Low anti-mullerian hormone decreased clinical pregnancy and increased risk of poor ovarian response in women over 35 years of age

To the Editor: Infertility affects 15% to 20% of couples worldwide, and assisted reproductive technology (ART) is the most effective treatment strategy. In addition to female age, many other factors, such as antral follicle count (AFC), basal follicle-stimulating hormone (FSH) level, and anti-Mullerian hormone (AMH) level, collectively known as biomarkers for ovarian reserve tests (ORTs), are widely used to predict ovarian responses and assisted-pregnancy outcomes.[1] Among them, AMH and FSH levels were discordant in 20% to 43% of women undergoing ART treatment, and AMH was proven to be superior in predicting live birth rate (LBR).[2] AMH and AFC have long been considered essentially interchangeable and have good performance in predicting ovarian response, while few studies have examined their discordance.[1] Recently, Zhang et al[3] showed that 16.76% (203/1211) of individuals with discordant AMH and AFC levels, and with low AFCs (<7) and normal AMH (≥1.1 ng/mL) level, showed a lower clinical pregnancy rate but a higher incidence of poor ovarian response (POR) than those with normal AFCs and decreased AMH level. Still, there are some limitations, such as a relatively insufficient sample size, failure to comment on LBR, and additional intermediate outcomes with consideration of potential cycle confounders. To determine whether AMH or AFC is associated with pregnancy and neonatal outcomes after ART when the test results are discordant, and to provide clues for optimal fertility counseling, we performed a retrospective cohort study on 44,069 patients who underwent their first autologous oocyte ART cycles. This study was approved by the Ethics Committee of Reproductive Medicine of Shandong University. Written informed consent was obtained from all the participants. Based on the selection process [Supplementary Figure 1, https://links.lww.com/CM9/B100], with an adequate sample size for study design, 44,069 patients with their first autologous oocytes in vitro fertilization (IVF)/intra-cytoplasmic sperm injection (ICSI) cycles from July 2013 to January 2019 in the Reproductive Hospital Affiliated to Shandong University were selected. The participants were categorized into four cohorts: (A) AFC <7, AMH <1.1 ng/mL; (B) AFC <7, AMH ≥ 1.1 ng/mL; (C) AFC ≥7, AMH <1.1 ng/mL; (D) AFC ≥7, AMH ≥1.1 ng/mL.[3,4] The LBR was the primary endpoint, which was defined as the delivery of at least one live-born infant after 28 weeks of gestation. The secondary outcomes included the POR rate (regarded as the number of oocytes retrieved <4 with only conventional protocols included),[4] oocyte utilization rate (defined as the proportion of embryos suitable for transfer or cryopreservation among retrieved oocytes), clinical pregnancy rate, and pregnancy loss rate. SPSS version 26.0 (IBM Corp, NY, USA) was used for data analysis. Continuous variables were analyzed using variance analysis or Kruskal–Wallis H tests. Categorical variables were analyzed using the chi-squared or Fisher's exact test. To adjust for the influence of confounding factors, binary logistic regression was performed using odds ratios (ORs) and 95% confidence intervals (CI). Statistical significance was defined as a two-tailed P value <0.05. The baseline characteristics are shown in [Supplementary Table 1, https://links.lww.com/CM9/B100]. There were 13.03% women (n = 5740) with discordant AMH and AFC levels; of these, 3.24% (n = 1427) showed less AFC but normal AMH (cohort B), while 9.79% (n = 4313) had normal AFC but low AMH (cohort C). Compared with individuals in cohort C, patients in cohort B had lower body mass index (23.32 ± 3.35 kg/m2vs. 24.03 ± 3.60 kg/m2, P < 0.001) and baseline FSH (7.51 ± 2.45 IU/L vs. 8.14 ± 3.17 IU/L, P < 0.001), as well as higher testosterone levels (22.37 ± 10.96 nmol/L vs. 21.21 ± 11.46 nmol/L, P = 0.005). ART cycle characteristics were compared between cohorts [Supplementary Table 2, https://links.lww.com/CM9/B100]. More agonist suppression and antagonist suppression protocols were applied in cohort B than in cohort C, while no differences were observed in the total gonadotropin dosage, days of ovarian stimulation, and a number of oocytes retrieved. However, the POR rate in cohort B was significantly lower than in cohort C (21.50% vs. 25.05%, P = 0.008). Comparable numbers of similar fresh embryo transfer rates, cryopreserved embryos, and oocyte utilization rates were found between these two cohorts. A total of 63.95% of women underwent fresh embryo transfer, resulting in an LBR of 47.50% per transfer cycle [Supplementary Table 2, https://links.lww.com/CM9/B100]. No differences were observed between cohorts B and C in terms of LBR (39.67% vs. 37.82%, P > 0.05), clinical pregnancy rate (49.03% vs. 46.15%, P > 0.05), ectopic pregnancy rate, and pregnancy loss rate. Furthermore, maternal and neonatal complications were equivalent among cohorts [Supplementary Table 3, https://links.lww.com/CM9/B100]. To evaluate the impact of age, patients with discrepant AMH and AFC were stratified into two age subgroups: <35 and ≥35 years. Confounding factors, such as serum FSH level, cycle type, and timing and number of embryos transferred, were adjusted [Table 1]. For women aged <35 years, LBR (50.35% vs. 46.93%, P-adjusted = 0.073), clinical pregnancy rate (57.19% vs. 54.45%, P-adjusted = 0.135), pregnancy loss rate (11.96% vs. 13.80%, P-adjusted = 0.384), and POR (16.84% vs. 16.67%, P-adj = 0.310) were