Human Reproduction Update Advance Access originally published online on November 29, 2004
Human Reproduction Update 2005 11(1):69-89; doi:10.1093/humupd/dmh052
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Fertility in female cancer survivors: pathophysiology, preservation and the role of ovarian reserve testing
Department of Obstetrics and Gynaecology, Reproductive Medicine Unit, Elizabeth Garrett Anderson and Obstetric Hospital, University College London Hospital, Huntley Street, London WC1E 6DH, UK
1 To whom correspondence should be addressed. Email: k.lutchman-singh{at}ucl.ac.uk
 |
Abstract |
|---|
TOP
Abstract
IntroductionThe improved long-term survival of adolescents and young women treated for cancer has resulted in an increased focus on the effects of chemotherapy on ovarian function and its preservation. These women may seek advice and treatment regarding their reproductive status, including ways of preserving their fertility and preventing a premature menopause—factors that can have a profound impact on their quality of life. This article comprehensively reviews ovarian reserve testing (ORT) in general. Special emphasis is placed on patients with cancer, including the pathophysiology of gonadal damage following chemotherapy, fertility preservation and the potential role of ORT. Baseline parameters of ovarian reserve [FSH LH, estradiol, inhibin B and anti-Müllerian hormone (AMH)] have not yet performed sufficiently well in predicting poor outcome in assisted reproduction, but biochemical markers of ovarian reserve appear to be better than chronological age. Inhibin B and AMH show potential for future use. Dynamic testing appears to show much promise, especially stimulated levels of inhibin B and estradiol. The most promising tests of ovarian reserve are the biophysical markers, where total antral follicle count was found to be most discriminatory followed by ovarian volume. Combination of biochemical, biophysical and clinical markers of ovarian reserve may also improve predictive capacity. However, there is a lack of data pertinent to ORT in cancer. As yet there is no single clinically useful test to predict ovarian reserve accurately. Patients with cancer represent a distinct cohort who have particular concerns about their future fertility and the possibility of a premature menopause, they can benefit greatly from knowledge of their functional ovarian reserve. Large, prospective, randomized, adequately controlled studies specific to different geographical areas are required in a control population of comparable reproductive age to determine the potential role of ORT in clinical practice.
Key words: cancer / chemotherapy / fertility / ovarian reserve tests / premature menopause
|
Introduction |
|---|
A review of cancer trends in England and Wales from 1950 to 1999 reveals an overall increase in relative survival for adult cancers (Office for National Statistics, 2004
). This includes breast cancer, the most common cancer affecting women in developed countries (Figure 1). In these patients, the decline in age-adjusted breast cancer mortality rates is almost certainly due to improvements in the earlier detection of smaller, palpable tumours and in adjuvant chemotherapy (Mettlin, 1999; Jatoi and Miller, 2003). It has been estimated that 25% of women diagnosed with breast cancer are pre-menopausal, with 7% being diagnosed before the age of 40 years (Hankey et al., 1994). Another estimate is that 10–20% of newly diagnosed breast cancers occur in women of childbearing age (Parker et al., 1997).
|
The remarkable success in improving childhood cancer survival is exemplified by the 5 year survival rate for all leukaemias approaching 80% (Office for National Statistics, 2004
). As such, it has been estimated that at the start of the 21st century, one in 1000 young adults in their third decade is a survivor of childhood cancer (Bath et al., 2002). Overall, much of the improvement in 5 year survival rates in adult and childhood cancer is attributed to improved treatment, of which chemotherapy plays a major role (McVie, 1999).
This success, however, is tempered by the fact that chemotherapy can result in ovarian toxicity. For these patients, who originally had expectations of a normal reproductive lifespan, the realization that they might suffer reduced fertility and a premature menopause can have a profound impact on their self-esteem and quality of life (Meirow and Nugent, 2001). Furthermore, women are delaying childbearing for domestic, professional and educational reasons (Friedlander and Thewes, 2003). In England and Wales the average age of mothers at childbirth has increased by 3 years since 1971, rising from 26.2 years to 29.1 years in 2000 (Office for National Statistics, 2004).
Ovarian reserve testing has become established in the reproductive medicine setting, where it is utilized as a counselling tool and a guide to treatment respectively for subfertility. Ovarian ageing is a concept intimately related to ovarian reserve, as it could potentially predict age at menopause, a measure which is known to be highly variable (Treloar, 1981). Biochemical markers also show considerable variability, with FSH levels not increasing permanently until 1–5 years after the onset of the peri-menopause (Burger et al., 1999). Inhibins and activins have also been explored (Santoro et al., 1999), and it has been suggested that there is a subtle decrease in inhibin B prior to the rise of FSH (Soules et al., 1998). Ultrasound determination of ovarian volume and antral follicle count (AFC) have also been utilized with some success (Flaws et al., 2001).
