Improving the Health of the Public: Part I Antioxidants and Heart Disease: An interview with Dr. Charles Hennekens
by Richard A. Passwater, Ph.D.
Dr. Charles Hennekens is Professor of Medicine and Ambulatory Care and Prevention at Harvard Medical School, and Chief of the Division of Preventive Medicine at Brigham and Women's Hospital. Dr. Hennekens received his MD from Cornell University Medical College in New York City, where he also trained in internal medicine at the New York Hospital. He served for two years in the U. S. Public Health Service as an Epidemic Intelligence Service (EIS) officer with the Center for Disease Control assigned to the Dade County Health Department in Miami, Florida, and then received an MPH, MS and DrPH in Epidemiology from the Harvard School of Public Health.
Dr. Hennekens is a diplomate of the National Board of Medical Examiners and the American Board of Preventive Medicine, a Fellow of the American College of Preventive Medicine (FACPM) and the American College of Epidemiology (FACE), and is the immediate past president of the Society for Epidemiologic Research.
Dr. Hennekens serves as editor-in-Chief of the American Journal of Preventive Medicine, and Co-Editor-in-Chief of the Annals of Epidemiology. He has been a member of the Editorial Board of Circulation, and the Board of Overseers of the American Journal of Epidemiology. The breadth and depth of Dr. Hennekens' contributions to public health are reflected in part by the number and nature of his publications. He has over 320 publications, including over 274 original papers, 44 reviews and two textbooks. He is a co-author of the textbook, Epidemiology in Medicine, which is in wide use in Medical and Public Health Schools.
During his career, Dr. Hennekens has elucidated much of what we know about causal and preventive risk factors for cardiovascular disease. He had the foresight to design and implement several epidemiologic studies over the last 20 years that have provided a great deal of knowledge about medical, behavioral, lifestyle and biochemical risk factors for the leading killer in the U.S. and most developed countries.
Dr. Hennekens also had the foresight to put together an outstanding research team, that includes as collaborators epidemiologists and highly respected clinicians as well as non-clinicians such as Drs. Julie Buring, J. Michael Gaziano, JoAnn Manson, Paul Ridker, Meir Stampfer and Walter Willett. One outstanding fact that follows Dr. Hennekens wherever he goes is his reputation for helping his colleagues and sharing the credit. He is a modest, down-to-earth man who is truly interested in helping all people at all levels.
Perhaps the most important -- and certainly the cheapest -- improvements in our health and longevity have come from the field of public health, and not medicine. A great stride in increased lifespan occurred when Semmelwise taught physicians to wash their hands after attending to each patient. As a result, child-bed fever and other deadly infectious diseases were greatly reduced. Aside from the discovery of sulfa-drugs and penicillin, no medical development has approached the life-saving significance of public health measures. Some of the most important health teachings have come from the studies of public health epidemiologists such as James Enstrom, Ph.D (see January 1993 WF) and the Harvard Group led by Charles Hennekens, MD, DrPH.
I have followed Dr. Hennekens' research for many years and have mentioned his beta-carotene and heart disease research in this column on several occasions. However, it wasn't until I met him that I learned of his scientific generosity. While we were discussing his research at the 1992 "Beyond Deficiency" conference that led to the famous Time article and many physicians taking a new look at the role of vitamins, I mentioned my 1976 study that indicated vitamin E was associated with reduced heart disease incidence. Dr. Hennekens immediately offered me five minutes of his twenty-minutes of lecture time to inform the entire conference of my study. He later, at the 1993 "Carotenoids in Human Health" conference, made the same offer to Dr. Matthias Rath of the Linus Pauling Institute to discuss his Lp(a) research.
We have had the opportunity to discuss much of his research at other meetings. I thought that you might like to share in some of our conversation.
Passwater: Please tell us something about the "Eureka" moment, when you realized that you needed more data and could possibly collect them from physicians, nurses and other health professionals.
