Utility of age-specific serum anti-Müllerian hormone concentrations
Article Outline
Abstract
This study assessed whether age-specific (as-) cutoffs for anti-Müllerian hormone (AMH) have higher specificity in reflecting ovarian reserve than non-age-specific (nas-) AMH values. as-AMH values were defined in 792 consecutive infertility patients by establishing as-quartiles of AMH within five age groups. Oocyte yields were then compared among women with AMH below or above the 25th centile at each age group. AMH continually decreased with advancing female age (P
<0.0001) and differed significantly in each of five selected age categories (P<0.001). In 442 women who reached IVF, as-AMH was predictive of lower oocyte yield if below as-25th centile and of higher oocyte yield if above as-75th centile. Combined normal and elevated as-AMH demonstrated 6.4 times (95% CI 3.9–20.6) odds of retrieving more than four oocytes than low as-AMH. Like as-FSH, as-AMH better reflects ovarian reserve than nas-ovarian reserve testing. However, in contrast to as-FSH, as-AMH defines risk for diminished ovarian reserve or high oocyte yields (i.e. ovarian hyperstimulation syndrome) and, therefore, may be a particularly useful ovarian reserve test in younger women in whom diminished ovarian reserved is most frequently overlooked, and who are at highest risk for ovarian hyperstimulation syndrome.
Anti-Müllerian hormone (AMH) is increasingly recognized for better specificity in reflecting ovarian reserve than FSH. Like FSH, AMH changes with advancing female age. Normal concentrations should, therefore, vary at different female ages. We, therefore, established age-specific (as-) AMH concentrations in five age groups and investigated whether oocyte number, obtained at IVF, differed on the basis of whether a patient’s as-AMH was in the normal as-range or below or above it. AMH demonstrated, once again, its better specificity in comparison to FSH by showing narrower normal ranges at all ages. Moreover, as-AMH allowed for discrimination of oocyte yields at all ages. This study confirms AMH as a consistent predictor of ovarian response. Moreover, AMH has the additional advantage of not only being able to predict diminished ovarian reserve and low oocyte yields but also high oocyte yields, and possible risk for polycystic ovarian syndrome and ovarian hyperstimulation syndrome. It, therefore, appears particularly suitable in the investigation of ovarian reserve in younger women.
Keywords: age-specific, AMH, anti-Müllerian hormone, infertility, ovarian reserve, pregnancy chance
Introduction
Functional ovarian reserve declines with advancing female age (Knauff et al., 2009); yet, ovarian function tests traditionally utilize cutoff values for normal ovarian function in age-independent ways. For example, with the most frequently utilized ovarian reserve test, FSH cutoff values of 10.0–12.0mIU/ml have traditionally been considered the upper limit of normal (Barad et al., 2007b).
More recently, anti-Müllerian hormone (AMH) has found increasing application in determining ovarian reserve (Broer et al., 2009, Carlsen et al., 2009, Ebner et al., 2006, Fleming et al., 2006, Knauff et al., 2009, Nelson et al., 2007, Nelson et al., 2009). It has been demonstrated that, while AMH and FSH correlate (Singer et al., 2009), AMH is superior to FSH in predicting oocyte yields (i.e. ovarian reserve) and IVF outcomes (Barad et al., 2009). Gnoth et al. (2008) suggested that a minimum concentration of 1.26ng/ml denotes diminished ovarian reserve (DOR) in women of all ages.
Barad et al. (2007a) previously pointed out that, in determining ovarian reserve, age-specific (as-) FSH concentrations are preferable to non-age-specific (nas-) cutoff values. as-FSH concentrations discriminate between better and poorer oocyte yields in association with IVF and allow for a more accurate diagnosis of DOR, especially in younger women under 38years of age. In a similar fashion, one can expect as-AMH concentrations to be superior to nas-cutoff concentrations. Such as-cutoffs have so far, however, not been defined. This study, therefore, analysed as-cutoffs in an infertility population of women and attempted to determine to what degree as-AMH concentrations could discriminate between women with better and poorer ovarian reserve, based on oocyte yields in IVF.
Materials and methods
An unselected initial study population included 1074 consecutive female patients with serum AMH determination between 2007 and 2009. This analysis was to test the added utility of AMH among a population of women already screened as having normal ovarian function by baseline FSH and oestradiol. Thus to define as-AMH concentrations, women with obviously elevated baseline FSH >12.0mIU/ml or oestradiol ⩾80pg/ml were eliminated, leaving 792 patients for statistical analysis. They were separated into five age categories: below 32years, 32–34years, 35–37, 38–39 and 40years and above. Within each age group, as-AMH concentrations were determined using first AMH sampling results at the study centre.
