been low.
In response to a question about linkage variables for the case ascertainment studies, Ms. Clerkin indicated that linkage typically is performed using the Social Security Number (SSN), date of birth, first and last name, and a combination of street address and telephone number. NAACCR and NCI are investigating ways to perform linkages without SSNs, which often are missing for young children and the oldest cohorts. Some institutions also refuse to provide SSNs. When SSNs are available, however, manual review is minimized.
Ms. Clerkin clarified that MatchPro is the standardized linkage software used for the VPR. Early testing revealed MatchPro to be the best software for producing reports and processing large cohort files. IMS developed MatchPro, which has been well received.
The VDW (Virtual Data Warehouse) developed within the Cancer Research Network (CRN) served as one model for the VPR. The VDW allowed all data to stay within the home institution’s firewalls while being analyzed across institutions.
Researchers cannot anticipate all potential partners who might be involved in a research project and, therefore, would like to be able to share linked cancer data with new partners. Researchers already encounter barriers to sharing National Death Index data with new collaborators. One goal of the DUA Task Force will be to develop a generic DUA for sharing de-identified data with any partner, which could be used by registries that agree to this approach. The Task Force also will work on further streamlining the registry data linkage process and plans to incorporate annual progress reports, notifications of publications, and other standard documents into the VPR.
Session VII: Plenary Session – Grants: Best Practices and Lessons Learned in Supporting Consortial Projects
Moderators: Drs. Joanne Elena and Marc Gunter
Dr. Joanne Elena highlighted the need for cohort investigators to look beyond NCI for expanded and continuing support of cohorts.
The Quest for the Holy Grail of Consistent Consortium Funding
Dr. Shelley Tworoger
Dr. Shelley Tworoger discussed efforts by the Ovarian Cancer Cohort Consortium (OC3) to complement ongoing data collection efforts. Investigators in the OC3 have spent about 3 years developing and negotiating DUAs and 4 years harmonizing data. Biospecimen data also were collected. Dr. Tworoger presented a list of OC3 funding applications to NCI, the Ovarian Cancer Research Program at the U.S. Department of Defense (DoD), and the American Cancer Society, which led to funding for three projects. DoD funding was received to support development of the OC3 Data Coordinating Center as well as an investigator-initiated grant on aspirin use and ovarian cancer risk. Support from DoD was also received to fund an OC3 spin-off consortium: The Prospective Early Detection Consortium for Ovarian Cancer (PREDICT). Dr. Tworoger highlighted the complexity of obtaining additional funding and mentioned the possibility of obtaining intramural support.
The OC3 collaboration has been successful in pooling resources in several ways. The cohort leaders and representatives are highly collaborative and quickly agreed to participate, usually using a fee-for-service model (which avoids the need for a subcontract). A boilerplate letter of support has facilitated this process. The odds of obtaining funding increase when many researchers use cohort resources. In spite of some loss in funding, the OC3 remains productive, largely due to a strong relationship between intra- and extramural investigators and the work of early investigators.
The OC3 has encountered obstacles during Study Section review of funding applications. Study Section members have expressed concerns about (1) harmonization of exposure variables and tumor-related data and the level of effort required for this task, (2) the consortium’s inability to accurately estimate sample size before collecting data, and (3) the lack of diversity in component cohorts. Data harmonization requires a high level of effort that is not fully understood by Study Sections. More papers describing the data harmonization process need to be published.
Dr. Tworoger recommended that cohorts carefully document methods and processes and develop standards and specific infrastructure to ensure continued funding. She would like to work with other cohort PIs to determine the best approaches for explaining how large sample sizes overcome misclassification concerns.
The Experience of the Pooling Project on Alcohol and Cancer Consortium
Dr. Pietro Ferrari
Dr. Pietro Ferrari discussed his data pooling experiences with the Pooling Project on Alcohol and Cancer (PPAC) Working Group. He presented alcohol use trends over time by nation and noted that alcohol is classified as a carcinogen for the oral cavity, larynx, pharynx, esophagus, liver, colorectum, and female breast. The PPAC will focus on cancer sites that do not have an established relationship with alcohol use yet, i.e., cancers of the upper aero-digestive tract, pancreas, prostate, thyroid, and kidney and non-Hodgkin lymphoma. The study presents an opportunity to examine reverse causality because analyses will also focus on alcohol intake during lifetime, i.e., a long time before participants’ recruitment.