Several hypotheses have been forwarded in an attempt to describe the process of ovarian ageing (Lobo, 2003; Nikolaou and Templeton, 2003) but the fact remains that reproductive ageing itself is highly variable (te Velde and Pearson, 2002). The basic doctrine which has prevailed to explain the concept of ovarian reserve and reproductive ageing assumes that the age-dependent loss of fertility in females is dictated by a continual process of follicle depletion (which is fixed from fetal life), leading to a reduction in both oocyte quantity and quality. Even this central doctrine has recently been challenged by the reported finding of the existence of proliferative germ cells that sustain oocyte and follicle production using a mouse model (Johnson et al., 2004b).
At present therefore, it is impossible to predict the functional life span of the chemotherapeutically damaged ovary and the reproductive potential of patients with cancer. There is a need for adequate assessment of functional ovarian reserve in pre-menopausal patients with cancer, as it may provide useful information regarding their fertility status and the prediction of premature menopause. The potential also exists for these tests to act as a guide to treatment with respect to fertility preservation, where an individualized approach would be ideal.
This article comprehensively reviews ovarian reserve testing (ORT) in general. Special emphasis is placed on patients with cancer, including the pathophysiology of gonadal damage following chemotherapy, fertility preservation and the potential role of ORT.
|
Pathophysiology of ovarian damage following chemotherapy |
|---|
The gonadotoxic effect of chemotherapeutic agents is well documented, although the prevailing mechanisms are not fully understood. In general, alkylating agents such as cyclophosphamide (C) which are non-cell cycle specific, are more cytotoxic to the ovaries than cell cycle-specific agents such as methotrexate (M) and fluorouracil (F), whose major effect is on ovarian follicle growth and maturation (Hensley and Reichman, 1998
). Ovarian biopsies in patients undergoing cyclophosphamide-based treatment reveal complete absence of ova or small numbers of inactive ova with fibrosis and no evidence of follicular maturation (Warne et al., 1973; Koyama et al., 1977). Animal studies have shown that exposure to cyclophosphamide causes follicular destruction in exponential proportion to increasing doses (Meirow et al., 1999). These events may affect sex steroid production, leading to disturbance of the hypothalamic–pituitary–ovarian (HPO) axis (Dnistrian et al., 1985; Dowsett and Richner, 1991; Mehta et al., 1992).
Ovarian damage following chemotherapy can present with a variety of symptoms that reflect varying degrees of damage, culminating in premature ovarian failure (POF). POF is associated with a wide array of effects including vasomotor symptoms (hot flushes and night sweats), genitourinary symptoms (vaginitis, dyspareunia, dysuria) and osteoporosis, which can lead to skeletal fractures (Ganz and Greendale, 2001). Ovarian failure after adjuvant chemotherapy has been shown to be associated with rapid bone loss in women with early stage breast cancer (Shapiro et al., 2001). The management of these symptoms is particularly difficult (especially in patients with breast cancer) owing to concerns about HRT (Rostom, 2001; Beral, 2003).
Chemotherapy-related amenorrhoea (CRA) is a term used to describe the occurrence of amenorrhoea following chemotherapy, the rate of which varies according to the diagnostic criteria used and length of follow-up (Bines et al., 1996). Its onset is mediated through ovarian failure, and is based on the observation that the hormone profile observed in pre-menopausal women treated with adjuvant chemotherapy for breast cancer who develop CRA is consistent with primary ovarian failure (Rose and Davis, 1977; Dnistrian et al., 1983; Dowsett and Richner, 1991). The risk of POF with polyagent adjuvant chemotherapy has been reported to range from 53 to 89% (Del Mastro et al., 1997). Generally, the risk of POF is related to the patient's age, treatment protocol and type of malignancy (Meirow and Nugent, 2001).
Age and time to CRA are inversely related, as well as cumulative dose required to produce CRA. The average incidence of CRA for CMF-based regimens is 40% for women aged <40 years and 76% for those >40 years (Bines et al., 1996). A cohort of pre-menopausal women with breast cancer, receiving either adjuvant CMF, CEF (E = epirubicin), tamoxifen, or no treatment was followed up prospectively for 1 year (Goodwin et al., 1999). Age and the use of systemic chemotherapy were found to be independent predictors of premature menopause. The use of CMF or CEF, whether in combination or not with tamoxifen, increased the risk of menopause in women aged 40 years from <5 to >40% (Goodwin et al., 1999). This potentiating effect of age on chemotherapy-induced gonadal damage has been reported elsewhere (Sanders et al., 1988; Moore, 2000).
Restoration of menstruation after CRA is possible. Again this is influenced by age and duration of follow up, and has been estimated at 39–55% in younger women (<40 years) and 0–11% in older patients (>40 years) (Bines et al., 1996). However, women who maintain normal menses throughout chemotherapy remain at risk for developing POF. This is evident from the high rates of POF seen in adolescents receiving alkylating agents for cancer (Byrne et al., 1992).
