Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.
OCR for page 136
Biosocial Surveys 6 The Women’s Health Initiative: Lessons for the Population Study of Biomarkers Robert B. Wallace Much of the discussion in this workshop has been directed to the study of biomarkers in cohort studies. My commentary addresses some lessons learned from the Women’s Health Initiative (WHI), a research program comprising a set of large, randomized clinical trials as well as a cohort study. These lessons derive from the intricate and complex management issues relevant to multisite collaborative studies, the challenges of establishing and maintaining a large biospecimen repository, the difficulties of providing support for many investigative groups, the response to the changing bioethical environment, and the potential for future biomarker applications in the health and social sciences. WHI began in 1992 and is ongoing. Its main goal has been to evaluate a series of interventions with the potential to prevent the onset of important chronic conditions in postmenopausal women, but, as discussed below, once the study data and biological specimens were in place, the opportunities for additional studies, some not previously planned, became substantial. In this report I briefly review the structure of WHI, describe the biomarker experience to date, discuss some of the strengths, weaknesses, and issues that arose in this very large, multicenter study, and offer general suggestions for further exploitation and investigation of stored specimens. In addition, I address some of the social and organizational aspects of managing biomarker studies and study investigators, as well as the large data sets, specimen collections, and scientific applications.
OCR for page 137
Biosocial Surveys STRUCTURE OF THE WHI WHI is comprised of a program office at the National Heart, Lung, and Blood Institute, National Institutes of Health (NIH), 40 clinical sites at which the participants are seen, a clinical coordinating center at the Fred Hutchinson Cancer Center in Seattle, and other supporting academic and laboratory units. The study design has been described in detail (Women’s Health Initiative Study Group, 1998). Basically, from 1993 to 1997, postmenopausal women ages 50-79 were recruited into a set of three large, preventive, randomized, placebo-controlled trials, listed as follows with the primary outcomes: (1) estrogen alone or estrogen plus progestin therapy to assess the effect on coronary heart disease; (2) low-fat diet to prevent incident breast cancer; and (3) calcium and vitamin D dietary supplements to prevent clinical fractures. The last trial, the calcium/vitamin D intervention, was a superimposed randomization on the participants of the other two trials and did not include separate participants. Of note, that makes the interpretation of the outcomes a bit more complex, despite the randomized experimental design, because the impact of the other intervention(s) must be accounted for in all analyses of outcomes. Recruitment to WHI was conducted by the 40 participating clinical sites, using such methods as media advertising, direct population mailing, community lectures and public service announcements, and celebrity endorsements. This yielded about 353,000 women who were evaluated for participation in WHI. About 68,000 enrolled in the trials. Of the remainder, those who were ineligible, unwilling, or unable to participate in the clinical trials were invited into a cohort study, the Observational Study (OS), in which no interventions took place; about 93,000 women began full participation in the OS. There were many specific reasons why potential trial participants opted for OS participation. Three of the most important were not being able to spend the time undergoing the trials’ rigors; unwillingness to be randomized with respect to the use of hormonal therapy; and having too low a percentage of dietary fat calories, making them ineligible for the dietary modification trial. Participants in the clinical trials as well as the OS had the same, extensive baseline data collection, including medical history, risk factors for important chronic illnesses, social and behavioral characteristics, mental health characteristics, and diet and dietary supplement use. In addition, and important to this discussion, virtually all participants had blood obtained through venipuncture, which was aliquoted and stored, and much of it is available today for additional studies, in addition to the many ancillary biomarker studies that have already taken place. Additional blood samples were obtained from some participants at selected intervals after the study onset. Of interest, aside from some planned safety
OCR for page 138
Biosocial Surveys studies related to the preventive trials, there were no specific plans on how the specimens might be applied or what determinations would take place. Yet many important issues have been addressed, related to modern and evolving scientific hypotheses, and some specimens have been used to better understand certain adverse effects of the trials that were not anticipated when they began. Although the major results of these trials have been published (Writing Group for the Women’s Health Initiative, 2002; Women’s Health Initiative Steering Committee, 2004; Prentice, 2006; Jackson et al., 2006), much more work is occurring, and WHI has resulted in over 100 publications to date. Of import to the application of biomarker studies, there are many secondary and ancillary scientific hypotheses being evaluated with WHI specimens. These studies have consumed some of the blood specimens acquired in WHI, an important issue in itself. The specimens collected in WHI continue to be used for many scientific purposes, under the heading of “ancillary studies,” which are described below. SPECIMEN ACQUISITION AND THE INSTITUTIONAL REVIEW BOARD Although many issues regarding the ethics of specimen collection are beyond the scope of this discussion, some issues and situations concerning the WHI experience are worthy of note. A general problem of multicenter studies is the requirement that all institutional review boards (IRBs) approve study protocols. At times that was true of WHI, in which there were 40 clinical site IRBs, the coordinating center IRB, and ethical review at NIH and perhaps other sites, such as at institutions performing biomarker determinations. Because WHI took place over many years, scientific advances occurring after the study was designed offered the possibility of new biomarker determinations on existing specimens. However, the initial informed consent for blood or urine or other specimen collection may not have covered some of the new, unexpected, and unspecified determinations that had not yet been conceived or invented. This can be solved in part by general consent form language that allows future, unspecified measures to be studied. Yet this may not cover all situations. An example is when a biomarker is developed that turns out to have the capacity to screen for a specific medical condition. For example, early studies of prostate-specific antigen (PSA) may have had biological import only, but as it became a candidate for clinical prostate cancer screening, studies conducting PSA determinations would need to have medical expertise available to dispose of clinically elevated levels. The study of biomarkers that turn out to have clinical import raises several important bioethical issues, requiring consid-
