The selective oestrogen receptor modulation: evolution and clinical applications

Introduction: the oestrogens

Steroids are widely dispersed throughout the biosphere where they fulfil a plethora of roles as bioregulators in both plants and animals. Indeed the simplest organism yet discovered to be utilizing them is a water mold which deploys a pheromone, oogoniol, which is a steroid. Within the animal kingdom, true hormonal use of the oestrogens appears to be restricted to the phyla Arthropoda and Chordata whereas their elaboration appears to have originated as early as ≈400 million years ago with the phyla Mollusca and Echinodermata which developed the cytochrome P450_arom necessary to convert androstenedione to oestrone and testosterone to oestradiol. Until relatively recently it was assumed that, in the human, the endocrine oestrogens were largely if not exclusively involved in reproductive physiology. They were known to be responsible, with other agents, for the complex signalling and signal transduction which prepares the female organism for the key events of ovulation, blastocyst implantation, maternal adaptation to pregnancy and lactation—all of which, being central to the survival of the species, are highly conserved. However, the last half century has seen a major expansion in our understanding of the range of oestrogenic activity in physiology. The physico-chemical characteristics of the oestrogens and their receptors provide a signal and signal transduction system of high utility which has found a role in many systems unrelated to reproduction.

Pathophysiology of osteoporosis: role of the oestrogens

It was just over half a century ago that Fuller Albright [1] breached the reproductive enclosure with a report that osteoporotic women frequently reported a history of surgical oophorectomy. This observation led to the investigations which have now placed the oestrogens at the very heart of bone physiology where they play a pivotal role in the maintenance of structural and functional integrity of the human adult female skeleton. This is perhaps best illustrated by summarizing the effect of their withdrawal at the time of menopause when climacteric ovarian failure, following functional oocyte exhaustion, leads to an order of magnitude decline in the level of plasma E₂ from ≈500 pmol l⁻¹ to ≈50 pmol l⁻¹. The latter is a level below that seen in many age-matched men and results in profound changes in bone homeostasis. The principal net effect of E₂ loss is observed at the Bone Multicellular Unit (BMU) which is responsible for bone turnover. In oestrogen replete women, the unvarying 5 month BMU cycle of bone resorption—pause—bone formation, results in new bone replacing old with no significant deficit in bone tissue. In other words, bone resorption and formation are securely coupled and oestrogen has a major role on maintaining this balance. This state of affairs is lost in oestrogen deficiency [2]. The amount of bone excavated by the osteoclast cooperative at each site is increased and, crucially, is not matched by osteoblastic replacement. The minute loss of bone mass at each BMU is subjected to a multiplier effect in that the absolute number of BMUs also rises thereby increasing turnover. The loss of oestrogen removes a restraint to the still unknown primary signal for osteoclasts to assemble at a bone surface to commence a BMU cycle (Figure 1).

Figure 1.

Unbalanced bone turnover.

Certain of the intermediate steps in the translation of oestrogen deficiency into derangement of bone remodelling have now been elucidated and involve parathyroid hormone [3], certain growth factors including insulin-like growth factor-I [4] and the cytokines IL-I and IL-6 [5].

In men also, oestrogen is now known to be essential to bone health. Precise documentation has been published of men who are profoundly oestrogen deficient, either through absence of the oestrogen receptor [6] or through the absence of cytochrome P450_arom [7]. These men exhibit giantism, unfused epiphyses and, in the case of aromatase deficiency, rapidly respond to oestrogen replacement with epiphyseal closure and bone density gain.

Other pathophysiological roles for oestrogens

The cardiovascular system was the next in which a physiological role for oestrogen was described. The Framingham study reported premenopausal women who had been deprived of endogenous oestrogen exhibited a significant excess of cardiovascular disease over oestrogen replete age-matched controls [8]. There followed a multitude of clinical and laboratory-based studies which have assigned to the oestrogens a significant role in cardiovascular physiology. In particular, they operate through direct vascular effects including stimulation of NO synthase and inhibition of endothelin-1 production, whose net effect is to restrain the process of atherogenesis [9]. This protective activity is aided by the favourable influence of oestrogen on the plasma lipid profile where they reduce total and LDL-cholesterol and increase HDL-cholesterol, a process whose clinical advantage is attested by observational [10] though not yet by interventional studies [11]. Data from randomized controlled trials such as the Women’s Health Initiative (WHI) in the US will be necessary to establish a causative association between hormone replacement therapy (HRT) use and the primary prevention of atherosclerotic disease.

