2012 Cohort Consortium Annual Meeting
The 2012 Cohort Consortium Annual Meeting was held October 24-26, 2012 at Natcher Conference Center on the NIH Campus in Bethesda, MD.
A Symposium was held on October 25, 2012 from 8:30 a.m. - 12:15 p.m. in the NIH's Natcher Conference Center in Bethesda, MD.
The Symposium, "When Bigger is Better: Combining and Designing Cohorts for Power" was jointly sponsored by the Epidemiology and Genomics Research Program (EGRP) of the NCI's Division of Cancer Control and Population Sciences (DCCPS), and the NCI's Division of Cancer Epidemiology and Genetics (DCEG).
Those who were unable to participate in person may view an NIH videocast of the Symposium.
View more information about the Symposium on October 25, 2012 including:
- Summary of the Symposium
- Related Materials
A summary is available from the Cohort Consortium Members Meeting, held on October 25, 2012 from 12:15 p.m. to 5:30 p.m. in the Natcher Conference Center on the NIH Campus in Bethesda, MD.View summary of Members Meeting
NCI Cohort Consortium Members Meeting Summary
Dr. Patricia Hartge, Deputy Director of the Epidemiology and Biostatistics Program at the National Cancer Institute's (NCI) Division of Cancer Epidemiology and Genetics (DCEG), welcomed cohort members. She presented an Outstanding Service award to Dr. James Cerhan of the Mayo Clinic for his leadership as Chairman of the Consortium Secretariat during the past year.
This year's meeting consisted of four sessions:
- Session I - a series of brief reports on Consortium project work during the past year,
- Session II - consisted of four Secretariat breakout groups,
- Session III - included the report backs from those groups, and
- Session IV - provided guidance for participation in the Consortium, including an update on available NCI funding, opportunities for linkages to Surveillance, Epidemiology, and End Results (SEER)-Medicare (SEERM) data, and a proposal to collaborate on a research project that addresses cancer in underserved populations.
- Session V - included an open Discussion and wrap up
Each breakout group discussed questions about promising research areas, the critical challenges faced by the cohorts and proposed solutions, and the lessons learned from participation in the Consortium.
Session I: Reports of Working Groups (WGs)
Second Cancers WG - Lindsay Morton, Ph.D.
Second cancers are a leading cause of morbidity and mortality among the more than 12 million cancer survivors in the United States. Risk factor data and the large patient populations in the Cohort Consortium could be of benefit in examining second cancer risk factors in prospective cohorts. The WG designed two proof-of-principle projects. The first, led by Dr. Todd Gibson (NCI), involves five cohorts and is a study of anthropometric factors and second cancer risks among colorectal cancer survivors. Stage was selected as a proxy to control for potential confounding by treatment. The pooled data includes 11,690 colorectal cancer survivors with body mass index (BMI) data, of which more than 1,000 patients developed a second cancer during follow-up.
Dr. Gibson and collaborators initially are investigating BMI and the risk of second primary obesity-related cancer, particularly post-menopausal breast, kidney, esophageal adenocarcinoma, pancreas, and endometrium cancers; 219 patients have such second cancers. Preliminary analyses show modestly increased risks with overweight/obesity, with higher risks for individuals with a family history of cancer. When the same outcomes were analyzed as first primary cancers, the risk estimates for overweight/obesity were similar, with no apparent effect modification by family history.
The second project, led by Dr. Meredith Shiels (NCI), examines the risks for multiple second primary tobacco- and alcohol-related cancers. The goal of this study is to determine the extent to which prevention measures should be tailored to survivors. Next steps include reviewing methodological issues, such as changes in exposure over time, sample size, and treatment data; and identifying collaborators to expand the study. Cohorts interested in participating in second cancer analyses should contact Dr. Morton Gibson of Virginia Commonwealth University; Drs. Shiels and Black of NCI; or Dr. Robien of George Washington University.
African American (AA) BMI and Mortality Pooling Project - Sarah Cohen, Ph.D.
Seven cohorts participate in this study. The project's goal is to assess the relationship between BMI all-cause mortality, cancer, and cardiovascular disease (CVD) mortality in AAs. The cohorts include approximately 240,000 eligible participants; analysis focused on approximately 109,000 healthy non-smokers at baseline. The mean age at enrollment was 52 and the mean follow-up was 12.5 years. Within this group, there were more than 13,000 deaths.
