Abstract
Keywords
A number of approaches have been undertaken to assess ovarian function or ovarian reserve in order to predict
This publication was singled out, as it is a recent study that examines many of the markers and protocols listed above. This study assessed the ovarian reserve in IVF patients by means of FSH challenge tests during the normal menstrual cycle prior to the treatment cycle. A comparison of ovarian morphology and a range of hormonal markers with both oocyte number and pregnancy outcome was then undertaken. The results of other studies with a similar design and outcomes have recently been published [2–6].
Protocol
The study consisted of 56 women undergoing IVF. The women were under the age of 38 years with day-3 serum FSH concentrations of less than 10 IU/l, normal 26–35 day menstrual cycles and no endocrine abnormalities. The bases of infertility were tubal and male factors in 26 patients, while the remainder were unexplained.
Follicle-stimulating hormone challenge test
These women were initially studied during the menstrual cycle prior to the IVF treatment cycle. Two FSH challenge tests were undertaken; the first test on days 3–4 of the cycle, the second between days 5–10 postovulation of the same menstrual cycle. Blood was taken 24 h after administration of human FSH (300 IU), prior to hormone measurements of serum estradiol, FSH, inhibin B and AMH.
Following the second challenge test, a gonadotropin-releasing hormone (GnRH) antagonist was administered to downregulate the pituitary secretion of gonadotropins. Ovarian follicular development was then stimulated by FSH treatment (150 IU/day, human menopausal gonadotropin [hMG] 150 IU/day). Ovulation was induced by human chorionic gonadotropin (hCG) treatment (10,000 IU) when three follicles larger than 17 mm were identified by ultrasound. Oocytes were collected 36 h later; a maximum of three oocytes were subsequently transferred back into the patient. Pregnancy rates were also assessed.
Methods: markers of ovarian function
The main outcomes investigated were the number of oocytes retrieved and number of pregnancies arising from the treatment. The objective of the study was to establish which of the markers below correlated best with these outcomes. The markers examined were: ovarian volume and antral follicle count, serum estradiol and FSH, serum inhibin B, and AMH.
Ovarian volume & antral follicle count
On days 3–4 of the menstrual cycle, ovarian volume and the number of antral follicles of 2–10 mm in size were determined by transvaginal ultrasound. The methodology and reliability criteria were not stated.
Serum estradiol & follicle-stimulating hormone
These end points were included as useful comparators for the other end points. It has been demonstrated that these markers correlate less well compared with inhibin B or AMH.
Serum inhibin B
Inhibins consist of α and either βA or βB subunits (αβA: inhibin A; αβB: inhibin B), which are produced by the granulosa cells of antral and preovulatory follicles, and by surrounding thecal tissue and corpora lutea. Inhibin A and B are primarily produced by preovulatory and small antral follicles, respectively. Inhibins are not expressed in primordial or atretic follicles. There are one, possibly two, identified roles of inhibin in women; first, inhibins act as potent endocrine feedback regulators of FSH production by the pituitary gland. Their ovarian production is stimulated by FSH [7]. Second, in mice and presumably in humans, inhibin acts a potent intraovarian clamp on follicular development [8]. Serum inhibin A and B are stimulated
Anti-Müllerian substance
AMH was originally identified during fetal development as an inhibitor of Müllerian duct development. AMH is produced by small antral follicles in adult mouse and human ovaries although its presence, albeit at reduced levels, has been detected in human primordial and preovulatory follicles [9,10]. In AMH-knockout mice, progression from primordial to antral follicle is inhibited, suggesting that AMH, like inhibin, is a suppressor of folliculogenesis [9]. Serum AMH is not sensitive to gonadotropin treatment and therefore does not respond in the FSH challenge test. Serum AMH has been proposed as a marker of ovarian response [3–6,9].
Results
Oocyte count, antral follicle number & serum FSH, estradiol, inhibin B & AMH
In Table 1, antral follicle number correlated equally well with serum inhibin B and AMH levels (r = 0.71–0.79), with poorer correlations with estradiol and oocyte volume. Oocyte number was significantly correlated with basal AMH and antral follicle number (r = 0.65–0.73). Similar observations were made when the FSH stimulation test was performed in either the follicular or luteal phases. Multiple regression analysis demonstrated that the best model for predicting the number of oocytes was antral follicle number (r = 0.66; p < 0.0001), and that the addition of the endocrine markers did not improve this prediction. Of the endocrine markers, the increase in FSH-stimulated inhibin B and AMH in the follicular phase provided the best prediction (r = 0.62; p < 0.0001).
