Multicenter evaluation of the Access AMH antimullerian hormone assay for the prediction of antral follicle count and poor ovarian response to controlled ovarian stimulation

Objective: To evaluate a new fully automated antimullerian hormone (AMH) assay for prediction of poor ovarian response (POR) to ovarian stimulation defined as four or fewer oocytes retrieved. Design: Prospective cohort study. Setting: Thirteen private and academic fertility centers in the United Slates. Patients(s): A total of 178 women undergoing their first in vitro fertilization (IVF) cycle eligible for the study were consented and enrolled, with data available from 160 women for prediction of POR and 164 women for AMH correlation with antral follicle count (AFC). Intervention(s): None. Main Outcome Measure(s): Cutoff point for AMH that predicts POR. Correlation of AMH with AFC, and cutoff point for AMH that correlates with antral follicle count >15. Result(s): The mean AMH among the poor responders was 0.74 ng/mL, compared with 3.20 ng/mL for normal to high responders. The AMH cutoff at 90% specificity for predicting POR with the use of the receiver operating characteristic (ROC) curve was 0.93 ng/mL, with an associated sensitivity of 74.1%. For prediction of POR, ROC analysis showed that AMH (area under the ROC curve [AUC] = 0.929) was significantly better than FSH (AUC - 0.615; P< .0001). AMH was positively correlated with AFC (Spearman rho = 0.756). The AMH at 90% sensitivity for AFC >15 was 1.75, with specificity of 59.1%. Conclusion(s): A fully automated AMH assay can be a useful biomarker for predicting POR in IVF cycles. Because AMH cutoff points vary depending on the assay used, future studies should continue to calibrate test results to clinically important outcomes. (C) 2018 by American Society for Reproductive Medicine.