Intriguing, but even less well established are the effects of oestrogen lack on the central nervous system. The excess of Alzheimer’s disease in women, the ability of oestrogen to promote cerebral blood low and neurotransmitter production have led to observational case-control and cohort studies which suggest that the incidence and clinical course of Alzheimer’s disease may be adversely affected by oestrogen deficiency [12]. Complementary in vitro data suggest that the elaboration of amyloid from amyloid precursor protein may be inhibited by oestrogen at nanomolar concentrations [13]. The list continues to grow. Other conditions in which quality observational evidence now suggests an aetiological role for oestrogen deficiency include adenocarcinoma of the colon [14], cataract [15] and certain auto-immune diseases such as rheumatoid arthritis and multiple sclerosis [16]. It must be emphasized, however, that observational studies, however, well conducted, can only show association and are subject to the play of chance, the intrusion of bias and the operation of confounding variables. Interventional studies with appropriate randomization blinding and rigorous controlling are the current gold standard for establishing causation.

For all the nonreproductive systems utilizing oestrogen for optimal function, the midlife climacteric with its functional and endocrine ovarian failure, poses a problem of adaptation. The postmenopausal reduction in circulating oestradiol brings in its train a set of symptoms which are well known and described and a set of metabolic changes which are not. High prevalence of menopause is a relatively new human female phenomenon. The present longevity of the female population—≈81.5 years in the UK—means that the prevalence of the postmenopausal state in developed countries is high and its clinical consequences are now of high relevance not only to individual welfare but also to the public health.

Oestrogen replacement

The realization that the symptoms reported by perimenopausal women were capable of amelioration by oestrogens has led, over half a century, to the present highly developed and researched field of hormone (oestrogen) replacement therapy (HRT). It is now beyond scientific debate that exogenous oestrogens will resolve such menopausal symptoms as hot flushes, night sweats, vaginal dryness and irritability. However, there is a major problem which centres on acceptability. Many women believe that the menopause is a natural phenomenon which should not attract medical intervention, whereas the counter argument, that the oestrogen deficiency state is itself profoundly unnatural in evolutionary terms, has failed to gain lay credence. Some women feel that postmenopausal stimulation of the oestrogen sensitive reproductive organs is not only unnatural but likely to be harmful in the longer term. To support this view it is accepted that unopposed oestrogen stimulation of the endometrium is clearly associated with an excess of endometrial adenocarcinoma and the incidence of breast cancer rises by a small but quantifiable amount for each year of HRT therapy [17].

A further major difficulty in the acceptance of HRT in the postmenopause is the return of cyclic or unscheduled bleeding cited by many women as a prime reason for a refusal to embark on HRT or to discontinue it. Conventional cyclic HRT regimes usually impose a withdrawal bleed which is either monthly or quarterly. The so-called ‘bleed-free’ preparations containing oestrogen+progestogen taken daily, avoid cyclic scheduled bleeding but may cause unscheduled bleeding especially in the first 6 months of treatment.

For these reasons it has been repeatedly observed that the acceptance of HRT is poor even among properly counselled women likely to benefit materially from its effects [18]. Hence as deliberate policy, a pharmacological research drive has now resulted in the first selective oestrogen receptor modulator (SERM) licensed for the prevention of bone loss, an effect apparently unaccompanied by the stimulation of such classical oestrogen-sensitive tissues as breast and endometrium. The title of SERM, cumbersome but accurate, describes an agent capable of effecting differential responses via the oestrogen receptor in specific tissues.

Evolution

The path leading to SERM development originates in 1896 when Beatson reported regression of advanced mammary carcinoma after oophorectomy in some but not all patients [19]. This differential response would eventually be tracked to the variability in the tumour of the oestrogen receptor (ER) which was described by Jensen & Jacobson in 1962 [20]. Four years earlier, however, Lerner [21] reported that a nonsteroidal antioestrogen MER-25 (ethamioxytriphetol) inhibited the action of oestradiol on endometrium without itself causing endometrial stimulation. Thereafter, and roughly contemporous with the discovery of the ER, the triphenytethylene clomiphene citrate was reported to be capable of inducing ovulation by blocking the negative feedback of oestrogen upon the hypothalamic-pituitary axis [22]. It thus promotes gonadotrophic secretion—a function which it performs therapeutically in certain anovulatory subfertile women to this day. This was followed in 1974 by the landmark paper by Jordan which showed that tamoxifen (ICI 46 474) inhibited the dimethylbenzanthracene (DMBA) induced mammary tumours in the rat [22]. That these agents had a more varied portfolio of action was shown by Beall et al. [23] using clomiphene and Jordan [24] using tamoxifen and raloxifene, the latter then known as keoxifen. These authors demonstrated that the two agents were capable of restraining bone loss in the oophorectomised rat model. These were remarkable observations since a pure antioestrogen would have been expected to promote rather than inhibit bone loss. Clomiphene is a racemic mixture of enclomiphene and zuclomiphene and hence tamoxifen must be accorded primacy in the SERM field. However, the discovery that tamoxifen was, in some patients, capable of endometrial stimulation leading to hyperplasia or neoplasia [26] precluded its development as a bone antiresorptive agent which would be bleed-free and, hopefully, breast-safe as well.

