Ideal value of serum anti-Mullerian hormone as a predictor of ovarian reserve and outcome in assisted reproductive technology

INTRODUCTION

Accurate assessment of ovarian function remains one of the big challenges in fertility practice, and hence a new marker for prediction of ovarian reserve and assisted reproductive technology (ART) outcome has been on search. The present study finds out whether anti-Mullerian hormone (AMH) predicts ovarian reserve and ART outcome.

MATERIALS AND METHODS

A prospective observational study carried out between August 2009 and August 2011. A total of 84 women who underwent in vitro fertilization (IVF)/intracytoplasmic sperm injection (ICSI) were included in the study regardless of their age or reproductive history. This study received institutional review board approval from the ethical committee of our university before it was executed. Each patient gave informed consent, authorizing the examination.

Analysis

A total of 84 women were recruited initially. After evaluation, 17 women did not come back for management and were taken as lost to follow-up, six were cancelled due to poor response and all of them had poor antral follicular counts, one was cancelled due to immature oocytes, and one due to high luteinizing hormone (LH) hormone values. Hence, finally 59 women who completed the ART cycles were analyzed [Table 1].

Table 1 Comparison between serum AMH and cycle cancellation

The statistical analysis was done by Fischer's exact test and P - value 0.010 was statistically significant, showing that cycle cancellation rate was higher with lesser AMH value.

As shown in Table 2 and Figure 1, median number of mature oocytes was retrieved in 59 patients; three in Group I, six in Group II, and seven in Group III. The statistical analysis was done using nonparametric Kruskal-Wallis test and P - value was 0.149, which was not statistically significant. Hence, there was no significant difference in the median number of mature oocytes between Groups II and III.

Close

Figure 1

Showing the number of mature oocytes in each group. AMH: Anti-Mullerian hormone

Export to PPT

Table 2 AMH and median number of mature oocytes

As shown in Table 3 and Figure 2, embryos obtained in 59 patients showed that the median number of embryos obtained were two in Group I, four in Group II, and four in Group III. The statistical analysis was done using nonparametric Kruskal-Wallis test and P - value was 0.067, which was not statistically significant. Hence it was inferred that even though more number of oocytes were obtained with higher AMH (Group III), there was no significant difference in the embryos obtained in Groups II and III.

Close

Figure 2

Showing the number of embryos in each group

Export to PPT

Table 3 AMH and median number of embryos

As shown in Table 4, out of 59 patients, 21 (35%) became pregnant. None of the patients in Group I with AMH <0.7 ng/ml became pregnant. Out of 40 patients in Group II (AMH 0.7-3.5 ng/ml), 15 patients (37.5%) became pregnant; and out of 17 patients in Group III (AMH >3.5 ng/ ml), six patients (33.3%) became pregnant. Pregnancy rate was almost similar in Groups II and III, whereas it was nil in Group I. The statistical analysis was done using Fischer's exact test and P - value was 0.796, which was not statistically significant.

Table 4 AMH and pregnancy rate

As shown in Table 5, seven patients with AMH <0.7 ng/ ml (Group I) had median FSH of 8.9, 54 patients with AMH 0.7-3.5 (Group II) had median FSH of 6.8, and 23 patients with AMH >3.5 (Group III) had median FSH of 6.4. The interquartile range of FSH was 7.8-26.5 in Group I. Hence, it was inferred that if the AMH value was less than 0.7, FSH value was higher. Thus, AMH and FSH were inversely proportional. The statistical analysis was done by Kruskal-Wallis nonparametric test and the P - value was 0.002, which was statistically significant.

Table 5 Comparison between AMH and FSH

The ROC curve for FSH and AMH for predicting the number of oocytes was drawn. As shown in Figures 3 and 4, the area under the curve (AUC) for FSH was 0.591 and for AMH was 0.790. Hence, it was inferred that AMH was a better predictor of number of retrieved oocytes than FSH.

Close

Figure 3

FSH and prediction of less than four oocytes. FSH: Follicle-stimulating hormone

Export to PPT

Close

Figure 4

AMH and prediction of less than four oocytes

Export to PPT

DISCUSSION

Diminished ovarian reserve is a major cause of infertility. Hence, ovarian reserve testing is crucial in ART. Traditionally, basal serum FSH levels were widely used to assess ovarian reserve, but not considered highly accurate due to intercycle variations. Currently, serum AMH measurements are considered best markers of ovarian reserve as AMH is not affected by pregnancy, oral contraceptive pills, and gonadotropin stimulation.^[3]

In our study, we found that serum AMH declined with age; which is consistent with the findings by Lee et al.^[4] David et al.,^[5] found that the median AMH levels decreased steadily with an increase in age from 24 to 50 years. The predictive value of AMH exceeds the performance of ovarian reserve testing with age, FSH, inhibin B, and estradiol.^[6]

As AMH is of ovarian origin, reduction in number of preantral and antral follicles will result in AMH reduction. Various studies showed strong association between serum AMH and ovarian pool.^[7,8,9] This is consistent with our study which showed decline in number of antral follicles, as serum AMH decreased. The AFCs were significantly less when serum AMH was <0.7 ng/ml. Majumderet al.,^[10] reported that serum AMH and AFC were comparable predictors of the quantity as well as the quality of ovarian responsiveness to exogenous gonadotropins. Plante et al.,^[11] found out that active smoking was associated with decreased AMH, confirming the effect of smoking on antral follicles.

