Recent concerns surrounding HRT

During the 1960s and 1970s, the increasing use of the combined oral contraceptive pill (OCP) heralded a major change in reproductive endocrinology. Women not only had easy access to highly effective contraception, but the OCP resulted in a regular and predictable light withdrawal bleed. In addition to family planning, the OCP allows control of menstruation and potentially the abolition of erratic cycles, menorrhagia and dysmenorrhoea. Acne and mild hirsutes can be improved with the use of low-dose cyproterone acetate as the progestogen and about one-third of pill-takers feel their symptoms of premenstrual syndrome improve on treatment. The concept of taking daily hormones to provide contraception and the additional benefits of cycle control have become routine for millions of women and many have taken the pill for one or two decades with no problems.

Thus many women may approach the climacteric with an understandable reluctance to suffer unpredictable bleeding and menopausal symptoms and a perception that the continued ingestion of exogenous hormones ensures continued well-being. This belief has been re-inforced by both lay media, and the publication of books such as Dr Robert Wilson's Feminine Forever, and by the scientific literature. The benefits of control of vasomotor instability, prevention of vaginal atrophy and bone preservation are well known. Magazines emphasize antiageing aspects and benefits for sexuality but, with the evidence from epidemiological studies that hormone replacement therapy (HRT) reduces the risk of osteoporotic fractures (Torgerson & Bell-Seyer, 2001), oestrogen replacement became the gold standard for the prevention of postmenopausal osteoporosis.

However, oestrogen treatment not only prevents the rapid bone loss that occurs in the 10 or so years after the menopause and maintains the epithelium of the urogenital tract, but observational studies suggested that HRT reduced the risk of heart disease. These putative benefits in relation to cardiovascular disease were extrapolated to other circulatory problems, with potential to reduce the risk of stroke and even the prevention of the development of Alzheimer's disease. Even the observation that oestrogen replacement therapy increased the risks of endometrial hyperplasia and endometrial cancer was not an obstacle to the wider use of HRT, as these risks could be lowered by the use of added progestogens. As a result, combined oestrogen–progestogen replacement therapy was introduced, with the progestogen added either on a sequential basis (12–14 days progestogen per monthly treatment cycle) or continuously (on each day of treatment). A wide range of medical practitioners including gynaecologists, rheumatologists, geriatricians, general practitioners and endocrinologists encouraged longer-term use of HRT to prevent chronic disease. For a while it really did seem like an ideal treatment and prophylaxis.

HRT is big business for the pharmaceutical industry: 20 million women worldwide (Western countries predominately) were estimated to be using a related product in the late 1990s (Beral et al., 1999). In the USA, approximately 38% of postmenopausal women use HRT at some time (Keating et al., 1999). In 2000, 46 million prescriptions were written for Premarin, making it the second most frequently prescribed medication in the USA and accounting for more than $1 billion in sales (Fletcher & Colditz, 2002). Not surprisingly, the industry has developed numerous different products using a variety of routes (tablets, patches, gels, vaginal pessaries, creams and rings, subcutaneous pellets and a nasal spray) and a bewildering variation in oestrogens and progestogens in single, sequential and continuous combined combinations. There are currently more than 120 products licensed for HRT use in the UK, and about a half are authorized for prevention of osteoporosis, implying longer-term use than control of menopausal symptoms alone.

To apply for and obtain a marketing authorization for HRT, prevention of osteoporosis, or any other indication, the product has to be shown in appropriate studies (usually randomized clinical trials) to be both clinically effective (i.e. provide benefit) and safe (i.e. with acceptable risk) for that dose in the relevant patient population and to be of acceptable pharmaceutical quality. If the trials have been restricted to a particular patient population, e.g. certain age groups or those with a mild (or severe) condition at the start of the trial, this may be reflected in the licensed indications or in the precautions that should be followed when prescribing. Conversely, concurrent medical conditions or treatments that would result in an unacceptable risk are contraindicated. However, as discussed below, whilst initial safety studies will highlight common side-effects of a given treatment, rarer or longer term side-effects may only emerge when larger numbers of patients are treated and/or for longer duration.

