Article Anti-Müllerian hormone as a predictor of IVF outcome

Transcription

1 RBMOnline - Vol 14. No Reproductive BioMedicine Online; on web 23 March 2007 Article Anti-Müllerian hormone as a predictor of IVF outcome Dharmawijaya Nayanananda Lekamge graduated from the Faculty of Medicine, University of Peradeniya, Sri Lanka. After completing his post-graduate degree in Obstetrics and Gynaecology in 1994 he joined the Department of Pharmacology at the University of Peradeniya, Sri Lanka in He is currently reading for his PhD in Reproductive Medicine in the Department of Obstetrics and Gynaecology, University of Adelaide. His main research interests are ovarian reserve, anti-müllerian hormone and the effects of oocyte-secreted factors in female fertility. Dr Dharmawijaya N Lekamge Dharmawijaya N Lekamge 1,2, Michael Barry 2, Michele Kolo 2, Michelle Lane 1,2, Robert B Gilchrist 1, Kelton P Tremellen 1,2,3 1 Research Centre for Reproductive Health, Discipline of Obstetrics and Gynaecology, School of Paediatrics and Reproductive Health, University of Adelaide, 5005 Adelaide; 2 Repromed, 180 Fullarton Road, Dulwich, 5065, Adelaide, South Australia 3 Correspondence: Fax ; kelton.tremellen@adelaide.edu.au Abstract Serum anti-müllerian hormone (AMH) concentration and antral follicle count (AFC) are two increasingly popular static measures used to predict ovarian reserve prior to IVF treatment. While they have been shown to be good predictors of oocyte yield during ovarian stimulation, their status as indicators of oocyte quality and pregnancy rates is currently uncertain. The present study measured baseline concentrations of serum AMH and FSH, and AFC from 126 women undergoing IVF treatment. These data were then related to IVF outcomes. As expected, patients with lower serum AMH and AFC produced a significantly (P < 0.001) lower number of oocytes compared with patients with higher serum AMH/AFC. Fertilization rates in patients with lower serum AMH were significantly inferior compared with patients with higher serum AMH, irrespective of whether IVF (P = 0.043) or intracytoplasmic sperm injection (P = 0.006) was used to achieve fertilization. These low AMH patients yielded fewer oocytes, had lower fertilization rates, generated fewer embryos, and had a higher incidence of miscarriage during fresh transfers, ultimately culminating in a halving of the pregnancy rate per IVF cycle compared with the high AMH group. Keywords: anti-müllerian hormone, antral follicle count, ICSI, IVF, ovarian reserve, pregnancy Introduction 602 The established predictors of reproductive potential during infertility treatment are maternal age (Faddy et al., 1995; Check et al., 1998), early follicular phase FSH concentrations (Scott t et al., 1989; Toner et al., 1991), and, less popularly, serum inhibin B concentration (Seifer et al., 1997). None of these parameters is a particularly reliable predictor of the number and quality of oocytes remaining within the ovary, or the likely probability of pregnancy from infertility treatment. Recently, interest in the use of anti-müllerian hormone (AMH) and antral follicle count to predict patient response to ovarian stimulation has been intense. Both of these static markers reflect the number of small follicles poised ready to be recruited by the supra-physiological concentrations of FSH used during IVF stimulation, making them potentially useful predictors of IVF response. AMH is a dimeric glycoprotein belonging to the transforming growth factor β (TGF β) super family (Josso, 1990). In the ovary, AMH expression is first observed in the proliferating granulosa cells in primary follicles, with the highest levels of expression in pre-antral and early antral follicles, with concentrations then falling in mural granulosa cells of large antral follicles (Weenen et al., 2004). Hence, it is thought that serum AMH concentrations are a reflection of the size of the growing cohort of small follicles (de Vet et al., 2002; Fanchin et 2007 Published by Reproductive Healthcare Ltd, Duck End Farm, Dry Drayton, Cambridge CB3 8DB, UK

2 al., 2003; Pigny et al., 2003), which in turn reflects the number of residual primordial follicles, or the ovarian reserve. AMH regulates follicular development by acting on two crucial stages of folliculogenesis. The first important function of AMH within the ovary is to inhibit the recruitment of quiescent primordial follicles into the growth phase (Durlinger r et al., 2001; Carlsson et al., 2006). The second important function of AMH is to attenuate the stimulatory effect of FSH on growing follicles (Durlinger r et al., 2001). Several investigators have observed a fall in serum AMH with increasing age (de Vet et al., 2002; Laven et al., 2004; Mulders et al., 2004; Tremellen et al., 2005), reflecting a drop in the number of growing follicles available for recruitment. This fall in AMH precedes the fall observed in more traditional markers of ovarian reserve such as FSH and inhibin B (Fanchin et al., 2003). In a clinical context, previous studies have linked serum AMH concentrations in the early follicular phase of a spontaneous cycle with numbers of oocytes retrieved during a subsequent IVF cycle (van Rooij et al., 2002; Muttukrishna et al., 2005; Penarrubia et al., 2005; Tremellen et al., 2005, Freour r et al., 2007) as well as pregnancy outcome (Hazout et al., 2004; van Rooij et al., 2006). Furthermore, it has been suggested that high serum AMH on day 3 is associated with superior clinical pregnancy rates (Penarrubia et al., 2005; Freour et al., 