OCR for page 139
Biosocial Surveys erable thought and medical consideration. Because of changing scientific interests and biomarker applications, WHI made a decision to reconsent surviving participants in 2005 for further, unspecified use of stored blood specimens, in order to modernize consent form language, address new tests being performed, and remind participants that the specimens were still in use. Both in collaborative and individual studies that maintain biospecimen repositories, oversight by IRBs is increasing and becoming more complex. For example, some universities require that all biorepositories have separate annual review, regardless of use. MANAGING BIOMARKER STUDIES The number of WHI investigators is large and may well become larger as other investigators take advantage of the biomarkers collected and stored by the study. In this section I discuss some of the logistical issues surrounding the management and disposition of blood biomarkers in WHI, with lessons for other long-term clinical trials as well as observational cohorts. Some context is needed when describing the WHI experience. WHI investigators early on developed priorities for specimen use. The first priority was for quality control and management of the study. Other biomarker applications, such as explaining WHI findings, pursuing industrial applications, and joining collaborative biospecimen groups, are discussed below. In addition, a major disposition for collected specimens is WHI investigator-initiated ancillary studies, those not part of the original WHI mission but that are deemed to be high-quality science and to have relevance to women’s health issues. Individual investigative groups in WHI may have collected additional specimens in local studies, but these were not provided to the parent WHI repository. The main WHI repository is gradually being made available to collaborating, non-WHI investigators. While it is not the intent here to make a detailed account of the techniques or logistics of specimen collection, the following is an overview of selected procedures and activities in WHI. In specimen collection, all participating staff were trained in the methods of specimen acquisition and initial processing, not only to standardize and preserve the accuracy of the determinations, but also to protect them from untoward exposure to biohazards. This was true even for staff who had substantial experience in prior studies or such techniques as venipuncture. Equal attention was paid to the timing of specimen collection, initial processing, shipping methods, and shipping duration in order to ensure that distortion of biomarker findings was minimized (Landi and Caporaso, 1997). Although not performed in WHI, if specimens for cell and molecular biological determinations were being performed, such as using cell cryopreserva-
OCR for page 140
Biosocial Surveys tion, additional steps would be needed (Holland, Smith, Eskenazi, and Bastaki, 2003). The same principles would be true for other types of biomarkers in population studies, such as trace elements in bodily fluids, microbiological specimens, or micronutrients (Wild, Andersson, O’Brien, Wilson, and Woods, 2001), each of which involves some separate processing issues.1 Whatever the scientific goals for specimen collection, there is a need to ensure laboratory quality control, both at the start of the study and over time for longitudinal studies (Gompertz, 1997; Tworoger and Hankinson, 2006).2 A corollary but critical issue is that the methods of specimen collection, handling, and storage should be well documented for use by future investigators. At the very least, such information should either be published in the indexed literature or made publicly available through protocol documents. There are several issues related to the burdens of managing specimen distribution after they are collected and stored. With a large number of specimens, there is a problem of registering and tracking the specimens—where they are, how much has been extracted, and how much is left for subsequent determinations. Not surprisingly, all of this requires a computerized database that is frequently updated and provides information on the type and quantity of specimens, the number of times the specimens have been thawed, if at all, and, ideally, specific reference to databases in which previous determination results of those specimens can be found. Creating and maintaining a repository database can be a costly and demanding activity and must be confronted early in study design. The issue of specimen stability and integrity over time is critical for biorepositories and their long-term applications. Scrupulous records must be kept on storage conditions, possible power outages, specimen retrievals, and other lapses in specimen maintenance. Even under optimal storage conditions, some potential and future biomarkers may not remain well preserved. In some cases, certain molecules are known to survive long periods, such as immune globulins and steroid hormones. However, other moieties, such as peptide hormones, may not. When proposing to use specimens that have been stored for long periods, the investigators should either cite the evidence for specimen stability or propose studies to demonstrate it. There are complex logistics with regard to distributing and retrieving stored specimens. Can they all be identified and retrieved in a timely 1 Additional useful information can be found at the website of the International Society for Biological and Environmental Repositories: http://www.isber.org. 2 The September 15, 2006, issue of Cancer Epidemiology Biomarkers and Prevention has several useful articles related to biomarker acquisition, processing, and storing in population studies.