Initial signals that oestrogen, either alone or in combination with progestogen, increased the risk of breast cancer were confirmed in a reanalysis of data from 51 epidemiological studies, published in The Lancet in 1997 (Collaborative Group on Hormonal Factors in Breast Cancer, 1997). This increased risk of breast cancer with duration of use coupled with evidence that there is a reduction in sensitivity and specificity of mammographic screening in women taking HRT (Kavanagh et al., 2000) led some to raise concerns about longer-term use of oestrogen (Dixon, 2001). However, many believed that if HRT reduced the risk of cardiovascular disease, then overall the benefits of reduced osteoporotic fractures and reduction in heart disease would outweigh the small increase in breast cancer.

Recent studies, however, provide evidence that the risk–benefit balance for the longer-term use of HRT is shifting. In July 2002, JAMA published the results following early termination of one arm of the Women's Health Initiative (WHI) randomized controlled trial (Writing Group for the Women's Health Initiative Investigators, 2002). Publicity around this caused widespread consternation amongst patients, doctors, researchers and, no doubt, pharmaceutical companies. Some of the publicity was scare-mongering, but the picture of benefit and risk in long-term use is changing. The study had found further evidence for an increased risk of breast cancer.

In the same issue of JAMA, another publication (Lacey et al., 2002) reported that ‘women who used oestrogen-only replacement therapy, particularly for 10 or more years, were at significantly increased risk of ovarian cancer.’ Extraordinarily, this study and the accompanying editorial (Noller, 2002) attracted little attention in the press, perhaps fortunately for the clinician attempting an in-depth interpretation of the literature for each individual woman. Yet the information regarding the increased risk of breast cancer and the lack of protection from coronary heart disease was not new, and had been highlighted in advice from the Committee on Safety of Medicines earlier in the year (Medicines Control Agency, 2002). The risks regarding ovarian cancer and stroke were not previously well known, although there have been a number of studies looking at these specific problems.

Short-term side-effects of HRT are well established, both from the randomized controlled trials (RCTs) used to develop and license different HRT products and from postmarketing surveillance once a product is launched. In contrast, longer-term risk is more difficult to establish and relies largely on observational studies that may be subject to bias. Good epidemiology tends to minimize bias, but certainty is impossible to achieve. This is true for all medicines, but the effort put into long-term study of hormones, mainly because of their use in essentially healthy women, has been greater than for other drugs. One of the difficulties in assessing the long-term risk and benefit balance is that many of the observational studies have included ‘generic’ HRT, without distinguishing between the routes of administration, the different products or even whether oestrogen alone or in combination with progestogen was used. In contrast, the findings on safety and efficacy from large long-term RCTs mainly relate to the use of one particular HRT regimen¹. This raises the question whether risks associated with one specific product can, or indeed should, be extrapolated to other products, i.e. is any increased risk a class effect or product-specific? Both the route of oestrogen administration (oral or transdermal) and the different progestogens and combinations will have different metabolic effects. Whilst these may affect the magnitude of any of the individual risks, this remains to be demonstrated using appropriate clinical endpoints.

Despite this limitation, the WHI study, and the HERS (Heart and Estrogen/progestin Replacement Study) and HERS II publications (Hulley et al., 1998; Grady et al., 2002) are landmarks in the understanding of the risk–benefit balance of HRT. The RCTs give us unbiased data not necessarily available from observational studies (Meade & Vickers, 1999). They also cover the spectrum of HRT users; HERS was in women with established coronary disease who were presumed to expect the greatest benefit of treatment, while the WHI study is a primary prevention trial. WHI was designed to help answer questions such as: ‘Does HRT preserve health and prevent disease? Is the risk–benefit profile consistent with a viable long-term intervention to prevent chronic disease?’ In the published report, oestrogen plus progestogen was prescribed to healthy postmenopausal women to assess the effect on overall health. The primary outcome for the trial was the impact on coronary heart disease (CHD), with breast cancer designated as a primary adverse outcome. Other secondary clinical outcomes, including the earliest occurrence of stroke, pulmonary embolus, endometrial cancer, colorectal cancer, hip fracture or death due to other outcomes, were included with CHD and breast cancer in an overall global index.