Correlation coefficients (r) between serum inhibin B, anti-Müllerian hormone and either antral follicle number or oocyte number when assessed in the follicular phase of the menstrual cycle prior to the treatment cycle.
p < 0.05
p < 0.001.
AMH: Anti-Müllerian hormone; FSH: Follicle-stimulating hormone.
Pregnancy rates
Basal pretreatment serum AMH levels were significantly higher (×1.77–2.03) in women who subsequently became pregnant compared with those who did not (Table 2). No other marker reached significance.
Serum inhibin B and AMH levels in pretreated women who were subsequently found to be pregnant or nonpregnant.
AMH: Anti-Müllerian hormone; FSH: Follicle-stimulating hormone.
Conclusions
It is concluded that antral follicle number and endocrine markers (AMH and inhibin B) are similarly effective as predictors of oocyte number. These findings are largely in agreement with other studies, which concluded that serum inhibin B, AMH and antral follicle number are similar in their ability to predict oocyte number [1–6]. In addition, serum AMH appears to be a better predictor of pregnancy outcome than serum inhibin B or other markers, in support of previous studies [6].
Reviewer's perspective
Depending on the study, different markers or combinations of markers have been highlighted as the best choices in predicting oocyte number and pregnancy outcome. Why these differences? There appear to be several issues: either the markers or protocols chosen are not appropriate, or the analyses are not optimal.
In consideration of the choice of markers, both AMH and inhibin B are products of granulosa cells primarily from small follicles and, in principle, ideal markers for monitoring follicle development. At this stage, it is believed that these markers are independent in action [8], although their presumed function in suppressing follicular maturation may be similar. Can one expect to find better markers? If so, what would one look for? The authors suggest insulin-like growth factor binding protein (IGFBP)-1 as a possible marker of pregnancy outcome. An oocyte factor produced once the follicle has left the primordial follicle pool may be ideal. If so, would this factor be detectable in serum? Is this assessment of oocyte quality best left to the assessment of embryos at a later stage in the treatment process?
Are there technical limitations? There may be. For example, in this study it is interesting to note that, in contrast to the serum hormone tests where the assay validity criteria were detailed, there was no such information provided for the antral follicle number. Up to 80 antral follicles were counted in one instance. How reliable was this number? If there is doubt, does this suggest that better methodologies in antral follicle count would lead to a tighter correlation and thus a better prediction of oocyte number? Similarly, in another study, the reliability of basal inhibin B measurements used in the FSH stimulation test was questioned, as these values were often near the level of detection of the assay [4]. Furthermore, in this and other studies, it should be noted that the sample numbers in the various groups are often low. Visual inspection of plots comparing, for example, antral follicle number with the other markers, does suggest that the statistical analyses are often influenced by outliers.
Serum AMH appears to be a better predictor of pregnancy outcome than serum inhibin B. As seen in Table 2, in contrast to inhibin B, the significant increase in serum AMH levels observed between pregnancy and nonpregnancy groups is related more to the lower standard deviation (reduced variance) in the AMH values than to the magnitude of the increase in AMH between groups. Presumably, the increased variance with the inhibin B levels is due to the effects of the FSH stimulation test compared with AMH, which is insensitive to FSH stimulation. These findings are somewhat surprising, as the sensitivity of inhibin B to gonadotropin treatment has been viewed as a good indicator of follicle quality and, thus, one would have expected it to be at least as responsive as AMH. Is this still a useful issue to pursue with tighter FSH stimulation protocols?
What is still unclear is to what extent these markers are useful in predicting either poor responders (low oocyte number) or pregnancy outcome. Using receiver operating characteristic curve analysis to identify poor responders, Muttukrishna demonstrated that serum AMH had the best sensitivity (87%) and specificity (64%) characteristics compared with serum inhibin B and FSH and antral follicle number [4]. However, these specificity values would suggest that serum AMH (and the other markers) as a single marker was of limited value. Even with a combination of markers (basal FSH, AMH, FSH-stimulated increased in estradiol and inhibin B, antral follicle count and age), the prediction of poor responders was still only 87% sensitive and 80% specific. Clearly an improvement, but unlikely to be cost effective.
This and other studies discussed do suggest that, depending on the treatment protocol, either serum AMH, inhibin B or antral follicle number provide some predictive value for detecting poor responders and, perhaps, pregnancy outcome. However, the prediction level observed is still limiting. It would appear to this reviewer that part of the approach to improve prediction may lie in better-defined protocols and assay methods.