Working Group on Future Research Opportunities in the Women’s Health (WHI) Extension to 2015
Rockledge 2 Building, Room 9100/9104, Bethesda
July 1, 2008
Welcome and Charge
Dr. Lauer outlined the charge to recommend areas of research that WHI can contribute to during an extension from 2010 to 2015, particularly in regard to understudied health issues in older women. By virtue of its origins, the scope of WHI research goes beyond that of most NHLBI programs. The advice of the working group on the science is an important initial step in the process as NHLBI considers renewal of the program. NHLBI will consider the proposed research in the light of its overall research portfolio, its strategic priorities, and available funding.
Background and Process for Renewal
Dr. Rossouw reviewed the objectives and design of the original WHI (1993-2005) program:
- To address the etiology and prevention of the major causes of morbidity and mortality in post-menopausal women
- Enrolled 68,132 women aged 50-79 in three clinical trials
--Hormone therapies to prevent CHD
--Dietary modification to prevent cancer
--Calcium and vitamin D supplementation to prevent fractures
- Enrolled 93,676 women in an observational study
He summarized the main findings from the 3 trials and the impact the hormone trials had on medical practice. The clinical trials provide the opportunity to evaluate long-term effects of the interventions on disease outcomes, and the observational study data provide an opportunity to compare associations in the observational study with findings in the clinical trials.
During the current WHI Extension study (2005-2010) 115,405 women have been re-consented, with high retention rates and near-complete adjudication of outcomes, yielding a one-third increase in number of outcomes to date. To date, the extension study has yielded about 50 publications and 7 new ancillary studies per year, in addition to core studies and 12 Broad Agency Announcement (BAA) studies. He outlined the resources created, to which is being added a medications inventory and CMS linkage. The most heavily studied outcomes include CHD, stroke, VTE, breast and colorectal cancer, and hip fracture, but even in those categories blood and DNA samples remain available for further study. A shift in focus to other outcomes during a further extension could take advantage of unused blood samples. Several avenues for sharing the resource are being utilized, including the BAA mechanism and making available genome-wide association data. Approximately one-half of ancillary studies and one-third of publications are led by non-WHI investigators.
The process for a potential renewal includes internal NHLBI meetings and external review by the Board of Extramural Experts and the NHLB Council prior to issuance of a solicitation in 2009.
Objectives for Renewal and Investigator Presentations
Dr. Stefanick presented the objectives for a proposed WHI extension to 2015 as being similar to those for the current extension in some respects:
- Enhance ability to study subgroups and rarer outcomes
- Serve as platform for additional studies (core, ancillary, BAA, consortia)
- Study longer term effects of clinical trial interventions
New research proposed for the extension to 2015 is focused on use of the cohort to investigate conditions that affect health of an aging population of women:
- Aging (functional status and frailty, healthy aging, osteoporosis and risk for fracture, cognitive function)
- Cardiovascular (heart failure, atrial fibrillation, venous thromboembolism)
- Cancer (longer term effects of interventions, cancer risk factors and pathogenesis, discovery and validation of biomarkers for early detection, breast cancer survivorship and management
- Overarching themes (improving dietary and physical activity methodology, applications of genomic and proteomic technologies, social and built environment).
- Maintain the current infrastructure of 39 field centers, a coordinating center, and study committees
- Maintain central database, data analysis capabilities, and bio-repository
- Further enhance the WHI resource (additional 5 years of outcome data, CMS and other linkages, medications inventory, ancillary studies)
- Conduct new research on aging (~100,000 women expected to participate, of whom ~80% will be >70 years old)
Dr. LaCroix presented an overview of the aging proposals, stating the overall question as “What are the factors that determine whether an older woman can live a long, healthy, relatively independent life?” She noted that by 2010 the WHI cohort is projected to include ~54000 women aged >70, and ~38000 aged >80 years. WHI has multiple measures of frailty, disability, and healthy aging including those that have been used in other studies. WHI investigators are currently conducting studies on the inflammatory pathway and incident frailty (drugs, biomarkers, candidate genes). Questions that WHI can address in regard to longevity include adult nutritional patterns using calibrated measures, biomarkers (including genomic and proteomic markers), use of medications, lifestyle and behavioral characteristics, race/ethnicity, exposure to trial interventions. WHI has a strong osteoporosis/fracture component and can address a similar variety of factors that relate to osteoporotic fracture in older women, and can also address novel osteoporotic phenotypes. She also stated that the cohort could be considered for potential randomized controlled trials of prevention strategies.