OCR for page 141
Biosocial Surveys manner? How are the fates of the remnant specimens monitored, and will these specimens be sent back for storage and forwarded to another investigator for other determinations? Is there a need for the parent study to monitor the quality of laboratory determinations that are done as part of ancillary research studies? Addressing this latter point is particularly important, because some research determinations may have direct implications for participant health status. If biomarker determinations are to be used for clinical alerts or those results are returned to participants for possible clinical use, then there may need to be appropriate certification of the laboratories providing the determinations. As is true of all biorepositories, someone has to pay for the management of specimens used for clinical alerts, and this should be considered when such studies are designed and budgeted. In this case, all storage and retrieval activity was sponsored for WHI by the National Heart, Lung, and Blood Institute. However, as new cohorts and other studies contemplating the acquisition of biomarkers are founded, budgets must include provision for these activities. A related but important concern is unapproved uses of distributed specimens to study investigators and their collaborators for ancillary research projects. If one gives an investigator specimens to do a certain number of tasks that have been scientifically endorsed, what if they use the specimens for another purpose, even if closely related to the approved application? For example, if an investigator is approved for determining certain candidate genes, is determining nearby regulatory genes or other, previously unspecified genes appropriate? In my view, this must be tightly controlled, because the specimens are commonly owned by all the investigators, both for their intellectual value and for their potential future applications. One must maintain the rules about how the specimens are used and about returning them when the assays are completed. Of course, it is possible to request additional applications of the specimens to the appropriate committee; in the case of WHI, it is the Ancillary Studies Committee. Another question is whether ancillary study findings related to biomarkers should at some point be released to the scientific community or the public. That is, most WHI investigators think it is important to provide the findings from local ancillary studies to the WHI data bank back for further study after the primary scientific reports have become public, allowing other investigators to use the biomarker determinations for other hypotheses and avoiding duplication of cost or accelerated specimen depletion. In the longer term, NIH policies are likely to demand that the data become available from large studies such as WHI, but rapid archiving of information will accelerate this scientific process. Box 6-1 is a partial list of selected biomarkers that have been proposed for or performed in ancillary studies in WHI, to give a sense of the scope
OCR for page 142
Biosocial Surveys BOX 6-1 Examples of Biomarkers from Women’s Health Initiative Safety and Ancillary Study Protocols Markers of Inflammation: IL-6, ultrasensitive C-reactive protein; tumor necrosis factor alpha Micronutrients: alpha and beta carotene; tocopherol; vitamin C; vitamin D Obesity-Related Hormones: adiponectin; ghrelin; leptin Gonadal Hormones: estrogens, progestins; hormone-binding globulins; markers of testosterone metabolism Pituitary Axis Hormones (free and bound): corticosteroids; growth hormone; IGF-1; thyroid and parathyroid hormones Clotting Factors: Factor VIII; Factor IX; elements of the complement cascade; thrombin; prothrombin; fibrinogen Lipids and Lipoproteins: LDL, HDL, and VLDL cholesterol and subfractions; lipoprotein particle size Tissue and Vascular Matrix Factors: E-selectin; matrix metalloproteinase factors of the biomarkers derived from investigator-initiated hypotheses that could be determined from frozen plasma, serum, and buffy coat/DNA. It is clear that many biomarkers of potential interest cannot be determined on these frozen blood materials. However, a potentially large number of determinations are available; an extensive discussion of organ- and disease-specific biomarkers is summarized in a recent compendium (Trull et al., 2002). As of this writing, all large NIH-funded projects require a data-sharing plan. WHI has received requests for data from persons outside the study, and the usual issues of maintaining confidentiality as well as allowing investigators to complete and publish their original studies are important here. Biomarker studies in general may have special issues of confidentiality, particularly for unique genetic coding sequences, and this must be considered when contemplating the release of such information. Whether distributing specimens to WHI investigators or to outside collaborators, the issue of developing procedures and priorities for specimen use is paramount. To begin, criteria are needed as to who is a suitable investigator to apply for the specimens and which investigators have priority. In the case of WHI, nonstudy investigators may apply only if they are affiliated with an existing WHI investigator or approved by the steering committee. Of course, the scientific promise and feasibility of a proposed ancillary study, and the relevance to the overall mission and themes of WHI, should be strong determinants of who receives speci-
