Validation study of the Access AMH antimüllerian hormone assay for the prediction of poor ovarian response to controlled ovarian stimulation

Objective To evaluate the diagnostic performance of the antimullerian hormone (AMH) level determined using the Access AMH assay for predicting poor ovarian response (POR) defined as ≤4 oocytes retrieved, including the validation of the predefined AMH cutoff of 0.93 ng/mL in both serum and plasma. Design Prospective cohort study. Setting Fifteen private and academic fertility centers (14 in the United States and 1 in Canada). Patient(s) Women aged 21–45 years planning controlled ovarian stimulation for in vitro fertilization. Intervention(s) None. Main Outcome Measure(s) Number of oocytes retrieved, categorized as POR and normal-to-high ovarian response (non-POR). The correlation of AMH level and antral follicle count. Result(s) Data were available for 472 participants who completed the study (74 with POR and 398 non-POR). The mean AMH serum level among those with POR was 0.99 ng/mL (median 0.76 ng/mL) compared with 2.83 ng/mL (median 2.36 ng/mL) among the normal-to-high responders. For confirmation of the 0.93 ng/mL AMH level cutoff as a predictor of POR, a receiver operating characteristic analysis gave an area under the curve of 0.852, with corresponding sensitivity and specificity of 63.5% and 89.2%, respectively. The associated positive predictive value was 52.2% and the negative predictive value was 92.9%. The AMH plasma values demonstrated a strong correlation with AMH serum values with an r value = 0.9980. The previously established AMH cutoff of 1.77 ng/mL for antral follicle count >15 resulted in a sensitivity of 83.8% (95% confidence interval [CI] 77.7–88.5) and a specificity of 59.9% (95% CI 54.2–65.4). Conclusion(s) This study validated the previously established AMH cut-point for the prediction of POR. Because this cut-point may vary depending on the assay used, the specific AMH assay should be reported in the literature whenever possible.