Raloxifene

This agent is a benzothiophene (Figure 2) which was first examined in the Indianapolis Laboratories of Eli Lilly as a potential inhibitor of breast tumour cell lines including MCF-7 [27]. Although performing well in vitro and proving inhibitory also to MCF-7 invasiveness, it was reported that raloxifene did not match the efficacy of tamoxifen in vivo. Similarly, raloxifene was shown not to stimulate endometrial tumours in nude mice [28] and also inhibited such tumours which had been stimulated by tamoxifen. With regard to bone, Sato et al. [29] again using the OVX rat model, showed that raloxifene prevented bone loss to a degree compatible with a parallel OVX group treated with oestradiol 17β. In the human, raloxifene has now been subjected to extensive phase III trials.

Figure 2.

Raloxifene [6-hydroxy-2-(4-hydroxyphenyl)-benzo[b])thien-3-yl][4-[2-(1-piperidinyl)ethoxy]pheny]methanone;keoxifene:[LY-139481].

Mechanisms of drug action

The ultimate site of raloxifene action is the oestrogen receptor. The receptor (ER) is a member of the steroid-thyroid receptor superfamily and is most concisely described as a nuclear ligand-inducible transcription factor. The family includes the receptor for tri-iodothyronine, cortisol, progesterone, testosterone and 9-cis retinoic acid and exhibits a modular construction with domains coded A to F as shown in Figure 3.

Figure 3.

Schematic representation of the oestrogen receptor with six domains (A–F).

The arrival of ligand leads to the discard of heat-shock proteins which cloak the inactive receptor which then undergoes conformational change. The ER dimerizes and its new configuration permits access to the AF-2 domain of coactivators which shepherd the activated receptor to the promoter sequences of those genes which are oestrogen responsive. The AF-1 domain is believed to be constitutive and to operate independently of ligand. The process then proceeds to the activation of RNA transcriptase and message is produced. The above is a necessary simplification but elucidates the essential steps which may be disabled by raloxifene.

As Brzozowski et al. have graphically demonstrated using electron crystallography (Figure 4). Raloxifene enters the binding pocket of the ER but the activation process is inhibited due to the very molecular construction of this ligand [30]. Raloxifene has an alkylaminoethoxy side chain which physically protrudes from the binding pocket. It has been shown by Levenson & Jordan [31] that a nitrogen atom in the side chain binds to AA residue 351 (aspartate) in the ligand-binding domain of the ER. This binding splints the key helix 12 which is thus unable to undergo its normal wild-type rotation to seal the pocket and configure the AF-2 domain for the adherence of coactivators. The process thus cannot proceed to transcription and raloxifene would hence be more precisely designated as an AF-2 inhibitor. Indeed, it is clear that mutation of aspartate 351 to a tyrosine residue will convert raloxifene from an oestrogen antgonist to an agonist. Thus, at those tissues such as breast and endometrium where intact AF-2 activity is required for oestrogen activity, raloxifene acts as an antagonist. Conversely, the other activation site, AF-1, proceeds with transcription in the presence of raloxifene. This is believed to be the basis for the drug’s action on bone and in lipid physiology. It may thus be speculated that a raloxifene inhibitory action should be anticipated at any other sites where AF-2 integrity is required. In addition to the above specific actions of raloxifene a more general selectivity may apply, in that the drug may be preferentially taken up by certain tissues and not by others. Further studies are required to finally delineate the means whereby the action of raloxifene acquires its tissue specificity.

Figure 4.