Dr. Shumaker outlined the suite of studies related to cognition and dementia in WHI that are all nested within the hormone trial. Strengths of the studies include their size, follow-up for more than 10 years, high retention rate, assessment of multiple cognitive domains using clinically sensitive measures, the inclusion of pre-clinical outcomes (mild cognitive impairment, vascular cognitive impairment, MRI) and careful ascertainment and adjudication of dementia. She noted that the adverse effects of hormone therapy persist during the follow-up. Future research directions that will be enhanced with continued follow-up of the cohort include risk factors for cognitive decline and impairment, relationships between cognitive function and physical function/disability, cognitive reserve, and biomarker discovery.
Dr. Manson emphasized that heart failure and atrial fibrillation are major and growing epidemics, with a disproportionate burden in women (because of the large number of older women). Venous thromboembolism also increases rapidly with age and remains understudied. Studies on CVD fit into the overarching theme of determinants of healthy aging.
Specifically, heart failure in women differs from that in men (more non-ischemic causes, preserved systolic function, co-morbidity, symptomatic and worse clinical outcomes, impairment of health-related quality of life and activities of daily living). WHI can contribute to understanding heart failure in older women and in minority women. WHI will need to resume adjudication of self-reported heart failure and standardize classification, perhaps using algorithms similar to those used in Framingham (for hospitalized cases) and the Atherosclerosis Risk in Communities study (for ambulatory cases). Research questions include whether risk factors for incident heart failure differ by age or race/ethnicity, preserved versus impaired systolic function, mortality rates by type of heart failure, relationships between heart failure type and functional status, cognition, other CVD, and HRQOL, biomarker and genetic, proteomic, and metabolomic studies. Geocoding could contribute information on air quality, particulate matter, social and built environment. CMS linkage could contribute information on cardiovascular service use and costs.
Atrial fibrillation is understudied in women in other cohorts. Ascertainment of atrial fibrillation will occur through a combination of methods: self-report followed by records review (including for clinical diagnosis, ECGs, imaging data), ECGs done serially in clinical trial participants, and CMS linkage. Research opportunities include improvement of risk prediction including traditional CHD risk factors, demographic factors, sleep disorders, biomarkers, genetic and proteomic predictors, dietary factors, medications use. The interrelationships of atrial fibrillation with heart failure, stroke, other CVD, physical and cognitive function, depression, and HRQOL could be studied.
WHI has the potential to become one of the largest prospective cohort studies of venous thromboembolism worldwide, with 1200 projected cases in the hormone trial cohorts by 2015, and could study genetic, biomarker, and lifestyle risk factors relevant to the pathways of risk for VTE. The relationship to air pollution/particulate matter and the interrelationships with frailty, other CVD, cancer, and all-cause mortality are other research opportunities.
Dr. Prentice focused on initiatives that cut across study diseases: dietary and physical activity epidemiology, use of genomic and proteomic technologies, and social and built environment, in addition to specific studies of longer term effects of clinical trial interventions, cancer risk factors and pathogenesis, early detection of cancer, and breast cancer survivorship and management. Nutrition and physical activity epidemiology: WHI has established comprehensive nutrition and physical activity resources (Food Frequency Questionnaire (FFQ), 4 Day Food Records (4DFR), 24-hour recall, physical activity questionnaire (PAQ), calibration studies of energy and protein consumption, and biomarkers of nutrient intake). Using regression coefficients from calibration models, the FFQ underestimates both energy intake and protein intake. Several cancers that appear not to be related to energy intake are significantly associated after calibration of FFQ data. During the current extension these FFQ-based association studies for energy and protein can be extended to CVD, diabetes, and weight change; also, analyses of the broader range of dietary instruments, expenditure studies, and biomarkers will be completed. During a further extension to 2015 the potential of the respiratory quotient from indirect calorimetry to calibrate fat and carbohydrate intake can be examined, and applied to FFQ and 4DFR to examine associations with cancer, CVD, diabetes, and weight change. Similarly, calibrated PAQ-based association for these same outcomes can be performed.
