|Home | About | Journals | Submit | Contact Us | Français|
Oxidative stress may contribute to the development of heart failure (HF). However, an increased risk of HF has been observed with antioxidant therapy in secondary prevention trials. No large clinical trials have addressed the role of antioxidant therapy in the primary prevention of HF.
We examined the effect of vitamin E and HF risk in 39,815 initially healthy women aged at least 45 years at baseline who were enrolled in the Women's Health Study, a randomized, double blind, placebo controlled trial of vitamin E (600 IU every other day). Over a median follow-up of 10.2 years, there were 220 incident HF events. In proportional hazards models adjusting for age, randomized aspirin and beta carotene treatment, vitamin E assignment did not significantly affect HF risk (HR 0.93, 95% CI 0.71–1.21, P=0.59). These results did not change with multivariate adjustment for other risk factors, including interim myocardial infarction. In a pre-specified subgroup analysis, vitamin E was inversely related to developing HF with normal ejection fraction (≥ 50%) with HR 0.59 (95% CI 0.38–0.92, P=0.02) but there was no statistically significant effect on the risk of developing systolic HF (HR 1.26, 95% CI 0.84–1.89, P=0.26).
In this population of apparently healthy women, vitamin E did not affect the overall risk of HF. The possible benefit on diastolic HF requires confirmation in larger populations.
Heart failure (HF) is a leading cause of cardiovascular morbidity and mortality, constituting a substantial and costly public health burden. Over 5.7 million people in the United States have HF, with 670,000 new cases diagnosed per year.1 The prevalence and incidence of HF will continue to rise due to aging of the population and increasing survival with associated risk factors such as hypertension and coronary artery disease. Furthermore, despite advances in therapy, prognosis remains poor with 50% of patients with HF dying within 5 years of diagnosis.1 Therefore, identification of therapeutic interventions and modifiable lifestyle factors that may aid in the primary prevention of HF is of critical importance.2
Antioxidant therapy has been identified as a promising intervention that may reduce the risk of HF, based on experimental data showing that oxidative stress may play an important role in HF pathophysiology,3,4 and that this risk may be ameliorated by antioxidant therapies.5–9 Despite these promising biologic data, long-term treatment with vitamin E was associated with an increased risk of HF and hospitalization for HF among patients with vascular disease or diabetes enrolled in the Heart Outcomes Prevention Evaluation [HOPE] trial.10 A similar, albeit non-significant, elevation in HF risk was also observed in patients with recent myocardial infarction enrolled in the GISSI-Prevenzione trial. In addition, vitamin E was associated with a significant 50% increase in HF risk in patients with left ventricular ejection fraction (LVEF) <50% at baseline.11
However, no study has examined whether vitamin E influences the risk of HF in a primary prevention population, particularly among women where a lower proportion of HF is due to systolic dysfunction.12–15 In addition, it is unclear whether vitamin E has a differential effect on HF risk depending upon the underlying primary pathophysiology of systolic or diastolic HF. In order to address these questions, we examined the overall effect of long-term vitamin E treatment on HF risk in the Women’s Health Study (WHS), a large-scale randomized clinical trial of vitamin E among apparently healthy women.16 We then examined whether the relationship between vitamin E and HF varied depending on the presence or absence of diminished systolic function, defined as an LVEF<50%.
Study subjects were participants of WHS, a randomized, double blind, placebo controlled, 2×2 factorial trial examining the benefits and risks of low dose aspirin and vitamin E in the primary prevention of CVD and cancer which was completed on March 31, 2004. Beginning in 1993, 39,876 female health professionals in the United States who were at least 45 years of age and free of CVD and cancer were randomly assigned to receive aspirin (100 mg every other day), vitamin E (600 IU every other day), both agents or placebo. A third arm of the trial, which tested beta carotene, was terminated early in January 1996 due to other trials of beta carotene which had null results or suggested possible harm among those at high risk for lung cancer.17 Written, informed consent was obtained from all participants. The study was approved by the institutional review board of Brigham and Women’s Hospital, Boston and monitored by an external data and safety monitoring board.
