The effects of estrogen deficiency on the skeletal system have been clearly established for more than 60 years. The molecular mechanisms accounting for estrogen-related bone loss have been known for a decade or longer, and the benefits of estrogen on the skeleton have been recognized for half a century. The adverse consequences of estrogen depletion and benefits of estrogen replacement have been reported by every clinical study evaluating estrogen use in the postmenopausal woman.
The benefits of estrogen therapy in the prevention of osteoporotic fractures were most recently confirmed in the largest clinical trial to date, the Women’s Health Initiative (WHI) (Cauley et al 2003, 1729-38). This study clearly demonstrates that therapy with continuous combined hormone therapy (0.625 mg conjugated equine estrogen + 2.5 mg medroxyprogesterone acetate ([CEE/MPA]) significantly decreases the risk of osteoporosis-related hip fractures and all other osteoporosis-related fractures in postmenopausal women. Despite these results, the authors concluded
“When considering the effects of hormone therapy on other important disease outcomes in a global index there was no benefit, even in women considered to be at high risk of fracture.” This conclusion is based on an analysis of stratification of fracture risk against the global index.
The statistics report the hazard ratio (HR) for the global index was not statistically significant for women in the lowest third (HR 1.20; 95% confidence interval [CI] 0.93-1.58) and upper third (HR 1.03 (95% CI 0.88-1.24) of the fracture risk scale (see below). The hazard ratio was statistically significant only for those patients in the middle third of the fracture risk scale (HR 1.23; 95% CI 1.04-1.46). The full text of the article states: “The interaction between treatment effect and summary fracture risk on the global index was not significant (p=0.54).Thus, there was no evidence of a net benefit, even in women at highest risk of fracture.” I readily admit to not understanding this sophisticated analysis but I believe that the data could equally well be reported that there was no evidence of net harm in those women at highest risk of fracture.
The global index was incorporated into the WHI as an overall measure of estrogen + progestin’s benefit vs. risk, which the investigators believe to be more significant than the evaluation of mortality alone. To my knowledge this innovative approach has not been used in other clinical trials, and has not been validated in any model. The WHI fracture risk scale is based on a validated model published by Black et al, which identified 20 different risk factors associated with fracture (Black 2001, 519-528). The differences between the WHI fracture risk assessment model and that described by Black et al are not trivial (e.g., the WHI did not include measurements of the hip bone mineral density for all patients, and did not assess peak adult height just to mention two items). The WHI fracture risk assessment model, like the global index, has not been validated. Thus, the conclusions based on the current WHI study are the result of an analysis of two nonvalidated models.
The global index, once validated, may indeed be an appropriate approach to look at overall benefits and risks of CEE/MPA. It becomes less appropriate when looking at individual outcomes. For example, there is abundant documentation of the devastation resulting from hip fracture including excess mortality in the first year, long-term morbidity and nursing home placement for those who survive the first year, the impact on the family of the affected patient, and the overall cost to the health care system. How can these serious complications be given equal weight to pulmonary embolism, which fortunately is fatal in only a small proportion of cases and is not associated with the long-term morbidity and health care costs that approach those of hip fracture? It is possible to perform a quality of life or cost analyses of each of the adverse outcomes included in the global index and take a very different approach to a benefit-risk analysis.
The above concerns notwithstanding, the WHI investigators have missed an important opportunity to add to our knowledge base. Epidemiologic studies have indicated that those patients who are at greatest risk for breast cancer are the least likely to develop osteoporosis and vice versa. None of the several articles documenting this (including at least two by one of the coauthors of this WHI paper) is cited in this manuscript! There appears to be little doubt that the skeleton is an estrogen-dependent tissue, as is the breast. It is not surprising that those at lowest risk for fracture (possibly determined by higher lifetime estrogen exposure) are also at greatest risk for breast cancer (by the same mechanism). What we should have learned from the WHI investigators is whether this relationship is true in their study population as well. 8102 women were assigned to placebo, and one-third of these were assigned to the low-fracture risk group. We are told that their annualized incidence of fractures was 1.3%, half the incidence seen in women assigned to the high-fracture risk group (2.7%). What was the breast cancer incidence in these groups in the placebo arm? There certainly was as much or more data available in this study than in any of the previous reports relating osteoporosis protection and increased breast cancer risk. If breast cancer did not contribute to the HR of the global index, then which adverse outcomes were noted in those on HT? We cannot ascertain that important question from this WHI manuscript.
Finally, there is a crucial element missing from the WHI, particularly as it relates to the skeleton. The most rapid postmenopausal bone loss occurs during the first 5 years, slowing down substantially thereafter. The WHI population included only 16% to 17% of women who were within 5 years postmenopause (there was no need to include these patients because the benefits of estrogen in the early menopause have been repeatedly demonstrated in numerous controlled clinical trials). Women using estrogen for management of the menopause do not need other medications for skeletal protection. This beneficial effect only lasts as long as the patient continues estrogen therapy; bone density and fracture risk are back at baseline within 1 year of discontinuing the hormone.
In clinical practice, we have known for decades that women tend to take a holiday from estrogen whenever a prescription runs out and do not refill the prescription if her menopausal symptoms do not recur. This long predates WHI and should not change post-WHI. Most women did not and do not take estrogen for as long as five years postmenopause. However, there has always been a substantial minority of women who do not tolerate this estrogen-free period, and quickly experience a recurrence of symptoms leading to the reinstitution of estrogen therapy.
In conclusion, I believe the following statements can be made:
* Estrogen + progestin is now the only osteoporosis therapy documented to decrease fracture occurrence even in women not specifically selected as being at increased risk of fracture.
* The benefits of estrogen in controlling the symptoms of menopause are not questioned as a result of the WHI study findings.
* Patients will be receiving skeletal protection as long as they use estrogen +/- progestin for control of menopausal symptoms.
* Other effective therapies that can protect the postmenopausal patient from fracture are available once she no longer experiences symptoms of menopause.