By Molly A. Brewer, DVM, MD, MS
Professor and Chair, Department of Obstetrics and Gynecology, Division of Gynecologic Oncology, Carole and Ray Neag Comprehensive Cancer Center, University of Connecticut Health Center, Farmington
Dr. Brewer reports that she receives grant/research support from the National Cancer Institute.
SYNOPSIS: The interpretation of hormone replacement therapy in postmenopausal women has varied dramatically before and after the publication of the Women’s Health Initiative study in 2002. New studies question the validity of the conclusions reached by the investigators.
SOURCE: Shapiro S, et al. Risks and benefits of hormone therapy: Has medical dogma now been overturned? Climacteric 2014;17:215-222.
In the past, hormone replacement therapy (HRT) was thought to prevent cardiovascular disease (CVD), the number one cause of mortality in women after menopause. Epidemiologic data had shown that the risk of CVD in women was lower than men prior to menopause but increased after menopause to become equivalent to men. Hormones associated with ovarian function, particularly estrogen, were believed to play a large role in the prevention of postmenopausal CVD. Therefore, researchers concluded that if postmenopausal women were given HRT, it would decrease their CVD risk.1-4 Over and above the protection against CVD, data also showed the preventive effects of HRT on osteoporosis,5 adding to the argument that postmenopausal HRT served multiple protective effects.
Since the publication of the Women’s Health Initiative (WHI) study in 2002,6 the use of HRT has changed dramatically. Up to this time, most clinicians caring for postmenopausal women thought that HRT reduced the risk of CVD and prevented osteoporosis. The majority of women were prescribed HRT, unless a woman had breast cancer. Following the release of the WHI manuscript (and the subsequent negative publicity), Dr. Rossoue, the lead author, stressed the importance of understanding how the risk to an individual woman can be low, but the risk to the population at large can be great:
The WHI results tell us that during 1 year, among 10,000 postmenopausal women with a uterus who are taking estrogen plus progestin, 8 more will have invasive breast cancer, 7 more will have a heart attack, 8 more will have a stroke, and 18 more will have blood clots, including 8 with blood clots in the lungs, than will a similar group of 10,000 women not taking these hormones. This is a relatively small annual increase in risk for an individual woman. Individual women who have participated in the trial and women in the population who have been on estrogen and progestin should not be unduly alarmed. However, even small individual risks over time, and on a population-wide basis, add up to tens of thousands of these serious adverse health events.7
The study was stopped early due to a 24% increase in the risk of breast cancer in the estrogen-progestin (E-P) arm. The results of the WHI study were widely publicized and women were universally warned through the media that taking HRT was damaging to their health.
Women either voluntarily stopped taking HRT or their physicians recommended they stop HRT as a result of this study. A 2003 publication determined that most regular users of HRT attempted to stop treatment based on the mass media discussion about the risks of HRT without really understanding the real individual risk of continuing their HRT.8 Aside from the individual risk of breast cancer and CVD, there were multiple other outcomes published from the WHI, most of which showed a detrimental effect of E-P on women’s health.9-11 This furthered the argument that postmenopausal HRT may be harmful to women.
A recent publication has challenged much of the data published from the WHI study. In a 2014 manuscript, Shapiro et al questioned both the study design and the conclusions reached by the WHI group.12 They suggested that the primary endpoints (i.e., CVD, breast cancer) changed multiple times throughout the study. This is worrisome since it implies that the authors were manipulating the data. The authors of the 2014 study said that the global index, which included CVD, breast cancer, stroke, pulmonary embolus, colon cancer, endometrial cancer, and fracture risk as a conglomerate score of risk vs benefit, is not valid because there was little similarity in these outcomes. Instead, they suggest that hazard ratios can be weighted to determine an overall risk or benefit. Multiple modifications of the index were made during the study, which the authors suggest biased the results. Women were informed of the risk of CVD and a large number of the E-P women were unblinded (44%), which would increase the surveillance for CVD and potentially diagnose more CVD. In addition, there was a 42% rate of nonadherence in the E-P group and 10% of the women in the placebo group crossed over to the treatment group. Analyzing women in the group to which they were randomized when there was such a high rate of nonadherence and a high rate of crossover introduced significant confounding factors, particularly if the risk is found to be elevated (i.e., causation will be inferred when no causation may exist). In addition, the hazard ratios are quite low for the risk of breast cancer (1.24), CVD (1.18), and stroke (1.39), the outcomes that were the most worrisome. The clinical significance of a hazard ratio < 2.0 is questionable. There is an epidemiologic concern that this low strength of association (low hazard ratio) is due to confounding factors and is not a real effect.13 The authors concluded that the WHI study, despite its wide impact, may not have demonstrated an increase in the risk of either CVD or breast cancer given the study design and statistical issues, and that as a result of these flaws, women have been harmed.
