John A. Goldman, MD
Division of Rheumatology and Immunology
Emory University School of Medicine
The Woman’s Health Initiative (WHI) was designed in 1991-1992 to define the risks and benefits of interventions to potentially reduce heart disease, breast and colorectal cancer, and osteoporotic fractures in postmenopausal women (1, 2, 3). After screening 373,092 women aged 50 to 79 years, 161,809 were enrolled in either an observational study or clinical trials of low dietary fat vs. self-selected diet, estrogen and progesterone (HRT) or estrogen (ERT) alone vs. placebo, or calcium and vitamin D supplementation vs. placebo. The HRT study randomized 8,506 women to active treatment and 8,102 to placebo and was terminated early by the Data Safety Monitoring Board (DSMB) due to an increase in breast cancers. The global index statistic (a combination of coronary heart disease (CHD), stroke, pulmonary embolism, colorectal cancer, endometrial cancer, hip fracture and death due to other causes) also indicated the risks exceeded the benefits (2, 4). It was estimated that for every 10,000 women taking HRT there would be eight more breast cancers, seven more cardiac events, eight more strokes and 18 more venous thromboembolisms. 23 more cases of dementia per 10,000 women per year were reported May 2003 (5). The report also showed six fewer colorectal cancers, five fewer hip fractures, and a 23% reduction in other osteoporotic fractures and a 24% decrease in total fractures. Fractures other than hip fracture were not included in the global index statistic. There were no differences in mortality or cause of death between the populations (1).
Goodson notes in a letter to the editor that the overall mortality for the HRT group was still lower than the placebo group (0.52% vs. 0.53% per year). He also points out that the mortality of colon cancer is greater than breast cancer and that the six-month mortality for hip fracture is greater than the 10 year mortality for stage 1 breast cancer (6). This would suggest that the mortality from improvement in colon cancer and hip fracture would offset the increase in mortality from breast cancer (6, 7). He wrote that this might contradict the need for early discontinuation of the study and the conclusion that “the absolute numbers of excess [adverse] outcomes [due to HRT] would increase…” if the study continued.
Further analysis of the risk of coronary artery disease in the WHI was reported in August 2003 and the number of excess events decreased from 7 to 6/10,000 women per year (8). There was a significantly increased risk for CHD in the first year, a trend toward less increase in subsequent years, and an increased risk in the placebo group by year six, which leads to an apparent risk reduction with HRT. The hazard ratios for CHD were actually less, 0.89 (not significant), for those in early menopause (< 10 years). Published at the same time was an article reporting no progression in coronary artherosclerosis by quantitative coronary angiography during therapy with 17ß-estradiol alone or 17ß-estradiol with medroxyprogesterone acetate (9). This article, reporting on 226 postmenopausal women enrolled in the Women's Estrogen-Progestin Lipid -- Lowering Hormone Atherosclerosis Regression Trial (WELL-HART), showed no evidence of reducing atherosclerosis with hormone therapy but no adverse effect either. These patients had established coronary artery disease and were all taking HMG-CoA-reductase inhibitors in addition to 17ß-estradiol. No comment was made on plaque stability or thrombosis as noted in the editorial by Herrington (10). Despite the well-known symptoms of menopause, which are often the major reason to initiate HRT and ERT, the WHI did not report menopausal symptoms in this article. Vasomotor symptoms occur in about two thirds of women and are distressing in 10 to 20 percent. Control of these symptoms may be a significant benefit in thousands per 10,000 women per year. If the effect of HRT on menopausal symptoms had been evaluated, there is little doubt that the number of women benefiting from treatment would have been substantially higher. In May 2003 an article titled "Effects of Estrogen plus Progestin on Health-Related Quality of Life" was published (11). This article concludes that HRT does not have a meaningful effect on health related quality of life compared to placebo. It evaluated the entire population of 16,608 women for only the first year. A subgroup of 1,511 women, 8.6% of the total study population, was randomly chosen for further evaluation at three years. The article does not state the number of women who did not enter the HRT trial because of severe menopausal symptoms. There was a 3-month washout period before the study started for potential participants who were on hormones. Women in this group who had moderate to severe menopausal symptoms were discouraged from participating. This obviously excluded people who may have had the greatest benefit from HRT. Despite this selection bias, there were favorable improvements in those on HRT for bodily pain, role of limitations due to physical problems and sleep disturbance after one year. The authors note these were small improvements and were not maintained after three years in the smaller group. Early menopausal women aged 50 to 59 with moderate-to-severe vasomotor symptoms at baseline (1,072 HRT, 974 placebo) showed improvement in hot flashes (76.7 % HRT, 51.7 % placebo, P<0.001) and night sweats (71% HRT, 52.8% placebo, P<0.001). There were 574 women aged 50 to 54 years. At one year these women were found to have improvement in sleep disturbance (P<0.02) and trends toward improvement in social functioning (P<0.06) and satisfaction with sex (P<0.06). The effect of HRT on osteoarthritis has also been overlooked. Studies have shown that patients on ERT had less worsening of osteoarthritis radiographic scores, with a moderate but not statistically significant effect in one study and a statistically significant effect in another (12, 13). This may have a major implication for patients with osteoarthritis who decide to discontinue HRT. The release of the WHI generated a tremendous outflow of hyperbole and conjecture in the media and yet the conclusions are far from clear. While the safety of patients participating in the WHI governed the DSMB decision to terminate the HRT arm, the risk to present and future patients whose therapy changed because of these conclusions remains largely unaddressed. Osteoporosis is under-diagnosed and under-treated and this study, as reported, underemphasizes the importance of fractures. Although patients are routinely screened for risk factors associated with heart attack, stroke and breast cancer, prevention and treatment of osteoporosis is often overlooked. There are three times more osteoporotic fractures than heart attacks, six times more osteoporotic fractures than strokes and eight times more osteoporotic fractures than breast cancer (14, 15, 16). The conclusions of the WHI undervalue the seriousness of osteoporosis and fractures. Since the need for therapy for postmenopausal women for both prevention and treatment of osteoporosis will increase with decreased use of hormone therapy, we are fortunate to have other options such as bisphosphonates, teriparatide, calcitonin, and raloxifene available. The WHI report clarifies some of the risks and benefits of HRT. However, by excluding information on menopausal symptoms, discounting the significance of osteoporotic fractures, and not addressing the potential effect on osteoarthritis the study did not provide enough information for patients and physicians to base their decisions. This can undermine public trust in scientific reporting. Hopefully, further WHI analyses will clarify what we knew or thought we knew in July 2002. We will anxiously await release of additional information from this study and the ERT arm to help us better understand the magnitude of risk and benefits of HRT and to help us better individualize treatment recommendations for our patients. (See references under Commentary by Mary K. Beard)
Last modified: December 17, 2012