Since 2007, the study centre assesses AMH routinely at time of a new patient’s initial blood draw. A total of 442 amongst the original 1074 women had reached IVF by the time of data analysis. This part of the analysis includes women with baseline parameters outside the FSH and oestradiol limits set for determining as-AMH. These patients were utilized to analyse oocyte yields in reference to the ovarian reserve parameters for as-AMH.
Patients signed an informed consent at the initial consultation, permitting use of data from their medical record for clinical research purposes, as long as the medical record remains confidential and the identity of the patient remains protected. Such record-based studies were then only subject to an expedited review process by the Institutional Review Board.
Race and ethnicity of patients were determined at initial consultation. Clinical circumstances and infertility diagnoses were periodically re-evaluated as new clinical and laboratory data were obtained. Selected clinical patient data were, aside from each patient’s medical record, also maintained in the centre’s electronic research database, which is written in Microsoft Access and is only accessible to authorized clinical investigators.
Log-transformed mean and 95% CI and the median and 25–75th centiles were calculated for each age group. Patients with AMH concentrations in the lowest quartile (<25th centile) were considered to have evidence of DOR and to be at risk for diminished ovarian function. Similarly patients with AMH in the highest quartile were considered to be at risk for unusually high oocyte yields and therefore possibly suffering from polycystic ovarian syndrome and potentially at risk of developing ovarian hyperstimulation syndrome. The study centre is continuing to collect data and will report further on this subject in the future.
The 442 women, who by time of data analysis had reached a first IVF cycle, were separately analysed from the whole study group. Like the complete study population, they were divided into the same age categories. Oocyte yields were then assessed within each age category, based on whether a patient demonstrated normal, low or high as-AMH.
As previously reported (Gleicher et al., 2010a), a commercially available enzyme-linked immunoabsorbent assay (ELISA) was utilized to assess AMH. In brief, this is the DSL-10–14400 active Müllerian inhibiting substance/anti-Müllerian hormone (MIS/AMH) ELISA (Diagnostic Systems Laboratories, Webster, USA), an enzymatically amplified two-site immunoassay that does not cross-react with other members of the transforming growth factor β superfamily, including TGF-β1, BMP4 and ACT (Kevenaar et al., 2006). The theoretical sensitivity, or minimum detection limit, calculated by interpolation of mean plus two standard deviations of eight replicates of the 0ng/ml MIS/AMH standard, was 0.006ng/ml. Intra-assay coefficient of variation for an overall average AMH concentration was ⩽10% (Kevenaar et al., 2006). Results are presented in ng/ml, with a conversion factor of ×7.14 to pmol/l (Ebner et al., 2006).
In addition to AMH, ovarian reserve was assessed via cycle-day-2–3 FSH and oestradiol concentrations, obtained in the cycle preceding IVF. Both hormones were assessed utilizing an automated chemoluminescence system (ACS 180; Bayer Health Care, Tarrytown, USA).
Initial ovarian stimulation protocols of patients are principally determined by their age, with secondary modifications made based on ovarian function assessments. With presumed normal ovarian reserve, women up to age 38years were routinely stimulated in a long agonist protocol with 150–300IU of human menopausal gonadotrophin daily. Above age 38, or if women were considered to suffer from DOR at even younger ages, routine stimulation called for a microdose agonist protocol with at least 450IU of gonadotrophin daily, mostly given as FSH but also 150IU human menopausal gonadotrophin, as previously reported (Karande and Gleicher, 1999). The ovulation induction protocol type was recorded and used as a factor in the subsequent statistical analysis.
IVF cycles were conducted in routine fashion. In brief, human chorionic gonadotrophin was administered after three leading follicles exceeded average diameters of 18mm and oocyte retrievals under ultrasound control took place approximately 34h later. Retrieved follicular fluids were immediately transferred to the embryology laboratory, where oocyte yields were determined.
All data are expressed as mean and 95% CI of the mean or as median and quartiles of distribution. Variables that did not conform to normality were log-converted and back-transformed. They are presented as means and 95% CI of the mean. A P-value <0.05 was considered statistically significant. Differences between normally distributed variables were tested with analysis of variance or co-variance. Differences between groups of variables, not conforming to normality, were log-transformed and then tested with analysis of variance. Non-parametric analyses were performed with the Mann Whitney U-test. All analyses were carried out utilizing Statistical Package for Social Sciences software for Windows, version 18.0 2009 (SPSS, Chicago, USA).