Details of the study design have been described previously.18,19 In brief, study participants received calendar packs containing active agents or placebo each year. Every 6 months for the first year and annually thereafter, study participants received follow-up questionnaires that inquired about compliance with study medications, adverse effects, risk factors and occurrence of endpoints. Compliance, defined as taking at least two thirds of the study capsules, averaged 75.8% and was similar between the active treatment and placebo groups. 16 Use of non-trial vitamin E supplements for at least 4 days per month was lower in the active treatment group than in the placebo group (8.5% vs 8.9%, P=0.07).16 Morbidity and mortality follow up were 97.2% and 99.4% complete, respectively. 16 The primary results for cardiovascular disease and cancer have been published previously.16, 20
For the purposes of this analysis, we excluded 13 women who had prevalent CVD (but reported after women had already been randomized into the WHS) and 48 who had evidence of HF prior to randomization, resulting in a study cohort of 39,815 women.
Covariates of interest were self-reported on the baseline questionnaire, and included age, race/ethnicity (which was self-reported by participants as white, black, Hispanic American, Asian American or other), blood pressure, height and weight; history of hypertension, diabetes, and hypercholesterolemia; lifestyle habits including smoking status, alcohol use, and physical activity; menopausal status and current use of hormone therapy; and multivitamin use.
Women reported the occurrence of nonfatal endpoints via annual follow-up questionnaires, letters, or telephone calls. Death were reported by family members or postal authorities, or ascertained through the National Death Index. After obtaining written consent, medical records were obtained and adjudicated according to predefined criteria by an endpoint committee of physicians blinded to treatment assignment. Myocardial infarction was confirmed if symptoms met World Health Organization criteria and if the event was associated with abnormal levels of cardiac enzymes or diagnostic electrocardiograms.19
On the 48 month questionnaire, study participants were asked to report any prior physician diagnosis of HF along with the month and year of diagnosis. On annual questionnaires thereafter, they were asked to report diagnoses of incident HF. Women who reported the diagnosis of HF were sent a supplemental questionnaire to collect additional information regarding symptoms, diagnostic evaluation, current medical therapy and functional status, and to request permission to obtain and review their relevant medical records, including physician and hospital records, laboratory and chest x-ray data, and results of diagnostic tests including electrocardiograms, echocardiograms, stress tests and cardiac catheterizations. If the diagnosis of HF was mentioned in medical records obtained for the purpose of confirming other endpoints, these records were also reviewed. For deceased participants, we contacted family members to obtain consent for medical records pertaining to the diagnosis of HF.
Medical records were reviewed by cardiologists blinded to treatment assignment and a diagnosis of HF was confirmed according to predefined criteria. Cases of incident nonfatal HF were confirmed if either the Framingham Heart Study (FHS) 21 or Cardiovascular Health Study22 (CHS) criteria were met. The FHS criteria rely on physical exam findings and radiographic data (Table 1), while the Cardiovascular Health Study criteria are predominantly based on the treating physician’s diagnosis of HF and use of specific medical therapy for HF (diuretic and digitalis or a vasodilator). “Definite” cases of nonfatal HF met both FHS and CHS criteria, and “probable” cases of nonfatal HF met either FHS or CHS criteria, but not both. 23, 24
Fatal HF cases included those not identified as a case of HF prior to death (N=2). “Definite” fatal HF was defined as meeting any of the following criteria from medical record review: pulmonary edema, visceral congestion, cardiomegaly on autopsy 21; death due to shock or low output syndrome 22; ICD-9 code 428.xx or 425.xx on the death certificate with next-of-kin or physician confirmation of HF as the immediate or underlying cause of death; or history of severe chronic HF, with the primary cause of death due to a related cause such as a ventricular arrhythmia.25 “Probable” fatal HF was defined as HF as the immediate or underlying cause of death on the death certificate, in the absence of available records or data fulfilling criteria for definite fatal HF.26
Information on left ventricular ejection fraction (LVEF) within 3 months of the diagnosis of incident HF was collected from medical record review and based on diagnostic tests performed closest to the date of diagnosis of incident HF. The mean duration (SD) between testing and incident HF diagnosis was 8.3 (15.3) days, with 90% being performed within one month of the diagnosis. The majority of the ejection fraction data were derived from echocardiography (68.8%) or left ventriculography (25.0%). Data were collected primarily as continuous variables, and if the report described ejection fraction as a range, the midpoint value was used. If the degree of impairment of systolic function was described qualitatively, we classified “none” or “mild” left ventricular dysfunction as having LVEF ≥ 50%, and “moderate” or “severe” systolic dysfunction as having LVEF<50%. We defined systolic HF as incident HF occurring in the setting of LVEF<50%, and HF with normal ejection fraction (or “diastolic” HF) as incident HF occurring in the setting of LVEF ≥ 50%. 27, 28
For this analysis, the primary endpoint was all incident HF, which included both definite and probable cases of nonfatal and fatal HF.23, 24 We also performed sensitivity analyses confining the endpoint to cases of definite HF. To examine whether the relationship between vitamin E and HF varied depending on left ventricular function, we performed pre-specified analyses using the endpoints of systolic or diastolic HF. Follow-up was censored at the end of the trial.