I have had significant concerns about the conclusions reached by the WHI study since its publication in 2002. The average age of the women enrolled was 63 years, an age that generally has not been associated with starting HRT, and 34% of women had a body mass index > 30 kg/m2, a known risk factor for breast cancer. With the large nonadherence in the E-P group (44%) and the 10% crossover from the placebo group, it is probable that at least 55% of the women did not receive the treatment to which they were allocated. Thus, the conclusion that HRT caused poor outcomes may, in fact, be spurious. As a result of this study, most women would rather suffer with severe atrophic vaginitis that interferes with their sexual and urinary tract function than take hormones because of their fear of breast cancer, CVD, and stroke. Many women will develop osteoporosis but will not take hormone replacement and instead opt for other osteoporosis drugs that are found to possibly have limited benefit. It is unfortunate that the questionable results of this study could have such a large impact on women.
Despite the fact that this was a randomized, controlled trial (the most powerful study design in terms of showing causation), the flaws in the study and the overinterpretation of the results have significantly affected women’s health. In addition, the widespread media coverage and the numerous commentaries by the authors have frightened both women and clinicians. There were six more CVD events, nine more breast cancer cases, nine more strokes, and nine more pulmonary embolisms per 10,000 women. It is not clear if this difference could have occurred more commonly in the women who did not take their HRT (the 44% nonadherence in the E-P group) because the increase is relatively low.
It is always important to weigh the risk and the benefits of any intervention. However, to do so, the results must be interpretable to patients and clinicians. As clinicians, we must always be guided by the caveat to first do no harm. In the case of the WHI, the debate continues as to the true value of this large and impactful study.
- Grady D, et al. Hormone therapy to prevent disease and prolong life in postmenopausal women. Ann Intern Med 1992;117:1016-1037.
- Rijpkema AH, et al. Effects of postmenopausal estrogen-progesterone therapy on serum lipids and lipoproteins: A review. Maturitas 1990;12:259-285.
- Stampfer M, Colditz G. Estrogen replacement therapy and coronary heart disease: A quantitative assessment of the epidemiologic evidence. Prev Med 1991;20:47-63.
- Adams MR, et al. Inhibition of coronary artery atherosclerosis by 17-beta estradiol in ovariectomized monkeys: Lack of an effect of added progesterone. Arteriosclerosis 1990;10:1051-1057.
- Weiss NS, et al. Decreased risk of fractures of the hip and lower forearm with postmenopausal use of estrogen. N Eng J Med 1980;303:1195-1198.
- Rossouw JE, et al; Writing Group for the Women’s Health Initiative Investigators. Risks and benefits of estrogen plus progestin in healthy postmenopausal women: Principal results from the Women’s Health Initiative randomized controlled trial. JAMA 2002;288:321-333.
- NHLBI Stops Trial of Estrogen Plus Progestin Due to Increased Breast Cancer Risk, Lack of Overall Benefit. Available at: https://www.nhlbi.nih.gov/news/press-releases/2002/nhlbi-stops-trial-of-estrogen-plus-progestin-due-to-increased-breast-cancer-risk-lack-of-overall-benefit. Accessed Dec. 3, 2015.
- Ettinger B, et al. Effect of the Women’s Health Initiative on women’s decisions to discontinue postmenopausal hormone therapy. Obstet Gynecol 2003;102:1225-1232.
- Chlebowski RT, et al. Estrogen plus progestin and lung cancer: Follow-up of the Women’s Health Initiative randomized trial. Clin Lung Cancer 2015;Oct 22. pii: S1525-7304(15)00236-3.
- Shumaker SA, et al. Estrogen plus progestin and the incidence of dementia and mild cognitive impairment in postmenopausal women: The Women’s Health Initiative Memory Study: A randomized controlled trial. JAMA 2003;289:2651-2662.
- Shumaker SA, et al. Conjugated equine estrogens and incidence of probable dementia and mild cognitive impairment in postmenopausal women: Women’s Health Initiative Memory Study. JAMA 2004;291:2947-2958.
- Shapiro S, et al. Risks and benefits of hormone therapy: Has medical dogma now been overturned? Climacteric 2014;17:215-222.
- Shapiro S. Bias in the evaluation of low-magnitude associations: An empirical perspective. Am J Epidemiol 2000;151:939-945.