all insignificant, even after adjustment for confounding factors. For women aged ≥35 years, however, patients in cohort B showed a higher rate of clinical pregnancy (38.82% vs. 35.56%, P-adj = 0.013, OR: 1.349, 95% CI: 1.066–1.707) and a lower rate of POR (27.32% vs. 35.04%, P-adj = 0.001, OR: 0.696, 95% CI: 0.559–0.866) compared with those in cohort C, while the LBR was still similar (26.32% vs. 26.17%, P-adj = 0.322, OR: 1.140, 95% CI: 0.879–1.479). Table 1 - Outcomes of ovarian stimulation and pregnancy in patients with discordant AMH and AFC. <35 years ≥35 years Variables Cohort B (n = 782) Cohort C (n = 2267) P-adjusted Adjusted OR (95% CI) Cohort B (n = 645) Cohort C (n = 2046) P-adjusted Adjusted OR (95% CI) AMH(ng/mL) 2.22 ± 1.08 0.73 ± 0.26 – – 1.95 ± 0.84 0.67 ± 0.27 – – AFC 5.13 ± 1.14 10.48 ± 3.36 – – 5.06 ± 1.16 9.20 ± 2.36 – – Poor response rate 16.84 16.67 0.310∗ 1.132 (0.891–1.437) 27.32 35.04 0.001∗ 0.696 (0.559–0.866) Rate of owning at least two frozen embryos 38.11 38.82 0.220∗ 0.896 (0.752–1.068) 21.86 18.72 0.212∗ 1.159 (0.919–1.461) Clinical pregnancy rate 57.19 54.45 0.135† 1.163 (0.954–1.418) 38.82 35.56 0.013† 1.349 (1.066–1.707) Pregnancy loss rate among clinical pregnancy 11.96 13.80 0.384† 0.839 (0.564–1.247) 32.20 26.39 0.282† 1.247 (0.834–1.866) LBR 50.35 46.93 0.073† 1.197 (0.983–1.456) 26.32 26.17 0.322† 1.140 (0.879–1.479) Data are expressed as mean ± standard deviation or percentage. Cohort B: patients with AFC <7, AMH ≥ 1.1 ng/mL, cohort C: patients with AFC ≥7, AMH <1.1 ng/mL.∗Adjusted P value after correcting for age, BMI, serum FSH level, total testosterone, and cycle type through binary logistic regression.†Adjusted P value after correcting for age, BMI, serum FSH level, total testosterone, cycle type, timing of embryos transferred, and number of embryos transferred through binary logistic regression.AFC: Antral follicle count; AMH: Anti-Mullerian hormone; BMI: Body mass index; CI: Confidence interval; FSH: Follicle-stimulating hormone; LBR: Live birth rate; OR: Odd Ratio. In our study, a 13.03% discrepancy was reported in 44,069 infertile individuals. Women aged ≥35 years with normal AFC but low AMH had credible evidence to have a decreased clinical pregnancy rate and increased risk of POR compared to those with low AFC but normal AMH, implying that AMH is more relevant to outcomes of pregnancy and ovarian response when its levels are discordant with that of AFC in women with advanced reproductive age. This result contradicts with the conclusion made by Zhang et al,[3] but is more convincing due to the larger sample size used. Recently, some randomized controlled trials have suggested that AMH is a strong predictor of oocyte yield. Normal AMH was proven to be associated with a higher probability of supernumerary frozen embryos than lower AMH.[5] Nonetheless, in our study, the number of cryopreserved embryos was similar in women with discordant AMH and AFC levels, even after stratifying patients by age. A possible explanation is that the approximate difference in the number of oocytes retrieved between the inconsistent cohorts might be insufficient to result in significant changes in embryo count, which was also supported by the comparable oocyte utilization rate (38.44 ± 26.01 vs. 40.07 ± 27.68). Further studies evaluating how much of the variance in retrieval outcomes could be explained by AMH vs. AFC in a predictive model are warranted, particularly in women aged >35 years. In addition, several studies have shown that the higher the number of oocytes retrieved, the higher the probability of achieving a live birth after the utilization of all cryopreserved embryos.[6] Although a similar number of frozen embryos and LBRs in the first embryo transfer cycle was observed, the comparison of cumulative pregnancy outcomes per start cycle warrants further exploration. This study has a few limitations. First, baseline characteristics, such as age and basal FSH level, were not comparable among the different groups. Although a binary logistic regression analysis was performed to adjust for confounding factors, intrinsic drawbacks could not be completely avoided. Second, as many studies have shown substantial ethnic/race-related differences in ovarian reserve markers, the cutoff thresholds for AMH and AFC to define abnormal ovarian reserve in the Chinese population remain controversial.[7] Our conclusion was based on the thresholds set at AMH ≥1.1 ng/mL and AFC ≥7 for normal individuals. In summary, our results revealed a 13.03% discordance between AMH and AFC levels in 44,069 autologous IVF/ICSI cycles. With the definition of normal ORT set at AMH ≥1.1 ng/mL and AFC ≥7, low AMH was associated with decreased clinical pregnancy and increased rate of POR in women aged >35 years when it discords with AFC, yet young women (<35 years) with discrepant AMH and AFC levels had comparable pregnancy outcomes. Hence, for women aged ≥35 years who have discrepant AMH and AFC levels, AMH levels may have better directive significance in clinical practice and personalized counseling. Acknowledgements Authors are grateful to all the patients who participated in this study. Funding This work was supported by grants from the National Natural Science Foundation of China (No. 81701406), Shandong Science Fund for Distinguished Young Scholars (No. JQ201720), Taishan Scholars Program for Young Experts of Shandong Province (No. tsqn20161069) and Projects of Medical and Health Technology Development Program in Shandong Province (Nos. 202005010520, 202005010523). Conflicts of interest None.