All analyses were performed on an intention to treat basis. Person-years of follow up were calculated from baseline to the date of the incident HF event, death, or the end of the trial, whichever came first. Kaplan-Meier curves were constructed to estimate the cumulative incidence of HF by randomized treatment group. The log rank test was used to compare incidence rates. Proportional hazards models were constructed to calculate the hazards ratios (HR) and 95% confidence intervals (CI) for vitamin E assignment compared to placebo and the risk of incident HF, adjusted for age and randomized aspirin and beta carotene assignment. These models were further adjusted for race/ethnicity, hypertension (defined as a self-reported history of hypertension, or systolic blood pressure 140 mmHg or higher, or diastolic blood pressure 90 mmHg or higher), hypercholesterolemia (defined as being on lipid-lowering therapy or having a total cholesterol of 240 mg/dL or greater), diabetes, body mass index (calculated as weight in kilograms divided by height in meters squared), smoking status, alcohol use, physical activity, menopausal status and current hormone therapy use, and multivitamin use. For the sensitivity analyses confining the endpoint to cases of definite nonfatal and fatal HF, those participants with probable HF were censored at the time of their HF diagnosis. The proportional hazards assumption was examined by including an interaction term of vitamin E with the logarithm of time in the Cox models for the endpoint of HF, and was not violated.
To examine if there was a differential effect of vitamin E on HF risk by left ventricular function, we performed a pre-specified series of proportional hazards models using systolic HF and HF with normal ejection fraction as the outcomes. For the endpoint of systolic HF, those participants with diastolic HF were censored at the time of their HF diagnosis, and vice versa.
We also performed several additional exploratory analyses. To evaluate whether the association between vitamin E and HF risk was mediated by coronary artery disease, we constructed another series of multivariate models that additionally controlled for interim myocardial infarction as a time-varying covariate. Effect modification of the association between vitamin E assignment and HF risk was examined using stratified analyses and testing for interaction using multiplicative interaction terms for vitamin E and relevant baseline characteristics using likelihood ratio tests, testing for trend when subgroup categories were ordinal.
All analyses were performed using SAS version 9 (SAS Institute Inc., Cary, North Carolina). A two-tailed P< 0.05 was considered statistically significant.
The 39,815 women in the study cohort had a mean (SD) age at randomization of 54.6 (7.04) years. Clinical characteristics were balanced and are described in Table 2. Over median follow-up of 10.23 years, there were 220 incident HF events. Of these, 106 had been randomized to vitamin E and 114 had been randomized to placebo. As shown in the Figure, there was no significant overall difference in the cumulative incidence of HF over the duration of follow-up between the vitamin E and placebo groups (log rank P=0.59). In proportional hazard models adjusting for age and randomized aspirin and beta carotene assignment, vitamin E assignment remained unassociated with HF risk (HR 0.93, 95% CI 0.71–1.21, P=0.59). These results did not substantially change after multivariate adjustment for other clinically relevant covariates and potential confounders, and were unaffected by additional control for interim myocardial infarction (Table 3). When these analyses were repeated in sensitivity analysis limited to definite HF cases (n=158, 72%), the results were similar. Compared to placebo, vitamin E assignment was associated with HR 0.90 (95% CI 0.66–1.23, P=0.50) for definite HF after full multivariate adjustment, including for interim myocardial infarction.