Study results for both the AA and whites pooling projects demonstrated increased risk of all-cause mortality among underweight individuals and those at higher BMI levels among men and women. The magnitude of association, however, is lower in blacks at the upper end of the BMI spectrum, which is consistent with published results from previous studies comparing whites and blacks.
The study includes stratification variables, including education as a proxy for socioeconomic status (SES). When comparing the most overweight individuals to those of normal weight, the hazard ratios (HR) for obese individuals compared with the healthy BMI referent were higher among participants with more than a high school education compared to those with less than a high school education. The study did not note any distinct patterns for all-cause mortality by region of the country. HRs for CVD mortality increased as BMI increased, while for all-cancer mortality, the HRs were close to the null across the BMI spectrum. The WG is preparing a manuscript that is expected to be submitted in 2013.
Ovarian Cancer Cohort Consortium (OC3) - Shelley Tworoger, Ph.D.
The OC3, led jointly by Dr. Tworoger of Harvard Medical School, Dr. Wentzensen of NCI, and Dr. Alan Arslan of New York University, received funding in September 2012 to build the OC3 infrastructure. Twenty-two cohorts, including the European Prospective Investigation into Cancer and Nutrition (EPIC), will participate, and ten cohorts have submitted their data. The WG expects to have between 7,000 to 8,000 cases with accompanying questionnaire data.
The OC3's primary aim is to evaluate OC risk factors by tumor heterogeneity and examine differences by histologic subtype, tumor dominance (i.e., location and spread of the tumor), and tumor fatality (i.e., identify more aggressive tumors). The second aim is to improve OC risk prediction models by accounting for heterogeneity of the disease associations in the model.
Next steps include cleaning current data that are being used for the risk prediction modeling project. OC3 anticipates receiving the remaining datasets by early 2013, with data cleaning and histological subtype analyses to be completed during that year.
Renal Cell Carcinoma WG - Mark Purdue, Ph.D.
Dr. Purdue provided an update on the expansion of a genome-wide association study (GWAS) of renal cell carcinoma (RCC) to a cohort within the Consortium. The project is led jointly by Dr. Purdue and Dr. Ghislaine Scélo at the International Agency for Research on Cancer (IARC).
A family history of kidney cancer can double one's risk for the disease, and several rare genetic disorders are affiliated with such risk. Investigators, however, are less informed about the role common genetic variants play in kidney cancer susceptibility. IARC and NCI investigators collaborated on a RCC GWAS that combined GWAS data on almost 4,000 cases and 8,500 controls from 11 studies. The initial GWAS investigations have revealed four risk loci to date (2p21 [EPAS1 gene], 11q13, 12p11.23 [ITPR2 gene]), and an additional locus that will be revealed in an upcoming article.
The group aims are to identify additional risk loci, explore gene-environment (GxE) interaction with established risk factors for kidney cancer, and conduct a GWAS of kidney cancer survival. It added 5,000 cases and 8,800 controls, will genotype soon, and identified 1,800 cases and 6,700 controls as a replication set. Next steps include publishing a manuscript by the spring of 2013 that describes their findings investigating the GxE interaction, and conducting a GWAS of RCC survival by early 2014.
Vitamin D Pooling Project of Breast and Colorectal Cancer - Stephanie Smith-Warner, Ph.D.
This Consortium includes 20 studies from three continents, with more than half of the studies being conducted in the United States. The Consortium's first aim is to evaluate the association of circulating vitamin D levels with breast and colorectal cancer risk; the group also will focus on high and low levels of the vitamin to evaluate the dose-response relationship. A second aim will examine whether associations vary by lifestyle factors (including dietary factors), other biomarkers, and common genetic variation in the vitamin D receptor.
Researchers also hope to determine whether associations vary by tumor subtype (e.g., by subsite for colorectal cancer and by hormone receptor status for breast cancer). The breast cancer analyses include 15 studies of more than 10,000 cases and 12,000 controls; 10 of their datasets previously measured vitamin D levels. The colorectal cancer analyses include 17 studies of approximately 5,800 cases and almost 7,400 controls, with nine datasets that include previously measured vitamin D levels. Vitamin D assays for the Consortium are nearly complete.