OCR for page 143
Biosocial Surveys mens. Also, as an extra precaution with regard to retaining some control over specimen disposition, it is probably important to release specimens only to institutions that have IRB approval mechanisms, institutional willingness to accept responsibility for the study, and possibly also an ongoing relation to NIH, so that some level of administrative communication and quality assurance remains. Whether a proposed biomarker study has funding in place at the time of application is a difficult conundrum. If repository specimens are promised and held for yet unfunded studies, the delay in acquiring funds may occupy specimens that otherwise could be put to an equally important use. However, applicants must demonstrate to a funding agency that the specimens will be adequate for their proposed goals (i.e., conduct pilot work) and also demonstrate ensured specimen access. The specimen distribution process would be simplified if the applicants had funding already in hand, but there is no easy solution to this dilemma. Another issue with respect to specimen allocation priorities is that, as ancillary studies progress and specimens are consumed, certain types of specimens may be in relatively short supply. Often, these are specimens from participants who had the outcomes of primary interest, conditions such as myocardial infarction, breast cancer, or bone fracture. In WHI, as in some other studies, planning for specimen use priorities among those with the greatest call may require special planning and political negotiation. Even when agreed-upon rules are developed, they may have to be revisited for important specimens that are near exhaustion. Despite the large amount of work and the committee process to create priorities for specimen allocation, there may be superimposed additional priorities that must be rationalized within the existing system. In the case of WHI, two major situations arose. One that commanded a special priority was the decision on the part of the investigators to explore some possible biological mechanisms related to the adverse events that occurred with the use of estrogen plus progestin. Substantial research using biomarkers is being conducted in WHI to determine possible causes of the adverse events as well as potentially to determine if specific high-risk groups can be defined. Laboratory investigations are being considered to explore other adverse effects that occurred with WHI interventions, such as an increased risk of dementia associated with hormone therapy (Shumaker et al., 2003) and a 17 percent increased risk of kidney stones associated among women taking calcium and vitamin D supplements. A second general priority was invoked when WHI was contacted by certain private companies, which inquired as to whether specimens and health outcome data might be used to search for genetic causes of the outcomes using high-throughput methods that otherwise would not be available to WHI. After considerable consultation with investigators
OCR for page 144
Biosocial Surveys having suitable expertise and several discussions by WHI investigators, it was decided to allow a portion of the specimens to be used for this activity. In some instances, complex negotiations with regard to ownership of the findings and their publication were necessary. Work on this partnership activity is continuing, and the possibility of defining genetic and other molecular risk factors for the hormone-related adverse effects may be important in the future. ADDITIONAL APPLICATIONS FOR BIOMARKER STUDIES IN WHI Participation in Large Scientific Consortia Emphasizing Biomarkers In addition to the priority dispositions of biomarker specimens generated by WHI leadership and investigators, larger population studies and clinical trials should consider partnering with collaborative studies that pool individual study data and findings, and thus have the capacity to explore important scientific questions that no one study can attain. As an example, WHI has agreed to participate in CGEMS, the Cancer Genetic Markers of Susceptibility project (find more information at http://cgems.cancer.gov). This is a three-year project sponsored by the National Cancer Institute that explores genetic alterations that lead people to be at greater risk of breast and prostate cancer. Genetic case-control studies are performed with large sample sizes, evaluating over 500,000 tag single-nucleotide polymorphisms. Although such participation may dilute individual scientific and authorship credit, it may allow scientific discovery that otherwise would not be possible. In addition, negotiations are under way to utilize WHI biospecimens for proteomic studies. Applications of WHI to the Social and Behavioral Sciences It is perhaps not adequately appreciated that WHI has many measures of social and behavioral function that might be of use to investigators interested in those domains, in addition to more disease-oriented themes. There are measures of general health status and health-related quality of life. Within the latter measures are subscales of physical functioning, role limitations in general and due to emotional problems, bodily pain, energy and fatigue, social functioning, and mental health. There are also cognitive measures, such as the Modified Mini Mental State Examination, a depression scale, measures of sleep disturbance, and items related to sexual satisfaction and preference. There are some basic assessments of physical functional status (activities of daily living) and, in a subsample, tests of physical performance. The WHI database can be a very rich source