Chatterjee and Kottaridis (2002) attempted to define this sequence of events by using a haematological model termed the gonadal insufficiency–premature gonadal failure syndrome (GI–PGF). As the name implies, the syndrome describes a heterogeneous picture whereby the gonadotoxic effect of these agents is not ‘all or nothing’ but can be both acute and cumulative, with the ovaries possessing a limited capacity for recovery. These authors made an attempt to grade the extent of damage using clinical, biochemical and biophysical parameters; however, it was noted that the clinical picture was variable and as such did not always comply with these parameters. The potential for ovarian recovery, as well as the response to therapy, was unpredictable (Chatterjee and Kottaridis, 2002).
While this model may not be directly applicable to all cancers, it remains an attractive one, which highlights the importance of developing tests that can accurately estimate ovarian damage, and correlate it with clinical features.
|
Effect of chemotherapy on fertility |
|---|
Young cancer survivors are concerned about their reproductive capability (both before and after treatment), which can have a profound impact on their self-esteem (Ganz, 2000
). Those patients who are not immediately rendered infertile are still likely to suffer problems with infertility and, ultimately, a premature menopause (Meirow, 2000). Approximately 50% of patients aged <35 years resume normal menses after completion of cytotoxic chemotherapy (Forbes, 1992), and are therefore potentially capable of becoming pregnant. However, although a regular menstrual cycle may serve as a convenient marker for ovarian function, a normal menstrual cycle is not synonymous with fertility. Similarly, irregular menses or amenorrhoea does not always imply infertility. This was shown in a retrospective series of women with breast cancer who received chemotherapy with FAC (5-fluoracil, anthracycline, cyclophosphamide) who were 
35 years at the time of treatment. In this series, 33 pregnancies occurred in 25 patients (21%), of whom only 64% continued to menstruate regularly during and after chemotherapy (Sutton et al., 1990).
Because of the possibility of spontaneous recovery, it has been suggested that ovarian function be reassessed periodically in patients with chemotherapy-induced gonadal damage. Conversely, if fertility is not desired, contraception should be used, and the combined oral contraceptive pill (COCP) can be used for HRT (Nasir et al., 1997).
A large retrospective survey of pregnancy outcomes after peripheral blood or bone marrow transplantation revealed that only 0.6% of patients conceived after autologous or allogenic stem cell transplantation (SCT), but the pregnancies were likely to have a successful outcome (Salooja et al., 2001).
Overall, the data available relating to pregnancy rates and outcomes following cancer is limited and relates only to distinct cohorts, which may not be representative of the entire population. Furthermore, follow-up is almost always limited in duration.
|
Options for preserving fertility |
|---|
Advances in reproductive technology have made fertility preservation techniques a real possibility for these patients. Decision-making in this area is particularly difficult because of the experimental nature of many techniques. Patients may benefit from knowledge of their functional ovarian reserve prior to embarking on these decisions.
Ovarian protection
This has been attempted by the administration of prior and concomitant treatment with GnRH analogues (Blumenfeld et al., 1999). ZoladexTM has been compared with CMF where it was found that goserelin offers an effective, well-tolerated alternative to CMF in pre-menopausal patients with estrogen receptor (ER)-positive and node-positive early breast cancer (Jonat et al., 2002). Concerns exist regarding the side-effects of goserelin, but are considered to be reversible (Nystedt et al., 2003). Despite encouraging results in animal models, few clinical studies have evaluated the effect of GnRH analogue co-treatment in preventing chemotherapy induced POF in cancer patients (Blumenfeld, 2003). The questionable presence of GnRH receptors in human gonadal tissue also challenges the plausible mechanism of action (Clayton and Huhtaniemi, 1982). A randomized prospective multicentre study is currently underway in the UK attempting to address this issue (Ovarian Protection Trial In Oestrogen Non-responsive Premenopausal Breast Cancer Patients Receiving Adjuvant or Neo-adjuvant Chemotherapy: ‘OPTION’; Scottish Cancer Network).
Embryo cryopreservation
The only established method available in clinical practice to preserve fertility in women being treated for cancer is the cryopreservation of embryos prior to chemotherapy (Revel and Schenker, 2004). There are significant drawbacks to its use, namely the need for a stimulated cycle and IVF. The procedure itself may be theoretically harmful to patients with hormone-sensitive tumours such as breast cancer. This risk has been circumvented in the past by employing natural cycle IVF (Brown et al., 1996). More recently, IVF and embryo cryopreservation after ovarian stimulation with tamoxifen has been employed with some success whilst possibly providing a safe alternative to ovarian stimulation in these patients (Oktay et al., 2003). Further considerations include the fact that a harvest of viable embryos cannot be guaranteed, and any complications or delay in the IVF cycle may delay the commencement of chemotherapy (Thomson et al., 2002). The fate of the embryos should be considered at the outset, given the fact that potential offspring may lose their mothers to the disease (Posada et al., 2001).
Oocyte cryopreservation
This is an attractive option for women without a partner. Drawbacks are analogous to embryo cryopreservation as a stimulated cycle is required. Since the first report of a birth from a frozen oocyte (Chen, 1986), the results of this procedure worldwide have been variable with a reported success rate of <2%, despite the improved success rate when combined with ICSI (Porcu et al., 1997; Fabbri et al., 2001). Despite this, the technique is still considered to have an important place in reproductive medicine, especially in patients with cancer (Van der Elst, 2003).
