Serum anti-Müllerian hormone and basal serum FSH as predictors of poor ovarian response in assisted conception cycles

ObjectiveThe aim of the study was to correlate serum AMH and serum FSH levels with ovarian response to stimulation in IVF–ICSI cycles.Design and settingsThis was a prospective observational study conducted in a private assisted conception unit.Subjects and methodsOne hundred and two patients were selected on their first IVF cycle. Basal serum FSH and serum AMH were measured one month before the stimulation cycle. A fixed dose GnRH antagonist protocol was used in all cycles transferring a maximum of three day-3 cleavage stage embryos. We defined poor ovarian response as retrieval of fewer than four mature oocytes in cycles requiring ⩾ 3000 IU of gonadotropins for stimulation or cycle cancellation due to poor response. The correlation between different parameters was expressed as a Spearman’s correlation coefficient. The clinical value of AMH and FSH as predictors of poor ovarian response as well as predictors of pregnancy was evaluated by constructing relevant receiver operator characteristics curves (ROC curves).ResultsOf these 102 cycles, 28 fitted our definition of poor response while the remaining 74 cycles all produced an adequate response to stimulation. There was a statistically significant difference between the adequate responders group and poor responders group regarding their mean age (31.5 versus 39.6, p < 0.001), the mean value of AMH (2.84 ng/ml versus 0.9 ng/ml, p < 0.0001) as well as the mean value of basal FSH (7.6 IU/ml versus 9.7 IU/ml, p < 0.0001). Serum AMH level had a positive correlation while serum FSH had a negative correlation with the number of oocytes collected while only serum AMH had a significant positive correlation with the occurrence of pregnancy. ROC curve analysis of our results showed that serum AMH with an optimal cut-off value of 1.2 ng/ml is a reliable predictor of poor ovarian response with an area under the ROC curve of 90.4%. Serum basal FSH with an optimal cut-off value of 8.9 IU/ml was of lower value than AMH as a predictor of poor ovarian response with an area under the ROC curve of 81.9%. However, neither serum AMH nor basal serum FSH was found to able to reliably predict the occurrence of pregnancy with an area under the ROC curve of 59.4% and 58.6% respectively.ConclusionOur results show that serum AMH level is more reliable than basal serum FSH as a predictor of poor ovarian response to stimulation with a cut-off value of 1.2 ng/ml shown to predict poor ovarian response with a sensitivity of 91.7%.