The group conducted calibration studies to adjust for differences in the assays, laboratories, and timing of the vitamin D measurements in the studies that previously measured vitamin D levels. These calibrations are necessary because the researchers want to examine Consortium-wide quantiles and common absolute cutpoints across studies for public health translation. The group also wants to compare its observed vitamin D levels with those in the VITamin D & OmegA-3 TriaL (VITAL)-a randomized clinical trial of vitamin D in fish oil-and with the "gold standard" LC/MS2 vitamin D assay.
The researchers expect to complete the assays by the end of 2012. Investigators will finalize harmonization of the data, continue to refine the analytic approaches, and conduct analyses in 2013. They also will continue investigating methods of incorporating information on additional vitamin D metabolism genes.
Male Breast Cancer Pooling Project - Michael B. Cook, Ph.D., Louise A. Brinton, Ph.D.
The stages of this study are to (1) identify case-control and cohort studies that can contribute data and possibly biologic specimens (serum, plasma, DNA) for pooled analyses of this malignancy; (2) harmonize questionnaire data and undertake analyses of such; (3) assess circulating hormones in relation to male breast cancer risk using pre-diagnostic blood samples; and (4) assess genetic polymorphisms in relation to risk of male breast cancer. The project includes 11 case-control studies and 10 cohort studies with approximately 2,400 cases combined.
The design and analytic strategy include detailed inclusion and selection criteria. For the cohort study, the group tested analytic strategies and selected a 40:1 ratio of controls per case. They conducted unconditional models followed by meta-analysis of the cohort and case-control pooled study type-specific estimates to use for the project. This method of analysis will allow for greater flexibility, is more inclusive, and allows researchers to examine how results differ by study design. For biologic analyses, controls were matched to cases in a 2:1 ratio; serum or plasma will be used to measure hormones and DNA to assess genetic predictors of risk.
The WG anticipates papers on hormonal-related factors, family and medical history, and tobacco smoking and alcohol consumption. Two manuscripts on molecular analyses are expected and include one restricted to cohorts using pre-diagnostic serum and plasma related to hormone analyses, and one on genetic analysis.
The group studied the association of the following exposures in relation to the disease: BMI, height, weight, diabetes, gynecomastia, Klinefelter Syndrome, fractures, osteoporosis, and ever had children.
Lung Cancer Cohort Consortium (LC3) - Paul Brennan, Ph.D.
LC3 was formed in response to the findings of a prior EPIC study. That study measured the prediagnostic serum of cases and control subjects for a panel of one-carbon metabolism (OCM) and found strong associations for vitamin B6, methionine, and, to a lesser extent, folate. The investigators found strong effects present in the never smokers and among those with long follow-ups, but no obvious confounding association or reverse causation to explain the results.
Although a causal role was not established, people with lower levels of several biomarkers had strong associations, indicating that this analysis could be useful for identifying high-risk groups in addition to smoking, and thus, LC3 was formed. Currently, 24 cohorts contribute almost 5,400 cases to LC3, with approximately two-thirds of cases from U.S., European, and Australian cohorts, and one-third from the Asian cohorts.
Approximately 1,900 are never smokers, 1,400 are former smokers, and 2,100 are current smokers. The study includes matched controls of 1:1 on smoking status, and the numbers are fairly balanced between men and women.
LC3 plans to measure 40 markers of OCM, have repeat samples on 1,000 controls to correct for regression dilution, and perform genetic analysis to determine if there are differences between cases and controls to attempt to replicate the results within the International Lung Cancer Consortium (ILCCO).
Biochemical analysis on the samples is in progress, and researchers anticipate that 90 percent of the analysis will be completed in Year 2. The group anticipates having preliminary risk analyses data available by next year's Cohort Consortium meeting. It also plans to use the infrastructure that was developed for the LC3 to perform additional lung cancer biomarker studies.
Liver Cancer Pooling Project (LCPP) and Biliary Tract Cancers Pooling Project (BTCPP) - Katherine McGlynn, Ph.D.
The LCPP was formed to address dramatically increasing rates of liver cancer in the United States since 1980. Fourteen U.S. cohorts provided questionnaire data and serum. LCPP researchers used the serum samples to determine the hepatitis B and C infection status of cohort cases; they also looked at inflammatory markers and at hormone levels. Currently, the LCPP has approximately 2,300 cases and 1.5 million non-cases with questionnaire data for prospective analyses. Approximately 300 of the cases and 700 matched non-cases have available serum for inflammatory and hormonal studies.
