Abstract
‘Studies in animals suggest that aromatase, an enzyme necessary to convert androgens to estrogen, could be a potential target for the inhibition of estrogen production.’
Ovulatory disorders (such as polycystic ovary syndrome [PCOS]) are commonly treated with clomiphene citrate (CC), with rates of ovulation ranging between 50 and 90%. However, even though the majority of pregnancies occur within the first three cycles of treatment [1], the overall pregnancy rates are disappointingly low, ranging from 20–40% [2–4], and the likelihood of miscarriage is relatively high [5].
This discrepancy between high rates of ovulation and low rates of pregnancy is believed to be a consequence of the antiestrogenic effect of CC on peripheral targets, such as the endometrium (affecting its thickness and maturation) and the endocervix (affecting the production and quality of cervical mucus). The problem is compounded by the long half-life of CC and the persistence of an isomer, zuclomiphene, both of which lead to an accumulating antiestrogenic effect with consecutive cycles of treatment [6]. In addition, due to the prolonged occupation of estrogen receptors by CC in the hypothalamus, the circulating amount of estrogen may rise to high levels from ongoing stimulation of the follicles by endogenous follicle-stimulating hormone (FSH) production. Such supraphysiological levels of estrogen may be deleterious to the developing oocyte, sperm and embryo [7]. Consequently, better methods for inducing ovulation are desirable.
Studies in animals suggest that aromatase, an enzyme necessary to convert androgens to estrogen, could be a potential target for the inhibition of estrogen production. Aromatase inhibitors (AIs) increased gonadotropin levels and ovarian weight in female rats [8], and have been found to produce multiple ovarian follicles in female primates [9]. The new generation of AIs (anastrozole, letrozole and vorozole) have more specific action, fewer side effects and lower toxicity in humans than their predecessors. These drugs are very potent and, at doses of 1–2.5 mg/day, are able to inhibit over 90% of total body aromatase action and, hence, estrogen production [7]. Their complete absorption after oral administration and rapid clearance, with a terminal half-life of only 45 h, are attractive properties of the drugs.
The fact that the ovary is the main source of estrogen production in premenopausal women suggests that aromatase inhibition in humans may be effective in inducing ovulation by releasing the hypothalamus from the negative feedback effect of estrogen. Furthermore, since AIs (in contrast to CC) do not bind to estrogen receptors, the central feedback mechanism remains intact, and the production of FSH can be regulated normally by the resulting production of estrogen. Consequently, supraphysiological levels of estrogen can be avoided and peripheral targets for estrogen action can respond appropriately and without any of the adverse effects noted with CC.
This theory was tested in small, observational studies in women with PCOS with a previous poor response to CC. Good rates of ovulation and pregnancy were observed in two small studies [10,11], but the rates in a larger study were less encouraging [12]. Subsequently, experimental evidence from a randomized trial comparing CC with letrozole in 60 women with CC-resistant PCOS, who had commenced treatment with metformin for 6–8 weeks prior to receiving the ovulation-inducing drugs, demonstrated a significantly higher ongoing pregnancy rate with letrozole (36.7 vs 10%; odds ratio [OR]: 4.82; 95% confidence interval [CI]: 1.04–29.88) [13]. Similar observations were documented in a randomized trial of 106 women with PCOS comparing CC with letrozole as first-line treatment, but without metformin [14].
In addition to the use of CC for ovulation induction, it is now well accepted that CC can be used for superovulation to release more oocytes in regularly ovulating women (with unexplained infertility) to enhance pregnancy rates. Using the same rationale, encouraging results were obtained with letrozole in a proof-of-principle study in women with unexplained infertility and suboptimal response to CC in previous cycles [10]. However, when CC was compared with letrozole in four randomized trials [15–18], comprising 264 women with unexplained infertility, the pooled OR for clinical pregnancy per woman was not significantly different (0.89; 95% CI: 0.46–1.65), even after excluding one trial [17] using a higher dose of letrozole (common OR: 1.08; 95% CI: 0.34–3.47). Although the numbers are not large enough for reliable inferences, these limited data fail to demonstrate a preference for CC or letrozole. As expected, estradiol levels were lower in the group receiving letrozole, but there was no difference in endometrial thickness. These findings are in contrast to those in women with PCOS.
Interestingly, a study performed in sexually mature female rats demonstrated similar results, with reduction in estradiol levels, but no difference in endometrial thickness [19], suggesting that the effect of CC on the endometrium in PCOS is different from that in unexplained infertility. In addition, the higher rate of ovulation with letrozole in women with PCOS may have contributed to the improved pregnancy rate observed in this group, an effect not observed in women with unexplained infertility in whom ovulation is not impaired. Despite the beneficial role for AIs in premenopausal women, there are several issues that require further discussion.
First, the optimal dose for AI use has not been identified. When CC was compared with letrozole in women with unexplained infertility, a higher dose of letrozole (7.5 mg/day) was associated with an apparent improvement in the effect of treatment compared with a lower dose (2.5 mg/day). Although it is possible that this effect is due simply to sampling error, a recent small, randomized trial in women with unexplained infertility, comparing two daily doses of letrozole (2.5 and 5 mg), suggests otherwise, as the number of follicles on the day of human chorionic gonadotropin (hCG) administration and clinical pregnancy rate were significantly higher with the higher dose [20]; in part, this latter difference can be explained by the somewhat lower pregnancy rate (5.9%) in the lower-dose group. Rates have ranged from 16 to 25% in other trials [15–18]. In addition, a Phase I study of anastrozole in premenopausal female volunteers demonstrated a higher mean number of follicles with higher doses of the drug [21]. Thus, given the fact that the doses used in infertility are derived from those given to postmenopausal women with breast cancer, the optimal dose of AI needs to be determined from dose–response studies.
