Editorials

Where are we now with hormone replacement therapy?

BMJ 2004; 328 doi: http://dx.doi.org/10.1136/bmj.328.7436.357 (Published 12 February 2004) Cite this as: BMJ 2004;328:357
  1. Klim McPherson, visiting professor of public health epidemiology (klim.mcpherson{at}obstetrics-gynaecology.oxford.ac.uk)
  1. Nuffield Department of Obstetrics and Gynaecology, Research Institute, Churchill Hospital, Oxford OX3 7BN

    It works for symptoms but is not good for future health

    Menopausal symptoms can be grim, and the desire to replace the hormonal “deficit” withexogenous hormones remains strong. Since the 1950s, hormone replacement therapy has been used increasingly,1 while evidence on the risks of unwanted side effects has accumulated. Twenty five years ago, the increased risk of endometrial cancer emerged, resulting in the addition of progestogen. Cohort studies had examined oestrogen alone and indicated important benefits, but, since the 1980s, combined preparations have dominated. Interpretations of the evidence were therefore confused, since whatever effects oestrogen or progestogen have on disease will differ. Eventually, evidence of increased risk of combined therapy on breast cancer, coronary heart disease, stroke,and venous thromboembolism from a randomised trial was reported.2 This trial was stopped early, after an average of five years' follow up among 17 000 women, during which around 40% stopped their trial drugs. The results from theoestrogen alone arm of the woman's health initiative study will reassess the role of combined therapy in 2005.

    Because it may double the risk of breast cancer for long term use3 and of heart disease in the first year of use, combined therapy is problematic. These risks are incommensurate with debilitating symptoms, but women need to be able to judge the risks for themselves. The balance now seems clearer than it was before we knew these risks; hormone replacement therapy works for symptoms but not for future health, which is not what had been widely promised.4

    What have we learned over these decades? Firstly, current biological theory does not predict the effects of hormones on cancer and cardiovascular disease well. Secondly, mass prescription requires large randomised clinical trials with long follow ups. Observational studies of the putative effect of drugs, for coronary disease especially,5 are unreliable. Thirdly, mass markets engender massive vested interests—personal, departmental, and corporate. Genuine care for women demands scientificindependence, since uncertainty allows expert, but illegitimate, conviction too easily.6 Beware of all such profitable bandwagons.

    Understanding the pathogenesis of coronary heart disease, knowing that combined hormone replacement therapy has an unambiguously beneficial effect on lipid concentrations but increases the risk of coronary heart disease, is challenging if oestrogen alone is cardioprotective.7 Menopausal hormones, especially progestogen, probably act by accelerating existing tumours in the breast. Once we get the life course biology right, the pharmacological role on disease prevention will mean massive opportunities (and mass markets). Hormone replacement therapy shows that good intentions can be seriously misleading; hence adequate, as opposed to merely plausible, science is required. Prof John Bailar of the University of Chicago reminds us, “We never know as much as we think we do.”8

    Meanwhile, women with menopausal symptoms have to make decisions. In today's BMJ a clinical net benefit analysis provides insight by balancing current benefitfor symptoms against future benefit and risk.9 It is one way of combining the risks and benefits of future events while ameliorating current discomfort. The latest risk estimates are combined with quality of life weightings associated with symptoms for women aged 50. This kind of analysis is essential but inevitably aggregated and somewhat static. However, the effects assumed for breast cancer and coronary heart disease are probably too conservative (because so many women in the trial stopped treatment). A relative risk of breast cancer of around two for combined treatment would considerably decrease the reported chance of net benefit compared with the 1.27 assumed. Similarly, a near doubling of risk of coronary heart disease in the first year would make things still worse, from an assumed overall relative risk of 1.08. These higher risk values would make the probability of net harm considerably greater for any woman.

    Combined hormone replacement therapy is not indicated for women who have no symptoms. Only with severe symptoms (the consequent reductions in quality of life are such that a woman would sacrifice three months in a year to eliminate them) do these analyses show the net benefit with hormone replacement therapy to be positive. Taking hormone replacement for moderate symptoms require special justification if an average women wishes to benefit in the long run. Menopausal symptoms and risk of breast cancer will always dominate women's decisions, while successfully preventing osteoporosis requires long term use of hormone replacement therapy and cannot be justified by the greater interim hazards of doing so.

    We will be increasingly invited to play up the risks of the diseases against which hormone replacement offers effective and downplay the increased risks. As is already the case, we are invited to believe that the women's health initiative study10 and the million women study are less relevant than is currently thought or flawed with biases. Subgroups will be cited for which the bad effects are not observed—the latest results on coronary heart disease from the women's health initiative study provide (unsafe) opportunities to cite subgroups for which the bad effects are not observed.11 This is a predictable bandwagon effect, not to be ignored—but such claims are often not plausible, never mind adequate. Reputations (and money) are at stake. Where is the scientific evidence for alternative inferences, more reliable than we now have? That the simple message above, intuitive to public healtha while ago, has had to wait to achieve credibility is pitiable.

    At least two further developments are now indicated. One is providing hormone therapy with which it is easier to cut the dose gradually. Women can test their individual metabolic balances with progressively lower doses and presumably thereby lower their risk of breast cancer and cerebrovascular disease. As it is, patches and pills can often be cut—butwhat is required is a product for achieving the lowest doses that can be found to combat symptoms with fewest side effects.

    Further, the increased availability of natural remedies that do not need licences requires care with efficacy and safety. If they work for some, fine, but evidence from trialswould be essential for women to be assured that they pose no greater risks than hormone replacement therapy. Safety data are vital for products whose constituents are not necessarily entirely known and that may contain, for example, phytoestrogens in large doses. What are the long term effects of these preparations, taken on the assumption that being natural they are safe? Will adequate research be done to ensure that we avoid another half century of uncontrolled experimentation on menopausal women? Women have greater expectations of menopausal remedies now—given the false promise of the hormone replacement therapy bandwagon.

    It can take decades to detect important and unanticipated side effects of medications reliably. Do the current regulatory provisions adequately provide for the sensible avoidance of more, tragic episodes? Tucker conservatively estimates an extra 1400 cases of breast cancer, 1200 cases of heart disease, and 1400 cases of stroke—against 860 fewer hip fractures and 1000 fewer cases of colorectal cancer per year in the United States alone.12 Regulators and legislators will be contemplating the implications, as they did after thalidomide, and hopefully we will not get the marketing so wrong again. Severe menopausal symptoms are rated as having worse effects on quality of life than having any of the diseases—pills for symptoms and prevention pose complicated public health problems. Hormone replacement therapy may be a mere example of what is to come—the opportunities remain enormous.

    Papers p 371

    Footnotes

    • Competing interests KM is a member of the Committee on the Safety of Medicines and its expert working group on hormone replacement therapy.

    References