Results
Table 1 summarizes patient characteristics separately for the total patient population, initially presenting 1074 women, and for 442 patients who reached IVF. Amongst the total population, 65% were Caucasian, 14.7% African and 18.9% Asian. The racial distribution amongst IVF patients was almost identical.
Table 1. Patient characteristics.
| All patients (n=1074) | IVF patients (n=442) | |
|---|---|---|
| Age (years) | 37.1 (36.5–37.1) | 37.4 (36.7–37.8) |
| Body mass index (kg/m2) | 23.6 (23.4–23.9) | 23.7 (23.1–25.1) |
| Race | ||
| African | 158 (14.7) | 65 (14.7) |
| Asian | 203 (18.9) | 93 (21.0) |
| Caucasian | 698 (65.0) | 281 (63.6) |
| Other | 15 (1.4) | 3 (0.7) |
| Diagnosis | ||
| Oocyte donors | 85 (7.9) | 48 (10.9) |
| Infertility patients | 989 (92.1) | 394 (89.1) |
| DOR/POA | 541 (54.7) | 221 (56.1) |
| Male infertility | 177 (17.9) | 70 (17.8) |
| Tubal infertility | 194 (19.6) | 68 (17.3) |
| PCO | 33 (3.3) | 17 (4.3) |
| Othera | 45 (4.6) | 18 (4.6) |
aOther |
Primary infertility diagnoses were also very similar in both patient groups. In the total population, this included the following: DOR and/or premature ovarian ageing (54.7%), tubal infertility (19.6%), male factor (17.9%), polycystic ovaries (3.3%) and ‘other’ (4.6%). IVF patients demonstrated an almost identical distribution (Table 1).
Figure 1 demonstrates the change in AMH and FSH as a continuous regression against age. Both FSH and AMH are plotted against an exponential curve to correct for normality of distribution. Adjusted r2±standard error for AMH was 0.26±0.56 (P<0.001) while that for FSH was 0.06±0.27 (P<0.001).
Figure 1.
Age-specific anti-Müllerian hormone (AMH) and FSH concentrations relating to age of 792 women with FSH <12
mIU/ml and oestradiol ⩽80pg/ml. Values are means and 95% CI.
With 792 individuals in the analysis, the 95% CI of the mean are too small to define a meaningful ‘normal’ group. However, when age categories are set for AMH, upper and lower quartile cutoff values, representing the 25–75th centile for any given age group, can be established. Figure 2 reveals box plots of AMH for women with baseline FSH <12mIU/ml and oestradiol ⩽80ng/ml. Table 2 shows AMH, FSH and oestradiol concentrations in the different age groups. Mean (log-transformed) as-AMH, as well as median and upper and lower cutoffs, decrease significantly between consecutive age groups (P<0.001).
Figure 2.
Definition of age-specific anti-Müllerian hormone (AMH) relating to age of 792 women with FSH <12
mIU/ml and oestradiol ⩽80pg/ml. Values are median (lines), 25–75th centile (boxes) and 95% CI (whiskers). AMH concentrations declined significantly with age (P<0.001). Table 2 summarizes lower and upper cutoff values at various ages. Circles=outliers; stars=extreme outliers.
Table 2. Hormone concentrations among 792 patients with baseline FSH <12mIU/ml and oestradiol ⩽80pg/ml.