The LCPP initially addressed reproductive variables among women, particularly the role of oral contraceptives in liver cancer risk in women, especially in populations that are not endemically infected with virus. The first results show no association with oral contraceptive use. The group also analyzed such variables as age at menarche and menopause and hormonal replacement therapy.
The group will examine the effects of diabetes and obesity on liver cancer and of non-steroidal anti-inflammatory drug (NSAID) use, suggested by a previous study to have a protective effect on liver cancer. Other effects being considered include physical activity; alcohol, folate, and tobacco use; and coffee and tea consumption. The group also might address the effect of pesticide exposures, selenium, and multivitamin use on cancer.
In 2012, BTCPP was created to examine rare biliary tract cancers. The project includes 17 cohorts from the LCPP and the BMI Pooling Project and presently is enrolling international cohorts. BTCPP's proposed studies include examining BMI and diabetes, smoking, NSAID use, and family history of cancer.
Parties interested in joining the LCPP should contact Dr. McGlynn or Dr. Campbell of the American Cancer Society (ACS). Those interested in joining the BTCPP should contact Dr. Koshiol of NCI or Dr. Campbell.
Lymphoid Malignancies Working Group (LMWG) - Brenda Birmann, Sc.D.
The Multiple Myeloma Cohort Consortium (MMCC) and the GWAS of Non-Hodgkin Lymphoma (NHL) are active biomarker-based projects in the LMWG. LMWG also collaborates with the BMI and All-Cause Mortality Pooling Project on analyses of anthropometric measures and risk of mortality from multiple myeloma (MM).
The MMCC studies include eight cohorts that collaborate on the serologic biomarker study and eight cohorts that collaborate on the genetic susceptibility study, which focuses on candidate genes. The NHL GWAS is a collaboration between LMWG and the InterLymph Consortium and involves case-control studies from InterLymph and several cohorts.
MMCC projects, motivated by emerging evidence that obesity is a risk factor for MM, are exploring several biologic pathways. Researchers designed serologic and genetic marker studies to examine potential dysregulation in growth factor and inflammatory pathways and the risk of MM. NHL GWAS investigators will attempt to identify novel loci and biologic pathways underlying NHL risk, possibly by histologic subtype.
MMCC's original study, which examined serologic markers in IGF-1, insulin, and IL-6 pathways, has completed main effects analysis and a manuscript is available online (Birmann BM, et al., Blood, 2012; PrePublished online Oct 16). Major findings included two markers, one in each pathway, that demonstrated a positive association with risk of a MM diagnosis within a relatively short time after blood collection (i.e., within 3 years of blood draw for one marker, and within 6 years for the other marker); no associations were noted with MM risk >6 years after blood collection. A new biomarker project is underway to explore other obesity-related pathways, particularly adipokines.
MMCC's genetic susceptibility studies are in early stages, with the sample size slightly larger than for the completed study. The group has completed genotyping for 124 markers in approximately 650 cases and 1,100 controls and is working on data cleaning and analysis. A manuscript is expected in 2013.
The NHL GWAS includes approximately 12,000 cases and more than 9,000 controls and focuses on the most common histologic subtypes of NHL. Preliminary main-effect analyses reveal several novel findings for several NHL subtypes, and manuscripts summarizing those findings are in preparation. A series of secondary analysis proposals have been approved and will focus on in-depth analysis of the genetic data and GxE interaction studies.
Sessions II and III: Attendance at and Report Back From the Secretariat Breakouts With Member Cohort Groups
Cohort members were divided into four groups. Group suggestions are summarized below.
- Are there some particular projects, or some broad types of projects, that you would like to see the Secretariat try to foster or stimulate in the year ahead?
- Include different areas for research focus.
- Focus on cancer subtypes, including common cancers.
- Address molecular subtyping of cancer and the new technologies that make it easier to address epigenetics and somatic mutations.
- Examine long-term cancer survivors and use existing cohorts to complement other studies.
- Address non-cancer diseases in cancer survivors and provide effective endpoint data from the Consortium to study these diseases.
- Follow up on recent results on multi-vitamin supplement data by expanding the project to include all cancer sites.
- Prioritize tissue sampling.
- Develop a tracking system for tissue sample collection by individual cohorts. Enhance cancer reporting forms to identify collection type (tumor blocks, tissue, or slides), cancer subtype, and whether treatment and stage information are collected.
- Encourage the Consortium and/or NCI to provide support for acquisition of tumor tissues, including use of new technologies to collect old formalin-fixed, paraffin-embedded (FFPE) tissues.
