Predictive factors of ovarian response and clinical outcome after IVF/ICSI following a rFSH/GnRH antagonist protocol with or without oral contraceptive pre-treatment

BACKGROUND: Prediction of ovarian response prior to the first controlled ovarian stimulation (COS) cycle is useful in determining the optimal starting dose of recombinant FSH (rFSH). However, potentially predictive factors may be subject to inter-cycle variability and many patients are pre-treated with oral contraceptives (OC) for scheduling purposes. Our objective was to determine predictive factors of ovarian response for patients undergoing COS with rFSH in a gonadotrophin-releasing hormone antagonist protocol and to determine the inter-cycle variability of these factors. METHODS: In this multinational trial, 442 patients were randomized to receive either OC treatment or no treatment prior to their first COS cycle. For candidate predictive factors, patient characteristics were collected at screening, and endocrine and sonographic data were collected during the early follicular phase of the two subsequent cycles. A treatment regimen of 200 IU rFSH and 0.25 mg ganirelix was applied during the second cycle. Predictive factors of ovarian response and of too low (< 6 oocytes) or too high (> 18 oocytes) ovarian responses were determined using stepwise linear regression and stepwise logistic regression, respectively. RESULTS: Anti-Mullerian hormone (AMH) and basal FSH were statistically significant predictors of the number of oocytes retrieved and of an excessive ovarian response. For low ovarian response, AMH was the only significant predictive factor. In the non-OC group, the predictive value was higher than in the OC group and higher at the early follicular phase of the stimulation cycle than of the previous cycle. The intercycle variation for AMH was low compared with the inter-cycle variation of other hormones. Between the two groups, there were no differences in the number or quality of embryos obtained or transferred, but the implantation rate was significantly lower in the OC group (24.1 versus 30.1%, P 0.03), resulting in an ongoing pregnancy rate of 26.3% compared with 35.7% in the non-OC group (P = 0.05). CONCLUSIONS: The best predictive model of ovarian response was in the non-OC group and included both AMH and basal FSH determined at the early follicular phase of the stimulation cycle. In the proceeding cycle, AMH alone had sufficient predictive value since it was not affected by inter-cycle variability or OC pretreatment. Clinical trial identifier: NCT00778999.