Association of antimullerian hormone with polycystic ovarian syndrome phenotypes and pregnancy outcomes of in vitro fertilization cycles with fresh embryo transfer

Objective The current study was undertaken to investigate the relationship between antimullerian hormone (AMH) and polycystic ovarian syndrome (PCOS) phenotypes and to determine whether AMH is associated with pregnancy outcomes in infertile women undergoing their first in vitro fertilization (IVF) treatment. Methods We performed a retrospective cohort study of 2973 infertile women, including 418 women with PCOS undergoing their first IVF treatment at a private fertility center from January 2014 to March 2018. Women were stratified into three groups using cutoffs defined by the 25(th) and 75(th) percentiles of the serum AMH level: 746 women had AMH <= 2.25 ng/mL; 1486 women had AMH between 2.25 to 5.71 ng/mL; and 741 women had AMH > 5.71 ng/mL. Endocrine characteristics, PCOS phenotypes, stimulation outcomes, pregnancy outcomes were compared among these groups. When there were any statistical differences (P < 0.05) among the three groups, Bonferroni test was performed as post-hoc tests to determine where the statistical differences existed. To assess the relationships between AMH and pregnancy outcomes in total patients and PCOS patients, logistic regression analysis, adjusted for potential confounding covariates, were performed. Results Women with high AMH had greater prevalence of hyperandrogenism (HA), polycystic ovarian morphology (PCOM) and amenorrhea than women with low or average AMH. The clinical pregnancy rate were significantly higher in the high-AMH group compared with low- and average-AMH groups (69.9% vs. 58.8% and 64.7% respectively; P < 0.001). The live birth rate was significantly lower in women with AMH <= 2.25 ng/mL compared with average- and high-AMH groups (47.6% vs. 55.2 and 59.5% respectively; P < 0.001). However, after controlling for maternal age, oocyte yield, as well as other confounders, AMH was no longer associated with a higher live birth rate (aOR 1.037, 95% CI 0.853-1.261, P = 0.717; aOR 1.099, 95% CI 0.858-1.408, P = 0.455, respectively) and clinical pregnancy rate (aOR 1.064, 95% CI 0.834-1.359, P = 0.617; aOR 1.181, 95% CI 0.875-1.595, P = 0.276, respectively). Moreover, pregnancy outcomes did not differ in PCOS women according to AMH quartiles. Conclusion Increased AMH levels associated with PCOS severity and greater ovarian stimulation. However, AMH was not associated with clinical pregnancy rate and live birth rate after controlling for other confounders in women undergoing IVF. Thus, AMH should not be used to alter clinical decisions and exclude patients based on a low or even undetectable AMH value.