- Coordinate record collection requests by the cohorts doing research in one state, which likely will lead to more cost-efficient and productive ways of encouraging hospitals to release greater numbers of samples to the cohorts.
- Facilitate discussions with hospitals concerning improved access to hospital tissue supplies. Discuss the option of collecting additional tumor samples now if such samples might be more difficult to collect later.
- Address best practices for using tissue samples and blocks.
- Consider these additional suggestions.
- Distribute a questionnaire to members prior to the annual meeting asking for the number of available cohort cases at their sites; include cases with non-cancer outcomes to aid in the design of future studies.
- Find alternative data sources and ensure that the cohorts have access to those sources (e.g., CMS, registries); streamline access as much as possible.
- Determine the minimum number of cases for participation in a Consortium project, based on scientific, statistical, and/or financial issues. Consider including this topic at next year's meeting.
- Build on prior experiences to develop more efficient ways of harmonizing data that could be used across projects. Avoid duplication of effort by centralizing data or the macros used to harmonize the data. Consider including this topic at next year's meeting.
- Include different areas for research focus.
- Last year, we decided to see if some projects with non-cancer outcomes could be conducted in the Consortium. This did not prove to be easy; should we continue to pursue that kind of pilot work?
Non-cancer research, although difficult to accomplish, continues to be of interest to Consortium members. Suggestions for increasing non-cancer outcome studies include:
- Provide valuable cases from the Consortium to non-cancer cohorts to study rare and often fatal diseases (e.g., EPIC ALS cases to the ALS Consortium). Providing access to the National Death Index might assist cohorts in obtaining outcomes for these diseases.
- Apply cohort cancer study design to other diseases.
- Examine the effect modification of outcomes on cancer development and survival.
- Examine rare environmental exposures and technical methods of assessing the exposures using non-questionnaire methods.
- Explain differences in race factor distribution, differences observed in incidence, and include international cohort sources.
- Demonstrate the ways in which cohorts jointly funded by multiple NIH Institutes can effectively collaborate on research. In addition, determine methods that cohorts can use to identify an Institute whose interests and funding opportunities align with new Consortium research.
- Alert NIH Institutes to the value of using cohorts for genetic, gene-environment, and other such studies that might offer data currently not available from the Institutes.
- Consider whether Consortium work in this area is compatible with the priorities of a cohort's parent organization.
- Develop strategies that showcase non-cancer research and highlight its importance to organizations that traditionally deal with cancer-related research.
- Investigate whether Early Detection Research Network (EDRN) data might be linked to cohort studies. EDRN researchers also might be interested in cohort data identifying biomarkers that appear shortly before the onset of the cancer.
- Address precision issues related to diagnoses for non-cancer endpoints.
- Last year, the Second Cancer/Survival WG continued pilot work, which will be reported later. Is this an area of current interest or activity to your cohort?
Suggestions regarding second cancer/survival pilot work include:
- Evaluate survival and related treatment data in cohort cases. The latter is beneficial to those reviewing proposals and journal articles addressing cancer.
- Consider whether cohorts are the optimal study design to use in survivorship research. Obtain advice from the NIH Office of Cancer Survivorship on whether cohorts, organizations such as registries, or methods such as clinical trials might be more effective.
- Determine if cohort data are of value for determining second cancer treatment options and survivorship.
- Consider metastasis as a secondary endpoint and determine methods to collect metastasis data in a standardized format.
- Use cohorts to investigate cancer survival. Cohort survival data could complement clinical trials and population-based cancer registry survivor studies.
- What are the major challenges your own cohort is facing? Are there activities the Consortium can undertake that could help?
Major challenges and, where mentioned, possible areas for Consortium assistance, include:
- Reduced funding opportunities.
- The need for cohort members to recuse themselves from cohort infrastructure funding application reviews, which could result in remaining reviewers who are less familiar with cohort studies.
- Lack of individual recognition in large collaborative projects, which could impede academic progression and the ability to obtain one's own research funding.
- Keeping participants engaged in pooled analyses if studies lack sufficient publicity.
- Researchers' time and effort in such studies. Coordination of effort between cohorts to obtain similar variables could reduce time constraints faced by researchers.
- Balancing dataset needs. Cohorts should consider how much data from previous projects are acceptable before requesting new data for a project.
- Balancing support between new and existing cohorts. Cohorts could be established that use new technologies and phenotyping mechanisms while maximizing the current research possibilities in existing cohorts.