Ovarian tissue cryopreservation
This has been a topic of renewed and intense interest as a viable alternative to restore both fertility and ovarian function to women with cancer (Falcone et al., 2004). Earlier studies provided a basis for the technique by showing that pregnancies and live births could be achieved in animals by reimplanting cryopreserved–thawed ovarian cortical strips in sheep (Gosden et al., 1994; Baird et al., 1999).
Whole ovary transplantation has been attempted with successful pregnancy following transplantation of frozen–thawed rat ovaries (Wang et al., 2002). An attempt to do the same in humans has recently been reported (Bedaiwy and Falcone, 2004).
A significant step was achieved when the first live birth after transplantation of ovarian tissue in non-human primates was reported (Lee et al., 2004). Progress in humans has also finally become possible after several years of efforts. (Oktay et al. 2004) recently described fertilization and formation of a 4-cell embryo using oocytes retrieved from reimplanted ovarian tissue from a woman aged 30 years with breast cancer. In another major breakthrough, Donnez et al. (2004) recently announced resumption of ovarian function followed by a live birth in a woman with Hodgkin's lymphoma, who underwent orthotopic autotransplantation of cryopreserved ovarian tissue almost 6 years after removal.
There are important concerns to consider regarding the technique. First, the tissue is obtained from cortical slices of the ovary obtained at laparoscopy, which exposes the patient to a surgical procedure requiring anaesthesia. Following storage, the options are regrafting the frozen–thawed ovarian cortex as an orthotopic (Radford et al., 2001) or heterotopic autograft (Oktay and Karlikaya, 2000). Concern about the possibility of grafting tissue which can potentially harbour malignant cells has been raised using a mouse lymphoma model (Shaw et al., 1996).
The advances in this field, which have only recently been announced, have created a new sense of enthusiasm and optimism for a technique that is still completely experimental.
In vitro oocyte maturation
This is considered an attractive goal as it would eliminate any risk of reimplanting residual cancer cells and could in theory produce more mature oocytes by avoiding follicle wastage created by ischaemia or normal atresia. Although not widely practised, pregnancy rates after IVM (using immature antral follicles) has been >30% in some centres, with an estimated 300 healthy infants born worldwide (Chian et al., 2004). Primordial follicle culture however, has proven to be particularly more challenging and is the subject of intense research interest (Gosden et al., 2002).
Although the advantages of this procedure in patients with cancer is readily apparent, there are immense ethical, moral and sometimes legal dilemmas associated with these procedures and as such should not be ignored (Robertson, 2000). With this in mind, continuing research in the field has created optimism for clinical use in the future (Chian et al., 2004). Until then, it should not be considered a viable alternative for patients who want fertility preservation if more established alternatives are an option (Revel and Schenker, 2004).
Summary of fertility preservation techniques
Fertility preservation techniques have evolved to the extent that there are now several options available to the patient with cancer. It is important to remember that despite several encouraging advances made in this area, the best option for patients who are suitable remains embryo cryopreservation. With the continual development of these techniques, however, the ideal scenario may develop where it might be possible to offer an individualized approach to management (Sonmezer and Oktay, 2004).
|
What is ovarian reserve? |
|---|
Fertility potential in the female patient is related to the total number and quality of the primordial follicles remaining in the ovaries and is referred to as ovarian reserve. The function of these follicles is 2-fold: gametogenesis, which governs fertility potential, and steroidogenesis, which governs the onset of a premature menopause.
|
Development of ovarian reserve tests (ORT) |
|---|
ORT have become established in the fertility clinic setting where the association of poor ovarian response due to diminished reserve, leads to cycle cancellations and reduced success rates in IVF (Pellicer et al., 1987
). Although natural fecundity and pregnancy rates decrease with increasing age (te Velde and Pearson, 2002), the need for ORT was clearly realized when it was established that chronological age and menstrual characteristics were unreliable in predicting reproductive age (Scott et al., 1995). Several models have now been described which accurately chart the progressive decline of primordial follicles in the ovary with a woman's age (Richardson et al., 1987; Faddy et al., 1992; Gougeon et al., 1994; Faddy and Gosden, 1996).
The most important aspect of diminished ovarian reserve and the associated decline in reproductive potential is that its onset is highly variable (Scott and Hofmann, 1995). Furthermore, the presence of regular menses does not establish the presence of adequate ovarian reserve, as the ovary has the ability to maintain a high frequency and number of ovulations despite continuously declining follicle number (Gosden, 1987). Hence the continual decline in follicle number and quality can best be described as a dynamic process, the mechanisms of which are not yet fully understood (te Velde, 1993).