How do the findings from these latest trials alter the picture of the long-term risks and benefits of HRT? How then are clinicians to advise their patients? It may help to look at the specific indications for HRT and the benefits and risks at differing sites.

Relief of menopausal symptoms

The classical symptoms of oestrogen deficiency relate to vasomotor instability, resulting in hot flushes and sweats, and to atrophy of the vaginal epithelium, resulting in discomfort, dyspareunia and the urethral syndrome. There are many other symptoms that may be attributable to the menopausal state, including fatigue, loss of concentration and confidence, aching joints, loss of libido, thinning hair, weight gain and depressive symptoms. Given that Professor Tony Weetman (2002) has told us that we should not treat our euthyroid patients suffering from fatigue, weight gain and thinning hair with thyroxine, because they do not get better, can we nevertheless expect women to respond to HRT if they are menopausal? As he emphasizes, a lot of people have these nonspecific symptoms, and endocrinologists would only expect symptoms to respond to hormone replacement when they are due to hormone deficiency.

Oestrogen replacement is highly effective at reducing (almost abolishing) vasomotor sweats and flushes, which are pathognomonic of oestrogen deficiency, and local oestrogen will improve vaginal epithelium and associated symptoms. When fatigue, poor concentration and depression are due to oestrogen deficiency with disrupted sleep patterns and recurrent daytime flushes, these nonspecific symptoms will respond. Arthralgia and stiffness (not deforming arthritis) may respond to oestrogen. Loss of libido is most commonly multifactorial, but tibolone is licensed for depressed mood and decreased libido and can be helpful.

These commonsense observations based on clinical experience have been substantiated in a recent publication looking at quality-of-life and depressive symptoms in postmenopausal women after receiving hormone therapy (Hlatky et al., 2002). Women in the HERS trial were assessed for physical activity, energy/fatigue, mental health and depressive symptoms. Those with flushing had significantly improved mental health and depressive symptoms on hormone therapy, but in those without flushing, hormone therapy led to faster declines in their physical activity and energy/fatigue scores. There was increased frequency in vaginal discharge, uterine bleeding and breast symptoms in association with hormone therapy. This study suggested that HRT does not have a general benefit for postmenopausal women; it improves quality-of-life only for women with menopausal symptoms due to oestrogen deficiency–exactly as endocrinologists would expect. If a disappointing response is seen, factors other than oestrogen deficiency should be sought.

Prevention of osteoporosis

Both men and women suffer age-related bone loss from around 40 years old, but the problem is exacerbated in women by a decade of accelerated bone loss between 50 and 60 years of age that is related to climacteric ovarian failure. Thus about one-third of women, compared to one man in 12, will develop osteoporosis during their lifetime. HRT regimens which deliver a day-to-day oestradiol level of > 250 pmol/l slow down the rapid bone turnover that is associated with oestrogen deficiency. This can be demonstrated by a decrease in excreted urinary markers of bone resorption, although these are not routinely measured in clinical practice in the UK. Clinically, the effectiveness of the treatment is generally determined by bone mineral density measurements [usually dual-energy X-ray absorbtiometry (DEXA) scanning], and an arrest of bone loss or a gain is demonstrated at hip or vertebrae. The clinical end point of interest is the impact on fracture rate. There are numerous observational reports of the benefits of oestrogen in maintaining bone mass and in significantly reducing the incidence of osteoporotic fractures at the key sites of the hip, distal radius and spine (Weiss et al., 1980; Cauley et al., 1994). RCTs have confirmed the epidemiological data with regard to vertebral fractures and the WHI trial confirmed a significant reduction in hip fracture (see Table 1).

However, the problem is that when oestrogen therapy is stopped, the accelerated bone loss begins again, and the evidence is that approximately 10 years after the cessation of HRT, the bone density is back to baseline, or below (Michaelsson et al., 1988). There are to date no prospective controlled data to show for how long the reduction in fracture risk is maintained after oestrogen treatment is stopped. Most clinicians believe that by the time a woman reaches her 80s, when the risk of fractured neck of femur becomes of major concern, any benefit from HRT given between the ages of 50–60 will have been lost. If long-term protection against osteoporosis were required, then treatment would have to be maintained long-term.