| <32 (n=193) | 32–34 (n=127) | 35–37 (n=139) | 38–39 (n=98) | ⩾40 (n=235) | |
|---|---|---|---|---|---|
| Age (years) | 27.0 (26.5–27.6) | 32.9 (32.7–33.0) | 36.0 (35.9–36.2) | 38.5 (38.4–38.6) | 42.3 (42.1–42.6) |
| FSH (mIU/ml) | 6.8 (6.5–7.1) | 7.1 (6.8–7.5) | 7.6 (7.2–8.0) | 7.8 (7.3–8.3) | 8.0 (7.7–8.3) |
| Oestradiol (pg/ml) | 46.2 (43.2–50.0)a | 47.8 (44.5–51.3) | 52.2 (47.6–57.2)a | 47.5 (42.3–53.4) | 57.1 (52.8–61.7)a |
| AMH (ng/ml) | 2.7 (2.4–3.1) | 1.94 (1.6–2.3)a | 1.34 (1.1–1.6)a | 0.98 (0.78–1.20)a | 0.56 (0.49–0.66)a |
| AMH (ng/ml) (median (25–75th centile)) | 3.1 (1.8–4.8) | 2.2 (1.4–3.3) | 1.6 (0.7–2.6) | 1.0 (0.5–2.0) | 0.7 (0.2–1.4) |
aBack-transformed after logarithmic transformation. |
The lower 25th centile as defined in Table 2 represents the as-cutoff for patients with low as-AMH. Likewise the upper 75th centile represents the upper cutoff and patients in each age group whose AMH concentration exceeded the 75th centile for their age group were considered to have high as-AMH. Using these defined cutoffs, among the 442 women who had an IVF cycle, 138 (31.2%) women demonstrated low as-AMH (<25th centile), 208 (47.1%) normal as-AMH (25–50th centile) and 96 (21.7%) high as-AMH (>75th centile) concentrations.
Figure 3 and Table 3 summarize oocyte yields in the five age categories. Oocyte yields declined overall with advancing female age (F=214, df=1; P<0.001). Figure 3, however, defines the subtleties of this decline after adjustment for age. Figure 3A demonstrates oocyte yields in the five age categories, depending on low or combined normal and abnormally high as-AMH. In all age categories, oocyte yields with abnormally low as-AMH were significantly lower than with normal and high AMH (P<0.001; for further statistical detail, see Table 3). The figure suggests that oocyte yields among women with normal and high as-AMH decline very little. In contrast for women with abnormally low as-AMH, oocyte yields appear to decline steadily until age 35–37, when the decline appears to flatten. Indeed, adjusting for age and ovulation induction protocol, the odds of retrieving more than four oocytes from women with normal and high as-AMH were 6.4 (95% CI 3.9–10.6; Wald 51.2; df=1; P<0.001) in comparison to those with low as-AMH concentrations.
Figure 3.
Oocyte yields at different ages and anti-Müllerian hormone (AMH) concentrations. (A) Oocyte yields among 442 women who underwent IVF with normal and high AMH (green) and abnormally low AMH (blue); ∗Significant differences within each age group are based on the Mann Whitney U-test (P
<0.05). (B) Oocyte yields in each age group based on abnormally low (blue), normal (green) and abnormally high (beige) as-AMH concentrations. Normal was defined as within the 25–75th centiles for a given age group; ∗Significant differences from normal within each age group (P<0.05). Table 3 gives further statistical details. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
Table 3. Oocyte yields among 442 patients reaching IVF.
| Age (years) | Low as-AMH | Normal as-AMH | High as-AMH | |||
|---|---|---|---|---|---|---|
| IVF cycles | Mean (95% CI) | IVF cycles | Mean (95% CI) | IVF cycles | Mean (95% CI) | |
| <32 (n=84) | 27 | 10.3 (8.0–12.5)b,c | 38 | 15.2 (12.9–17.6)a | 19 | 16.1 (11.9–20.3)a |
| 32–34 (n=65) | 21 | 8.3 (5.4–11.2)c | 24 | 10.4c (8.3–12.6)c | 20 | 19.2 (13.8–24.7)a,b |
| 35–37 (n=69) | 21 | 3.5 (2.6–4.6)b,c | 31 | 8.3 (6.7–10.2)a | 17 | 11.0 (8.5–13.5)a |
| 38–39 (n=60) | 22 | 3.0 (2.0–4.4)b,c | 26 | 7.0 (5.4–9.2)c | 12 | 16.5 (10.6–22.4)a,b |
| ⩾40 (n=164) | 47 | 2.2 (1.7–2.6)b,c | 89 | 4.4 (3.8–5.2)a,c | 28 | 12.0 (10.3–13.8)a,b |
Figure 3B further defines these data by separating oocyte yield in each age category by abnormally low, normal and abnormally high as-AMH. In four out of five age groups, as-AMH statistically differentiates low yields with abnormally low as-AMH from higher yields with normal and even abnormally high as-AMH. In three out of five of the age groups (marked with asterisk) excessively high as-AMH did define a high-risk group for significantly higher-than-average oocyte yields (P<0.05). Table 3 summarizes the statistical details.