OCR for page 145
Biosocial Surveys on the health, behavior, and aging of postmenopausal women, including the sociology and psychology of aging. Other Potential Research Directions One area of biomarker application that can be exploited in such studies as WHI, in which both the sample size and geographic coverage are large, is to address the association of disease occurrence or biomarker levels with variation in the ambient environment. For example, if a new biomarker is becoming available that reflects an exposure to the physiochemical environment, such as an airborne pollutant, studies like WHI can test whether such pollutant levels are perturbed by a general set of population characteristics, such as smoking, alcohol consumption, or variation in diet, as well as age, race, ethnicity, and many other potential confounders of interest. Once the properties of these emerging biomarkers are understood, it is possible to apply them to scientific hypotheses that will lead to more specific associations. Variation in geographic distributions of certain environmental pollutants measured though biomarker determinations can then also be explored, as well as relating these markers to physiological and disease outcomes. Another use of the biomarkers has been to validate questionnaire responses. In the past, there have been substantial work on the validation of self-reports of cigarette smoking by determining nicotine metabolite levels. An additional example is the validation of dietary questionnaires. Self-report of diet has always been problematic, and while most dietary instruments used in population studies have some demonstrated validity, there may be problems with accuracy and precision. For example, accurate measures have been critical to understanding the outcomes of the dietary modification trial conducted by WHI. Measuring of micronutrient biomarkers has been one way to address this problem. WHI is addressing the accuracy of dietary intake instruments in other ways as well: a doubly labeled water study has been performed to obtain an objective biomarker of energy intake, and verification of protein intake was conducted by comparing dietary reports with urinary creatine levels. Biomarkers can thus be used in selected situations to validate other kinds of data collection. It may be possible to use populations such as in WHI to explore health disparities, particularly those that may be manifest in biomarker determinations. With the range of ethnic and racial groups represented, the substantial representation of both rural and urban women, and the variation in health and functional status, this study in longitudinal context may provide substantial insight into health outcomes related to biological measurement. Available outcomes, such as institutionalization
OCR for page 146
Biosocial Surveys and mortality, are likely to be of interest in disparities research. It should be noted, however, that WHI did not provide general medical care and required that medical insurance for certain clinical outcomes be provided by the participants. Thus, there were fewer uninsured persons in WHI than in the general population. Finally, in addition to research applications, it is possible that population study biomarker repositories may provide a set of services for the participants, if that is desired, related to the clinical uses of the specimens. Archived blood components, for example, may have several applications, such as providing preinfection antibody levels if there is a question of an incident infectious process, in which diagnostic assistance is needed. Furthermore, there may be serological evidence of prior environmental toxicant exposures that could help in diagnosis. If a participant is deceased, stored DNA may be helpful to living family members if there is an issue of genetic transmission leading to a risk of a chronic illness. The use of such services may be infrequent, but their availability may be of public relations value to establishing and maintaining cohorts. STRENGTHS AND WEAKNESSES OF BIOMARKERS STUDIES USING CLINCAL TRIAL POPULATIONS It is important to note the issues that population scientists face when attempting to study clinical trial populations such as WHI and its companion observational cohort. There are many strengths to such an approach, including the very large sample size, the broad age distribution, the large number of minority participants, the broad range of social and behavioral measures available for study, the large and multidisciplinary group of investigators who can provide collaboration and consultation, the exquisitely