Cancers

Vascular disease

Coronary heart disease

Cardiovascular disease is of great public health importance, because it is the leading cause of death in both men and women and CHD is its major component. There are a number of metabolic and vascular effects of oestrogen that make a potential benefit on the risk of vascular disease biologically plausible. The sex difference in mortality from CHD decreases with increasing age, which has been interpreted as suggesting a protective effect of oestrogen in premenopausal women, but this interpretation may be mistaken (Tunstall-Pedoe, 1998; Lawlor et al., 2002), as there is no change in rate in CHD around the menopause (Heller & Jacobs, 1978). Oestrogen deficiency is associated with adverse effects on lipids and lipoproteins, including increases in total cholesterol, triglycerides, low-density lipoproteins (LDL) and Lp(a) and decreases in HDL. Cardiovascular tissues express oestrogen receptors, which may mediate effects on vasomotor tone and the endothelial response to injury. Rapid, transient vasodilatation occurs with oestrogen and oral oestrogen results in reduction in LDL cholesterol and increases in HDL cholesterol. Oestrogen therapy has a beneficial effect on other risk factors such as fibrinogen, plasma viscosity, insulin sensitivity and markers of platelet aggregation and endothelial cell activation. Conversely, other effects of oestrogens, such as increasing triglyceride levels and increasing C-reactive protein (CRP) levels, in keeping with a proinflammatory effect, could adversely effect the development of cardiovascular disease. When oestrogen therapy is combined with progestogens, there may be loss or attenuation of the beneficial effects on lipids, although progestogens have differing effects on lipids and lipoproteins depending on their androgenicity.

These surrogate markers, coupled with observational studies showing a reduced risk of CHD, has led to HRT being considered for use as preventative therapy in postmenopausal women with established CHD risk (Stevenson, 2001). However, the clinical relevance of these surrogate markers is of uncertain significance. Observational studies may have been biased by lifestyle and general health status of those women choosing to take HRT. Indeed, in 1997, the British Medical Journal published a meta-analysis of data pooled from 22 randomized trials of HRT, which showed no cardiovascular benefit from therapy (Hemminki & McPherson, 1997). There was a small but nonsignificant increase in cardiovascular events and confirmation of the increased risk of thromboembolic events. This conclusion was regarded as extremely controversial at the time. Nevertheless, there has been a widespread belief that HRT would confer protection against heart disease, and product information for a number of HRT products referred specifically to the beneficial effect on lipids and a potential beneficial effect on the risk of CHD in postmenopausal women, even though no HRT product has ever been licensed in the UK for either the primary or secondary prevention of cardiovascular disease.

The potential of HRT to prevent CHD led to the establishment of randomized trials to test the hypothesis that HRT would reduce the incidence of CHD in both healthy women and women with established CHD. Unexpectedly, HERS (Hulley et al., 1998) suggested an adverse effect on coronary disease in the first year of treatment, with a trend towards possible benefit over 4·1 years of follow-up. In April 2002, the Committee on Safety of Medicines issued advice and re-iterated that HRT does not have an indication for the prevention of cardiovascular disease and that no benefit had been shown in recently conducted RCTs (Current problems in Pharmacovigilance, 2002). This advice was in line with the recent view of the American Heart Association (Mosca et al., 2001). The advice from CSM was questioned with respect to cardiovascular disease in a leader in the British Medical Journal by Stevenson & Whitehead (2002), who considered that the lack of benefit on CHD applied only to the particular combination of hormone therapy, namely conjugated equine oestrogen (CEE) 0·625 mg and medroxyprogesterone acetate (MPA) 2·5 mg. They re-iterated the belief that observational studies have suggested a major health benefit of postmenopausal HRT in reducing heart disease.