Table 3 demonstrates that, among 442 women who reached oocyte retrieval, those with low as-AMH at all ages yielded fewer oocytes than women with normal as-AMH. A generalized linear model found the mean oocyte yield (adjusted for age and ovulation induction protocol) for the low as-AMH group to be 5.5 (95% CI 4.6–6.5; P<0.001) oocytes and for the high as-AMH group to be 14.0 (95% CI 12.8–15.1; P<0.001) compared with normal which was 8.9 (95% CI 9.1–9.6). Table 3 summarizes the statistical details within each age group. Women with abnormally low as-AMH had clearly different oocyte yields compared with all other patients (normal and high as-AMH combined; F=214.2, df=1; P<0.001).
Discussion
Challenges in assessing ovarian reserve correctly and limitations of currently available methodologies have recently been the subject of a number of insightful publications (Broer et al., 2009, Fleming et al., 2006, Knauff et al., 2009, Sun et al., 2008). Unanimity appears to confirm that AMH in many ways may represent a more specific marker of DOR than the historically utilized hormone, FSH (Barad et al., 2009, Ebner et al., 2006, Hazout et al., 2004).
However, with few exceptions (Barad et al., 2009, Ebner et al., 2006, Gnoth et al., 2008, Singer et al., 2009), the literature so far does not offer cutoff values that may delineate between normal and abnormal ovarian reserve. Moreover, the literature, with one exception (Barad et al., 2009), does not comment on potential differences in the utility of AMH at different female ages, as observed for FSH (Abdalla and Thum, 2004, Toner, 2004).
Based on FSH data, Barad et al. (2007a) suggested that as-ovarian reserve assessments may be superior to nas-testing in predicting DOR and production of lower oocyte yields in association with IVF. Sun et al. (2008), who pointed out the importance of differentiating between age-dependent (physiological) and non-age-dependent (premature) ovarian ageing, recently suggested a similar concept. Considering such evolving concepts, it appeared important to investigate whether as-AMH concentrations, like as-FSH, may offer improved specificity in detecting DOR over nas-AMH testing. The current study has done that and, as previously reported for FSH, the AMH data strongly re-emphasize that, judged by oocyte yields in IVF, as-ovarian function tests appear superior to nas-testing.
The study population to a large degree represented infertility patients with significant DOR (Table 1). To avoid extremes, DOR patients with elevations of baseline FSH or oestradiol were eliminated in establishing as-AMH. The effectiveness of this approach is well documented by the fact that the final study populations still demonstrated the well-described declines in AMH (and rises in FSH) with advancing female age (Singer et al., 2009; Figure 1).
This study, however, offers important additional insights. It demonstrates that the range of AMH is at all ages narrower than that of FSH (Figure 1). Since narrower testing ranges reflect more specificity, it is not surprising that AMH has been found to be more specific in reflecting ovarian reserve than FSH (Barad et al., 2009).
Figure 1 also demonstrates, that both hormones, AMH and FSH, demonstrate the narrowest ranges of as-concentrations at approximately age 35. This observation would suggest that at this age both of these ovarian reserve parameters are probably at their best in reflecting ovarian function (i.e. demonstrating highest specificity). Below and above that age, normal ranges widen and hormone concentrations, therefore, are likely to become less specific. This, of course, should not be surprising abnormally high FSH concentrations at younger ages have been reported as less predictable of poor treatment outcomes (Abdalla and Thum, 2004, Toner, 2004) and Barad et al. (2009) previously reported that, although superior to FSH, AMH loses specificity at more advanced female ages. Ovarian reserve evaluations by FSH and AMH, and even their as-values, therefore, may have to be viewed differently at different ages.
The current study thus sheds further light on the value of AMH testing at different ages. As Figure 3B demonstrates, even as-AMH, while still superior to nas-AMH, appears to offer its best diagnostic specificity only at ages <32–39, correlating well to the narrowest range of as-AMH at approximately age 35 (Figure 1). Within that age range, as-AMH provides good discrimination in regard to oocyte yields at both extremes of AMH: abnormally low concentrations are statistically associated with abnormally low oocyte yields, while abnormally high AMH is predictive of high oocyte yields.
While abnormally low oocyte yield is often defined as four or less retrieved oocytes, such an age-independent definition does not make physiological sense since expected oocyte yields, of course, are much higher at younger than older ages (Singer et al., 2009). In practical terms this means that four or less oocytes will always represent a low count in younger women but may represent an excellent retrieval result in older women. In the same way, seven or eight oocytes, clearly above this widely utilized cutoff value, may still represent a low yield in a 22-year-old. The relativity of oocyte yield, based on female age, therefore, needs to be considered when results of this study are assessed.