2007), in addition to being associated with greater numbers of mature oocytes and embryos (Hazout et al., 2004; van Rooij et al., 2006). However, while some researchers have confirmed the utility of early follicular phase serum AMH concentration as a predictor of oocyte number in an IVF cycle, others have not found it predictive of pregnancy outcome (Ficicioglu et al., 2006). Measurement of serum AMH around the time of oocyte retrieval has revealed positive correlations between AMH concentrations and follicular health (Fanchin et al. 2005a,b). Furthermore, Silberstein et al. (2006) reported that high serum AMH concentrations at the time of human chorionic gonadotrophin (HCG) administration were associated with a greater number of retrieved oocytes, better embryo score and superior implantation rate. It is not possible to compare directly AMH measurements from IVF cycles with those taken in the early follicular phase of a spontaneous cycle, as it has been well described that serum AMH concentrations fall significantly during gonadotrophin stimulation (Feyereisen et al., 2006). While both serum AMH concentrations taken during an IVF cycle and the early follicular phase may predict IVF outcomes, values taken prior to an IVF cycle are more likely to be clinically useful, as they give an insight into predicted IVF outcome remote from the index cycle of treatment. In this study, attention has therefore been focused on early follicular phase serum AMH concentration as a predictor of IVF outcome. Antral follicle count has also been used alone (Chang et al., 1998; Bancsi et al., 2002; Erdem et al., 2002; Loverro et al., 2003; Elter et al., 2005a,b; Muttukrishna et al., 2005) or in combination with FSH/inhibin B (Bancsi et al., 2003) to predict ovarian response during ovarian stimulation. Like AMH concentration, antral follicle count (AFC) appears to have relatively good power in predicting the number of oocytes retrieved during IVF. However, a recent meta-analysis of 17 studies examining the link between AFC and pregnancy outcome found that AFC counts were not predictive of pregnancy during IVF treatment (Hendriks et al., 2007). This study has concentrated on AMH as a marker of reproductive performance, as it appears to be the most robust clinical measure of the number of small antral follicles ready for ovarian stimulation recruitment during IVF. The aims of this study were three-fold. Firstly, it was desired to confirm the correlation between AMH and numbers of oocytes retrieved following ovarian stimulation. Secondly, the study investigated whether AMH was related to markers of oocyte quality; specifically fertilization rates, embryo quality and miscarriage rates. Finally, the study analysed whether AMH concentrations were predictive of pregnancy during IVF treatment. Materials and methods A total of 126 patients attending the clinical infertility service of the University of Adelaide (Repromed) from January 2005 to December 2005 were included in this retrospective observational study. Institutional Ethics Review Board approval to analyse the link between markers of ovarian reserve and IVF outcome had previously been obtained from the Women s and Children s Hospital, Adelaide, South Australia. The criteria for inclusion were: (i) first cycle of IVF treatment in patients who had undergone basal AMH and AFC measurements as part of their routine assessment; and (ii) no evidence of endocrinological disorders (normal testosterone, prolactin, thyroid stimulating hormone). Women with polycystic ovarian syndrome (PCOS) were excluded, as it is known that PCOS is associated with significantly elevated basal AMH and AFC concentrations (Pigny et al., 2003; Laven et al., 2004). Women with basal FSH concentrations exceeding 10 IU/l were excluded from the study, as these patients clearly have reduced ovarian reserve and chance of pregnancy on traditional diagnostic criteria (Toner, 1993; Sharif et al., 1998). All patients had regular menses and normal ovulatory function according to mid luteal progesterone concentration. Serum FSH, LH, oestradiol, AMH and AFC measurements were taken on days 3 5 of a spontaneous menstrual cycle within 12 months of the index IVF cycle. Serum samples were separated within 1 h of collection and frozen at 20 C until assayed. All hormone measurements, with the exception of AMH (see below), were conducted using the automated ADVIA Centaur chemiluminescent immunoassay system obtained from Bayer Australia (Pymble, NSW, Australia). The FSH assay, calibrated against the WHO second International Standard, IS 94/632, has an analytical sensitivity of 0.3 IU/l with inter-assay and intraassay coefficients of variation of <4.0%. Two sonographers carried out the AFC measurements on days 3 5 of the menstrual cycle. A 7.5 MHz trans-vaginal probe and GE scanner were used in all examinations. Follicles between 2 and 5 mm in diameter were recorded in both ovaries and combined to give a total AFC. The IVF protocol used in this study was a traditional long down-regulation cycle. Briefly, this consisted of commencing pituitary down-regulation in the midluteal phase of the preceding menstrual cycle by the use of nafarelin acetate (Synarel- Pharmacia, Rydalmere, NSW, Australia). After a minimum of 14 days down-regulation, recombinant FSH (Puregon; Organon, Lane Cove, NSW or Gonal F; Serono, Frenchs Forrest, NSW, Australia) was commenced according to an agerelated formula (Borini and Dal Prato, 2005). Women under 603