Hennekens: I am not sure that there was a single moment, but rather a growing recognition that populations of health professionals might in fact serve as ideal groups from which to recruit large samples of participants who could provide high quality data over long periods of time. What had become very clear to me by the mid-1970's was that the next generation of epidemiologic studies had to be an order of magnitude larger in sample size than the traditional observational cohort studies or randomized clinical trials conducted to date. For the types of exposures or interventions that we were beginning to explore, the magnitude of the beneficial or harmful effects were most likely to be smaller in size than previously thought. It was unlikely that we would be evaluating another penicillin, a drug that reduced mortality from broncho-pneumonia from 90% to about 15-30%, or even relationships like the 15-20 fold increased risk of lung cancer due to heavy cigarette smoking. To evaluate this type of exposure/disease relationship, large-scale observational studies such as the Nurses' Health Study of 121,7000 women would clearly suffice. On the other hand, if antioxidant vitamins, for example, did reduce risk of cardiovascular disease, the most likely benefit would be on the order of 10-50% - clinically very worthwhile for this common and serious disease, but very hard to detect reliably statistically. To do so, then, would require not just large sample sizes, but randomized trials designed in a simple, streamlined and cost-efficient way.
Thus, in the 1970's I first began to collaborate with Dr. Frank Speizer to assemble a cohort of women who could be followed prospectively over time with respect to a wide variety of risk factors and disease outcomes. In fact, our pilot study was conducted among doctors' wives, and was originally designed to evaluate whether oral contraceptives increased risk of breast cancer. I determined that a sample of about 18,000 women could have been assembled from this group, but I was interested in particular on risk factors for cardiovascular disease, which is far less common than breast cancer in women under 60. As a result, this sample size would not have been adequate for these endpoints, and thus I proposed to Dr. Speizer to change the population to female nurses. In doing so, we were able to enroll over 120,000 subjects who were 30-55 years of age in 1976 and who are still being followed today - which enabled us to test a much wider variety of risk factors for both cancer and cardiovascular disease.
By the late 1970's, I turned my attention to the possibility of conducting large, simple randomized trials. I began to collaborate with Sir Richard Doll and Richard Peto of Oxford, England, who were testing the effects of aspirin in the prevention of heart disease in approximately 5000 British male physicians. Seeing that the question needed a larger sample size to answer definitively, I conducted pilot studies and later submitted a proposal to the National Institutes of Health which was funded to evaluate the role of low dose aspirin and beta-carotene in the primary prevention of heart disease and cancer in a large group of healthy men. This randomized, double-blind, placebo-controlled trial, the Physicians' Health Study, is being conducted among 22,071 U.S. male physicians, aged 40-84 in 1982. To address these important prevention questions in women, my colleague Dr. Julie Buring and I were funded in the early 1990's to conduct the Women's Health Study, a randomized trial of the risks and benefits of low-dose aspirin, beta-carotene, and vitamin E on cardiovascular disease and cancer. This trial is being conducted among 40,000 U.S. female health professionals over the age of 45. And most recently Drs. JoAnn Manson, Julie Buring, and myself have just been funded to conduct a trial of the antioxidant vitamins beta-carotene, vitamin E and vitamin C in the treatment of cardiovascular disease among over 8,000 female health professionals who were ineligible for the Women's Health Study because of a prior history of cardiovascular disease.
My 17 years of experience with the 122,000 female nurses in the Nurses' Health Study, 13 years of experience with the 22,000 male physicians in the Physicians' Health Study, and the now 2 years of experience with the 40,000 female health professionals in the Women's Health Study have all amply demonstrated the willingness and ability of these populations to provide high quality information over long periods of time.
While health professionals were chosen as the study populations for our randomized trials for a number of scientific and practical reasons, what it also meant is that for the first time, these large-scale trials could be conducted entirely by mail, at a small fraction of the cost of a usual trial. While most trials that require clinic visits or in-depth workups cost from $3000 to $15,000 per randomized participant per year, for the first five years of the Physicians' Health Study, the cost per participant was a total of $80 per year. Thus, not only has the use of health professionals allowed us to obtain high quality information over long periods of time, but allowed us to do so in a very cost-effective way.
Passwater: What data were you specifically looking for then, and what "bonus" information have you found hidden in among the data?
