Over the past five years, since the first major publication of the Women's Health Initiative (WHI), the clinical use of postmenopausal hormone therapy has significantly reduced.
This has been reinforced by the findings of the Million Women Study (MWS). Sensational reporting of associated risks has made women and their medical advisers much more cautious about, and even fearful of, its use.
Benefits for symptom relief have not been denied and those for bone were strengthened by the WHI results. On 4 April 2007, JAMA reported a further analysis of the data from the WHI.1 This appeared to contradict some of the previously reported cardiovascular risks. Only two weeks later, the MWS published new data regarding risks of ovarian cancer associated with the use of hormone therapy.2 Once again, there is mass confusion. What do these new findings mean in the real world and what difference will they make to clinical practice?
Women's Health Initiative
The WHI was established in 1991 to test the hypothesis that postmenopausal hormone replacement had a role in primary prevention of cardiovascular disease.
This followed observational data from large cohorts of women that appeared to show a protective effect, with women taking hormone therapy showing a much lower incidence of CHD. This was of major clinical and public health importance because CHD is the top cause of death among women, exceeding breast cancer by a factor of 10 and gynaecological cancer by a factor of 25.
In the WHI, 16,608 women were randomised to a combined oestrogen and progestogen regimen of 0.625mg conjugated equine oestrogens (CEE), with 2.5mg medroxyprogesterone acetate (MPA) or placebo. The global index was a predetermined 'basket' of significant outcomes that comprised CHD, venous thromboembolism (VTE), stroke, breast cancer, colorectal cancer, hip fracture and total mortality.
This trial was stopped after 5.2 years due to a rise in the global index and to everybody's surprise, was reported as showing a headline figure of a 29 per cent increased risk of CHD.
In a second trial, 10,739 women who had had a hysterectomy were randomised to 0.625mg CEE or placebo. This trial, which did not show the same rise in global index, continued for a further two years, publishing its initial report in 2004.
A number of factors need to be appreciated. First, the population of randomised women had an average age of 63 (range 50-79). Second, many significantly symptomatic women were excluded on the basis that relief of oestrogen deficiency symptoms would unblind them.
Third, 36 per cent were being treated for hypertension and 34 per cent had a BMI >30kg/m2. It has been appreciated that this population was different from those symptomatic perimenopausal women in whom we would initiate hormone therapy for symptom relief in clinical practice.
Subgroup analysis within the two trials suggested a reduction in CHD risk in women aged 50-59 for the CEE trial and in women less than 10 years after menopause in the CEE + MPA study. The small numbers of younger women and low incidence did not allow a significant outcome to be established.
Therefore data from the two groups have been amalgamated in this study.
Details were included of 24,317 women in whom menopausal age was either known or could be assumed. The WHI investigators have reanalysed the data about CHD, stroke, total mortality and global index and looked at the influence of age and years since menopause.
Their figures show a non-significant reduction in cardiac risk (HR 0.76 95% CI 0.50-1.16) for women starting hormone therapy less than 10 years after menopause, with a trend to increase over time to reach a significant 28 per cent increase at more than 20 years (HR 1.28 95% CI 1.03-1.58). The trend has significance at the p = 0.02 level. Significance is generally accepted at p = 0.05 (5 per cent risk of the result being due to chance).
However, The Wall Street Journal reported that after the submission of this paper, the JAMA reviewers requested that a higher burden of proof should be required. Significance required is usually predetermined, not modified post hoc.
These hazard ratios translate to an absolute risk reduction of -6 per 10,000 person years for women less than 10 years after menopause.
Risk increases of +4 for women 10-19 years and +17 for women 20 years and more after menopause (per 10,000 women years) were derived.
Relating this to age rather than time difference creates absolute risks of -2 per 10,000 at age 50-59 and -1 at age 60-69 (both decreases) and +19 at age 70-79 (an increase). This must be very reassuring for our patients and deserves greater emphasis.
The study did not contain sufficient numbers of women to calculate for the narrower and more typical age bands of 50-54 or less than five years after menopause.
It did demonstrate an increased risk of stroke of 32 per cent (HR 1.32, 95% CI 1.12-1.56), which does not vary significantly with age or years since menopause. So screening for stroke risk is an important component of the risk analysis at any age, although the additional attributable risk for healthy younger women will be very small. Total mortality is not increased (HR 1.02, 95% CI 0.90-1.15), although there is a non-significant trend with reduction in younger women (HR 0.70 at 50-59, 1.05 at 60-69, 1.14 at 70-79 and trend p = 0.06).
Window of opportunity
The effect of oestrogen on coronary arteries appears to be a trade-off between its beneficial effects on the endothelium and lipids and its negative effects on coagulation and inflammation.
Once vessels are diseased, regression does occur, but there may be a preventive benefit when started early. This has been termed the 'window of opportunity' theory and is consistent with observational data, animal models and laboratory studies. The study provides no information about long-term use of HRT.
Further confusion
The MWS has reported an increase in ovarian cancer diagnosis and death in women taking HRT. This observational trial has been subject to considerable criticism and its breast cancer conclusions have appeared to overestimate risk in comparison with the randomised trials.
To date, the issue of ovarian cancer risk remains unproven and controversial.
The study now reports an estimated additional incidence of one ovarian cancer per 2,500 women taking hormone therapy for five years, leading to one additional death in approximately 3,300 women. Even if correct, this is a very small risk, but patients should nonetheless be told. Women deserve up to date and accurate information, but it should be imparted in a format that is comprehensible and personalised (see box 1), rather than sensational and frightening.
| BOX 1: WHAT YOU CAN TELL PATIENTS |
| Hormone therapy used for four to five years (the average exposure in this trial) shows:
|
Where are we now?
CHD is the greatest cause of death to women. Risk reduction would be very welcome, but even the confidence that there is no risk increase is much appreciated. Risks should be related to the age and health of our patients and put into the context of their everyday lives. Benefits for symptoms, quality of life and bone protection, an absence of cardiac risk and potential risks of stroke, VTE, and breast, endometrial and now ovarian cancer must be explained. An informed decision can then be made.
- Dr Sarah Gray is a GP specialist in women's health in Truro, Cornwall
References
1. Rossouw JE, Prentice RL, Manson JE et al. Postmenopausal hormone therapy and risk of cardiovascular disease by age and years since menopause. JAMA 2007; 297: 1465-77.
2. Beral V, for Million Women Study collaborators. Ovarian cancer and hormone replacement therapy in the Million Women Study. Lancet 2007; doi: 10.1016/S0140-6736(07)60534-0