Electron crystallographic map of the ligand binding domain of the oestrogen receptor with a) oestradiol and b) raloxifene shown in situ. From: Brzozowski AM, et al. Nature 1997; 389: 753–758: with permission.

Clinical trial data

As noted above, the central perturbation of bone physiology after menopause is an increase in BMU recruitment with the decoupling of bone formation and resorption in favour of the latter. Thus, useful data were presented by Heaney & Draper [32] who conducted calcium kinetic studies using the stable calcium isotope 45 Ca in postmenopausal women. These authors found that both raloxifene 60 mg day⁻¹ and a parallel group receiving oestrogen (as conjugated equine oestrogen 0.625 mg day⁻¹) exhibited a significant shift to baseline calcium balance which was apparent at 28 days and maintained at 217 days. The external cause of the positive balance shift was a reduction in urinary calcium which was itself tracked to a ≈15% reduction in bone resorption. Interestingly, neither raloxifene nor oestrogen affected calcium gut absorption and raloxifene, unlike oestrogen, did not reduce bone formation at the terminal analysis. The authors concluded that the basic activity of raloxifene was to suppress bone turnover principally by inhibition of osteoclastic bone resorption.

Several of the early trial reports of raloxifene reported on the use of dosages up to 400 mg. However, since the standard licence due has been set at 60 mg day⁻¹ this report will concentrate on studies utilizing this regimen.

The largest randomized controlled trial published to date was reported by Delmas and colleagues [33]. Some 601 healthy women were recruited who were between 2 and 8 years postmenopausal and whose bone mineral density (BMD) was above the WHO agreed threshold for osteoporosis which is set at 2.5 standard deviations (s.d.) below that of the young normal mean. In other words, none were osteoporotic. Raloxifene 60 mg day⁻¹ produced evidence of a fall in bone turnover with measurements of bone formation (osteocalcin) and resorption (Type I Collagen C-telopeptide) being significantly depressed from baseline by 3 months. This suppression of turnover was accompanied by significant percentage gains in means (s.d.) in BMD of 2.4 (+0.4) at spine, 2.4 (0.9) at total hip ad 2.0 (0.4) at total body. All these gains were significant when compared with placebo treated controls (all P<0.001). No fracture data were reported in this study.

Subsequently, Lufkin and colleagues [34] reported a 12-month randomized controlled trial in 143 osteoporotic postmenopausal women who manifested, at entry, at least one vertebral fracture. Again, markers of bone formation (osteocalcin and bone-specific alkaline phosphatase) fell, as did a number of bone resorption (urinary Collagen I C-telopeptide). No change was found in HDL-cholesterol or in triglycerides. There were significant gains in BMD at total hip (P<0.05) and at the ultradistal radius (P<0.002). These authors did not find significant gains over untreated controls in BMD at spine or total body and at 60 mg day⁻¹ there was no reduction in incident vertebral fractures using either a 15% or 30% reduction in vertebral height. Again, the treatment was well tolerated with no excess, over control, of breast or uterine side-effects. These data suggested a quasi-oestrogenic activity for raloxifene, though of a lower power than that expected with standard antiresorptive conventional oestrogen regimens. The lack of effect is even more striking when compared to the effects observed with full oestrogen implant therapy [35]. Interventional study data have been published [36], which indicate that the relative risk (95% CI) of new vertebral fracture in raloxifene treated osteoporotic women compared with placebo-treated controls was 0.70 (0.50–0.80).

Cardiovascular effects

With respect to the cardiovascular system, raloxifene 60 mg day⁻¹ resulted at 24 months in a significant mean percentage (s.d.) fall of 6.4 (1.1) in total cholesterol and a 10.1 (1.4) fall in LDL-cholesterol (both P<0.05) [33]. A slight reduction in plasma HDL-cholesterol failed to achieve significance and triglycerides were unchanged. No clinical end-point data have yet been published.

