The Women's Health Initiative 2004 - Review and Critique

John A. Goldman, MD, FACP, FACR, FASLMS

In This Article

Coronary Heart Disease

CHD was a primary outcome. Further analysis of the risk of CHD in the HT WHI arm was adjudicated centrally, and when reported in August 2003, the CHD risk was decreased from 7 (as reported in July 2002) to 6 per 10,000 women per year.[8] There was a significant increased risk for CHD in the first year and a smaller, non-statistically significant increase in subsequent years. However, it was also noted that there was a trend toward decreasing relative risk over time that was statistically significant (z score = minus 2.36 [P = .02]). CHD risk actually increased in the placebo group by Year 6, which leads to an apparent although not statistically valid risk reduction with HT.[8] The HR for CHD was smaller for women in early menopause (< 10 years since menopause) as compared with older postmenopausal women (HR = 0.89), but this was not significant. The authors stated that the low number of events during the course of the study, the decreased adherence to the study medications over time, and the fact that the results were confined to women at risk for their first coronary event should warrant caution with respect to interpretation of the CHD data.[8] The HRs for adverse CHD events with time since menopause did not show an increased risk, but rather a decreased risk[8] ( Table 2 ). Notably, in the 2003 publication, the CHD data were stated to be nonsignificant after the first year. In addition, as was found for the Heart and Estrogen/progestin Replacement Study (HERS) trial, in the WHI the women on HT who used both statins and HT did not have an increased risk of CHD.[17,18]

The media attention that accompanied the release of the results of the HT arm of the WHI to physicians and the public should have been repeated when the new CHD data were released. That there was a significant trend for decreasing relative risk is an important point -- one that has not been emphasized to treating physicians. The ET arm of the WHI shows a mild, nonsignificant, early increase in CHD; however, no increase in CHD was observed for the treatment group in total, and in fact there was a decrease with ET in the younger age group (age 50 to 59 years; HR = 0.56, nCI = 0.30- 1.03).[11]

Table 3 illustrates some of the contrasting results of the 2 WHI studies regarding CHD. Note the increase in CHD in the HT WHI study and the decrease in the ET WHI. Note also that Table 2 compares the HT WHI cardiovascular study details vs the ET WHI general paper.[8,11] Other authors have used data comparing both general papers.[8,11,19] The release of further cardiovascular data from the ET WHI is necessary for a more comprehensive analysis and comparison. Perhaps then we will be able to explain the paradoxical outcomes in the ET and HT arms.

The WHI authors also reported that the results of the ET arm did not support those of earlier observational trials that found ET to be protective for CHD; yet, if one looks at the data presented in the article, there is a decrease for the women in the younger age group (50 to 59) who were randomized to ET.[11] It may be that the authors will address this in future articles. Fasting lipids were measured in 8.6 % of the ET WHI participants. The women in the treatment group had a greater decrease in low-density lipoprotein cholesterol (-13.7% vs -1.0%, P < .001), a larger increase in high-density lipoprotein cholesterol (15.1% vs 1.1%, P < .001), comparable reductions in total cholesterol, and a comparable increase in triglycerides (25% vs 3%, P < .001). The systolic blood pressure, but not the diastolic blood pressure, was elevated by a mean of 1.1 mm Hg, with standard error 0.4 mm Hg. The initial data from the ET WHI also show a modest benefit of ET, as noted by the authors, with a decrease in CHD at 7 years for the entire group. Remember, there was a nonsignificant decrease in CHD after 6 years with HT, although it was neither hyped nor heralded by the WHI investigators or the publishing journals.

Published at the same time as the HT WHI cardiovascular data in the same issue of The New England Journal of Medicine (NEJM) was an article reporting no progression in coronary atherosclerosis by quantitative coronary angiography during therapy with 17beta-estradiol alone or 17beta-estradiol with medroxyprogesterone acetate, which are different hormone preparations from those used in the WHI.[20] The latter article's authors, the Women's Estrogen-Progestin Lipid-Lowering Hormone Atherosclerosis Regression Trial (WELL-HART) Research Group, monitored 226 postmenopausal women in a controlled study and found no evidence of reducing atherosclerosis with hormone therapy, but neither was there an adverse effect. (No comment was made on plaque stability or thrombosis, as noted in the editorial by Herrington[21]). Notably, these patients with established coronary artery disease were all taking HMG-CoA reductase inhibitors along with the 17beta-estradiol.[18] Because there was no increase in CHD in the participants of the WHI who were on statins -- a finding that was also reported in HERS -- this might suggest that the use of cardioprotective therapy such as statins or low-dose aspirin should be considered as an adjunct therapy to HT in patients with risk factors for CHD. The HT WHI authors note that HT seemed to have "less adverse coronary effect on women who were taking statins," but further remarked that until clinical trials are done, there are not enough data to support this recommendation.[8] The data on statin use are only briefly noted for those with CHD and are not discussed for the other participants in this paper.[8] This is an area where evidence-based medicine needs to be developed, but for the present the clinician will need to make a decision on a case-by-case basis.