He summarized the WHI involvement (and some preliminary data) in genome-wide association studies (GWAS) and confirmation studies on CHD, stroke, breast cancer, hip fracture, pancreatic cancer, and SHARe. Some of these GWAS/confirmation studies are done with core resources, others are ancillary studies, BAA studies, collaborations with NCI, or funded directly by NHLBI in the case of SHARe. There is considerable potential for further studies, including studies of gene-environment interaction, given the large numbers and rich phenotyping of the cohort (including trial exposures). WHI has several proteomic initiatives, again funded through a variety of mechanisms including collaboration with NCI. Most of these are at the stage of discovery prior to validation with ELISA. The results from a core study of proteome changes in response to hormone therapy could be validated in several instances by ELISA and are biologically plausible.
Longer-term effects of interventions will benefit from further follow-up; interesting (and somewhat concerning) initial data on cancer during the first several years of follow-up of the E+P trial have been published. The post-intervention data can help interpret national trends breast cancer following changes in patterns of hormone therapy use. In the DM trial a low-level maintenance of the intervention is being implemented and it is expected that any effects on cancer will take a decade or more to emerge. Numbers of incident cancers are expected to be 80-100% greater by 2015 compared to 2005 and will include substantial numbers of less common cancers (e.g., ~1200 for ovary and NHL; ~700 for bladder, pancreas, leukemia, kidney) and for less common subtypes of cancer (e.g., ~1800 for ER-, PR- breast cancer) or for less common exposure subtypes (e.g., ~450 lung cancers among never-smokers). The outcome and exposure data can be used for a wide array of risk factor and etiologic studies, including the development of improved prediction models. Work on discovery and validation of cancer early detection biomarkers will continue, using quality-controlled pre-diagnostic serum and plasma. He gave examples of studies of breast cancer survivorship and management that could be done. He proposed that WHI collect tumor blocks to allow for improved tumor profiling by gene expression and other molecular profiling of tumor tissue.
Discussion and Prioritization by Working Group
For this part of the discussion, two investigators (Drs. Stefanick and Prentice) were invited to remain as observers and resources; the Chair had the option of calling for a closed session.
The working group discussed each of the research areas proposed, and prioritized the areas (this activity was finalized by a brief questionnaire completed after the meeting). The group also commented on the general structure of any follow-up.
The comments, recommendations, and priorities summarized are those of the working group and do not reccessarily reflect the views of NHLBI. These and other inputs will be considered by NHLBI through internal processes, with eventual further reviews by external experts.
--There was a dichotomy of opinion as to how the cohort could best be leveraged; one group favored specific hypotheses to be explored in subsets of the cohort (or a smaller overall cohort), while the other viewed the cohort as an invaluable resource for high-dimensional studies requiring large numbers of well-phenotyped outcomes and therefore recommended following all participants.
--The CT cohort was seen as more valuable than the OS cohort because of the specific randomized exposures, which for example make it ideal for studies of gene-environment interactions.
--The group acknowledged the collaborative record of the WHI group, and encouraged training workshops for younger investigators in the use of WHI data, and special programs for minority investigators.
--Some recommended streamlining of outcomes data collection, e.g., further centralization of outcome data collection, or passive data collection, in lieu of field centers.
--Do not need precise diagnosis for studies of health resources/cost, do need for etiologic studies.
--If interviewer-administered questionnaires, physical measurements or blood collection are needed for specific substudies, home visits by research staff may be more cost-effective than clinic visits (there are companies that offer this service on national basis)
--Need to bring in additional expertise in areas where WHI investigators are not expert (e.g., use of CMS data would need outcomes researchers, heart failure, atrial fibrillation research).
--Encourage efforts to re-enroll women who have dropped out.