- Consent issues related to electronic health records (EHRs). The Cohort Consortium could do pilot studies to examine consent issues and the quality of EHRs. The Consortium should request that NIH leaders help it meet these challenges. The Consortium also should think collectively about how to have an impact and propose provocative questions for NIH.
- What are the key lessons from your participation in Consortium projects?
- The Consortium provides excellent opportunities for scientific research in collaborative projects and provides additional funding opportunities.
- Is the request for new projects 3 times per year too much, too little, or just right? Do the project proposals provide you with sufficient information to make a decision on participation?
- Two groups reported that three requests a year is acceptable.
- Timing of deadline alerts is important; researchers need time to obtain letters of approval and to prepare for discussions of potential proposals prior to in-person meetings with parent organization representatives.
- One group suggested more frequent requests would be helpful. An ongoing mechanism for submitting proposals might encourage members to seek advice and feedback from the Secretariat and create greater support for future studies.
Additional comments included during the report back included:
- The breakout session was beneficial and a longer session should be included during next year's annual meeting.
- NCI and NIH could encourage organizations with federally funded research to share tissue samples with the cohorts.
- The number of international participants in the Consortium has increased, yet most Consortium projects are funded by NCI and other U.S. funding organizations. The Consortium should engage funding agencies from other parts of the world and invite them to next year's annual meeting.
- NIH and the European Union (EU) have funding reciprocity agreements and, therefore, the EU could provide funding for cohorts that collaborate globally. NIH, however, requires additional review for foreign cohorts that participate in a U.S. study.
SESSION IV: Looking Ahead
NCI Cohort Infrastructure Program Announcement Update - Daniela Seminara, Ph.D., M.P.H.
The Epidemiology and Genomics Research Program (EGRP), of NCI's Division of Cancer Control and Populations Sciences (DCCPS), currently provides support to more than 60 national and international cancer epidemiology cohorts (CECs), including the NCI Cohort Consortium. It supports biobanks, phenotypic and exposure data harmonization, and incorporation of new technologies for structural and functional molecular characterization in the consortia. In addition, EGRP promotes the translation of consortia results into practice. Cohorts also are involved in other ways in many EGRP-funded grants.
EGRP-supported CECs include international participants of all ages and represent many cancer types; current plans include increasing minority representation. CECs have collected DNA from 51 percent and blood from 35 percent of biospecimens. CECs are a beneficial resource for EGRP-supported GWAS, and NCI's largest post-GWAS study relies primarily on the CECs infrastructure. CECs also are used for a number of EGRP-supported whole exome sequencing and whole genome sequencing projects.
Fifteen EGRP-funded cohorts are part of the NCI Cohort Consortium. EGRP, in conjunction with DCEG, also provides additional support, such as participation in and leadership of collaborative projects, coordination of meetings, and funding of contracts related to the cohorts.
In 2011, EGRP launched the "Core Infrastructure and Methodological Research for Cancer Epidemiology Cohorts" (PAR-11-167) as a source of long-term support for CECs. The short-term goals included establishing, maintaining, or upgrading new or existing CECs core functions and performing methodological research to validate or evaluate new or existing methods of improving core functions.
The first grant was awarded in 2011 and the final receipt date for this issuance is scheduled for November 2013. Six applications were funded during the first three review rounds; these funded cohorts include more than 400,000 participants. The cohorts include non-cancer outcome studies and have high-quality biospecimen banks. The award success rate for this PAR is approximately 14 percent, which is similar to large grant mechanisms previously used to fund such research.
EGRP facilitates collaborative data sharing across cohorts. These cohorts are required to develop a cooperative agreement with a robust data sharing plan, provide descriptive data for a web-based database, and participate in cross-cohort data harmonization if possible. In addition, EGRP has launched three initiatives to facilitate cross-cohort collaborations and data sharing: (1) establishment of a descriptive public database of CECs accessible from the EGRP website and integrated with the consortia database, (2) pilot harmonization across CECs, and (3) establishment of data coordinating centers for cancer consortia.
Medicare Linkages-Opportunities and Needs for Cohorts - Joan Warren, Ph.D.
Dr. Warren, of DCCPS' Applied Research Program (ARP), discussed different techniques to obtain linked SEERM data, or Medicare data only, and benefits to cohorts that use the two data sources.