There are other noteworthy facts about the WHI: There was a high dropout rate -- 42% in the HT group and 38% in the placebo group. In addition, the number of subjects in many of the subgroups was small (although it should be noted that, compared with other studies, the numbers in some subgroups are sizable), and there were a number of comparisons made between the subgroups, with approximately 36 tests for interaction. In the ET arm, 53.8% of the participants dropped out by stopping their medicines.[11] The authors caution that these factors can have an effect on the statistical data and thus, the data should be interpreted with caution.[1,8] In particular, Bailer[22] notes that the loss to follow-up affected the ratios of loss of patients to follow-up to the number patients with a cardiac event. The ratio is 57:99 in WELL-HART and 541:335 in WHI, so that completion of follow-up for 57 missing subjects in WELL-HART or 541 missing subjects in WHI might add enough endpoints to alter the findings.[22] The WHI involved 16,608 women at 40 participating centers with as many as 12,000 independent treating physicians. This study was a gigantic undertaking, and a high dropout rate would not be surprising; indeed, the investigators should be given major credit for keeping as many participants as they did. Even the subanalyses of the WHI represent large studies when compared with other hormone studies. The patients who were lost to follow-up may also simulate the on/off nature of estrogen usage in the real world.

Kleerekoper[23] has advised that we be cautious with our interpretations of the cardiovascular data collected in WHI. Why did the study confirm previous observational studies for the risk of breast cancer and thrombosis but not the bulk of the cardiovascular studies? He emphasizes the need for careful experimentation and observation to see the effect of estrogen on the cardiovascular system rather than taking the WHI as the last word.[23] The release of the ET WHI data showing no increase or decrease in CHD seems to lend support to this point.[10,11] The ET WHI study does not show an increase in CHD -- does this mean that progestin added to CEE can increase the risk? There was a large decrease in the risk for CHD in the younger age group (50 to 59 years) with an HR of 0.56 -- does this vindicate the earlier observational trials that showed cardioprotection with HT or ET? A full report of the ET WHI cardiovascular endpoints should help us to answer these questions.

In addition, it must be noted that the WHI studies used a single type and dose of estrogen, namely 0.625 mg CEE supplied by Wyeth Pharmaceuticals. Would similar results occur with other hormonal products? Another question one ought to consider is whether a patient without an intact uterus has different risk factors for CHD.[24] The ET WHI authors offer possible explanations for the difference in outcomes in the HT and ET arms, including the presence of progestin, different baseline risk factors in the populations of the HT and ET WHI studies, duration of intervention, follow-up time, and the role of chance.

Grodstein, Clarkson, and Manson[25] have suggested there may be a variety of explanations for the difference found between randomized trials and observational trials. The studies may have methodologic differences, including confounding ("healthy user") bias, compliance bias, and incomplete capture of early clinical events. There may be different biologic effects that depend on formulation and dose of the hormone regimen; or differences in characteristics of the study population, including endogenous estrogen level, time since menopause, and stage of monitored endpoint such as atherosclerosis. The WHI data diverge from observational studies with respect to CHD, but with breast cancer there is concurrence between the WHI and observational studies. The WHI showed an increase in CHD with HT, HERS showed the same risk as placebo with HT, and observational trials showed a decrease in CHD with HT. And, the WHI ET showed no increase in CHD.

It has been suggested that different hormone regimens may have yielded a different outcome and that using HT earlier, ie, in younger menopause (on average the women in the WHI were 10 years postmenopausal) and before coronary atherosclerosis is established, would not have led to an increase in CHD. It has been stated that by design the WHI was 10-fold underpowered to show cardioprotection by estrogen and progestin.[26] Thus, observational trials that have shown cardioprotection are the only reliable guides for now until randomized clinical trials are done.[26] Clarkson and Appt[27] have considered the question of whether progestins, such as MPA, attenuate ET's beneficial cardiovascular effects. Their conclusions are that, currently, experimental, clinical, and observational data do not find that progestins attenuate the cardiovascular benefits of estrogen.[27] This kind of information also needs to be considered by the clinician who ultimately makes the therapeutic decisions.

In light of the fact that the trial enrolled more than 166,000 participants and that a total of at least 24 articles are predicted to come out of the WHI, we must keep in mind that the meaning of the results presented in the individual papers on cardiovascular risks cannot be fully appreciated and interpreted until we get a complete view of the whole study. What if a detailed profile of risk factors for the patients participating in the HT and ET arms differs from what we have deduced on the basis of what has been published so far in terms of conclusions regarding who is at risk? This is something that physicians and patients who are earnestly trying to make long-term clinical decisions may have to confront. The WHI observational study, which was also published in JAMA,[28] suggests that use or nonuse of HT was less important as a predictor of cardiovascular risk than were baseline levels of either CRP or IL-6. This article thus gives us a different view and perspective. Another observational WHI study[29] indicates depressive symptoms are significantly related to increased risks of cardiovascular disease (CVD) death and all-cause mortality, even when controlling for CVD risk factors. Taking antidepressant medications did not alter the depression-associated risks. This clearly emphasizes the role of depression and highlights this as another CVD risk factor to treat. This study is curious, however, because antidepressants did not alter the associated risks; perhaps other factors of therapy may affect the outcome, such as type of antidepressant or dose or even timing of therapy. Also the type of depression may need to be clarified.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.