--Consider layered consent, e.g., women could agree to participation in studies involving matching to administrative files/obtaining surgical specimens/questionnaires.
--Generalizability to US population can be improved by weighted analyses.
--Better integration of ancillary study data into study database.
--Invest in getting more phenotypes in women who were studied in more detail, e.g., GWAS substudy, DXA substudy.
--Uncertain about use of cohort for new trials. Consider high-dose vitamin D trial.
CMS linkage—very high priority
--CMS linkage supported for case-finding (annual cycles), ascertainment of other outcomes e.g., Parkinson’s, and co-morbidities. For health services research data can be obtained in batches, less frequent than annually.
--CMS data could be used at regional differences in medical practices and outcomes, health services outcomes associated with cognitive impairment, congestive heart failure.
--Need to involve experts—data are “dirty”.
--CMS data increasingly important in aging cohort, who have high likelihood of suffering an event or death before annual self-report.
--Consider substituting for ascertainment of most outcomes to reduce costs
Medications inventory—very high priority
--Support collecting data on medication use more proximal to event. Note polypharmacy in elderly, potential for medications to be associated with adverse outcomes. Useful for disease classification, e.g., self-report of T2DM plus medication, self-report of Parkinson’s plus medication.
Aging—very high priority overall
--Ideal cohort to study aging. Aging multidimensional, need systems biology approach, thus needs large sample size and measures that fit a specific paradigm. Pick aging as focus, need to integrate initiatives into overall concept. All diseases are related to aging, try to coordinate biomarker measurements relevant to all outcomes. Integrate studies of physical and cognitive function into studies of diseases of old age.
--However, the lack of physical measures would impair the ability to do aging work, as would the lack of data on cognitive function are lacking in the majority of women.
--Aging as continuum (health---disability---death) vs. separate trajectories.
--Focus on products, e.g. WHI risk calculators.
Functional status and frailty—very high priority
Successful aging and longevity—very high priority
--Rapid fall-off due to death in women after 85 makes large cohort necessary to study extreme longevity.
--Weakness—no continued measurements of mental health, social support, gait/walking/mobility—perhaps assessment by proxy, or telephone administered measures could be implemented.
--One member recommended “aging research cohort” comprised of CT participants.
--While there are many other studies of frailty, WHI offers exceptional opportunity to study successful aging, paradoxical phenotype (have risk factors but do not get disease).
--Recommend continued data collection on obesity and diabetes in follow-up, others have experience in self-report of body size measurement (weight, height, waist circumference).
Cognition—very high priority
--Need more information on effects of HT on cognition in women starting close to the menopause. Suggest study broader age range in all CT cohort.
--Information on co-morbidity needs to be built into study of aging and cognition.
--For applications of genomics/proteomics technologies to cognition WHI cohort is unique.
--Need functional measures during follow-up.
Osteoporosis and fracture—moderate priority (however, note osteoporosis expert could not attend meeting).
--Other cohorts are studying this, perhaps follow CT only, or only in Asian/Native American where little data.
Cardiovascular disease—moderate to high priority overall
Continued ascertainment of CHD, stroke—high priority.
--Opportunities to study CHD, stroke subsyndromes, most studies in men. Construct stroke risk model.
Heart failure—high priority only if can be better characterized
--Important and understudied disease, especially heart failure with preserved systolic function. Difficult to study, but somebody has to do it.
--Problem with characterization is common to all epidemiologic studies of heart failure. Potential solution: can we get digital record of echo at time of hospitalization to read centrally, quantify parameters in standard fashion. Not convinced Framingham heart failure algorithm works for obese people, does not work for diastolic dysfunction. Get BNP from chart.
--Good opportunity for studies by race/ethnicity, health services disparities.
--Need good phenotyping for etiologic studies, not necessary for studies of health care utilization/costs.
--Additional expertise would be needed.
Atrial fibrillation—high priority only if can be better characterized
--Incident AF difficult to document.
--Additional expertise would be needed.
Venous thromboembolism—moderate priority
--Factors underlying increased risk with aging and with hormone therapy are not well documented
--Relationships to physical activity, renal function, hemostasis, functional outcomes of VTE, relationship of VTE to future CVD not well studied.