At younger ages (<32years), as-AMH still discriminates risk for low oocyte yields, but appears insufficiently specific to discriminate high oocyte numbers from normal yields. This study has a fairly large sample of women aged 40 and above and so is able to assess the value of as-AMH in this group. as-AMH appears to be able to predict both low and high response even in the oldest age group. Still these results must be interpreted with caution since most of these women aged 40 and above have been treated with dehydroepiandrosterone (DHEA) supplementation. Interestingly Gleicher et al. (2010b) previously reported that in women with DHEA supplementation, occurrence of pregnancy was almost exclusively linked to improvements of AMH.
Combined, these data may suggest that with advancing female age, AMH may find a place in defining who, amongst older women aged 40 and above, may benefit from infertility treatment.
The accurate diagnosis of DOR, therefore, appears important at all ages. This may seem counterintuitive to current clinical practice, which largely assumes that younger women, even if afflicted by DOR, still possess adequate ovarian reserve to conceive (Gleicher and Barad, 2006a) and that, therefore, a timely and more accurate diagnosis of DOR in young women may be less of a priority than in older women.
While this study does not contradict this argument, accurate and timely diagnosis in young women may be even more important than in older patients since DOR is less clinically obvious at younger ages, frequently unsuspected and overlooked and often leads to an inappropriate diagnosis of so-called unexplained infertility (Barad et al., 2007b, Gleicher and Barad, 2006b). Younger women, therefore, may actually be the best targets for as-AMH testing. Once suspected of DOR, they then can be followed closely and can consider time adjustments to family building efforts or fertility preservation treatments.
Comparing here the reported differences in as- and nas-AMH to previously published FSH data (Barad et al., 2007b), the discriminatory abilities of as-AMH in predicting oocyte yields, and therefore insipient DOR, appear superior at all ages. These findings, of course, relate well to the better specificity of nas-AMH over nas-FSH (Barad et al., 2009). Whether combining as-FSH and as-AMH further improves assessments of DOR and expected oocyte yields is currently under investigation.
These conclusions are similar to those of previously published work by Austrian colleagues: Ebner et al. (2006) not only suggested that AMH appears superior to FSH in predicting oocyte numbers and their quality, but actually defined a nas-AMH range for maximal oocyte quality between 1.7 and 4.5ng/ml. This range, almost perfectly relates to the normal as-AMH range defined in this study for women up to age 32 (Table 2). It now would be interesting to determine whether this ideal AMH range with regards to egg quality also changes with advancing female age or whether the obviously superior quality of young age is not recoverable at later ages.
Historically, AMH has been primarily utilized to rule out the presence of DOR. AMH, however, is also potentially useful at the other end of the ovarian reserve spectrum, when a diagnosis of polycystic ovarian syndrome is contemplated (Carlsen et al., 2009, Nelson et al., 2007, Nelson et al., 2009). AMH cutoff values in polycystic ovarian syndrome patients have, so far not been well defined, and where attempts at definition were made, nas-testing was utilized (Nelson et al., 2009).
While this study does not define diagnostic as-AMH concentrations for polycystic ovarian syndrome, it very clearly demonstrates that the high as-AMH, in most of the older age groups, discriminates between normal and high oocyte yields with IVF (Table 3 and Figure 3B).
Nelson et al. (2009) suggested that a nas-AMH of above 15pmol/l (2.1ng/ml) already denotes risk for ovarian hyperstimulation. Such an AMH concentration, on as-basis, is near the lower cutoff point in women under age 32 and in the middle of the normal range of women aged 32–34 and 35–37 (Table 2). As it would include a majority of young women with normal ovarian reserve, it does not appear to be specific enough to identify patients at risk for ovarian hyperstimulation. This nas-AMH cutoff value would be clinically impractical as a screening tool and irrational for changes in stimulation protocol, as suggested by Nelson et al. (2009).
In the current study, 4.8ng/ml represents the upper limit of normal in the youngest and therefore in the highest risk group for ovarian hyperstimulation. While none of the women did develop significant clinical hyperstimulation, some produced excessively high oocyte yields (Table 3). More importantly, in all age groups above age 32, the upper as-AMH cutoff clearly defined a patient population that produced significantly more oocytes than women in normal as-AMH range.
