The relationship between anti-Müllerian hormone (AMH) levels and pregnancy outcomes in patients undergoing assisted reproductive techniques (ART)

Patients

Fifty women (n = 50), aged 20–45 years were recruited from Centre of Assisted Reproduction and Endocrinology (C.A.R.E) Clinic in Westville, Durban, South Africa who were undergoing IVF treatment. This study was approved by Ethical Committee of the Durban University of Technology (Project reference 128/16) and Research Committee, C.A.R.E. Clinic, Durban, South Africa. After approving the study by the research ethics committees, written informed consents were obtained from all the patients.

GnRH antagonist protocol

A gonadotrophin-releasing hormone (GnRH) antagonist protocol with recombinant FSH (GONAL-f, Merck Serono, Darmstadt, Germany) was used as downregulatory (Park et al., 2015). The second approach was followed by administering 0.25 mg/day Cetrotide (Merck Serono). When at least 3 or more follicles reach a diameter equal or above 17–18 mm, the endometrial thickness reached at least seven mm by ultrasound and E2 levels were about 1,500–1,800 pmol/L then Human chorionic gonadotropin (hCG) was administered. All patients received 5,000–10,000 IU hCG (Ovitrelle^®, Merck Serono). Oocyte retrieval was performed 36 h after the administration of the hCG. Conventional ICSI was performed according to previously described protocols.

Sample collection

Blood samples were collected every 3–4 days on commencement of the treatment. The blood samples were centrifuged at 3,000 rpm for 10 min using a Biofuge centrifuge (Biofuge Primo–Heraeus) to obtain the blood serum. AMH and FSH levels were recorded, upon the first visit. Estrogen(E₂) and LH levels were monitored throughout the program until a peak E₂ and LH level were reached.

Hormone assays

Gen II ELISA (Beckman Coulter Inc., USA, catalog number A79765/A79766, unmodified version). (Beckman Coulter, USA) kit was used to estimate hormone levels (FSH-Cat. No. 33520 Access hFSH reagent, 100 determinations, 2 × 50 tests); E2 (Cat. No. B84493 Access Sensitive Estradiol Reagent, 100 determinations) and AMH (Cat. No. B13127 Access AMH Reagent, 100 determinations, 2 × 50 tests) from the blood serum according to manufacturer’s instructions. Insemination and intra cytoplasmic sperm injection (ICSI), oocyte retrieval, culture, fertilization, embryo culture, and transfer were carried out as previously described by Gardner et al. (2001).

Inclusion criteria

The population of the study included female patients ranging between the ages of 20–45.

Exclusion criteria

Patients undergoing cancer therapy and patients on immune suppressant drugs were excluded from study.

Statistical analysis

The data were analysed using IBM SPSS software (Chicago, IL, USA). Pearson’s correlation was used to determine the direction, strength, and significance of the correlation between X and Y variables between the different semen parameters. A parametric multiple linear regression analysis was used to evaluate the relationship between AMH and other available endocrine markers. ROC curves were used to assess predictive value for E₂ and AMH and evaluating cut off values to optimise sensitivity and specificity. A p value of < 0.05 was considered statistically significant.

Institutional Review Board approval

This study was approved by Ethical Committee of the Durban University of Technology (Project reference 128/16) and Research Committee, C.A.R.E. Clinic, Durban, South Africa and was performed in accordance with the Helsinki Declaration of 1975 (as revised in 1983).

Pearson correlation

Pearson correlation coefficients were calculated to determine if any statistical significance exists between AMH on a quantitative scale and age, E₂ and FSH (Table 6). The Pearson Correlation co-efficient of 0.151 indicates that a very weak positive relationship existence between E₂ and AMH, which is not statistically significant (p = 0.341). Furthermore, Pearson correlation coefficient between the AMH and age had a coefficient of −0.028 thus showing no statistical significance p = 0.859 (p > 0.05). The Pearson Correlation between AMH and FSH produced a coefficient of −0.185 thus indicating no statistical significance p = 0.240 (p > 0.05). Pearson correlation coefficient showed no significant association between AMH and number of oocytes (p = 0.191), number of mature oocytes (p = 0.300) and number of oocytes fertilized (p = 0.146). The number of oocytes, mature oocytes and oocytes fertilized all presented a no statistically significant correlation with AMH (0.206, 0.164, and 0.228, respectively).