--Potential for gene-environment interaction studies.
--Perfect cohort for study of particulate matter and VTE.
--Paucity of studies looking at VTE, important cause of impaired QOL, health care utilization, and mortality.
--Not all members were enthusiastic, many of the hypotheses could be explored using existing data.
Cancer—moderate to high priority overall
Longer term intervention effects—high priority
--Not unique cancer cohort, except for trial exposures—only WHI can do latter.
--Ethical imperative to study long term effects, given harm observed in hormone trials.
--Question how real the difference is in fat intake post-trial—need biomarkers. Study carotenoids.
--Body weight needed in follow up of trials to better interpret findings in the DM trial.
Cancer risk factors and pathogenesis—moderate to high priority
--Support getting tumor blocks—tumor genetics are going to be much more informative than blood genetics. Tumors in CT, specifically for breast, colon, endometrial highest priority.
--Create resources for proteomics, genomics (including sequencing, confirmatory studies), metabolomics. Needed to pool 15 cohorts for study of pancreatic cancer, need large studies like WHI.
Discovery and validation of biomarkers—very high priority
--Availability of pre-diagnostic bloods, good phenotyping.
--Opportunity to look at long-term prediction, lifetime risks using earlier phenotyping and biological specimens.
Breast cancer survivorship and management—moderate priority
Applications of genomic and proteomic technologies—very high priority
--Potential for studies of copy number variants, epigenetics, gene-environment interactions (in trials).
--Large numbers needed for GWAS and studies of GEI.
--WHI ideal for replication studies, subphenotyping.
--Good record for opening resource, high quality samples, large number of cases for validation.
--Divergent views on genetics in older persons: some think not important in old survivors, while others think multiple common genes continue to play a role.
Nutrition and physical activity epidemiology—moderate priority
--Diet and PA affects or modulates many outcomes of interest. Support biomarkers to calibrate reported energy, protein, potassium intake, respiratory quotient for fat and carbohydrate energy.
--New dietary measure—web based 24 hr. recall used by NCI.
--Problems with reported physical activity as severe as for diet, maybe use accelerometers, heart rate variability.
--Methods research can greatly contribute to knowledge of measurement error and correction for it.
--Continue to collect dietary and PA data on DM trial cohort.
--Continue supporting analyses of current FFQ and PA instruments.
Social and built environment—moderate priority
--Opportunities by linkage to existing datasets, assess whether independent predictors.
--Need to tailor to demographics, elderly may be different.
Additional Research Areas:
--Parkinson’s—self-report+meds has good positive predictive value
--Neuropathy (questionnaire and monofilament exam)
--Sensory impairment, hearing loss, cataracts, pneumonia, other infections
--Health services research
--Epigenetics—collect DNA during follow-up
--Death rates from liver disease could inform the issue of screening for genetic hemochromatosis in women
--Rates of surgery, mortality for abdominal aortic aneurysm could inform issue of screening in women
--Mitral valve disease, bypass surgery.
Working Group Members
Deborah Grady, MD, MPH (Chair)
Javed Butler, MD, MPH
Tim Byers, MD, MPH
Mary Cushman, MD, MSc
Luigi Ferruci, MD, PhD
Tamara Harris, MD
Robert Hoover, MD
Tom Pearson, MD
Diana Petitti, MD, MPH
MaryFran Sowers, PhD (call-in)
Sudhir Srivastava, PhD, MPH
Amy Subar, PhD, MPH, RDH
Eric Whitsel, MD, MPH
Lawton Cooper, MD, MPH
Peter Kaufmann, PhD
Michael Lauer, MD
Shari Ludlam, MPH
George Papanicolaou, PhD
Jacques Rossouw, MD
Joan McGowan, PhD
Neil Buckholtz, PhD
Andrea LaCroix, MPH, PhD
JoAnn Manson, MD
Ross Prentice, PhD
Sally Shumaker, PhD
Marcia Stefanick, PhD
Visit the Women's Health Initiative Web site.
^ Back to top