Among the 220 HF events, ejection fraction data were available in 176 participants (80%). Of these, 95 participants (54%) had LVEF<50% while 81 participants (46%) had LVEF ≥50% around the time of HF diagnosis. As shown in Table 3, vitamin E assignment did not affect the risk of systolic HF in this population of healthy women. However, vitamin E assignment resulted in a significant 40% reduction in the risk of developing HF in the setting of normal ejection fraction (P=0.02). This risk estimate remained essentially unchanged after multivariate adjustment.
We also explored our data for potential effect modification of the association between vitamin E and HF by important clinical characteristics. As shown in Table 4, there was the suggestion of effect modification by postmenopausal status and hormone use (P for interaction 0.02), as well as by multivitamin use (P for interaction 0.04). There was no evidence of effect modification by randomization to aspirin or beta-carotene, age, race/ethnicity, history of other medical conditions such as diabetes, hypertension, or hypercholesterolemia, body mass index, smoking status, alcohol intake, or physical activity level.
In this large randomized clinical trial, vitamin E assignment had no association with the risk of incident HF in initially healthy women. However, while vitamin E did not influence the risk of developing systolic HF, it was associated with a significant 40% reduction in the risk of developing HF with normal ejection fraction. To the best of our knowledge, these are the only data available regarding the influence of vitamin E on HF risk in a primary prevention population.
Our findings differ somewhat from prior data from secondary prevention trials which raised concern about vitamin E supplementation increasing the risk of HF. In the HOPE trial,10 9541 patients (mean age 66 years, 26.7% women) with vascular disease or diabetes were randomized to vitamin E (400 IU/d) or placebo; over 4.5 years of follow-up. Vitamin E was associated with a higher risk of hospitalization for HF defined as having clinical and radiologic signs of congestion (RR 1.21, 95% CI 1.00–1.47, P=0.045), and a higher risk of developing HF regardless of the need for hospitalization (RR 1.13, 95% CI 1.02–1.26, P=0.03). Similar increases in HF risk were seen in the 7040 participants followed for an additional 2.6 years in the HOPE-TOO extension trial. In GISSI-Prevenzione,11 an open label trial of vitamin E (300 mg/d) without placebo control, 8415 patients (mean age 58.4 years, 13.9% women) who had a myocardial infarction in the prior three months were followed for 3.5 years for the development of HF, defined as the need of hospitalization for HF management or death from HF as classified by an endpoint validation committee. Similar to HOPE, vitamin E was associated with a 20% increased risk of HF (RR 1.20, 95% CI 0.92–1.56, P=0.18) although this was not statistically significant. When stratified by tertiles of ejection fraction at baseline, patients with LVEF<50% had a significantly increased risk of HF with vitamin E compared to controls who did not receive vitamin E (RR 1.50, 95% CI 1.03–2.20, P=0.0345).