The PPAC received funding from the National Institute on Alcohol Abuse and Alcoholism (NIAAA) in 2016. Dr. Ferrari noted that the NIAAA was interested in studies making secondary use of existing data. Other attributes of the funding application that facilitated funding included the fact that the project was built upon an existing project (the Diet and Cancer Pooling Project coordinated by Stephanie Smith-Warner), together with the infrastructure provided, particularly data centralization and harmonization capabilities. The funded study examines lifetime alcohol consumption and patterns of intake with a focus on non-smokers. The study has generated preliminary findings indicating that alcohol consumption is related to UADT cancers in all smoking groups, as well as among never smokers. Dr. Ferrari also is involved in another study recently funded by NCI to examine alcohol, diet, and body fatness and stomach cancer, which includes pooling of international prospective studies funded by the Australian NH-MRC (PI: Harindra Jayasekara).
The Cancer Research Data Commons has a goal of seamless data aggregation across cohorts. Participants emphasized the need to share lessons learned across cohorts and agreed on the importance of communicating the level of effort involved in harmonizing data along with the value of that effort. Dr. Tworoger now is beginning to harmonize questionnaire data for her study. She noted some reasons investigators reharmonize data, such as the fact that each cancer risk factor has unique characteristics that require different harmonization approaches. Maelstrom and Data Commons can help investigators understand how other cohorts have posed questions, identify common data elements, and determine how data from existing questions can be harmonized along the common data elements. Dr. Tworoger recommended documenting data decisions in real time as much as possible, but efficient systems for documenting data handling need to be developed.
Work on the Teachers Study taught investigators the value of granularity. The Cohort Consortium could support projects by identifying the most granular data elements, because investigators always can aggregate granular data elements to form common data elements. Identification of granular data elements would accelerate data sharing. A coordination data center for projects within the Cohort Consortium also would support data sharing and pooling. The Cohort Metadata Repository is a step toward a data coordination center. Participants noted that legal agreements often interfere with data sharing between institutions. At the metadata level, however, projects should be able to share data easily. Dr. Ferrari added that the Mica-Opal platform (Maelstrom) can handle metadata. Participants recommended conducting a demonstration project to test ways to harmonize all variables for a specific condition.
The Cancer Research Network (CRN) created a common data model for EHR data, demonstrating that a common data model and repository might be feasible. Other data repository approaches should be examined, along with the tools used to consolidate and query those data. For example, the Maelstrom platform and Opal software are used to share data across the Canadian cohorts. All data reside in each cohort, but can be accessed and, to some extent, analyzed using the Maelstrom platform. This approach allows investigators to perform preliminary tests to determine sample size and what data they want to request for a planned study. Maelstrom originally was used to create data inventories.
The CRN uses a tool that allows simple analyses of data from other centers. The U.S. Food and Drug Administration’s Sentinel Network also has a system that combines claims data from big insurers and smaller, integrated health care systems. In addition, investigators at Harvard are developing methods for distributed data analysis that generate results similar to those resulting from data pooling.
Participants cautioned against over-harmonization, which can produce broad, uninformative findings. They also noted that developing common data elements can be extremely difficult across research areas and diseases. This undertaking could be expensive, which raises the question of who would pay for the creation of a centralized data system. The Cohort Consortium might want to evaluate the costs and benefits of such an effort. As a minimum, cohorts can share approaches for harmonizing different types of variables. One participant suggested beginning by creating federated metadata tables that can be queried.
Wrap Up of Main Meeting
Drs. Celine Vachon and Lynne Wilkens
Dr. Vachon concluded the meeting by thanking the meeting organizers. Dr. Lynne Wilkens announced the winners of the poster session competition and the Best Early Career Fellow Awards.
- Poster Session Competition winner
- Dr. Aurora Perez-Carnago – “Biomarkers and risk of prostate cancer death: A collaborative analysis of longitudinal participant data from 13 prospective studies”
- Best Early Career Fellow Award winners
- Dr. Lauren Hurwitz – “Toward a consensus definition of truly aggressive prostate cancer for epidemiologic research”
- Ms. Suzanne Dixon-Suen – “Estimating the burden of cancer attributable to physical inactivity in the United States and Australia”