Hennekens: Since 1976, the Nurses' Health Study has provided and is continuing to provide extremely valuable data concerning the risks and benefits of a large number of risk factors for cardiovascular disease and cancer. As I mentioned, originally the study was designed to evaluate whether oral contraceptives increased the risk of breast cancer. The very large sample size facilitated the demonstration of the increased risk of coronary heart disease associated with oral contraceptives, especially among older women who smoke, and provided reassuring evidence concerning duration of use and risk of breast cancer. But the Nurses' Health Study has also evaluated a wide range of exposure/disease relationships, from hair dyes and lung cancer; to postmenopausal hormones and coronary heart disease, uterine cancer, and breast cancer; to aspirin and coronary heart disease; to the effects of smoking, obesity and diabetes on heart disease. In 1980, Dr. Walter Willett added a dietary questionnaire to the Nurses' Health Study, which has provided a crucial and timely dimension to the study, allowing evaluation of a wide variety of nutrients, including fat, fiber and antioxidant vitamins.
Similarly, the Physicians' Health Study was designed to evaluate aspirin and beta-carotene in the primary prevention of heart disease and cancer. The randomized aspirin component of the trial was ended prematurely in January, 1988 by the external Data Monitoring Board, due primarily to the finding of a highly significant 44% reduction in risk of a first myocardial infarction among those receiving active aspirin. The termination of the randomized aspirin component has given us the opportunity to explore a number of other hypotheses, including the effect of aspirin on the development of angina; the role of aspirin in the prevention of a number of conditions including cataract, migraine and pulmonary embolism; as well as aspirin and colon cancer. While the randomized beta-carotene component of the trial is still ongoing and is scheduled to end in 1995, we have been able to explore in a preliminary way the effect of beta-carotene in the subgroup of physicians with evidence of cardiovascular disease at baseline in the form of unstable angina or a prior revascularization procedure.
Passwater: The beta-carotene data caught many by surprise. It seemed as if even the scientists expecting to see benefit from beta-carotene were thinking mainly of cancer, not heart disease. How did the relationship between beta-carotene and reduced risk of heart disease evolve?
Hennekens: It is true that in 1980, when we were designing the Physicians' Health Study, the excitement about beta-carotene was primarily as a potential chemopreventive agent. But within the next few years, evidence began to accumulate from a number of sources that there might be a relationship with heart disease. The 2x2 factorial design of our trial provided us the unique opportunity to examine beta-carotene and cancer, as well as beta-carotene and cardiovascular disease in the same study.
As has been the case with antioxidant vitamins and coronary heart disease, advances in our medical knowledge proceed on several fronts, optimally simultaneously. Basic researchers provide explanations or mechanisms to explain why a particular intervention prevents human illness; clinicians provide enormous benefits to patients through advances in diagnosis and treatment and in addition, formulate most of the best and brightest ideas or hypotheses from their own clinical experiences - that is, their case reports and case series; and epidemiologists and biostatisticians, optimally working with clinicians, formulate hypotheses from descriptive studies and test hypotheses in case-control, observational cohort or randomized trial designs to determine whether a particular intervention prevents human illness. Each discipline, and every research strategy within a discipline, contributes importantly relevant and complementary information to a totality of evidence upon which rational clinical decision-making and public policy can be safely based.
The synergistic contributions of basic research, astute clinical observations, and epidemiology - descriptive, observational analytic, and randomized trials - have contributed to an emerging body of evidence concerning the increasing atherogenicity of, first, total cholesterol, then low density lipoprotein or LDL cholesterol, and most recently oxidized LDL. This, in turn, has raised the possibility that antioxidant vitamins - such as beta-carotene, vitamin E and vitamin C - might reduce cardiovascular risks by inhibiting oxidation of LDL, the particularly atherogenic form of cholesterol.
Epidemiologic studies that have explored the antioxidant vitamin hypothesis in humans have included descriptive and cross-sectional studies, analytic investigations using case-control and prospective cohort designs, and several small randomized trials. The findings have not been totally consistent, but they generally have supported the hypothesis that antioxidant vitamins reduce the risk of cardiovascular disease.
We began by exploring the association between antioxidants and cardiovascular disease in several of our observational prospective cohort studies, including the Nurses' Health Study, the Massachusetts Elderly Cohort Study and the Health Professionals Follow-Up Study.