3 years were started on 150 IU/day, women aged on IU/day and women aged 40 years or older commenced on 300 IU/day. After 7 days of gonadotrophin stimulation, a pelvic ultrasound and oestradiol measurement was taken to assess ovarian response. The dose of gonadotrophin was then adjusted according to the response. The patient was scheduled for a trigger injection of 5000 IU of HCG (Profasi; Serono) once two or more lead follicles had reached mm in size. Trans-vaginal oocyte retrieval was performed 36 h later under light sedation. Routine insemination of oocytes using density gradient purified spermatozoa was employed if good quality spermatozoa were available. If the male partner s normal sperm morphology was <15% (normal range for normal morphology >20%) or there were fewer than motile spermatozoa available, intracytoplasmic sperm injection (ICSI) was used to achieve fertilization. Light microscope examination was used to verify fertilization after 16 h. Embryos were graded excellent (grade 1), good (grade 2), poor (grade 3) and very poor (grade 4) according to the usual morphological criteria for grading cleavage stage embryos (cell number, fragmentation, nuclear appearance). Grade 1 and 2 embryos were preferentially transferred and only grade 1 and 2 embryos were cryopreserved. Embryo transfer was performed under ultrasound guidance between 2 and 5 days after oocyte retrieval. Luteal support was provided with vaginal progesterone (Crinone; Serono) supplemented with one 500 IU dose of HCG on day 6 postretrieval. In this study, clinical pregnancy was defined as the ultrasound observation of fetal heart movements at 7 8 weeks of gestation. AMH assay Serum AMH concentrations were measured using the new high-sensitivity Immunotech immunoenzymetric assay (Beckman Coulter, Marseille, France). The analytical sensitivity of this assay is 0.7 pmol/l. Inter-assay and intraassay coefficients of variation were 14.2 and 12.3% respectively. The results of this study are expressed in pmol/l. Conversion of the commonly reported AMH measurement of ng/ml to pmol/l can be made using the multiplier 7.14 (i.e. 1 ng/ml AMH = 7.14 pmol/l AMH). Previous work within the authors unit (Tremellen et al., 2005) has shown that an early follicular phase AMH of 8 pmol/l was the best cut-off in predicting a poor ovarian response to ovarian stimulation. Since this publication, the manufacturer of the AMH assay has altered the amplification system in order to improve sensitivity (Fanchin et al., 2005b; Freour et al., 2007). Therefore, the current AMH concentrations do not correspond directly to a previous publication that used the standard sensitivity AMH assay. However, comparative measurements of frozen thawed blood samples using both assays suggest that the high sensitive assay results are double the previous assay results (data not shown). Data analysis and statistics The Mann Whitney U-test was used for non-parametric data and Student s t-test was used for all normally distributed data. Results are expressed as mean ± SEM. A P-value <0.05 was considered statistically significant. Receiver operating characteristic (ROC) curves were plotted to determine the diagnostic value of different parameters of ovarian reserve. Finally, a multiple regression analysis was used to generate a predictive equation from basal markers AMH and AFC. All data were analysed using Sigma Stat 2.03 or SAS 9.1. Results A strong Spearman s correlation coefficient (P < ) between serum AMH concentration and oocyte number confirmed previous observations that baseline follicularphase serum AMH is a good predictor of oocyte response in IVF. A serum AMH concentration of 14 pmol/l was found to offer the best sensitivity (73%) and specificity (73%) of predicting poor response to ovarian stimulation according to ROC analysis. Therefore, the study group was then divided into two cohorts based on their serum AMH concentration (low AMH, 14 pmol/l; high AMH, >14 pmol/l). This result is consistent with a previous publication linking AMH values below 16 pmol/l (equivalent to an AMH concentration <8 pmol/l using the standard sensitivity assay) with a low oocyte response during IVF (Tremellen et al., 2005). Using the above-mentioned AMH classification, the study cohort consisted of a total of 54 patients in the low AMH group and 72 patients in the high AMH group. Table 1 depicts the patient baseline characteristics of the two groups. There were no significant differences in mean age, body mass index (BMI) or FSH concentrations between the two groups, or in their causes of infertility. The mean AFC in the low AMH group was 4.3, compared with 12 in the high AMH group (P < 0.001), as would be expected. Patients in the high AMH group had on average 12 oocytes collected after ovarian stimulation, more than double the number from patients with lower AMH concentrations (P < 0.001), despite these patients receiving substantially more FSH (P < 0.001) (Table 2). There was no significant difference in the proportion of low and high AMH patients having ICSI to achieve fertilization (53.7 versus 72.2% respectively), as would be expected given that the rate of male infertility did not differ significantly between the two AMH groups. Fertilization rates in the low AMH group were significantly inferior to those in the high AMH group, irrespective of whether routine IVF (41 versus 52%, P = 0.043) or ICSI (52 versus 66%, P = 0.006) (Table 2) was performed. These results suggest that low AMH not only predicts low oocyte number, but also poor oocyte quality, as signified by the inferior fertilization rates despite good quality spermatozoa (IVF cohort) or mechanical assistance (ICSI cohort). Day 2 cleavage-stage embryos were graded 1 4 by the embryology team according to established morphological criteria, with good quality embryos (grades 1 2) being preferentially transferred rather than those of poorer quality (grades 3 4). The quality of cleaved embryos was not significantly related to the patient s AMH status (65% grade 1 2 embryos in low AMH group versus 62% in high AMH group; Table 2). A total of 115 embryos were generated in the low AMH group and 447 in the high AMH group. On average, a patient in the high AMH group generated significantly more embryos than a low AMH patient (6.2 versus 2.1 embryos respectively (P <

