HERS II: Follow-Up Report of the HERS Trial

Abstracts & Commentary

Synopsis: There was no cardiovascular benefit associated with hormonal treatment of women with coronary heart disease during 6.8 years of observation, and hormone treatment increases the rates of venous thromboembolism and biliary tract surgery.

Sources: Grady D, et al, JAMA. 2002;288:49-57; Hulley S, et al. JAMA. 2002;288:58-66.

Of the 2763 postmenopausal women in the heart and Estrogen/progestin Replacement Study (HERS), 2321 (93%) agreed to be involved in additional follow-up evaluation. The original study1 lasted 4.1 years, the average extended follow-up equaled 2.7 years, for a mean total of 6.8 years. At the beginning of the follow-up period, and the average age of the participants was 71 (67 at baseline and 74 at closure). The HERS investigators could detect no significant differences in the rates of coronary events or secondary cardiovascular events comparing the treated group with the placebo group. There was no statistical trend for a beneficial effect of hormone therapy with longer duration of treatment. Because of the absence of a difference, the follow-up period, scheduled to last 4 years, was terminated early.

The original HERS report indicated a 2-3-fold increase in deep vein thrombosis (DVT) and pulmonary embolism in the hormone-treated group. In the follow-up period, there was no longer a statistically significant increase in DVT. There was no reduction in pulmonary embolism, but the number of events was too small to provide accurate assessment. The event rates for venous thrombosis were 5.9 per 1000 women per year of treatment and 2.8 in the placebo group. There were 3 deaths from pulmonary embolism in the treated group. Aspirin attenuated the risk associated with hormone therapy.

Overall, there was a 48% increase in risk for biliary tract surgery in the treated group—6 more cases per 1000 women per year compared to placebo.

There were no statistically significant differences in cancer rates, including breast cancer, comparing the treated and placebo groups. Likewise there was no difference in fractures, but this study was not designed to have sufficient power to study fractures (for example, the treated and placebo groups were not matched for fracture risk factors, bone density, or use of drugs that affect bones), and x-rays were not performed to detect spinal fractures.

The HERS investigators concluded that there was no cardiovascular benefit associated with hormonal treatment of women with coronary heart disease during 6.8 years of observation, and that hormone treatment increases the rates of venous thromboembolism and biliary tract surgery.

Comment by Leon Speroff, MD

The original HERS trial was interpreted as demonstrating an early adverse effect of hormonal therapy and an emerging later benefit. This follow-up report emphatically does not support the presence of an emerging benefit with longer duration of treatment.

The additional follow-up period was unblinded; patients and physicians could choose to continue, discontinue, or initiate hormonal or other therapy. Hormone use in the original treated group in HERS declined from 81% after 1 year to 45% during the 6th year (and 11% were using preparations other than the original 0.625 mg conjugated estrogens and 2.5 mg medroxyprogesterone acetate). During the 6th year, 8% of the placebo group were now receiving hormone therapy. Raloxifene or tamoxifen had even been initiated, 3% in the hormone group and 4% in the placebo group. The HERS investigators recognized this problem conceding that their power to detect an increasing benefit was eroded by the changing treatments; however, their analysis indicated an ability to detect at least an 18% reduction in cardiac risk. What about statins?

At baseline, the use of statins (and aspirin) was essentially equally prevalent in the treated and placebo groups (about 40% of the subjects used statins and 80% used aspirin). However, more women in the placebo group began treatment with statins, so that by the end of the follow-up period, the 69% vs. 65% difference-comparing placebo with treatment was statistically significant. The HERS investigators addressed this potential confounder by adjusting for the difference in statin use (as well as other confounders) and concluded that the adjusted analyses were essentially identical to the original analyses. However, no mention is made of the fact that the percentage use of statin use is impressively high. What if any beneficial effect of estrogen is lost because of the effect of statin therapy? Indeed, in a primary prevention trial, inhibition of atherosclerosis with estrogen treatment was observed only in women not receiving statins.2 The HERS investigators compared coronary heart disease events in the hormone group with the placebo group in women not using statins or aspirin and found no difference. However, this very important possible explanation for the lack of a beneficial effect of estrogen in HERS cannot be answered by the analysis of the HERS data because statin and aspirin treatment were not randomized, and the number of events in women not on statins or aspirin was very small. Statin drug treatment reduces the risk of coronary events by approximately 30% (greater in higher risk individuals) in both men and women, exerting both primary and secondary prevention.3,4 The increasing use of statins, especially in older women who are not hormone users, makes it difficult to have a true placebo group in studies of coronary heart disease.