By far the largest study to evaluate antioxidant vitamins and cardiovascular disease is the Nurses' Health Study, the prospective cohort study I discussed earlier of 121,700 U.S. female nurses aged 30-55, begun in 1976. Of these, 87,245 nurses were free from coronary heart disease, stroke or cancer and provided complete dietary data in 1980. Dr. Meir Stampfer led the analyses on vitamin E, and Dr. JoAnn Manson those on beta-carotene and vitamin C.
Women who were in the highest quintile (fifth) of dietary intake of beta-carotene were found to have a statistically significant 22% lower risk of developing coronary heart disease than women in the lowest quintile. There was also a 34% reduced risk of coronary heart disease associated with the highest intake of vitamin E, but the benefit was almost entirely seen among those who used supplements of vitamin E, rather than from dietary intake alone. For vitamin C, there was no significant pattern of reduction in risk of coronary heart disease associated with either diet or supplements. Finally, the overall association of antioxidant vitamin intake and coronary heart disease was assessed by a total antioxidant vitamin score summing the intakes of beta-carotene, vitamin E and vitamin C. In this analysis, there was a 46% reduction in risk for those with the highest intake of antioxidant vitamins.
Dr. Michael Gaziano has taken the lead in our second prospective dietary intake study, which was conducted among a cohort of 1299 elderly men and women who were residents of Massachusetts in 1976. After controlling for the potential confounding effects of age, sex, smoking, alcohol consumption, cholesterol intake, and functional status, those with the highest intake of beta-carotene had a 43% lower risk of cardiovascular death and a 68% lower risk of fatal myocardial infarction.
These hypotheses have also recently been explored in the Health Professionals Follow-Up Study, a prospective cohort study of 51,529 U.S. men, aged 40-75. This study was begun in 1986 by Dr. Walter Willett, and the antioxidant analyses were led by Dr. Eric Rimm. The analyses were based on four year follow-up data from the study among 39,910 men with no previous history of cardiovascular disease or conditions that were deemed likely to have resulted in significant changes in dietary habits.
Those in the highest quintile of intake of beta-carotene had a 25% reduced risk of coronary heart disease. For vitamin E, there was a 39% reduction, and although the relation was strongest for supplement users, an association of borderline statistical significance was still apparent for dietary vitamin E intake alone. Vitamin C intake appeared unrelated to risk of coronary heart disease.
The intakes of beta-carotene in these studies that appeared to be associated with decreased risks of cardiovascular disease are well within the range that can be achieved through diets high in fruits and vegetables. However, the benefits of vitamin E levels, while apparent to some degree from diet, were greatest among those who used supplements.
Passwater: I spoke with Dr. Meir Stampfer in June of 1992, just after the numbers were crunched in the Nurses' Health Study for vitamin E and heart disease. I believe he was shocked at the strength in the numbers and that other confounding factors could not be shown to explain the results. Were you surprised at the strength of the relationship between vitamin E and reduced incidence of heart disease in both the Nurses' Health Study and the Health Professionals Follow-Up Study?
Hennekens: Not really - to me the data from these observational studies simply add support to the hypothesis of possible cardiovascular disease benefits of antioxidant vitamins, and reinforce the need for randomized trials to answer the questions definitively. Because of the limitations that are inherent in all observational studies, the actual magnitude of the association observed, while certainly promising, may well be an overestimate due to unmeasured or unmeasurable confounding factors. In other words, regardless of the number or sample size of such investigations or even the consistency of their findings, observational studies are unable to control for potential effects of confounding variables not known to the investigator or not collected. It may be, for example, that greater dietary intake of antioxidant vitamins, measured by blood levels or a diet assessment questionnaire, is only a marker for some other dietary practices that are truly protective. It is certainly plausible that intake of antioxidant-rich foods is indeed protective, but that the benefits may result not from their antioxidant properties, but other components these foods have in common. It is also possible that intake of antioxidant vitamins from food or supplements is correlated with other unmeasured or unknown non-dietary lifestyle behaviors. Thus, when searching for small to moderate effects, the amount of uncontrolled confounding in all observational studies is likely to be as large as the most plausible hypothesis being studied. When the most plausible effect size is 10 to 50%, a small amount of uncontrolled confounding could mean the difference between a relative risk of 0.8, a protective effect, 1.0, a null effect, or even 1.2, which suggests an increased risk.