Potential explanations for the contrast between our findings and those of the HOPE and GISSI-Prevenzione trials include differences in the populations studied. Perhaps most importantly, WHS was a primary prevention trial that examined the effect of vitamin E in a cohort of apparently healthy individuals, whereas HOPE and GISSI-Prevenzione were secondary prevention trials conducted in patients with pre-existing vascular disease or its equivalent, or prior myocardial infarction, respectively. In addition, we could not exclude differences in the effect of vitamin E based on sex. In contrast to WHS, the HOPE and GISSI trials involved relatively small populations of women (26.7% and 13.9%, respectively), although the HOPE investigators reported no differences in subgroup analysis by sex. The doses and dosing regimens of vitamin E differed amongst these trials, with WHS testing a dose of 600 IU on alternate days whereas HOPE and GISSI-Prevenzione used doses of 400 IU daily and 300 mg daily, respectively. Finally, the incidence of HF in our cohort was lower than in HOPE and GISSI-Prevenzione, which is not unexpected given that WHS is a younger, primary prevention population. However, the smaller number of endpoints may have limited our statistical power to detect an association between vitamin E and HF risk. Based on our confidence intervals (95% CI 0.71–1.21), we cannot rule out a modest effect of vitamin E on HF risk, i.e., up to a 29% reduction or a 21% increase in risk, the latter of which would be similar in direction and degree to the HOPE and GISSI-Prevenzione findings
While we observed an overall null effect of vitamin E on HF risk, in a pre-specified subgroup analysis we found a protective effect of vitamin E assignment on the risk of developing HF with normal ejection fraction. This finding should be interpreted with caution, since it was a subgroup analysis based on a relatively small number of HF cases, but it nevertheless merits further study. Heart failure with normal ejection fraction, where the primary abnormality is related to diastolic dysfunction or impaired relaxation, represents approximately half of all cases of HF.12–15, 27, 28, 29 Diastolic HF is increasing in prevalence and is more common in women than in men.12–15, 29 Although associated with similar biologic derangements and adverse prognosis as systolic HF,12–15, 27–29 fewer effective therapies exist. Recent studies in animals have demonstrated that oxidative stress may play an important role in diastolic dysfunction and that antioxidant therapies may improve diastolic performance in conditions including hypertension, insulin resistance and diabetes, and hypercholesterolemia.30–34
Strengths of this study include the fact that WHS was a large prospective randomized clinical trial of vitamin E with long duration of follow-up. In addition, the incident cases of HF were validated using established epidemiologic criteria. Our study also has several limitations. Because WHS participants were healthy, middle aged and predominantly Caucasian female health professionals, our findings may not be generalizable to other populations. The incidence rate of HF in WHS (0.56 per 1000 person-years) is lower than typically reported in other epidemiologic studies, although limited data exist on HF incidence in younger populations such as WHS. For example, in the community-based Atherosclerosis Risk in Communities Study (ARIC), the HF incidence rate in Caucasian women was 3.4 per 1000 person years, adjusted to the mean age at baseline of 54 years.35 The lower incidence rate of HF in WHS may be explained by the substantially lower rates of risk factors such as hypertension, diabetes, smoking and obesity in WHS compared to ARIC; in addition, in ARIC, participants with prevalent CHD at baseline were not excluded from their analysis. An additional limitation is that we could not assess if vitamin E has a differential effect on HF risk in women with or without valvular heart disease, as there were only 15 cases for whom severe valvular disease could be determined to be the primary etiology. We did not otherwise attempt to assign a primary etiology to the HF cases in WHS, due to the frequent coexistence of causes of HF (e.g., coronary artery disease, hypertension and diabetes) as well as clinical variability in the use of diagnostic testing such as coronary angiography. Ejection fraction data were derived from medical record review of tests performed in the clinical setting, as opposed to assessment by a standardized protocol, which may result in variability in these measurements. Heart failure events were self-reported by WHS participants; however, the confirmation rate in WHS was 67%, comparable to those in other epidemiologic studies36, 37 and the reports were validated by application of standardized epidemiologic criteria. Our findings apply only to vitamin E in the formulation and dose tested in WHS and cannot be generalized to other antioxidant therapies. Finally, although this is the largest study of vitamin E and HF risk in a primary prevention population, due to the limited number of events, we cannot completely exclude a small to moderate effect of vitamin E on HF risk based on these data.
In conclusion, long-term treatment with vitamin E did not affect the overall risk of incident HF in this randomized trial of initially healthy women. These results underscore the importance of focusing on other primary prevention measures proven to reduce the risk of future HF, including effective control of blood pressure and the primary prevention of coronary artery disease.2 If the inverse association between vitamin E and diastolic HF is confirmed in other prospective studies, then future randomized trials of antioxidant therapy in patient populations at high risk for diastolic HF may be warranted. However, at the present time, the cumulative evidence to date does not support the use of vitamin E supplementation to reduce the risk of cardiovascular diseases.
This work was supported by the Elizabeth Anne and Karen Barlow Corrigan Women’s Heart Health Program at Massachusetts General Hospital, the Donald W. Reynolds Foundation, and grants HL-043851 and HL-080467 from the National Heart Lung and Blood Institute and CA-047988 from the National Cancer Institute. We thank the 39,876 dedicated participants of the Women’s Health Study.