Overall, most of the data pertaining to ORT are limited by a lack of prospective, controlled data outside a reproductive medicine setting. Therefore, until appropriate validation occurs, there is as yet no clinically useful predictive test to accurately assess ovarian reserve.
|
Ovarian reserve tests |
|---|
In general, ORT are either biochemical or biophysical (Table I). Depending on the basis of the test being used, this may be further classified as being direct or indirect, or alternatively as being static or dynamic in nature.
|
|
Biochemical markers |
|---|
FSH
Basal serum FSH (bFSH) is considered an indirect estimate of ovarian reserve and depends on the presence of an intact HPO axis. It is one of the longest established parameters for estimating ovarian reserve.
The first reports on the usefulness of bFSH measurements showed FSH to be predictive of ovarian response and pregnancy outcome in IVF cycles (Muasher et al., 1988; Scott et al., 1989), and to be more predictive than the patient's age (Toner et al., 1991). Subsequent studies have supported this, including a large retrospective analysis of IVF patients where elevated FSH was associated with high pregnancy loss and poor live birth rates, regardless of age (Levi et al., 2001).
However, there are now several studies that challenge the predictive capabilities of bFSH, especially when a patient's age is considered. For example, in patients undergoing their first IVF cycle, bFSH concentration was a better predictor of cancellation rate than age, yet age was a stronger predictor of pregnancy rate (Sharif et al., 1998; Creus et al., 2000). Some studies suggest that young age exerts a protective effect over the deleterious effects of reduced ovarian reserve (Hanoch et al., 1998), while others reveal no difference in outcome between younger and older women with elevated bFSH (El Toukhy et al., 2002). A retrospective cohort study of IVF patients revealed that increasing age, but not bFSH, was associated significantly with reduced implantation and pregnancy rates (Chuang et al., 2003). In a recent prospective study, regularly cycling women aged >40 and <40 years, with elevated bFSH levels undergoing their first IVF cycle, were compared for ovarian response, ongoing pregnancy rates and implantation rates. Although younger women with elevated bFSH levels had higher cycle cancellation rates, implantation and ongoing pregnancy was superior to their older counterparts (van Rooij et al., 2003). This has prompted the distinction that when a comparison is made, age better reflects oocyte quality whereas FSH reflects oocyte quantity (Toner, 2003). This study has been criticized, however, for a lack of adequate controlling and statistical power (McDonough, 2003).
Threshold levels for bFSH are also an issue. In one study, which is unique for its duration of follow-up, ongoing pregnancy was seen in as much as 28% of regularly cycling women with FSH levels of 15–20 IU/l. Only when the FSH level was >20 IU/l was there a clear fall in ongoing pregnancy rate regardless of age (van Rooij et al., 2004). A high threshold level of FSH to achieve acceptable prediction for treatment failure has been reported elsewhere (Bancsi et al., 2000). It seems therefore that younger women with moderate elevations of FSH should not be ignored (Toner, 2004). Younger women (age <41 years) with elevated bFSH levels who have a poor first response to IVF, however, may represent a specific group which should be counselled against further attempts, as was shown in a recent retrospective study (Klinkert et al., 2004). Which threshold level of bFSH to use is the question, as some investigators have shown that moderately elevated levels of FSH (defined as 10–11.4 IU/l) are difficult to interpret as these may be confounded by a poor response to gonadotrophin stimulation (Esposito et al., 2002).
As most of the data regarding the performance of bFSH in predicting ovarian reserve comes from studies in an IVF population, it is significant that a recent meta-analysis concluded that the performance of bFSH for predicting poor ovarian response in this group was moderate, while prediction of non-pregnancy was poor (Bancsi et al., 2003). Information pertaining to the general subfertile population is lacking, but a nested case–control study revealed no statistically significant difference in cumulative pregnancy rates between patients with elevated bFSH levels and controls (Van Montfrans et al., 2000). Others have shown elevated bFSH to be of limited value in patients with regular cycles, in that it should not lead to exclusion of treatment (van Rooij et al., 2004).
Questions remain regarding the reproducibility of bFSH measurements (Sharara et al., 1998). One of its main limitations is the significant intercycle variability of FSH, hence limiting the usefulness of a normal value (Scott et al., 1990). A recent randomized prospective study revealed that women with limited ovarian reserve exhibited strong intercycle variability of bFSH and FSH response to clomiphene citrate (Kwee et al., 2004). A study performed on normally ovulating women aged <35 years revealed a large inter-individual variation in bFSH (Schipper et al., 1998). Intra-cycle variability seems to be less of an issue, with some flexibility being apparent in sampling from day 2 to day 5 of a regular menstrual cycle (Hansen et al., 1996; Klein et al., 1996a).
Another concern is the establishment of normative data within individual laboratories due to the likelihood of interassay variability (Hershlag et al., 1992). The establishment of threshold values within individual assay systems is also important to prevent errors in interpretation of results, especially when clinical outcomes are being correlated. This is more likely to happen in institutions where the clinical volume of patients is insufficient to allow these threshold values to be established (Scott, 2004).