For all these reasons, only randomized trials of sufficient sample size and duration of treatment and follow-up are able to detect reliably small to moderate treatment effects. If the trials are large enough, the randomization process will evenly distribute among treatment groups known as well as unknown confounding variables. In addition, very large trials will be necessary to avoid the possible uninformative null results of no benefit when in fact a small to moderate benefit truly exists.
Passwater: Do you, then, have any randomized trial data that addresses this question?
Hennekens: To date, no large-scale randomized trials have been completed of antioxidant vitamins in either the treatment or prevention of cardiovascular disease. However, we do have some very intriguing data from a subgroup analysis within our Physicians' Health Study testing beta-carotene among 333 physicians with a history of angina or myocardial revascularization- either CABG or PTCA - at baseline.
As I mentioned, the Physicians' Health Study is being conducted among 22,071 male U.S. physicians, aged 40-84 at baseline, and free from myocardial infarction, stroke, transient ischemic attack and cancer. However, 333 doctors did enter the trial with chronic stable angina or a prior coronary revascularization procedure - either CABG or PTCA. Based on the emerging hypotheses concerning antioxidant vitamins and cardiovascular disease, the availability of the data base from the completed aspirin component afforded us the opportunity to conduct an analysis of beta-carotene and cardiovascular disease in this small subgroup of high-risk participants who were treated and followed for an average of 60.2 months at the time the aspirin component was terminated. Drs. Michael Gaziano and JoAnn Manson took the lead in these analyses.
We defined two combined endpoints, major coronary events and major vascular events. The first, major coronary events, was comprised of nonfatal myocardial infarction, fatal coronary heart disease, and coronary revascularization. The second endpoint, all major vascular events, included nonfatal and fatal stroke in addition to the major coronary events.
Those assigned beta-carotene (50 mg every other day) had a statistically significant 51% reduction in risk of major coronary events. For major vascular events, there was a statistically significant 54% reduction.
Since the postulated mechanism of a benefit of antioxidant vitamins relates to an effect on progression of atherosclerosis, we hypothesized that one might not expect an immediate effect of antioxidant vitamin supplements on cardiovascular disease. We therefore analyzed the relative risk of major vascular events by year of follow-up. In this analysis, there was indeed a suggestion of a delayed effect. There was no apparent effect of carotene during the first year of treatment. The apparent benefit began to appear during the second year of treatment, and persisted thereafter.
In addition, since the postulated benefits of aspirin are acute and of beta-carotene chronic, the 2x2 factorial design enabled us to evaluate their possible interaction. We looked at the development of first myocardial infarction among these 333 men with chronic stable angina or myocardial revascularization at baseline according to both aspirin and beta-carotene treatment assignment. Of the 27 incident myocardial infarctions, 7 were in the aspirin group and 20 in aspirin placebo. With respect to beta-carotene, 10 occurred in the active group and 17 in placebo. Of most intriguing interest, however, among those randomized to both aspirin and beta-carotene there were no myocardial infarctions, a finding that, although based on small numbers in a subgroup analysis of a randomized trial, is compatible with the differing mechanisms proposed for each of these two agents.
While such data raise the possibility of a benefit in patients with existing cardiovascular disease, firm answers on this question require larger trials, and such findings do not address whether beta-carotene will decrease risks of cardiovascular disease in apparently healthy individuals without prior cardiovascular disease. This question will be answered definitively over the next several years with the continued dedicated and conscientious efforts of the participating doctors in the Physicians' Health Study. As a randomized subject myself I can tell you it takes some effort to comply with the daily pill taking and completion of the forms. I consider my fellow colleagues true collaborators who are providing very valuable contributions to the health of the general public.
Passwater: Yes, and we are all grateful to all of the volunteers. Thanks for your explanations of your findings. In the next column, let's discuss the strength of the evidence and what needs to be done to further strengthen it.
All rights, including electronic and print media, to this article are copyrighted to Richard A. Passwater, Ph.D. and Whole Foods magazine (WFC Inc.).