The most award winning
healthcare information source.
TRUSTED FOR FOUR DECADES.
Abstract & Commentary
Synopsis: The WHI and the WELL-HART clinical trials report that hormone therapy does not reduce the risk of coronary events or slow the progression of atherosclerosis in older women.
Manson and colleagues reported the final results on the risk of coronary heart disease (CHD) from the cancelled estrogen-progestin arm of the Women’s Health Initiative (WHI). The results reflect central adjudication of the cardiac diagnoses in contrast to the initial report that relied upon local diagnoses. The final report covers an average of 5.6 years of follow-up, compared with 5.2 years in the initial report. Based on these data, there would be an increase of 6 cases of CHD per 10,000 women per year in the treated group (see Tables 1 and 2).
Hodis and colleagues reported the results of a secondary prevention, 3-year trial (the WELL-HART study) assessing whether unopposed estradiol or a sequential regimen of estradiol and medroxyprogesterone acetate could slow the progression of atherosclerosis. The double-blind, placebo-controlled trial involved 226 postmenopausal women with an average age of 63.5 years (range, 48-75), who already had at least 1 demonstrated coronary artery lesion. The results were based on follow-up angiograms in 59 women in the placebo group, 54 in the estradiol group, and 53 in the estradiol-medroxyprogesterone acetate group. A reduction of LDL-cholesterol to less than 130 mg/dL was achieved by dietary intervention, but coronary angiography to measure the change from baseline in the percent stenosis failed to demonstrate a difference among the groups receiving placebo, unopposed, daily 1 mg estradiol, or daily 1 mg estradiol and 12 days each month of 5 mg medroxyprogesterone acetate. There were also no differences in the cardiovascular events during treatment. The results are consistent with previous secondary prevention trials in older women, and this indicates that different estrogens behave similarly. The results further indicate that medroxyprogesterone acetate (at least administered in a sequential regimen) is not associated with an adverse cardiovascular effect (Manson JE, et al. N Eng J Med. 2003;349:523-534; Hodis HN, et al. N Eng J Med. 2003;349:535-545).
Comment by Leon Speroff, MD
The diagnosis of coronary heart disease in the WHI included myocardial infarction that required overnight hospitalization, death due to coronary artery disease, and silent myocardial infarction identified on electrocardiography. Acute myocardial infarctions and deaths due to coronary heart disease were adjudicated by a central facility unaware of the treatment groups. Local and central reviews disagreed with 10% of the diagnoses for myocardial infarction and 3% for death due to coronary heart disease. Notice that this small degree of disagreement changed the strength of the conclusions comparing the initial report1 with the current report. Indeed, the overall results by definition do not achieve statistical significance, and only the first year results are statistically significant in the year-by-year analysis.
The intent-to-treat analysis was adjusted for multiple (7) outcomes, the Bonferroni adjustment. All adjusted results were not statistically significant. I still struggle to understand the clinical meaning of this manipulation, but currently I believe that this indicates a slightly lower mathematical conclusion than presented in the non-adjusted data. This, of course, would further weaken the power of the reported results.
The characteristics of the participants are now well known: average age 63.3 years, an average of 18 years distant from menopause, 42% dropout rate in the treated group and 38% in the placebo group, and 6.2% drop-in rate (began hormone use) in the treated group and 10.7% in the placebo group. Women with significant menopausal symptoms were excluded from the study to avoid an exceedingly high dropout rate in the placebo group. Women who had been on hormone therapy (about 25% of the participants) and then underwent a 3-month "washout" period and experienced menopausal symptoms were discouraged from participation (about 12.5% of the participants reported vasomotor symptoms upon entry, but were willing to be assigned to placebo, and, therefore, their symptoms were unlikely to have had a major disturbing effect). This exclusion means that only a small number of women in the WHI were close to their age of menopause (about 16.5% of the participants were less than 5 years since their menopause). For example, there were only about 250 women in the treated group and 225 in the placebo group who were aged 50-54.2
The WHI investigators argue that the high dropout rate could lead to an underestimation of adverse effects; however, this would not be the case if longer duration of treatment exerts a beneficial effect. Analysis of an 8.6% subsample revealed that the treated group had greater reductions in total cholesterol, LDL-cholesterol, glucose, and insulin levels; and greater increases in HDL-cholesterol and triglyceride levels, and it is tempting to link these findings with the test for trend that revealed a decreasing relative risk over time, which was statistically significant. However, this analysis was hampered by decreasing numbers over time and an increasing dropout rate; the conclusion is not a strong one.
Subgroup analyses (demographic and clinical characteristics, inflammatory and clotting markers, lipid profile) used a nested case-control study with 205 cases of myocardial infarction or cardiac death and 513 controls. No pattern could be identified distinguishing a difference between the 2 groups, except that women with elevated baseline levels of LDL-cholesterol had a higher risk of coronary events with treatment (the test for interaction was significant; the actual hazard ratio did not achieve statistical significance).