4 0.001). This difference reflects the higher number of oocytes collected and superior fertilization rates in the high AMH group. Consequently, the high AMH patients had on average 2.4 embryos available for cryopreservation per cycle, while the low AMH group had only 0.6 embryos (P <0.001; Table 2). In the low AMH group, 28% (15/54) of the patients did not reach embryo transfer (14 due to no fertilization and one because only one grade 4 embryo was available), while only 4% (3/72) of the high AMH group had no embryo transfer, all due to ovarian hyperstimulation syndrome (P = 0.001) (Table 3). There was no significant difference in clinical pregnancy rates between the low and high AMH groups when analysed only for fresh embryo transfer (25.6 versus 30.4% respectively; Table 3). However, the high AMH group s superior oocyte number and fertilization rate led to a significantly greater number of embryos being available for later transfer in frozen thawed embryo transfer cycles. Although there was no difference in pregnancy rates between the two patient groups using frozen embryos (22.2% low AMH versus 24.3% high AMH group), when the resultant pregnancies from these cryopreserved embryos were taken into account, the cumulative clinical pregnancy rate (Table 4) resulting from fresh plus frozen embryos from one stimulated cycle of IVF was significantly higher in the high AMH group (42 versus 22%; P < 0.036). This number is in fact likely to be an underestimate, as the high AMH group still has 119 embryos cryopreserved for later transfer for a second attempt to achieve a pregnancy, whereas the low AMH group has only 18 remaining cryopreserved embryos. If future pregnancies are predicted based on the observed pregnancy rate from thawed embryos (24%), it would be expected that once all embryos are transferred, either fresh or frozen, there would be a 73.6% pregnancy rate per cycle commenced in the high AMH group compared with only 27.8% in the low AMH group (Table 4). This is a very meaningful observation, as the principal financial and personal cost of an IVF cycle is the initial stimulation procedure. Women with high AMH are at least twice as likely to become pregnant from the use of all resultant embryos of one IVF cycle, compared with those in the low AMH group. Finally, the proportion of women pregnant following a fresh embryo transfer who then went on to miscarry was significantly higher in the low AMH group compared with the high AMH group (33 versus 5%). This suggests that the embryos transferred in the low AMH group may be of inferior quality. Predictive value of AMH and AFC The diagnostic value of an early follicular phase AFC and serum AMH measurements in predicting a positive outcome in terms of oocyte number and pregnancy outcome was assessed using ROC curve analysis. A total of four or fewer oocytes were assigned a negative outcome, whereas five or more oocytes were assigned a positive outcome. The area under the curve (AUC), using AMH score as the test variable to predict positive or negative oocyte outcome, was calculated to be (Figure 1), while for AFC it was (Figure 2). This indicates that AMH and AFC are both comparable excellent predictors of oocyte yield in an IVF cycle. In order to develop a mathematical model for the number of oocytes retrieved during an IVF cycle, a multiple regression model was applied. This found that AMH and AFC were both significant predictors of oocyte yield, while age and day 3 5 FSH were not. This was not surprising, given that women with elevated basal FSH concentrations were excluded from the study. When considered separately, early follicular phase serum AMH concentrations (P < 0.001, adjusted r ) was found to be a slightly superior predictor of oocyte yield in IVF, compared with AFC (P < 0.001, adjusted r ). The predictive formula for oocyte yield during IVF is as follows: oocyte number = AFC AMH. As the predictive ability of this equation did not improve by adding maternal age and basal FSH concentrations, those two parameters were not included in the formula. ROC calculations were also undertaken to assess the usefulness of serum AMH measurements as a predictor of pregnancy resulting from either a fresh embryo transfer (AUC of 0.603) or combined fresh and thawed embryo transfers (AUC of 0.642; Figure 1). When AFC was used as a predictor of cumulative pregnancy resulting from combined fresh and thawed embryo transfers, the AUC was (Figure 2). These values indicate that AMH as well as AFC have only a moderate predictive value for pregnancy. Table 1. Baseline characteristics of the patients in the low anti-müllerian hormone (AMH) and high AMH study groups. Characteristic Low AMH group High AMH group P-value (n = 54) (n = 72) Serum AMH (pmol/l) 14 >14 - Age in years (range) 36.6 ± 0.6 (19 40) 34.3 ± 0.8 (23 41) NS Body mass index 26.6 ± ± 0.6 NS Day 3 FSH (IU/l) 6.7 ± ± 0.2 NS Antral follicle count 4.3 ± ± 1.0 <0.001 Values are mean ± SEM. NS = not statistically significant. 605