Other reasons for an elevated bFSH should be considered (Lambalk, 2003). Laboratory errors can occur in the presence of heterophilic antibodies, which can interfere with the FSH immunoassay (Cahill and Thomas, 1992; de Koning et al., 2000a). Increased levels and pulsatility of FSH in the follicular phase are seen in mothers of hereditary dizygotic twins (Lambalk et al., 1998). FSH receptor variants have now been identified (Perez et al., 2000; Sudo et al., 2002). This may cause slightly diminished receptor function, resulting in higher levels of FSH to achieve adequate receptor response whilst not necessarily affecting ovarian reserve (Bukulmez and Arici, 2004).
The sensitivity of bFSH in identifying women who will not become pregnant with IVF has been calculated to be only 8%, whereas it is considered to be a specific test, in that 98% of women who achieved pregnancy had normal findings (Barnhart and Osheroff, 1998). The positive predictive value (the probability of not becoming pregnant given a positive test result) of the test is difficult to interpret, especially in younger patients or the general subfertile population where the prevalence of non-pregnancy is lower. As such there is a danger of overinterpreting its predictive value (Barnhart and Osheroff, 1999).
The reasons for these conflicting reports are not entirely clear, but are certainly related to the varying methodologies employed by different investigators, especially with regard to different threshold values for FSH, study groups, outcome measures and their definitions, duration of follow-up and data analysis (Table II).
|
Efforts to increase the sensitivity of prediction have been attempted by combining FSH with LH as a ratio of FSH:LH, with limited success (Mukherjee et al., 1996
; Kim et al., 1997). Despite these limitations, more studies have been performed, and hence much more is known, about the predictive value of FSH than any other marker of ovarian reserve. This, combined with the relative practicality, patient tolerability and low cost of performing the test, is likely to result in bFSH remaining one of the most commonly performed tests of ovarian reserve for some time to come.
Estradiol
The condensed follicular phase length in older women may be as a result of a more advanced follicular recruitment by cycle day 3. This early dominant follicle selection is expressed by high serum estradiol (E2) concentrations (Licciardi et al., 1995).
It has been shown in an assisted reproduction treatment population (where GnRH analogues were not administered) that increasing day 3 estradiol concentrations are associated with decreasing oocyte numbers and pregnancy rates (Licciardi et al., 1995), a correlation which has been repeated elsewhere (Smotrich et al., 1995). In patients with normal FSH levels, basal estradiol (bE2) has been shown to predict high cancellation rates and low oocyte yield in IVF (Evers et al., 1998). In another study, cancellation rates did correlate with bE2 levels but did not correlate with pregnancy outcome in those patients who were not cancelled (Frattarelli et al., 2000). Pregnancy rates have also been shown to be higher in a group of women undergoing in vitro maturation (Mikkelsen et al., 2001).
The predictive ability of bE2 is improved in patients of advanced reproductive age, especially when combined with bFSH (Buyalos et al., 1997). However, these observations have not been confirmed by others (Lee et al., 1988; Scott et al., 1989) (Table III).
|
No data are currently available regarding basal estradiol levels in the general subfertile population (Bukman and Heineman, 2001
). Further studies including data on day 3 estradiol values and fecundity in spontaneous cycles are required before evaluating this parameter further. Inhibin B
Inhibins and activins are glycoproteins that belong to the transforming growth factor ß (TGFß) family. Activins are dimers of ß subunits and act as functional antagonists of inhibin to stimulate pituitary FSH synthesis and secretion (Muttukrishna and Knight, 1991). Inhibins are heterodimers consisting of two dissimilar subunits (
and ß) linked by disulphide bridges, and are secreted by the ovarian granulosa and luteal cells during the menstrual cycle (Lockwood et al., 1998).
Inhibins are a part of the HPO axis, specifically having an inhibitory effect on pituitary FSH synthesis and secretion (Muttukrishna and Knight, 1990) (Figure 2). Serum dimeric inhibin B is regarded as a direct measure of ovarian reserve as it is mainly secreted by pre-antral follicles (Klein et al., 1996b), whereas inhibin A is produced primarily during the late follicular phase by the mature follicle and by the corpus luteum (Roberts et al., 1993) (Figure 3).
|
|
Inhibin B rises from early in the follicular phase to reach a peak coincident with the onset of the mid-follicular phase decline in FSH concentrations, it then declines during the luteal phase apart from a peri-ovular peak, which may represent release of follicular inhibin B from the rupturing follicle into the circulation (Muttukrishna et al., 1994
; Groome et al., 1996). Inhibin B has also been shown to respond to exogenous FSH administration in the follicular phase (Burger et al., 1998).
The close temporal relationship between changes in levels of inhibin B and FSH in the mid-follicular phase suggest that the release of inhibin B by the pre-ovulatory follicle critically regulates pituitary FSH secretion, with inhibin B exhibiting a distinct periodicity in normal women (Lockwood et al., 1998). Studies in older women reveal significantly lower levels of inhibin B in women aged 40–50 years with raised FSH compared to women with normal FSH (Klein et al., 1996b; Muttukrishna et al., 2000). Additionally, concentrations of inhibin A and B have been shown to be lower in women with elevated FSH and regular menstrual cycles (de Koning et al., 2000b).