Since then, HERSII and WHI have reported. HERSII reported the unblinded follow-up, for a further 2·7 years, of women randomized in the HERS study (Grady et al., 2002). The study concluded that, over 6·8 years of follow-up, there was no significant decrease in the rates of primary or secondary cardiovascular events; the trend seen at first was not continued with longer follow-up. The prematurely terminated combined HRT part of the WHI trial, found no reduction in risk of CHD over 5·2 years follow-up, and like HERS, a nonsignificant increased hazard ratio (1·29) for CHD. The study estimated seven excess CHD events per 10 000 person-years for the oestrogen plus progestogen combination. The WHI has both a clinical trial and observational study components. Prospective analysis of this large, initially healthy, cohort of postmenopausal women demonstrated that baseline levels of CRP were significantly higher among women who developed incident CHD, and that women on HRT had significantly higher median CRP levels (Pradhan et al., 2002). The findings regarding risks associated with elevated CRP have been replicated in the recent report from the Women's Health Study (Ridker et al., 2002). Both WHI and HERS/HERSII used the same HRT, a specific combination of continuous oral CEE 0·625 mg plus MPA 2·5 mg, marketed as Prempro. This exact combination is not available in the UK, although the closest products are Premique (CEE 0·625 mg plus MPA 5 mg), or Premique Cycle (CEE 0·625 mg plus MPA 10 mg for 12 days a month). Comparable long-term safety data for other HRT combinations are not available and it is not clear how data from one particular HRT regimen can or should be extrapolated to other products. There are no comparable data yet that suggest product differences in CHD clinical outcomes.

The oestrogen-only arm (CEE 0·625 mg) of the WHI trial still continues (due to end 2005), and this will result in valuable data for healthy women with hysterectomy, but oestrogen-only treatment has no place in the management of women with an intact uterus. ESPRIT-UK (Oestrogen in the Prevention of Re-infarction Trial) is using oral oestradiol valerate only in women following first myocardial infarction. WISDOM (Women's International Study of Long Duration Oestrogen after the Menopause) has recently been terminated early, but was randomizing women to the same CEE and MPA preparation used in WHI. It was not stopped on safety grounds, but for practical reasons of recruitment and consideration that little new information would emerge from the trial. Therefore we are not likely to get more data on combined HRT using different products in large RCTs in the near future.

The possibility that the lack of benefit relates only to the use of the one particular HRT regimen used in WHI and HERS and that different oestrogens and progestogens, and other routes of administration may have different metabolic effects is widely acknowledged. However, the hope that other products may still show cardiovascular benefit is weakened by the recent publication of a RCT from Papworth Hospital (Clarke et al., 2002). Importantly, the PHASE study (Papworth HRT Atherosclerosis Study) used transdermal 17β-oestradiol in women with a hysterectomy, and transdermal oestradiol and norethisterone in women with an intact uterus. Treatment with oestrogen-alone or combination therapy did not reduce the incidence of acute coronary events, with a nonsignificant increase in cardiac events in the HRT-treated group.

Critics may argue that the initial harm did not reach statistical significance in any of the studies. However, what is clear from these studies is, regardless of possible harm in the short term, there is no evidence of benefit for preventing CHD in either healthy women or in women with pre-existing disease (Beral et al., 2002). Claims for other HRT products, based on observational data, that HRT substantially protects against coronary heart disease should now be discounted (Beral et al., 2002).

Cognition and Alzheimer's disease

A number of observational studies have shown a reduction in risk of dementia among postmenopausal women taking HRT. Most of these studies were rated by the US Preventive Task Force as being of poor methodological quality (USPSTF, 2002). More recent observational studies (based on relatively few exposed cases) have suggested little, if any, beneficial effect when confounding has been taken into account (Lindsay et al., 2002; Zandi et al., 2002). Some small randomized trials, also surveyed by the USPSTF, showed some benefit on cognition. Some small trials have used HRT as a treatment for Alzheimer's disease but these have not shown benefit overall. The USPSTF concluded there was uncertainty whether HRT had benefit for the prevention of dementia.