Logistic regression analysis

A logistic regression model was used to determine the possible predictor variables for the pregnancy outcome. The model was fitted to the data with the result of the pregnancy namely, “Positive” or “Negative” as the binary dependent variable and age, E₂ LH, Basal FSH, Basal AMH and number of oocytes fertilized as independent variables (Table 7). As shown in Table 7, LH has p = 0.042 (p < 0.05) and E₂ has p = 0.065, which is statistically not significant at a 5% level. The SPSS output for the model is given in Table 1 (Data S1) signifying that overall, 73.8% of the cases were correctly classified, while 5∕12 = 0.417 or 41.7% of the positives were correctly classified, and 86.7% of the negative cases were correctly classified.

Area under the curve

The ROC curves of the serum AMH concentrations and E₂ for the prediction of the clinical pregnancy are depicted in Fig. 2. The areas under the curves (AUC) for E₂ were 0.725 and for AMH (AUC = 0.497). E₂ is therefore a better single predictor of pregnancy outcome when compared to AMH. It has been shown that E₂ can better predict the number of oocytes obtained.

Correlation between basal AMH and E₂

Pearson analysis between E₂ and AMH presented a Pearson Correlation co-efficient of 0.151 with p = 0.341 (p < 0.05) which indicates that a weak significant relationship exists between E₂ and AMH. Most previous studies (Ramalho de Carvalho et al., 2012; Ubaldi et al., 2005) have shown a relationship between a raised basal E₂ level and a reduced ovarian response using different values to express elevated estrogen levels which replicated the findings in this study therefore showing that a low AMH can result in low estrogen levels. Also, it can be concluded that a poor AMH value results in a poor ovarian reserve indicating follicles produced will not be correlated to a raised estrogen level, therefore indicating poor follicle growth, thus reducing the number of oocytes produced. However, it was determined that poor response to stimulus in IVF, indicative of a lower ovarian reserve, is associated with declined baseline serum AMH concentrations (Van Rooij et al., 2004). Consequently, when women have regular ovarian reserve and decent retort, disappointment of IVF must look for additional infertility reasons, e.g., male specific issue i.e., Y chromosome microdeletion. Furthermore, this conclusion is reinforced by the data of woman undergoing IVF which indicated that male factor infertility resulted in an unsuccessful cycle. Although E2 levels in these cases were above those of controls, they are still within the range of 25–100 pg/ml (Sahmay et al., 2014), suggesting that E2 single-handedly is not capable of predicting the female reproductive potential.

Correlation between basal AMH and age

Pearson correlation between AMH and age (Table 6) presented a co-efficient of −0.028 thus displaying a weak, negative association with a p = 0.859 (p > 0.05). A stronger relationship between these two variables was expected as it is known that as age increases, AMH should decrease. This contrary association corresponds as reported by Van Rooij et al. (2004), where serum AMH levels decline with age in normal women with proven fertility. Also, it is suggested that serum AMH is identified as the improved endocrine marker to measure the reproductive capability in advanced age.

Correlation between basal AMH and FSH

Basal FSH is one of the primary endocrine markers presented into ART program. The Pearson correlation amid AMH and FSH (Table 6) had a coefficient of −0.185 thus displaying a weak, negative relationship and with a p = 0.240 (p > 0.05). This study specifies a negative correlation, i.e., the higher the FSH the higher the chances the patient can present with a poor ovarian reserve and early menopause. This was strategic on the awareness that these women will respond well to ovarian stimulation while the basal FSH level is lesser at the beginning of the cycle. The outcomes of this study revealed that woman who were poor respondents or had a reduced ovarian reserve had a poor outcome and frequent testing is of no worth. Women who had a history of high FSH level must undergo treatment without further delay. By postponing treatment for these patients can be detrimental as they get older and fast approaching menopause (Uzumcu & Zama, 2016).