5 Table 2. IVF treatment parameters in the low anti-müllerian hormone (AMH; 14 pmol/l) and high AMH (>14 pmol/l) study groups. Parameter Low AMH group High AMH group P-value (n = 54) (n = 72) Mean number of oocytes retrieved 5.0 ± ± 0.8 <0.001 Total gonadotrophin administered (IU) 3420 ± ± 104 <0.001 Peak oestradiol (nmol/l) 3.9 ± ± 0.5 <0.001 Fertilization rate (%) ICSI 77/146 (52.7) 294/447 (65.8) IVF 38/93 (40.9) 153/293 (52.2) Embryo quality (ICSI + IVF) (%) Good quality (G1 and G2) 75/115 (65.2) 277/447 (62.0) NS Poor quality (G3 and G4) 40/115 (34.8) 170/447 (38.0) NS Mean number of embryos created per patient <0.001 Mean number of embryos frozen per patient <0.001 Number of embryos used for FET Number of embryos remaining for future FET FET = frozen thawed embryo transfer; G1 G4 = grades 1 4; ICSI = intracytoplasmic sperm injection; NS = not statistically significant. Table 3. Pregnancy outcome from fresh embryo transfer in the low anti-müllerian hormone (AMH; 14 pmol/l) and high AMH (>14 pmol/l) study groups. Parameter Low AMH group High AMH group P-value (n = 54) (n = 72) No. of cycles commenced No. of patients with ET (%) 39 (72.2) 69 (95.8) <0.001 Mean no. of embryos transferred NS Positive β-hcg/transfer (%) 15/39 (38.4) 22/69 (31.0) NS Miscarriages (%) 5/15 (33.3) 1/22 (4.5) Clinical pregnancy/transfer a (%) 10/39 (25.6) 21/69 (30.4) NS a Clinical pregnancy defined as fetal heart beat at 8 week ultrasound. ET = embryo transfer; β -HCG = β-human chorionic gonadotrophin; NS = not statistically significant. Table 4. Cumulative pregnancy outcome (fresh and thawed embryo transfers) in the low anti-müllerian hormone (AMH; 14 pmol/l) and high AMH (>14 pmol/l) study groups. Parameter Low AMH group High AMH group (n = 54) (n = 72) P-value Positive β-hcg per IVF cycle commenced (%) 17/54 (31.5) 36/72 (50.0) NS Positive β-hcg per embryo transfer cycle (%) 17/48 (35.4) 36/106 (34.0) NS Miscarriages per cycle transfer (%) 5/48 (10.4) 6/106 (5.7) NS Miscarriages per positive β-hcg (%) 5/17 (29.4) 6/36 (16.7) NS Clinical pregnancy per cycle commenced (%) 12/54 (22.2) 30/72 (41.7) Projected cumulative pregnancies per cycle commenced a (%) 15/54 (27.8) 53/72(73.6) a Projected pregnancy = cumulative clinical pregnancy to date plus expected pregnancy from remaining frozen embryos. β-hcg = β-human chorionic gonadotrophin; NS = not statistically significant. 606

6 Figure 1. Receiver operating characteristic (ROC) curves analysing the value of basal serum AMH in detecting poor responders (ROC AUC = 0.834) and cumulative clinical pregnancy (ROC AUC = 0.642). Continuous line showing ROC for poor responders ( 4 oocytes). Dotted line showing ROC for cumulative clinical pregnancy. Figure 2. Receiver operating characteristic (ROC) curves analysing the value of antral follicle count (AFC) in detecting poor responders (ROC AUC = 0.849) and cumulative clinical pregnancy (ROC AUC = 0.616). Continuous line showing ROC for poor responders ( 4 oocytes). Dotted line showing ROC for cumulative clinical pregnancy. 607

7 608 Discussion In summary, the results of this study clearly indicate that both AMH and AFC are good predictors of oocyte yield during IVF. Furthermore, the results suggest that serum AMH is predictive of oocyte quality, as evident by the lower fertilization rates and increased pregnancy wastage in the fresh embryo transfer cycles of women with low serum AMH. So far as is known, this study is the first to describe the link between early follicular phase serum AMH concentrations and oocyte quality. Finally, the results strengthen the body of evidence supporting the use of serum AMH measurements for the prediction of pregnancy outcome during IVF treatment. The results of the current study provide multiple lines of evidence that patients with high baseline serum AMH concentrations have higher reproductive potential, compared with low AMH patients. These high AMH patients yielded more oocytes, had higher fertilization rates, generated more embryos, and had a lower incidence of miscarriage during fresh transfers, ultimately culminating in an approximate doubling in pregnancy rate per IVF cycle commenced. This is particularly noteworthy in light of the fact that these patients were otherwise indistinguishable from the lower AMH group in terms of routine baseline characteristics such as age and FSH concentrations. Furthermore, low AMH is not only associated with lower oocyte yield during IVF treatment, but also diminished oocyte quality as assessed by fertilization rates. Previous researchers had linked severely diminished ovarian reserve, as determined by elevated FSH concentrations, with reduced oocyte fertilization (Bancsi et al., 2003; van Rooij et al., 2003). Recently, Ebner et al. (2006) have related the morphological appearance of dark granular cytoplasm in oocytes, a reported feature of poor quality oocytes, to low follicular phase serum AMH concentrations. One previous study has suggested that day 3 serum AMH concentration is a strong predictor of subsequent oocyte/ embryo yield and clinical pregnancy rates during IVF treatment (Hazout et al., 2004). However, another study reported that while day 3 serum AMH was a sensitive indicator of oocyte yield during IVF, it did not predict pregnancy success (Ficicioglu et al., 2006). These conflicting results may be explained by a lack of statistical power in the smaller (Ficicioglu) study. In this study, there