documented nature of many of the health outcomes, including those that are less common and not part of the initial study design or primary outcomes, and the advanced planning for specimen storage and retrieval. Yet some weaknesses of this approach to secondary data analysis associated with primary exploration of the blood specimens are apparent. Most importantly, all of the cohorts in WHI are comprised of volunteers meeting several inclusion and exclusion criteria; these individuals are not likely to be referent to any particular geographically defined population. Among the most important criteria is the absence of disabling chronic illness and known preterminal or institutionalizable conditions. The OS is even further selected, since these are participants who had originally sought participation in the clinical trials. This lack of generalizability can be difficult to overcome if directly relevant to a particular set of scientific hypotheses, but nonetheless there may be suitable variation in
OCR for page 147
Biosocial Surveys study measures to allow detailed exploration of many important scientific hypotheses. Other potential impediments include differing propensities of the 42 IRBs to allow ancillary studies and the potential IRB demand for reconsenting of participants for new, previously unplanned determinations on archived biological specimens. In addition, the blood or blood components for specific participants may be dwindling or not available, and that may thwart certain types of studies of participants with specific characteristics. The specimens most likely to be exhausted are those belonging to participants with diseases and conditions comprising the primary outcomes of the study. CONCLUSION Although WHI is perhaps a larger study than most investigators will ever encounter, there are important lessons for all population investigators about the acquisition, management, and disposition of biomarker collections. It is hoped that clinical trial populations will be considered as worthy populations for study. In many instances, the distinct strengths of this approach to large group scientific inquiry should more than counterbalance the lack of population reference. ACKNOWLEDGMENTS The author wishes to thank the many investigators and staff at the study sites, the Clinical Coordinating Center, the participating centers that provided additional study management and laboratory determinations, and the National Institutes of Health, where staff scientists were so instrumental in the genesis and stewardship of the Women’s Health Initiative. In addition, the author is grateful to the over 161,000 participants who gave their time, energy, and dedication to exploring the frontiers of women’s health and elder health. REFERENCES Gompertz, D. (1997). Quality control of biomarker measurement in epidemiology. IARC Scientific Publications, 142, 215-222. Holland, N.T., Smith, M., Eskenazi, B., and Bastaki, M. (2003). Biological sample collection and processing for molecular epidemiological studies. Mutation Research, 543, 217-234. Jackson, R., LaCroix, A., Gass, M., Wallace, R., et al. (2006). Calcium plus vitamin D supplementation, and the risk of fractures. New England Journal of Medicine, 354, 669-683. Landi, M.T., and Caporaso, N. (1997). Sample collection, processing, and storage. IARC Scientific Publication, 142, 223-236.
OCR for page 148
Biosocial Surveys Prentice, RL. (2006). Research opportunities and needs in the study of dietary modification and cancer risk reduction: The role of biomarkers. Journal of Nutrition, 136, 2668S-2670S. Shumaker, S., Legault, C., Rapp, S., Thal, L., Wallace, R., Ockene, J., Hendrix, S., Jones, B., III, Assaf, A., Jackson, N., Kotchen, J., Wassertheil-Smoller, S., and Wactawski-Wende, J. (2003). Estrogen plus progestin and the incidence of dementia and mild cognitive impairment in post-menopausal women. The Women’s Health Initiative Memory Study: A Randomized Controlled Trial. Journal of the American Medical Association, 289, 2652-2662. Trull, A., Demers, L., Holt, D., Johnston, A., Tredger, M., and Price, C.P. (2002). Biomarkers of disease: An evidence-based approach. Cambridge, England: Cambridge University Press. Tworoger, S., and Hankinson, S. (2006). Use of biomarkers in epidemiological studies: Minimizing the influence of measurement error in the study design and analysis. Cancer Causes and Control, 17, 889-899. Wild, C., Andersson, C., O’Brien, N., Wilson, L., and Woods, J. (2001). A critical evaluation of the application of biomarkers in epidemiological studies on diet and health. British Journal of Nutrition, 86, (Suppl.)1, S37-S53. Women’s Health Initiative Steering Committee. (2004). Effects of conjugated equine estrogen in post-menopausal women with hysterectomy. Journal of the American Medical Association, 291, 1701-1712. Women’s Health Initiative Study Group. (1998). Design of the Women’s Health Initiative clinical trial and observational study. Controlled Clinical Trials, 19, 61-109. Writing Group for the Women’s Health Initiative. (2002). Risks and benefits of estrogen plus progestin in healthy post-menopausal women: Principal results from the Women’s Health Initiative. Journal of the American Medical Association, 288, 321-333.