SEERM data are obtained from two large, linked population-based data sources - NCI's SEER Program and CMS' Medicare Program - and currently include data from 1.8 million cancer cases. SEER consists of cancer registries in geographically defined areas that collect cancer incidence data from within those areas; currently approximately 28 percent of the U.S. population lives in the SEER areas.
For persons who are eligible for Medicare, CMS collects claims for all health services from the time of eligibility until a person's death if the person has fee-for-service coverage. Most people attain Medicare coverage at age 65. Medicare also covers people under age 65 who have long-term disabilities and individuals with end-stage renal disease (ESRD).
Medicare data for persons with fee-for-sevice coverage include a broad spectrum of covered services from multiple sources, such as hospital care, physician visits, outpatient clinic, and home care. Medicare includes a unique patient ID number, which simplifies the creation of an individual composite health care profile. SEERM data also contain Census data that have information about the socioeconomic status of the area where a patient resides.
This information can help to control for potential underlying differences in study populations. The SEERM data also provide information about care providers and can be linked to files with information about the availability of health resources within a county.
The SEERM database linkage is updated every 2 years and currently includes cases diagnosed from 1991–2009 on all Medicare-eligible people in SEER. A non-cancer cohort of Medicare patients in SEER areas is available. The non-cancer cohort is derived from a 5 percent random sample of persons residing in SEER areas who have not been diagnosed with cancer. The cohort then is used by researchers to create comparison groups or a true 5 percent sample for specific cancer comparisons.
SEERM covers the full research spectrum across the cancer control continuum. Research areas that are suited for SEERM data include patterns, quality, and costs of care; health disparities; cancer risk factors that are reported on insurance health claims, and late effects of treatment that require a procedure/surgery or physician care. Research topics that might not benefit from inclusion of SEERM data are treatment effectiveness, recurrence, and disease progression.
Cohorts could benefit from using SEERM as the data are longitudinal and include large numbers of cases that cover a broad spectrum of health services, which are provided in a variety of clinical settings. The data represent diverse geographic areas and are population-based, thus reflecting "real world" practice versus clinical trials. SEERM also include data on multiple disease conditions and a cancer-free control group that allows researchers to evaluate healthcare among the elderly, rather than cancer patients only.
Limitations of using SEERM include: the generalized data include only individuals with disabilities or ESRD in the under 65 age group, newest data often are not current, some services such as long-term care at the U.S. Department of Veterans Affairs' facilities, and Medicare patients in managed care plans are not covered. In addition, SEERM is not a clinical database, offers no information on why tests were performed, and provides no test results.
Researchers might consider linking a cohort to Medicare data directly. This linkage can be completed with the assistance of the Research Data Assistance Center (ResDAC), operated at the University of Minnesota. ResDAC is funded by CMS to assist researchers who want to acquire and use Medicare data. Use of Medicare data could increase case ascertainment, identify people with conditions of interest, and allow researchers to contact people directly. Medicare data also provide comorbidity and longitudinal information regarding outcomes.
In answer to a question about the cost of obtaining SEERM data, Dr. Warren noted that NCI and CMS have an interagency agreement, and NCI bears the cost for linking the data. The SEERM Website outlines the cost to researchers, which is less than obtaining Medicare data directly from CMS.
Cancer in the Underserved: Addressing a Critical Research Need - Electra D. Paskett, Ph.D.
Dr. Paskett (Dr. Tim Rebbeck from the University of Pennsylvania, the other Principal Investigator (PI)) presented a proposal on behalf of the Cancer Disparities Research Network (CDRN). The CDRN-formed by 15 institutions in the United States that represent regions and populations with differing disparity challenges-proposes to study basic biology, lifestyle/immigration patterns and GxE interactions on multiple levels and the ways in which interactions of those multiple levels can lead to poor health outcomes.
CDRN's proposal links underserved populations in the Cohort Consortium with underserved populations in the CDRN network. To qualify as underserved, individuals must meet at least one of the following criteria: live in a medically underserved area as defined by the Health Resources and Services Administration (HRSA), be of low literacy, low income, or be uninsured.
Individuals of this underserved cohort will be 40-75 years of age. CDRN hopes to enroll 100,000 participants in the cohort and estimates that an additional 30,000 participants must be accrued to reach that goal.