Studies in an assisted reproduction population demonstrated that women with a low cycle day 3 inhibin B concentration (<45 g/ml) had a poorer response to ovulation induction and decreased likelihood of achieving pregnancy compared with women who had higher day 3 inhibin B levels (Seifer et al., 1997). Decreased inhibin B was also found in women with normal FSH levels implying that decreased inhibin B precedes a rise in FSH (Seifer et al., 1999). Doubt has been expressed, however, regarding the quality of the inhibin B assays used in these earlier studies (Bancsi et al., 1997).
Correlation with other markers of ovarian reserve has been shown (Tinkanen et al., 2001). There is evidence that early follicular inhibin B levels correlate with follicle cohort size (Elting et al., 2001) and oocytes retrieved following controlled ovarian stimulation with FSH (Eldar-Geva et al., 2000, 2002), and with an improved sensitivity and specificity compared to other basal markers of ovarian reserve (Ficicioglu et al., 2003). Another study which also showed good correlation with oocytes retrieved did not show a correlation with pregnancy (Fried et al., 2003). Inhibin B levels on cycle day 5 have been shown to be predictive as an early indicator of response during ovarian stimulation as well as outcome (Penarrubia et al., 2000; Fawzy et al., 2002).
The predictive ability of basal inhibin B with regard to pregnancy has not been confirmed by several authors, leading to doubt regarding the clinical usefulness of day 3 inhibin levels in clinical practice (Corson et al., 1999; Hall et al., 1999; Creus et al., 2000; Dumesic et al., 2001). Studies regarding basal inhibin A have also shown a link with ovarian function and follicular development, but limited value in predicting IVF outcome (Hall et al., 1999; Casper et al., 2001) (Table IV).
|
Inhibin B has been evaluated in predicting response as a component of dynamic testing (Ficicioglu et al., 2003
). These include the clomiphene citrate challenge test (CCCT) (Hofmann et al., 1998), GnRH agonist test (GAST) (Ravhon et al., 2000) and the exogenous FSH stimulation test (EFORT) (Dzik et al., 2000; Eldar-Geva et al., 2002). These studies are all limited by the fact that although they show correlation with ovarian response, predictive value was not assessed (Welt, 2002). More recently, however, studies suggest that the incremental increase in inhibin B following ovarian stimulation may be more predictive of assisted reproduction treatment outcome than basal levels of inhibin B. In a prospective study, the inhibin B response to exogenous FSH correlated well with the antral follicle count (AFC) and the number of oocytes received (Yong et al., 2003). Kwee et al. (2003) undertook a prospective randomized study, which compared endocrine markers of ovarian reserve using the CCCT and EFORT to predict ovarian response, and revealed a high correlation with inhibin B in EFORT for number of follicles after ovarian stimulation, with increments in estradiol and inhibin B providing prediction of ovarian capacity.
The reasons for these conflicting reports are not clear and may reflect inter-assay variability and differing reference points and standards between different institutions, resulting in reduced specificity of the inhibin assay. Given the expense of performing the test, further prospective evaluation and confirmation of its predictive capacity is required before inhibin B can assume a routine place in the clinical assessment of women with reduced ovarian reserve. Until then, its measurement in the field of reproductive medicine should only be as part of a research protocol (Tong et al., 2003).
Anti-Müllerian hormone (AMH)
In females, the granulosa cells of the ovary produce AMH, which is a member of the transforming growth factor ß (TGFß) family. Levels can be almost undetectable at birth (Rajpert-De Meyts et al., 1999), with a subtle increase noted after puberty (Hudson et al., 1990). Serum levels on day 3 of the menstrual cycle show a progressive decrease with age, which correlates with antral follicle counts (de Vet et al., 2002).
The exact physiological basis of AMH is poorly understood in humans. The main physiological effect is the regression of Müllerian ducts in male fetuses (Josso et al., 1998). Based mainly on in vitro work and experiments performed in rats, AMH seems to play a pivotal role with regard to follicle recruitment, in that it inhibits the recruitment of primordial follicles into the pool of growing follicles and also decreases the responsiveness of growing follicles to FSH (Durlinger et al., 2002). There is evidence that AMH is mainly expressed in pre-antral and early antral follicles (Baarends et al., 1995). AMH levels are also seen to decline gradually during multiple follicular maturation as part of controlled ovarian stimulation (COS). This may be accounted for by the increased number of mature follicles expressing less AMH (Fanchin et al., 2003a). These findings suggest that AMH is solely produced by antral follicles capable of growing, and as such serum levels of AMH may represent both the quantity and quality of the ovarian follicle pool (te Velde and Pearson, 2002).