was a trend towards a superior pregnancy rate in the high AMH group compared with the low AMH group. It is unclear if these researchers included cumulative pregnancies resulting from thawed embryo transfers in their study analysis. As their high AMH group had at least six more oocytes than the low AMH group, it is likely that several more cryopreserved embryos would have been created in the high AMH group. Inclusion of these pregnancies may make the observed non-significant trend become statistically significant. This study is the first to observe that low serum AMH predicts poor fertilization, irrespective of fertilization technique (IVF, ICSI) or sperm quality. It is postulated that these poor fertilization rates are due to oocyte cytoplasmic deficiencies such as mitochondrial dysfunction in the low AMH group (Eichenlaub-Ritter et al., 2004). When oocytes from the low AMH group were subjected to IVF (i.e. good quality spermatozoa), the fertilization rates were inferior to those obtained with oocytes from the higher AMH group (41 versus 52%, P = 0.043). The observed difference in fertilization could be due to poor oocyte cytoplasmic competence, defective cumulus function or sperm zona interaction in the low AMH oocytes. Interestingly, when oocytes from the low AMH group were subjected to ICSI, effectively nullifying any male contributions and zona/cumulus barriers to fertilization failure, the fertilization rates were still inferior to the high AMH group (53 versus 66%, P = 0.006). This suggests that intrinsic deficiencies in the oocyte cytoplasm are principally responsible for the diminished fertilization rate in oocytes retrieved from women with low AMH. Based on the observation that fertilization rates were apparently better with ICSI than IVF (52 versus 40%) in the low AMH group, it may be worth considering using ICSI in all women with low AMH concentrations, irrespective of their partner s semen quality. This, of course, would need further verification by larger studies before full clinical implementation. The hypothesis that low AMH may be associated with poor oocyte quality is supported not only by diminished fertilization rates, by also by post-fertilization embryonic wastage. Whilst the biochemical pregnancy rate was comparable between the two groups, it was striking to note that 33.3% of women in the low AMH group experienced miscarriage following their fresh embryo transfer, while only 4.5% miscarried in the high AMH group. This observation again provides evidence to support the hypothesis that low AMH also reflects poor oocyte quality, and supports the notion that reproductive ageing manifests as loss of oocyte numbers concomitant with deterioration of oocyte quality. To date, it is unclear precisely in what manner oocytes deteriorate with reproductive ageing, although it is apparent that oocyte mitochondrial activity is impeded, subsequently adversely affecting embryonic development (Wilding et al., 2001, 2005; Eichenlaub-Ritter et al., 2004). In this study, it is important to note that cumulative pregnancies have been significantly underestimated in the high AMH group. The number of embryos cryopreserved in the high AMH group after the first cycle was 174. At the time of preparation of this manuscript only 55 of these embryos had been thawed and transferred, leaving 119 cryopreserved embryos for future use (Table 2). If it is considered that 80% of these cryopreserved embryos will survive thawing, 95 embryos will remain for future transfer. The results suggest that the chances of a thawed embryo producing a clinical pregnancy are about 24%. Therefore, it would be expected that these 95 thawed embryos would have the potential of creating an additional 23 pregnancies for the high AMH group, bringing the cumulative total pregnancies to 53. This equates to a cumulative pregnancy rate per cycle of up to 74% in the high AMH group. When considering similar calculations for the 18 cryopreserved embryos remaining in the low AMH group, it is predicted that there will be only an additional three pregnancies and a resulting final cumulative pregnancy rate of 28%. With the current trend of delaying pregnancy until later in life, diminished ovarian reserve is more commonly encountered as a significant problem during IVF treatment. Ovarian reserve tests such as AMH and AFC should not be used to exclude poor prognosis patients from IVF treatment, as reasonable

8 pregnancy rates can still be achieved even when these static markers of ovarian reserve are abnormal. However, AMH and AFC measurements are still very useful in an IVF setting for two principal reasons. Firstly, they allow for adjustment of the starting dose of gonadotrophin stimulation. Currently, most IVF clinicians determine starting dose of gonadotrophin in the first cycle of IVF based principally on age and follicular phase FSH concentrations (Borini and Dal Prato, 2005). The regression analysis showed that AMH and AFC were superior predictors of oocyte yield compared with age. It is difficult to find associations between basal FSH concentration and oocyte yield in this study, as patients with high FSH (>10 IU/l) were excluded, thereby weakening any association. If a patient s AMH concentration predicts that their oocyte yield will be less than ideal, the clinician then has the opportunity to increase the starting dose of gonadotrophin, thereby minimizing the risk of a disappointing outcome in the first cycle. Obviously adjustment of starting dose will principally be of use in young women who would not