Predictive Value of AMH, FSH and AFC for Determining Ovarian Response in Vietnamese Women Undergoing Assisted Reproductive Technologies: A Prospective Study

Abstract

Background: There are comparatively few data on the value of different ovarian response predictors in conjunction with a gonadotropin-releasing hormone (GnRH) antagonist controlled ovarian stimulation (COS) protocol. This study assessed the predictive value of AMH, FSH and AFC for determining poor and high ovarian responses in Vietnamese patients (n=820) undergoing GnRH antagonist protocol COS. Methods: Poor, normal and high response were defined as ≤ 3, 4-15 and >15 oocytes retrieved, respectively. AMH, FSH and AFC were assessed on cycle day 2. Cut-off predictive values were identified, and poor and high response models developed. Results: AMH had the highest accuracy for predicting both poor and high ovarian response (cut-off values ≤1.25 and >3.57 ng/mL, respectively) and was significantly better than AFC (cut-offs ≤5 and >12), and both AMH and AFC were significantly better than FSH (cut-offs >8.94 and ≤7.36 IU/mL). For prediction of poor ovarian response, a model including AMH+AFC (AUC 0.93, 95%CI 0.91, 0.96) was equivalent to one using AMH only (AUC 0.92, 95%CI 0.90, 0.95; p=0.131), and both were better than AFC alone (AUC 0.89, 95%CI 0.86, 0.92; p<0.001). For high ovarian response, AMH+AFC (AUC 0.90, 95%CI 0.88, 0.92) was significantly better than AMH alone (AUC 0.89, 95%CI 0.87, 0.91; p=0.03), and AMH+AFC and AMH were better than AFC alone (AUC 0.86, 95%CI 0.83, 0.89; p<0.001). Conclusions: In Vietnamese women undergoing GnRH antagonist COS, the best single biomarker was AMH, while a model including AMH+AFC had the highest predictive value for high ovarian responses.