This study, therefore, suggests that as-AMH at all ages allows for discrimination of oocyte yields, but does so differently at different female ages. Whether the here-utilized methodology of defining normal ranges as between the 25th and 75th centiles represents the best methodology remains to be seen. The data presented here, however, suggest that as-AMH testing offers clear advantages over nas-testing and that, whatever ultimate cutoff values shall be chosen to define risk towards abnormal ovarian responses, they should be age specific.
By also defining risk towards high oocyte yields, as-AMH demonstrates yet another distinct advantage over as-FSH, which has only predictive value for abnormally low oocyte yields (Barad et al., 2007b). The prevention of ovarian hyperstimulation syndrome has, to a degree, remained elusive (Nelson et al., 2007) and is especially important in younger women, where risks are the highest since oocyte production is the largest (Engmann et al., 2008).
The as-AMH ranges presented here, need to be viewed with caution. As previously demonstrated for FSH, as-hormone concentrations will be dependent on study populations (Barad et al., 2007a). Since the 95% CI for age will vary dependent on the percentage of women with DOR in each age group, different as-FSH concentrations between IVF centres, dependent on their respective patient populations, were reported, as one would suspect. More favourably selected patients in one centre, therefore, will demonstrate lower FSH and higher AMH cutoffs than less favourable patients at another centre. In considering the design of this study, this experience was taken into account and women with possible DOR, defined by FSH concentrations ⩾12.0mIU/ml or baseline oestradiol ⩾80pg/ml, were not included.
Considering the adverse patient selection and high prevalence of premature DOR at this centre (Table 1, Barad et al., 2007a), the selected patient population may, nevertheless, still be more affected by DOR than patients at many other IVF centres. Extrapolating from previously published FSH data (Barad et al., 2007a), it therefore seems likely that the as-AMH cutoff concentrations reported here will be conservative for a majority of infertility centres. Fertility centres with less adversely affected patients may, therefore, have to utilize slightly higher as-AMH cutoffs at all ages. Preferably, fertility centres should establish centre-specific as-cutoff values until, ideally, universal cutoff values have been reported for normal, fertile populations, which would be applicable to all women.
In conclusion, this study demonstrates the advantages of as-ovarian reserve testing in comparison to current widely practised nas-testing. By having demonstrated advantages to as-FSH and now as-AMH testing over traditional nas-testing, it seems increasingly likely that as-testing is, in general, superior to nas-testing when it comes to ovarian reserve assessments.
Utilizing multiple ovarian reserve parameters, Verhagen et al. (2008), based on a meta-analysis of published studies, recently argued that the use of more than one ovarian reserve test can currently not be supported. After reviewing multivariate models for prediction of ovarian reserve and occurrence of pregnancy with IVF, the authors concluded that predictive values of various models, utilizing different tests, did not vary significantly from the accuracy of antral follicle counts as a single test. While the data presented here cannot address the benefits of multiple ovarian reserve tests over the utilization of single tests, it is important to note that the meta-analysis by Verhagen et al. (2008) involved only nas-testing. It seems possible, and maybe even likely, that as-testing of ovarian reserve may give different results.
Like Verhagen et al., 2008, Barad et al., 2009 recently compared predictive values for ovarian reserve and pregnancy, based on receiver operating characteristic curves and demonstrated a significant advantage of AMH over FSH in predicting both outcomes. Those comparisons, like those of Verhagen et al. (2008) were based on nas-testing. Now that significant advantages of as- over nas-testing for FSH and AMH have been demonstrated, it seems increasingly likely that only as-use of all ovarian reserve tests will further improve sensitivity and specificity of such testing. This means that whether ovarian reserve testing is performed using FSH, AMH, antral follicle counts or other testing procedures, all should utilize as-, rather than nas-cutoff values.
Age-specific ovarian reserve tests appear to offer distinct benefits at all ages and may be especially beneficial for younger women in whom a diagnosis of DOR is rarely suspected and, therefore, often overlooked (Barad et al., 2007b) and who are at highest risk for ovarian hyperstimulation syndrome.