Currently, CDRN is developing joint demonstration projects with Consortium cohorts that might be best suited to this proposal. They hope to submit a UM1 grant application in March 2013. CDRN would like feedback from Consortium members regarding the proposal, including:
- who would like to participate and the populations and sample sizes available,
- the optimal pilot project that will demonstrate collaboration between groups without exchanging data across sites, and
- the most effective funding model that will work for Consortium members. CDRN will pay for pooling data in this cohort.
If CDRN expects to include populations from other countries, it was suggested that the criteria for qualifying as underserved be modified.
In response, Dr. Paskett noted that CDRN recognizes that countries have different definitions for underserved and is interested in further discussions with international groups that might consider establishing parallel cohorts.
SESSION V: Open Discussion and Wrap Up
The OC3 is considering obtaining updated data in addition to baseline data and asked for guidance from cohorts that have done this. Issues to consider are whether people are willing to contribute the extra data and how the statistical analyses would work in different cohorts.
During discussion of this issue, the following points were made:
- Follow-up data are good for rebaselining the cohort, but mortality fall-off results in reduced response rates and the risk of bias being introduced into assessments.
- Several cohort members noted that their cohorts have frequent serial blood collection. Follow-up data collection could be discussed on a conference call.
- A group of cohorts have significant experience with heart outcomes and cancer and might have a conference call to discuss this issue.
New Cohort Members
Dr. Piet van den Brandt is the PI of the Netherlands Cohort Study (NCS), which has been in existence for 25 years but recently joined the NCI Cohort Consortium. NCS has a large number of cancer cases with follow-up data dating from1986. The NCS cohort includes approximately an equal number of men and women from the general population of the Netherlands. Follow up is by linkage with the cancer registries and a national pathology registry. NCS collected questionnaires and toenail samples (90,000 samples), which have been an effective source for DNA genotyping.
Dr. Hilde Langseth represents the Janus Serum Bank (JSB) of Norway, which recently joined the NCI Cohort Consortium. Dr. Giske Ursin is the PI of the project and the Director of the Cancer Registry of Norway. The JSB cohort is a population-based biobank that includes samples from 320,000 individuals in Norway, which were collected during the 1970s through the 1990s.
The cohort includes 55,000 cancer cases obtained from annual linkage to the cancer registry, including a sizeable pool of rare cancer cases, and questionnaire data from the National Institute of Public Health. JSB participates in the Consortium's Vitamin D Pooling Project and the Male Breast Cancer Project, as well as the AGRICOH Consortium of agricultural cohort studies sponsored by IARC and NCI.
Dr. Lyle Palmer is the Executive Scientific Director of the Ontario Health Study. The cohort includes samples from 230,000 ethnically diverse adults between the ages of 18 and 99 who were recruited from Ontario. Biases include more women (60%) than men and a slightly better education and higher income than average. Currently, the cohort includes 20,000 cancer patients, and recruitment is expected to continue for at least 5 more years.
In response to a request for clarification on how data are harmonized in the Consortium, it was noted that harmonization in the NCI Cohort Consortium is an evolutionary process that usually begins with one project that others use as an example. The Consortium does not have a standardized, centralized harmonization system, and no central coordinating center exists. In cancer research, the United States has followed a decentralized research model, although groups have shared code books and codes as each project moves forward.
Dr. Cerhan, who is stepping down from the Secretariat, noted that the office attempts to be flexible to the needs of the Consortium in its pursuit of scientific goals. For example, the Secretariat has been working on infrastructure and the Social Security Death Index closure. NCI and other institutions are collaborating to develop methods to make data more accessible and useful.
The Secretariat recognizes that members do not want additional electronic communication, and three requests a year seems sufficient. The Secretariat also tries to limit email communication to PIs and cohort members to no more than once a month; if additional key pieces of information are needed for proposal submissions, members should alert the Secretariat. The portal will continue to be updated. The annual meeting is a main communication tool, and members were encouraged to complete the meeting evaluation survey.
During last year's meeting, members suggested examining non-cancer outcomes. Although two pilot projects were attempted (Vitamin D and ALS), non-cancer research by the Consortium has been difficult. Discussions concerning second cancers have increased in recent years, and the presentation from the Second Cancers WG provides impetus to determine additional opportunities in that research area.
In closing, Dr. Cerhan acknowledged the dedication of Dr. Tricia Hartge and the managerial support of Ms. Nonye Harvey of DCCPS' EGRP and officially turned over the chairmanship of the Secretariat to Dr. Zeleniuch-Jacquotte.