Endometrium

The effect of raloxifene on the histology of the human endometrium was examined in a study of healthy postmenopausal women in which the SERM, given in doses of 200 mg or 600 mg day⁻¹ for 8 weeks, was compared to control groups receiving standard oestrogen and placebo [37]. A meticulous scoring system was deployed to delineate oestrogenic effects on endometrial glandular cell shapes, nuclear/cytoplasmic ratio and mitotic rate. Similarly, stromal density and mitotic rate were recorded. The raloxifene treated groups, receiving up to 10 times the clinical daily dose, failed to manifest any oestrogenic effect on these histological sites of oestrogen activity. This was a short study but in the clinical trials the lack of vaginal bleeding at 2 years [33, 34] tend to suggest that the endometrium is truly insensitive to raloxifene and that patients may be advised that their risk of uterine bleeding is not above background. A study in premenopausal women found that raloxifene given for 28 days from day 3 of the cycle did not prevent ovulation and exerted little effect on cycle length or on endometrial histology. No clinical utility in this population is apparent and indeed the licensed indication restricts the use of raloxifene to the prevention of osteoporotic fracture in postmenopausal women. Clinical trials of raloxifene, however, will continue to acquire data on uterine behaviour, in particular on endometrial depth as measured by ultrasound and MRI. Any patient on prescribed raloxifene who reports postmenopausal bleeding should be fully investigated to exclude genital tract malignancy.

Breast

Since raloxifene is pharmacologically, though not chemically, related to the triphenylethylenes such as tamoxifen and since its initial observed action was an inhibition of mitosis in MCF-7 breast cancer cells, there has been considerable clinical interest in the incidence of breast cancer in treated patients. Symptomatically, patients on raloxifene did not report any excess of breast swelling or mastalgia when compared with controls [33]. At the American Society of Clinical Oncology Meeting in 1998, Cummings reported from a randomised placebo-controlled trial of 7705 osteoporotic postmenopausal women assigned in a ratio of 2:1 to raloxifene:placebo [38]. After a median 30 months of study, 11 breast cancers had occurred in raloxifene treated women and 21 in those receiving placebo. These data have now been fully published [38]. The relative risk (95% CI) in the raloxifene group was 0.24 (0.13–0.44); p<0.001. At the same meeting Jordan et al. presented data from a range of raloxifene:placebo RCTs encompassing some 14 800 woman years of follow-up. In this group of studies there had been 49 cases of breast cancer, 23 on raloxifene, 26 on placebo. The relative risk (95% CI) of breast cancer in raloxifene treated patients was 0.42 (0.25–0.73). There was a trend for the new cases to decline after 12 months on raloxifene possibly due to the elimination of individuals in whom subclinical breast cancer was present at randomization. Unpublished data indicate that protection against breast cancer is confined to ER+tumours with no apparent effect on the ER-variety. Caution is required in implementing the above results since the overall number of cancers is low but the occurrence of the expected number of tumours among the controls, and the significant reduction only in ER+tumours in treated groups, gives grounds for cautious optimism. It should be noted that an osteoporotic population carries no excess risk of breast cancer and a study is now recruiting in the US which will prospectively compare the preventative effects of raloxifene and tamoxifen in a large and higher risk population.

Safety

The drug was well tolerated and in particular there was no excess of mastalgia in the raloxifene 60 mg day⁻¹ group nor of vaginal bleeding when compared with placebo treated controls. It should be noted that breast and uterine related side-effects are the principal means for noncontinuation of conventional HRT therapy [18].

Side-effects with raloxifene have been generally few and transient and include hot flushes, leg cramps and peripheral oedema. In this context it is worthy of remark that raloxifene does not ameliorate the symptoms of the menopause. Treatment of such symptoms will continue to require the use of conventional HRT therapy. The most serious adverse event report to date is an excess of venous thromboembolic episodes (VTE) including pulmonary embolism of which the RR in treated patients is quoted [Summary of product characteristics, Eli Lilly & Co. 1998] as 2.49 (1.23–5.02). This is an attributable risk similar to that reported by Jick [39] and by Daley [40] with conventional HRT. The overall absolute risk of a VTE with raloxifene is of the order of 4–5/10 000 per annum. The favourable side-effect profile led overall to a ≈90% continuation rate at 3 years in raloxifene treated patients in the clinical trials.

Conclusion

As with the adrenoreceptor and the histamine receptor, eludication of complexity in structure and function of the oestrogen receptor is a gateway, not a barrier, to therapeutic opportunity. The SERM are at present are assuredly not HRT replacement therapy. Until they are capable of resolving menopausal symptoms and of precise mimicry of the action of oestradiol, this native oestrogen will continue to play a major role in postmenopausal management.

With the rapid elucidation of the isoforms of the ER and the realization that different cell types within the same tissue may respond to specific isoforms, it is possible to envisage a range of SERMs becoming available. These will specifically provide oestrogenic support for those tissues where it is required—in an individual patient—while retaining the common ability to disable the reproductive-site receptors of breast and uterus where postmenopausal quiescence has been for so long desired by patient and clinician alike.