References
- . An elevated basal FSH reflects a quantitative rather than qualitative decline of the ovarian reserve. Hum. Reprod. 2004;19:893–898
- . Validation of age-specific cutoffs for ovarian function assessment by FSH levels via a second patient population: inherent differences in patient populations between IVF-centers. Fertil. Steril. 2007;88:S279
- . Age-specific levels for basal follicle-stimulating hormone assessment of ovarian function. Obstet. Gynecol. 2007;109:1404–1410
- . Comparing anti-Mullerian hormone (AMH) and follicle-stimulating hormone (FSH) as predictors of ovarian function. Fertil. Steril. 2009;91:1553–1555
- . The role of antimullerian hormone in prediction of outcome after IVF: comparison with the antral follicle count. Fertil. Steril. 2009;91:705–714
- . Anti-Mullerian hormone concentrations in androgen-suppressed women with polycystic ovary syndrome. Hum. Reprod. 2009;24:1732–1738
- . Basal level of anti-Mullerian hormone is associated with oocyte quality in stimulated cycles. Hum. Reprod. 2006;21:2022–2026
- . The use of gonadotropin-releasing hormone (GnRH) agonist to induce oocyte maturation after cotreatment with GnRH antagonist in high-risk patients undergoing in vitro fertilization prevents the risk of ovarian hyperstimulation syndrome: a prospective randomized controlled study. Fertil. Steril. 2008;89:84–91
- . Dynamics of FSH-induced follicular growth in subfertile women: relationship with age, insulin resistance, oocyte yield and anti-Mullerian hormone. Hum. Reprod. 2006;21:1436–1441
- . ‘Ovarian age-based’ stimulation of young women with diminished ovarian reserve results in excellent pregnancy rates with in vitro fertilization. Fertil. Steril. 2006;86:1621–1625
- . Unexplained infertility: does it really exist?. Hum. Reprod. 2006;21:1951–1955
- Gleicher, N., Weghofer, A., Barad, D.H., 2010a. Anti-Mullerian hormone (AMH) defines, independent of age, low versus good live-birth chances in women with severely diminished ovarian reserve. Fertil. Steril.
- . Improvement in diminished ovarian reserve after dehydroepiandrosterone supplementation. Reprod. Biomed. Online. 2010;21:360–365
- . Relevance of anti-Mullerian hormone measurement in a routine IVF program. Hum. Reprod. 2008;23:1359–1365
- . Serum antimullerian hormone/mullerian-inhibiting substance appears to be a more discriminatory marker of assisted reproductive technology outcome than follicle-stimulating hormone, inhibin B, or estradiol. Fertil. Steril. 2004;82:1323–1329
- . A rational approach to the management of low responders in in-vitro fertilization. Hum Reprod. 1999;14:1744–1748
- Serum anti-mullerian hormone levels reflect the size of the primordial follicle pool in mice. Endocrinology. 2006;147:3228–3234
- Anti-Mullerian hormone, inhibin B, and antral follicle count in young women with ovarian failure. J. Clin. Endocrinol. Metab. 2009;94:786–792
- . Serum anti-Mullerian hormone and FSH: prediction of live birth and extremes of response in stimulated cycles – implications for individualization of therapy. Hum. Reprod. 2007;22:2414–2421
- Anti-Mullerian hormone-based approach to controlled ovarian stimulation for assisted conception. Hum. Reprod. 2009;24:867–875
- . Correlation of antimullerian hormone and baseline follicle-stimulating hormone levels. Fertil. Steril. 2009;91:2616–2619
- . A new approach to ovarian reserve testing. Fertil. Steril. 2008;90:2196–2202
- . Modest follicle-stimulating hormone elevations in younger women: warn but don’t disqualify. Fertil. Steril. 2004;81:1493–1495discussion 496–7
- . The accuracy of multivariate models predicting ovarian reserve and pregnancy after in vitro fertilization: a meta-analysis. Hum. Reprod. Update. 2008;14:95–100
Dr Barad, MD, MS, FACOG, is director of assisted reproductive technology and clinical research at Center for Human Reproduction. He graduated from Rutgers Medical School, completed his residency in obstetrics and gynaecology at Sloane Hospital for Women of Columbia Presbyterian Medical Center and completed a Fellowship in reproductive endocrinology and infertility at Brigham and Women’s Hospital. He is a diplomate of American College of Obstetrics and Gynecology with subspecialty certification in reproductive endocrinology and Associate Clinical Professor at Albert Einstein College of Medicine in both the departments of Obstetrics & Gynecology and Women’s Health.
PII: S1472-6483(10)00787-X
doi:10.1016/j.rbmo.2010.12.002
© 2010 Reproductive Healthcare Ltd. Published by Elsevier Inc All rights reserved.