normally be commenced on a maximal dose of gonadotrophins. Secondly, measurement of AMH prior to IVF treatment allows the clinician to give the couple a realistic expectation of their chances of pregnancy from one cycle of IVF. Creating realistic expectations is very important in maintaining patient satisfaction. Personal experience suggests that when a couple are well prepared for a possible low embryo yield prior to their first IVF cycle, they are less likely to abandon IVF treatment after only one cycle because of unexpected disappointment. Measurement of AMH as a marker of ovarian reserve has many advantages over other static markers. Serum AMH concentrations begin to fall well before serum FSH concentrations become abnormal, improving test sensitivity. Secondly, unlike FSH, serum AMH concentrations do not fluctuate significantly during the menstrual cycle (Hehenkamp et al., 2006), making ovarian reserve assessment possible on any day that the patient presents to the clinic. Finally, unlike AFC measurements, serum AMH assays are not observerdependant, resulting in less test variability. Since the results suggest that AMH can predict pregnancy outcome during IVF treatment, measurement of AMH is likely to be of clinical benefit to all women under 37 years of age if taken prior to initiating IVF treatment. At the present time, one of the principal impediments to the more widespread use of AMH to predict IVF outcome is the lack of standardization of AMH assay results. There is no international reference standard for AMH, which has resulted in widely variable AMH measurements made by the various AMH assays, even when using the same clinical sample (Fanchin, 2005b; Freour et al., 2007). For example, AMH measured on the Immunotech/ Beckman Coulter enzymelinked immunosorbent assay (ELISA) are 4.6-fold higher than that measured by the DSL kit (Oxford Bio-Innovations, UK). Similarly, AMH measurements made using the ultra-sensitive version 2 of Immunotech s AMH ELISA (marketed from late 2005 onwards) are double those of the standard sensitivity version 1 assay (pre-2005). Laboratories must keep this factor in mind when using published data to generate their own normal ranges. It is advisable for all IVF laboratories to develop their own method-appropriate reference ranges and cut-offs to avoid incorrect clinical decisions guided by erroneous AHM determinations. Acknowledgements We would like to thank Thomas Sullivan of the Department of Public Health, University of Adelaide, for his contribution on the statistical analysis of the data. Dharmawijaya Lekamge was supported by an International Post-Graduate Research Scholarship (IPRS), Adelaide University Scholarship and Colin Mathews Research Grant (2005) from the Research Centre for Reproductive Health, University of Adelaide. Finally we would like to thank Lesley Ritter, Samantha Schulz and the staff at Repromed for their friendly assistance. References Bancsi LF, Broekmans FJ, Mol BW et al Performance of basal follicle-stimulating hormone in the prediction of poor ovarian response and failure to become pregnant after in vitro fertilization: a meta-analysis. Fertility and Sterility 79, Bancsi LF, Broekmans FJ, Eijkemans MJ et al Predictors of poor ovarian response in in vitro fertilization: a prospective study comparing basal markers of ovarian reserve. Fertility and Sterility 77, Borini A, Dal Prato L 2005 Tailoring FSH and LH administration to individual patients. Reproductive BioMedicine Online 11, Carlsson IB, Scott JE, Visser JA et al Anti-Müllerian hormone inhibits initiation of growth of human primordial ovarian follicles in vitro. Human Reproduction 21, Chang MY, Chiang CH, Chiu TH et al The antral follicle count predicts the outcome of pregnancy in a controlled ovarian hyperstimulation/intrauterine insemination program. Journal of Assisted Reproduction and Genetics 15, Check JH, Peymer M, Lurie D 1998 Effect of age on pregnancy outcome without assisted reproductive technology in women with elevated early follicular phase serum follicle-stimulating hormone levels. Gynecologic and Obstetric Investigation 45, de Vet A, Laven JS, de Jong FH et al Antimullerian hormone serum levels: a putative marker for ovarian aging. Fertility and Sterility 77, Durlinger AL, Gruijters MJ, Kramer P et al Anti-Müllerian hormone attenuates the effects of FSH on follicle development in the mouse ovary. Endocrinology 142, Ebner T, Sommergruber M, Moser M et al Basal level of anti- Müllerian hormone is associated with oocyte quality in stimulated cycles. Human Reproduction 21, Eichenlaub-Ritter U, Vogt E, Yin H, Gosden R 2004 Spindles, mitochondria and redox potential in ageing oocytes. Reproductive BioMedicine Online 8, Elter K, Kavak ZN, Gokaslan H, Pekin T 2005a Antral follicle assessment after down-regulation may be a useful tool for predicting pregnancy loss in in vitro fertilization pregnancies. Gynecological Endocrinology 21, Elter K, Sismanoglu A, Durmusoglu F 2005b Intercycle variabilities of basal antral follicle count and ovarian volume in subfertile women and their relationship to reproductive aging: a prospective study. Gynecological Endocrinology 20, Erdem A, Erdem M, Biberoglu K et al Age-related changes in ovarian volume, antral follicle counts and basal FSH in women with normal reproductive health. Journal of Reproductive Medicine 47, Faddy MJ, Gosden RG 1995 A mathematical model of follicle dynamics in the human ovary. Human Reproduction 10, Fanchin R, Louafi N, Mendez Lozano DH et al. 2005a Per-follicle measurements indicate that anti-mullerian hormone secretion is modulated by the extent of follicular development and luteinization and may reflect qualitatively the ovarian follicular status. Fertility and Sterility 84, Fanchin R, Mendez Lozano DH, Louafi N et al. 2005b Dynamics of 609