Intention-to-treat analysis compares all individuals in the treated group to all in the placebo group, regardless of individual compliance or completion of the study. Proponents argue that this is the best method of analysis for clinical trials because it reflects the full effect of randomization. Opponents contend that this method is intuitively wrong; how can the long-term benefit of a treatment be assessed if subjects receiving treatment for only a short period of time are included? HERS II performed an "as-treated" analysis, focusing on women with 80% or more compliance and found relative hazards (like relative risk) similar to those in the intent-to-treat analysis. However, the relative hazard for primary coronary heart disease events in HERS II was lower, although not statistically significant (RH = 0.81; CI = 0.52-1.32). Events were fewer in the as-treated analyses because only 37% of the events qualified. Adjustment for statin use was performed only in the intent-to-treat analysis ("only a trivial effect on the findings"). Therefore, do the HERS II results reflect intent-to-treat analysis (with a difficulty in detecting a long-term effect) and few events in the as-treated analysis (because of compliance and drop out problems)?

Every effort thus far to find a defining characteristic that would identify susceptible patients for the possibly increased risk of cardiac events in the first year of treatment has failed to identify a high-risk group of women.5 It remains an unanswered question whether reported increases in cardiovascular events early after the initiation of hormone therapy reflect a true risk of hormone therapy or the effect of reduced events in the placebo groups because of new onset treatment with statins (and aspirin).

It seems to me that the cardiovascular results over the last few years are supporting an emerging theme. The theme is: you need healthy endothelium to respond to estrogen. Experimental evidence in the monkey indicates that the beneficial effects of hormonal treatment are progressively diminished with increasing atherosclerosis.6 In postmenopausal women, the vasodilatory effects of estrogen dissipate with increasing age.7 By that time, the endothelium is involved with atherosclerosis, and it is too late for estrogen to exert a beneficial effect. Therefore the recent results are not so surprising.

The results of HERS II indicate that the increased risk of venous thromboembolism associated with hormone therapy is concentrated in the first 2 years of use (in fact, the increase in HERS was statistically significant only in the first year). They also support the conclusion that low-dose aspirin and statin treatment protect against this risk.8 When considering the venous thromboembolism rates in HERS, remember that these women were at high risk for this complication, reflected by a high event rate in the placebo group.

And let’s not ignore the encouraging finding of no significant increase in breast cancer in these older women treated for 6.8 years! In addition nonsignificant (due to small numbers) decreases in colon cancer and endometrial cancer were observed in the hormone-treated group.

The recent trial results are reasons to be conservative regarding hormone therapy for older women with evidence of coronary heart disease. Certainly we should not promote estrogen as a first-line drug to prevent further clinical events in women with coronary artery disease, especially in women who have had a recent myocardial infarction. Multiple clinical trials have established that treatment with statins is very effective in preventing clinical cardiac events. The results also indicate that there is no need to avoid the use of medroxyprogesterone acetate, because there has been no difference observed comparing women treated only with estrogen to those treated with estrogen and progestin. The recent reports make an argument that the optimal approach to postmenopausal hormone therapy is to start treatment close to the menopause, avoiding a significant period of exposure to low estrogen levels prior to beginning therapy. And there continues to be good reason (a combination of biologic data and uniform agreement in a large number of observational studies) to believe that hormone therapy has a beneficial role in the primary prevention of coronary heart disease. 

Dr. Speroff is Professor of Obstetrics and Gynecology, Oregon Health Sciences University, Portland.


1. Hulley S, et al. JAMA. 1998;280:605-618.

2. Hodis HN, et al. Ann Intern Med. 2001;135:939-953.

3. LaRosa JC, et al. JAMA. 1999;282:2340-2346.

4. Vaughan CJ, et al. J Am Coll Cardiol. 2000;35:1-10.

5. Furberg C, et al. Circulation. 2002;105:917-922.

6. Mikkola TS, Clarkson TB. Cardiovasc Res. 2002;53: 605-619.

7. Herrington DM, et al. Arterioscl Thromb Vasc Biol. 2001;21:1955-1961.

8. Grady D, et al. Ann Intern Med. 2000;132:689-696.

Attention Readers

American Health Consultants is happy to announce that we are opening up our Primary Care Reports author process to our readers. A biweekly newsletter with approximately 5000 readers, each issue is a fully referenced, peer-reviewed monograph.

Monographs range from 25-35 Microsoft Word document, double-spaced pages. Each article is thoroughly peer reviewed by colleagues and physicians specializing in the topic being covered. Once the idea for an article has been approved, deadlines and other details will be arranged. Authors will be compensated upon publication.

As always, we are eager to hear from our readers about topics they would like to see covered in future issues. Readers who have ideas or proposals for future single-topic monographs can contact Managing Editor Robin Mason at (404) 262-5517 or (800) 688-2421 or by e-mail at robin.mason@ahcpub.com.