Results from the HERS trial (Grady et al., 2002) did not show any benefit on cognitive function in women with coronary heart disease, as measured on six standardized tests. The only statistically significant result (P = 0.02) was that women assigned to HRT did worse on the verbal fluency test than women assigned to placebo. Recently (May 2003) results from the WHI Memory Study (WHIMS) have been published (Rapp et al., 2003) based on the largest preventative trial to date. The combination HRT Prempro did not prevent cognitive impairment in the group studied (4895 women aged 65 years or older) and there was a small increased risk of dementia in women receiving combination therapy compared with placebo, although this was only statistically significant in women over the age of 75 years.

Thus the overall results from trials do not confirm the apparent benefits seen in some observational studies. This is reminiscent of studies on CHD. Currently women should regard the idea that HRT will prevent dementia with great caution.

Conclusions

Most of the known risks associated with HRT increase with increasing duration of use. Therefore for short-term use to alleviate menopausal symptoms of oestrogen deficiency, for which HRT is uniquely effective, the benefits may still be considered as outweighing the risks for most women who use it. Longer-term use of HRT is authorized only for the prevention of osteoporosis but patients should be made aware of the increased risks with HRT use over time. There are alternative preventative and therapeutic options for osteoporosis (most notably bisphosphonates) that do not seem to be associated with the risks of treatment with oestrogen.

HRT is not authorized for the prevention or treatment of any vascular disease, and there are many strategies proven to reduce the risk of vascular problems, including aspirin and control of cholesterol levels and blood pressure. Currently, a core SPC has been developed for HRT products across Europe, and all references to beneficial effects on cardiovascular disease have been deleted. In the USA, the safety information for Prempro, Premphase and Premarin has now been updated to state specifically that the products are not indicated and should not be used to prevent CHD. The Food and Drug Administration is considering changing the term for the class of drugs from ‘hormone replacement therapy’ to ‘hormone therapy’ or ‘menopausal hormone therapy’. The decision to use HRT should be discussed with each woman on an individual basis, taking into consideration her history, risk factors and personal preferences. As many of the risks increase with increasing duration of use, a woman's risks and benefits should be regularly reappraised with continued HRT use. However, it must be emphasized that the absolute incidence of any of the adverse outcomes is low and the risk to an individual woman is small, but small risks accumulate over time. Table 1 gives a summary of the best estimates of benefits and risks associated with using HRT.

Women with a premature menopause represent a particular challenge to endocrinologists. Patients with hypopituitarism or premature ovarian failure could require treatment for decades, depending on the age of presentation. The impact on risk factors such as the development of breast cancer or of cardiovascular events associated with long-term administration in these young women is not known. Extrapolation from studies in older women (extending their lifetime exposure to oestrogen) to the correction of premature oestrogen deficiency is inappropriate. Most physicians will prescribe oestrogen replacement to prevent osteoporosis and symptoms of deficiency up to the age of 50 years and then re-evaluate the risk and benefit of continued use as in any other menopausal woman. The choice of formulation should be made on an individual basis, but a combined oral contraceptive pill provides oestrogen for only 3 weeks out of 4, and delivers supraphysiological levels of synthetic oestrogens and a ‘menopausal’ preparation may be safer and more appropriate.

Clinicians will continue to prescribe hormones in the short term to benefit millions of women with symptoms of oestrogen deficiency. They should make it clear that on current evidence the risk–benefit balance to overall health is not usually favourable for long-term prophylaxis, but that many unanswered questions remain. The role of selective oestrogen receptor modulators (SERMS) such as raloxifene and their impact on chronic disease remains uncertain. The metabolites of the synthetic steroid tibolone have oestrogenic, androgenic and progestogenic properties and this may result in a different risk profile in the long term; however, this has yet to be established. We await the results of trials of oestrogen-alone compared with placebo. Local (vaginal) oestrogen preparations may help with the chronic problems of urogenital deficiency, but current products are not licensed for long-term use. The relative merits of continuous combined treatment compared with cyclical preparations need confirmation.

The collective term HRT covers many different products on the market and the outstanding question remains: what is the implication of the WHI study for the long-term safety of other hormone therapy products, doses and routes of administration? However, data on every possible combination are not ever going to be available, and it cannot be concluded without evidence that other products are safer. Careful analysis and interpretation of the evidence to date must inform good prescribing and patient care.