9 610 serum anti-mullerian hormone levels during the luteal phase of controlled ovarian hyperstimulation. Human Reproduction 20, Fanchin R, Schonauer LM, Righini C et al Serum anti-mullerian hormone is more strongly related to ovarian follicular status than serum inhibin B, estradiol, FSH and LH on day 3. Human Reproduction 18, Feyereisen E, Mendez DH, Taieb J et al Anti-Müllerian hormone: clinical insights into a promising biomarker of the ovarian follicular status. Reproductive BioMedicine Online 12, Ficicioglu C, Kutlu T, Baglam E, Bakack Z 2006 Early follicular antimullerian hormone as an indicator of ovarian reserve. Fertility and Sterility 85, Freour T, Mirallie S, Bach-Ngohou K et al Measurement of serum anti-mullerian hormone by Beckman Coulter ELISA and DSL ELISA: comparison and relevance in assisted reproduction technology (ART). Clinica Chimica Acta 375, Hazout A, Bouchard P, Seifer DB et al Serum anti-mullerian hormone/mullerian-inhibiting substance appears to be a more discriminatory marker of assisted reproductive technology outcome than follicle-stimulating hormone, inhibin B, or estradiol. Fertility and Sterility 82, Hehenkamp WJ, Looman CW, Themman AP et al Anti- Müllerian hormone levels in the spontaneous menstrual cycle do not show substantial fluctuation. Journal of Clinical Endocrinology and Metabolism 91, Hendricks DJ, Kwee J et al Ultrasonography as a tool for the prediction of outcome in IVF patients: a comparative metaanalysis of ovarian volume and antral follicle count. Fertility and Sterility, advanced online publication Josso N 1990 Anti-Müllerian hormone: hormone or growth factor? Progress in Growth Factor Research 2, Laven JS, Mulders AG, Visser JA et al Anti-Müllerian hormone serum concentrations in normoovulatory and anovulatory women of reproductive age. Journal of Clinical Endocrinology and Metabolism 89, Loverro G, Nappi L, Mei L et al Evaluation of functional ovarian reserve in 60 patients. Reproductive BioMedicine Online 7, Mulders AG, Laven JS, Eijkemans MJ et al Changes in anti- Müllerian hormone serum concentrations over time suggest delayed ovarian ageing in normogonadotrophic anovulatory infertility. Human Reproduction 19, Muttukrishna S, McGarrigle H, Wakim R et al Antral follicle count, anti-müllerian hormone and inhibin B: predictors of ovarian response in assisted reproductive technology? British Journal of Obstetrics and Gynaecology 112, Penarrubia J, Fabregues F, Manau D et al Basal and stimulation day 5 anti-mullerian hormone serum concentrations as predictors of ovarian response and pregnancy in assisted reproductive technology cycles stimulated with gonadotropin-releasing hormone agonist gonadotropin treatment. Human Reproduction 20, Pigny P, Merlen E, Robert Y et al Elevated serum level of antimullerian hormone in patients with polycystic ovary syndrome: relationship to the ovarian follicle excess and to the follicular arrest. Journal of Clinical Endocrinology and Metabolism 88, Scott RT, Toner JP, Muasher SJ et al Follicle-stimulating hormone levels on cycle day 3 are predictive of in vitro fertilization outcome. Fertility and Sterility 51, Seifer DB, Lambert-Messerlian G, Hogan J et al Day 3 serum inhibin-b is predictive of assisted reproductive technologies outcome. Fertility and Sterility 67, Sharif K, Elgendy M, Lashen H, Afnan M Age and basal follicle stimulating hormone as predictors of in vitro fertilisation outcome. British Journal of Obstetrics and Gynaecology 105, Silberstein T, MacLaughlin DT, Shai I et al Mullerian inhibiting substance levels at the time of HCG administration in IVF cycles predict both ovarian reserve and embryo morphology. Human Reproduction 21, Toner JP 1993 The significance of elevated FSH for reproductive function. Bailliere s Clinical Obstetrics and Gynaecology 7, Toner JP, Philput CB, Jones GS, Muasher SJ 1991 Basal folliclestimulating hormone level is a better predictor of in vitro fertilization performance than age. Fertility and Sterility 55, Tremellen KP, Kolo M, Gilmore A, Lekamge DN 2005 Anti-mullerian hormone as a marker of ovarian reserve. Australian and New Zealand Journal of Obstetrics and Gynaecology 45, van Rooij IA, Broekmans FJ, Hunault CC et al Use of ovarian reserve tests for the prediction of ongoing pregnancy in couples with unexplained or mild male infertility. Reproductive BioMedicine Online 12, van Rooij IA, Bancsi LF, Broekmans FJ et al Women older than 40 years of age and those with elevated follicle-stimulating hormone levels differ in poor response rate and embryo quality in in-vitro fertilization. Fertility and Sterility 79, van Rooij IA, Broekmans FJ, te Velde ER et al Serum anti- Müllerian hormone levels: a novel measure of ovarian reserve. Human Reproduction 17, Weenen C, Laven JS, Von Bergh AR et al Anti-Müllerian hormone expression pattern in the human ovary: potential implications for initial and cyclic follicle recruitment. Molecular Human Reproduction 10, Wilding M, Di Matteo L, Dale B 2005 The maternal age effect: a hypothesis based on oxidative phosphorylation. Zygote 13, Wilding M, Dale B, Marino M, et al Mitochondrial aggregation patterns and activity in human oocytes and preimplantation embryos. Human Reproduction 16, Received 2 January 2007; refereed 31 January 2007; accepted 26 February 2007.