Can an anti-mullerain hormone cutoff level predict ovarian response among women with poor ovarian reserve?

Abstract

OBJECTIVE: The purpose of our study was to determine if the addition of AMH levels can predict ovarian response in IVF cycles among known poor responders undergoing maximal stimulation protocols. AMH, a marker of ovarian reserve that remains relatively consistent throughout the cycle, has been associated with the number of small antral follicles. Recently, levels of ≤1.26ng/ml have been associated with diminished ovarian reserve. However, an AMH level that predicts ovarian response among known poor responders has not been established. DESIGN: Retrospective cohort study. MATERIALS AND METHODS: We randomly selected 90 cycles between 2005 and 2006 in patients undergoing IVF using non-suppressed protocols with maximal stimulation. Patients either underwent a short protocol (n=44) or a GnRH antagonist protocol (n=46). Variables evaluated were: AMH levels, Day 2 FSH, antral follicle count, age and BMI. Outcomes of interest were: total number of oocytes retrieved, number of oocytes mature, number of oocytes fertilized and clinical pregnancies. All subjects were defined as poor responders based on a prior poor response to aggressive gonadotropin stimulation. Data were tabulated and analyzed using SPSS software. RESULTS: Mean maternal age was 39.6 (±0.4) years, mean day 2 FSH level was 7.1 (±0.3) mIU/ml, and mean day 2 antral follicle count was 8 (±0.4). Logistic regression analysis revealed that AMH and antral follicle count were significant predictors of very poor response (< 4 oocytes). FSH level in this group was not found to be a significant predictor of number of oocytes. The mean AMH level of all patients was 0.7 (±0.1). We established a cut-off level of 0.38ng/ml based on the mean AMH level among those defined as having a very poor ovarian response. That is, 81% of those with an AMH above the cut-off level had greater than four oocytes retrieved (p<0.0001). We also calculated a positive predictive value of 0.89, suggesting that this cutoff was 89% accurate in identifying those who, although overall poor responders, could still have greater than four oocytes retrieved. CONCLUSIONS: AMH is an additional predictor of response to gonadotropin stimulation. Using a cutoff level 0.38ng/ml we can counsel patients, prior to stimulation, if they are likely to produce more than 4 oocytes. As chances for pregnancy are related to ovarian response, this information would be useful to patients in determining whether or not to proceed with ART or to use gamete donation.