Correlation between age and FSH and number of oocytes

The Pearson Correlation between FSH and age (Table 6) displayed no statistical significance, p = 0.583 (p > 0.05). For most of Pearson Correlation analysis, no significant relationships were found with most of the p-values, being greater than 0.05. This may be due to the small sample size used in this study of 42 patients. AMH being compared to age and number of oocytes showed a slightly negative correlation which is expected as it is shown in previous studies that AMH and number of oocytes decrease with maternal age (Van Rooij et al., 2004) and Gobikrushanth et al. (2018). This inverse relationship is in agreement by Van Rooij et al. (2004), who reported that serum AMH levels deteriorate with age in normal women with proven fertility. Additionally, serum AMH indicates the simplest endocrine marker to measure the age-related decline of reproductive competence. AMH levels, in our group who were high respondents were over 3.0 ng/ml, normal respondent over 1.0 ng/ml and low respondents found to be below 0.9 ng/ml. Oocytes were still recovered even with low AMH levels. Neither fertilization rate nor embryo quality can be assessed using basal AMH levels. This contrasts with the findings reported by Vaegter et al. (2017), where embryos had superior morphology and cleavage performance in patients with AMH levels >2.7 ng/ml as compared with patients with values below this threshold.

Correlation between AMH and the number of oocytes, number of mature oocytes and number of oocytes fertilized

Our second objective of the study was to examine if AMH levels affected oocyte quality. In this study, the Pearson Correlation test (Table 6) showed no significant relationship between AMH and number of oocytes (p = 0.191), several mature oocytes (p = 0.300) and number of oocytes fertilized (p = 0.146). The number of oocytes, mature oocytes and oocytes fertilized all showed a weak positive relationship to AMH (0.206, 0.164, and 0.228, respectively). These findings are in agreement with that reported by La Marca & Sunkara (2014), La Marca et al. (2010) and Dehghani, Tayebi & Asgharnia (2008), where mean amount of oocytes was lower in poor responding patients than in normal patients attending IVF programs. This therefore led to the inference that ovarian response can be regarded as a reflection of the ovarian reserve. The Chi-square test for Independence was done to determine whether there is an association between the number of oocytes collected and the AMH category (Table 8). A Chi-squared value of 21.246, degrees of freedom = 8, with a p = 0.007 was observed. There was a significant relationship between the numbers of oocytes collected versus AMH category (p < 0.05). The Chi-square test for Independence was performed to see whether there is an association between the number of oocytes fertilized and the AMH category (Table 9). A Chi-squared value of 18.5, degrees of freedom = 12, with a p = 0.10 was found. There was thus no statistically significant relationship between the numbers of embryo’s fertilized versus AMH category (p > 0.05). This is anticipated as AMH has been used an indicator of oocyte reserve in previous studies whereas the resulting fertilized or transferred embryo’s may be due to a chance process based on many various factors such as the quality of the oocyte and sperm (Yarde et al., 2013). Ebner et al. (2006), demonstrated that AMH serum levels were related with oocyte quality in stimulated cycles. The quality of the embryos was not assessed using baseline AMH which agrees with our findings. However, the fertilization rate was not correlated with the serum AMH which varied with the results of the present study.

AMH category and positive pregnancies

Embryo quality has been suggested to be of paramount importance to predict the occurrence of pregnancy after IVF. In a regression model E₂ has a p = 0.017 (p < 0.05) and LH has a p = 0.035 (p < 0.05). Both variables are significant and age and basal AMH play a role in the pregnancy outcome and the model is thus adjusted for these two variables

In this study, AMH value for predicting pregnancy outcomes does not exist because oocyte quality is not accounted for by ovarian reserve markers. As demonstrated in this study the clinical pregnancy rate for patients 20–24 years was 100%, 25–29 was 50%, 30–34 years was 18%, 35–39 years was 27% and 40–44 years was 25% (Fig. 1). Patients presenting with a low AMH did not vary from those women presenting with higher AMH concentrations in same age group. A positive pregnancy outcome was logged across all age groups regardless of the AMH level. These results advocate that low ovarian reserve is not correlated with low oocyte quality in patients and the prediction remains the similar despite low AMH concentrations. Kini et al. (2010) stated the role of AMH in foreseeing cumulative pregnancy outcome during IVF treatment. It was recognized that serum AMH concentration on day 6 of stimulation was suggestively higher in participants who resulted in an ongoing pregnancy in IVF compared to those who did not. Serum AMH is a suitable indicator of ovarian hyper-response. In a metanalysis study conducted by Yao et al. (2015) to evaluate role of serum AMH role in forecasting the pregnancy outcome in IVF/ICS, it was concluded that there is positive correlation between serum AMH and pregnancy. Nevertheless, association between serum AMH and non-pregnancy cannot be ruled out either.