The analysis revealed no interaction with age, and the interaction with years since menopause was not statistically significant. The presence or absence of hot flashes did not change the results. Other factors that did not influence the results included previous use of hormone therapy, use of aspirin or statin treatment at baseline, body mass index, past use of oral contraceptives, or physical activity. There was no increase in angina or congestive heart failure. Despite these negative statements, it is striking that only the women who were 20 or more years distant from menopause had a statistically significant increased risk of coronary heart disease. How many of the women who experienced an event in the first year of treatment (the only statistically significant increase in the year-by-year analysis) were in the 20+ years group? (See Table 3.)
The WHI concludes that hormone therapy is "not a viable intervention for primary prevention," basing this conclusion on the overall WHI results including coronary heart disease, stroke, venous thrombosis, and breast cancer. I am comfortable in accepting the conclusion that postmenopausal hormone therapy does not reduce or slow the progression of established coronary disease. However, the WHI did not study the appropriate population in the appropriate time period to establish that hormone therapy does not exert a primary preventive effect on the risk of coronary heart disease. I disagree with the statement in the accompanying editorial (Herrington DM, Howard TD. N Engl J Med. 2003;349,519-521) that ". . . many people suspended ordinary standards of evidence concerning medical interventions and concluded that hormone therapy was the right thing to prevent heart disease . . ." There was a very large body of case-control, cohort, and biological evidence to support this conclusion, and indeed, similar evidence continues to be reported. Not every medical decision has the luxury of clear-cut, randomized clinical trial evidence to provide unassailable support. I would argue that the WHI does not provide irrefutable, indisputable evidence. That is not to say that the case for primary prevention of coronary heart disease by postmenopausal hormone therapy does not merit controversy. But at the same time, the issue is not settled, and perhaps there never will be a large trial of the appropriate population. Again I would argue with the editorialists and say that sometimes pharmacologic interventions do not have the benefit of "proof." The case for the causation of lung cancer by smoking is based upon observational and biological studies!
It is worth re-emphasizing that the strength of the conclusions is not overwhelming. A shift of a few cases can change the findings. This updated report, after adjudication, produced even weaker confidence intervals than the initial report. Results in those lost to follow-up by dropouts and drop-ins could alter the conclusions. The only statistically significant increase in coronary events occurred in the first year of the trial, a conclusion based on a difference of only 19 cases. Does this small difference represent a genuine side effect (is it clinically real), or is there another explanation?
Rather than simply accepting the findings because they are derived from a clinical trial, it seems to me that a more useful and intellectual approach would be to try to understand the reasons for the differences between the WHI and previous reports. This requires critical analysis. For example, there remains an incredibly important analysis that has not been performed or reported. Statin treatment (and perhaps aspirin treatment) lowers the risk of coronary heart disease and slows the progression of atherosclerosis. Statin-induced stabilization of atheromatous plaques is achieved relatively quickly, within 3 months.3 And estrogen therapy has been documented to have no effect on atherosclerosis when the studied subjects are already on statin treatment.4,5 The HERS trial analyzed statin use and discovered that new statin use was more common in the placebo group.6 There was neither an increase nor a decrease in risk for coronary events after adjustment for new statin use, indicating that the overall null result with hormone therapy was not due to new statin use. However, statin use attenuated the increase in events associated with hormone therapy in the first year, and it is possible that the small difference being reported in the WHI (only a 19-case difference between the user and placebo groups in the first year of the WHI) reflects statin use in the placebo arm. An analysis of statin use in the WHI needs to be performed, but even this has a problem because in both the HERS trial and the WHI, statin use was not randomized, and there was no standardization of statin drugs and dosage. The appropriate study would require randomization and standard treatment.
The late Trudy Bush was fond of saying that the objective of both basic and clinical research is to know the truth, and every study, no matter how good or how large, gives only one view of the truth. She always cautioned that it takes many views to come close to seeing the truth. The WHI is only one view of the truth.
Dr. Speroff is Professor of Obstetrics and Gynecology Oregon Health Sciences University Portland.
1. Writing Group for the Women’s Health Initiative Investigators. JAMA. 2002;288:321-333.
2. Hays J, et al. N Eng J Med. 2003;348:1839-1854.
3. Son JW, et al. Int J Cardiol. 2003;88:77-82.
4. Hodis HN, et al. Ann Intern Med. 2001;135:939-953.
5. Espeland MA, et al. Am J Epidemiol. 1995;142: 1011-1019.
6. Herrington DM, et al. Circulation. 2002;105: 2962-2967.