La Marca et al.,^[12] reported the AMH level of 0.7 ng/ml had a good sensitivity and specificity of identifying 75% of poor responders. In other studies,^[13,14] AMH >3.5 ng/ml predicted hyper response and OHSS. In our study, AMH range of 0.7-3.5 ng/ml had 83.3% sensitivity and 25% specificity for predicting the number of oocytes. Fouda et al.,^[15] found out in a study that AMH was a reliable predictor for cycle cancellation. Our study also inferred that cycle cancellation rate was higher with lesser AMH value.

Eldar-Geva et al.,^[16] reported that an additional increase in AMH levels were observed in polycystic ovary syndrome (PCOS) patients due to abnormal activity of granulosa cells, hyperandrogenism, and obesity. In our study, we found out that a significant number of follicles were obtained when AMH value increased.

Nakhuda et al.,^[17] suggested that AMH was correlated with number of oocytes retrieved; and peak estradiol and AMH appears most useful in prediction of gonadotropin sensitivity, allowing individualization of dosing protocol. Also; Fouda et al.,^[15] and Ebner et al.,^[18] claimed a positive association between AMH level, oocyte number, and quality. Similarly our study also found out that more number of oocytes were retrieved with higher AMH (>3.5 ng/ml) and few oocytes with lesser AMH value (<0.7 ng/ml); whereas, an optimal number of oocytes were obtained with AMH between 0.7 and 3.5 ng/ml.

In our study, even though more number of oocytes were retrieved with higher AMH (>3.5 ng/ml), there was no significant difference in the number of mature oocytes and top quality embryos between Group III (AMH >3.5 ng/ml) and Group II (AMH 0.7-3.5 ng/ml). But the total number of oocytes, mature oocytes retrieved, and top quality embryos were significantly less in Group I (AMH <0.7 ng/ml). Hence, the optimal range of AMH was 0.7-3.5 ng/ml for predicting ovarian reserve and optimal ovarian response to gonadotropin stimulation. La Marca etal.,^[12] also stated that normal level of AMH was between 0.7 and 3.5 ng/ml. The levels between 3.5 and 5 ng/ml and levels >5 ng/ ml showed borderline high fertility and high fertility, respectively.

Nakhuda et al.,^[19] reported AMH as a good predictor of OHSS and Tremellen et al.,^[20] reported less AMH levels with poor ovarian reserve. Our study showed that 61.1% of patients with AMH >3.5 ng/ml (Group III) had OHSS; all were of the mild variety. Only 10% in AMH range0.7-3.5 ng/ ml (Group II) and none of them with AMH <0.7 ng/ ml (Group I) had OHSS. Thus, serum AMH strongly predicted OHSS.

Few studies^[21,22] showed that higher AMH levels were associated with a greater number of mature oocytes, a greater number of embryos, and ultimately a higher pregnancy rate. Hazeout et al.,^[23] were the first to demonstrate that AMH was found to have great value in predicting outcome of ART cycles and greater prognostic value than age, serum FSH, inhibin B, or estradiol. Barad et al.,^[24] concluded that when AMH and FSH were compared, AMH was clearly superior in predicting IVF outcome; also adding FSH to AMH, did not improve results in predicting pregnancy.

In contrast, our study did not show significant association of pregnancy rate and serum AMH levels. Nelson et al.,^[21] similar to our study concluded that basal AMH had a very good correlation with the number of oocytes retrieved, but like basal FSH, did not seem to predict clinical pregnancy. Similarly some other studies^[15,17,23] also showed no significant difference in AMH level between pregnant and nonpregnant cases.

Gleicher et al.,^[25] compared the concordance between FSH and AMH and found that women with normal FSH and abnormal AMH had reduced oocyte yield; whereas, women with normal FSH and normal AMH had the best oocyte yield, showing that AMH was better marker than FSH for predicting ovarian response. We had noted, with AMH value less than 0.7 ng/ ml, FSH value was higher with an interquartile range of 7.8-26.5 mIU/ ml. Thus, AMH and FSH were inversely proportional.

In this study, ROC curves for FSH and AMH for predicting number of oocytes and pregnancy were drawn. The AUC for FSH was 0.591 and for AMH was 0.790 for predicting the number of oocytes and the AUC for FSH was 0.457 and for AMH was 0.556 for predicting pregnancy. AMH was better predictor of number of retrieved oocytes than FSH and both AMH and FSH were not good predictors of pregnancy. These findings were in accordance with those of earlier studies.^[15,25]

ROC curve analysis^[16] showed that estimation of AMH levels before IVF/ICSI was a fair test for discrimination between cancelled and completed cycles with highly significant value (AUC-0.747), but it was a poor test for discrimination between nonpregnant and pregnant cases (AUC-0.659). Our study could not establish a definite cutoff value to predict pregnancy due to overlapping values in pregnant and nonpregnant cases.

Broer et al.,^[26] performed a review of role of AMH in ART outcome and concluded that AMH was an excellent predictor of ovarian response to controlled ovarian stimulation, but could not predict pregnancy after ART.

CONCLUSION

Serum AMH accurately predicts ovarian reserve and oocyte retrieval number in ART. AMH is a better predictor of ovarian reserve than FSH. The optimal range of AMH to get good ovarian response is 0.7-3.5 ng/ ml. AMH predicts the development of OHSS; whereas, AMH like FSH is a poor predictor of pregnancy. Higher was the cancellation rate when AMH was very low.