Second, it is not clear whether AIs should be administered in multiple daily doses for 5 days (as is the case with CC) or in a single dose on day 3 of the menstrual cycle. A single dose of 20 mg should last approximately 5 days and be completely cleared from the body by day 13, a time when ovulation is expected to occur [22]. Limited data from a small, nonrandomized study in women with unexplained infertility, showing no significant differences in endometrial thickness, estradiol levels, numbers of follicles and clinical pregnancy between single- and multiple-dose regimens, support the single-dose regimen as a reasonable option if randomized trials are able to confirm these findings [22].
Third, AIs were registered for use in postmenopausal women with breast cancer. Their current use in premenopausal women with infertility represents ‘off-label’ use of the drug. Although such use of drugs is not unusual, concerns regarding letrozole in early pregnancy were raised recently by a Canadian study [23]. Among 150 babies born to women treated with letrozole for unexplained infertility or PCOS, there were seven cases (4.7%) of serious malformations, a rate much higher than the 1.8% observed in a control group of 36,050 births in women considered to have a ‘low-risk’ pregnancy (OR: 2.9; 95% CI: 1.4–5.9) [101]. Following this report, Novartis, the manufacturer of letrozole, reviewed its safety database and found two reports of children with birth defects among 13 pregnant women who had received the drug [102]. On November 17, 2005, after discussions with Health Canada (the healthcare agency for Canada), officials from Novartis Pharmaceuticals Canada Inc. issued a letter advising Canadian healthcare professionals that, “Femara® (letrozole) is contraindicated in women with premenopausal endocrine status, in pregnancy and/or lactation due to the potential for maternal and fetal toxicity and fetal malformations”. This letter was posted on the Health Canada website on November 28, 2005 by the Health Products and Food Branch, alerting all stakeholders of the concerns regarding such use of letrozole [103]. A spokeswoman from Novartis Pharma (Switzerland) indicated that letters were to be sent to fertility doctors worldwide to reiterate this warning [102]. Although the US FDA had not taken any action at that time, the matter was under review [102]. Interestingly, this issue surfaced much earlier (in 2003) in India, where the Drug Controller-General had commissioned an inquiry into media reports that a Mumbai-based pharmaceutical company had been promoting and marketing letrozole as a treatment for infertile women without the required regulatory approval [24].
The investigators of a recent observational study, undertaken to evaluate outcomes among infertile women rather than comparing outcomes in treated infertile women with those of women in the general population conceiving spontaneously, refuted the concern regarding congenital anomalies. No differences in the overall rates of major and minor congenital anomalies were observed [25]. However, it should be noted that these rates are generally higher than those expected in the general population, making it unclear whether the higher rates are due to the treatment or the infertility disorder.
In clinical medicine, it is not unusual to encounter examples of new uses for existing drugs. Evidence gathered, initially from observational studies and later from experimental studies, is sanctioned by the scientific peer-review process that permits its publication despite the lack of regulatory approval for the use of the drugs in this off-label manner. For example, CC is only indicated for the treatment of ovulatory dysfunction in women desiring pregnancy, yet its use in inducing superovulation in women with unexplained infertility (and who have no ovulatory dysfunction) is now an acceptable standard of care. Another example is the use of the gonadotropin-releasing hormone agonist, which was registered for the treatment of prostate cancer but is now routinely used in the ovarian stimulation regimen for assisted reproduction. We now face the same challenge with the use of AIs that have the potential to provide benefit in treating some aspects of infertility but are not registered for use in this manner.
What should be the strategy to address this issue? One option is to stop using the drug, as has happened in Canada in response to the safety advisory notice issued by the manufacturer. This option will not permit the realization of the full potential value of the drug. Another option is to convince the manufacturer to file an application with the regulatory authorities to change the label to include infertility treatment. This is a challenging goal, since many factors have to be considered in developing a business case for such an application, including the financial costs of gathering the required evidence to support the application, the return on the investment, the length of time for profits to be realized and the potential size of the market. A more realistic option is to identify conditions for the appropriate use of AIs in infertility. This approach would minimize risk and enhance benefit to patients, but requires gathering better-quality evidence on its efficacy and potential adverse effects. Furthermore, issues such as the optimal dose and duration of treatment, the use of a single-dose or multiple-doses, and which patients who will benefit from its use, have yet to be resolved.
‘Infertile women should be advised that pregnancies achieved with both clomiphene citrate and aromatase inhibitors are associated with higher rates of congenital anomaly compared with the general population’
So far, the limited evidence suggests that a single-dose regimen may be sufficient and treatment is likely to be more beneficial than CC only in women with PCOS. It is essential to rule out pregnancy before embarking on treatment so that inadvertent exposure during pregnancy to the drug can be avoided. In contrast to CC, with its longer elimination time from the body and likelihood of accumulation over successive cycles, AIs are more appealing for use in ovulation induction since they are more rapidly cleared, and the single-dose regimen is even more likely to ensure that no residual drug is left in the body at the time when the embryo is beginning to implant and establish its circulatory system. Interestingly, CC is well accepted for use in infertility even though it is labeled as a Category-X drug in terms of risk to pregnancy, whereas the AIs, letrozole and anastrozole, are labeled in a relatively safer rank as Category-D drugs. Infertile women should be advised that pregnancies achieved with both CC and AIs are associated with higher rates of congenital anomaly compared with the general population.
It is time for the scientific community to address these issues in a systematic and methodologically sound manner so that a potentially effective treatment is not excluded form our repertoire of choices for infertility management. The collective medical experience with the responsible and safe use of CC can be easily applied to the use of AIs.