Recent studies have shown an association between reduced baseline serum AMH and poor response to IVF (van Rooij et al., 2002), and that initial AMH is associated with ovarian response in IVF patients who have normal FSH levels (Seifer et al., 2002). The relationship between AMH and other markers of ovarian reserve has been assessed, where AMH was found to correlate more strongly with the antral follicle count (AFC) than basal inhibin B, E2, FSH and LH in an infertile population (Fanchin et al., 2003b) (Table V).
|
Serum AMH levels have been measured at different times during the menstrual cycle, suggesting minimal fluctuation (Cook et al., 2000
). Hence AMH is relatively convenient to determine, especially as it seems to exhibit a relatively stable expression during the menstrual cycle, making it an attractive determinant of ovarian reserve (Gruijters et al., 2003). On this basis further validation of the test seems justified.
Normative data have been published (Cook et al., 2000; Laven et al., 2004) but are lacking with regard to day-to-day fluctuations, pulsatility and inter- or intra-cycle variability. The assay is expensive, and although several sensitive assays have already been developed (Long et al., 2000), the problem of inter-assay variability and the setting up of standardized reference values must be addressed. More research is required into the relationship between AMH and ovarian follicle dynamics, and as a marker of ovarian reserve in the general population before allowing its entry into clinical practice.
|
Dynamic ovarian reserve tests |
|---|
The recognized limitations of biochemical markers of ovarian reserve led to the development of a number of dynamic ovarian tests in an attempt to ‘unmask’ patients whose diminished ovarian reserve may have been missed using these markers (Scott and Hofmann, 1995
). GnRH agonist stimulation test (GAST/G-test):
This test evaluates the estradiol serum concentration change from cycle day 2 to day 3 after administration of a GnRH agonist, the latter causing a temporary increase in pituitary secretion of FSH and LH (Sharara et al., 1998).
Early studies were able to show a limited ability for the early rise in estradiol concentration to differentiate between normal and reduced ovarian reserve, albeit with different threshold values (Padilla et al., 1990). Others have looked at the sum of stimulated and bFSH concentrations with some success, but was this not superior to bFSH (Galtier-Dereure et al., 1996). In a significant study, the actual increase in estradiol concentration from day 2 to day 3 (
E2) following GnRH agonist administration on day 2 was a better predictor of ovarian reserve than age, bFSH and FSH:LH ratio (Ranieri et al., 1998). The same authors were also able to demonstrate that such an assessment would also allow effective drug regimen selection for IVF (Ranieri et al., 2001).
Because inhibin B is considered a direct marker of ovarian reserve, combining it with the GAST may have improved the predictive value of the test for IVF outcome. In fact, changes in the concentrations of inhibin B and estradiol have both been shown to correlate highly with the ovarian response to stimulation for IVF treatment (Ravhon et al., 2000) (Table VI).
|
The GAST appears to have much promise as a dynamic test of ovarian reserve but has not yet been validated outside an assisted reproductive setting. Furthermore the cost of performing the test has limited its acceptability in the general infertility population.
Clomiphene citrate challenge test (CCCT)
First described by Navot et al. (1987), this test involves the administration of 100 mg clomiphene citrate on cycle days 5–9, and the determination of FSH concentrations on days 3 and 10. In women with normal ovarian reserve, the overall increase in estradiol and inhibin production by the developing follicles should be able to overcome the estrogen antagonist effect of clomiphene on the HPO axis, and suppress FSH levels back into the normal range by day 10.
The evidence for this physiological suppressive effect of inhibin was supported by a study which showed reduced inhibin B levels following the CCCT in women with reduced ovarian reserve (Hofmann et al., 1998). In this way, the CCCT may be more indicative of oocyte quality than quantity. Nevertheless, a quantitative relationship between the CCCT and follicle density has been shown (Gulekli et al., 1999).
Several studies have now been published showing that women with a normal test respond better to ovarian stimulation, whereas an excessive response predicted a poor outcome to COS for IVF (Loumaye et al., 1990; Tanbo et al., 1992; Hofmann et al., 1996; Csemiczky et al., 2002). Correlation with biophysical markers of ovarian reserve (ovarian volume and AFC) has been shown in an infertile population (Erdem et al., 2004). The CCCT also predicts for improved pregnancy rates compared with age alone in an IVF population (Scott and Hofmann, 1995) or general subfertile population (Scott et al., 1993).
A retrospective study of women aged <40 years revealed an inverse relationship between the likelihood of successful pregnancy and CCCT findings, with no definite threshold being identified for FSH beyond which no pregnancy could be achieved (Yanushpolsky et al., 2003). The patient's age should not be ignored, however, as pregnancy rates still diminish in patients with advancing age despite a normal CCCT (Scott et al., 1995). This limitation renders the test, although highly specific, of low sensitivity (26%), albeit higher than that of bFSH (Barnhart and Osheroff, 1998). Furthermore, the exact relationship between elevated day 3 FSH levels and day 10 levels following the CCCT remains to be elucidated (Scott et al., 1995). Until this question is answered from further prospective study, the value of performing the test in the presence of a raised bFSH must remain in doubt (Table VII).
|