P-543 Predictive factors for ovarian response in a corifollitropin alfa/GnRH antagonist protocol for controlled ovarian stimulation

Study question Can predictors of low ( 18 oocytes retrieved) be identified in patients undergoing controlled ovarian stimulation with corifollitropin alfa in a gonadotropin-releasing hormone (GnRH) antagonist protocol? Summary answer Anti-Mullerian hormone (AMH), antral follicle count (AFC), and age on stimulation day 1 were prognostic for both high and low ovarian response in a corifollitropin alfa/GnRH antagonist protocol, in addition to follicle-stimulating hormone (FSH) for high ovarian response and menstrual cycle length for low ovarian response. What is known already The majority of studies on predictors of ovarian response have analyzed patients treated with recombinant(r) FSH in long GnRH agonist protocols. Systematic reviews have identified AMH, AFC, and basal FSH as predictors of low ovarian response and AMH and AFC as predictors of high ovarian response in these protocols. There is, to date, no information on predictors in a corifollitropin alfa/GnRH antagonist protocol. Study design, size, duration In this retrospective analysis, statistical model building for high and low ovarian response was based on the corifollitropin alfa treatment group of the Pursue trial (n = 694). Multivariable logistic regression models were constructed in a stepwise fashion (P Participants/materials, setting, methods Infertile women aged 35-42 years received a single injection of 150 µg corifollitropin alfa, followed by ≤300 IU/d rFSH starting on stimulation day 8 if needed. 0.25 mg/d ganirelix started on day 5 until final oocyte maturation with 250 µg recombinant human chorionic gonadotropin (rhCG). Main results and the role of chance 14.1% of subjects were high, and 23.2% were low ovarian responders. The regression model for high ovarian response included 4 independent predictors, with high AMH levels and AFCs increasing the risk while high FSH levels and increasing age decreased risk. The apparent area under the curve (AUC) of the receiver operating characteristic (ROC) curve for this model was 0.888. Sensitivity and specificity were 84% and 80%, respectively. The regression model for low ovarian response also included 4 independent predictors. Older age increased, while higher AMH, higher AFC, and longer menstrual cycle length decreased the risk of low ovarian response. The apparent AUC of the ROC curve for the complete model was 0.886. Sensitivity and specificity were 77% and 87%, respectively. Limitations, reason for caution This study was restricted to women aged 35-42 years, with a body weight of ≥50 kg and body mass index ≥18 and ≤32 kg/m2; therefore, its generalizability may be limited. Wider implications of the findings In assisted reproductive technology, both very low and very high ovarian responses are associated with increased cancellation rates and lower pregnancy rates. A high ovarian response also increases the risk for development of ovarian hyperstimulation syndrome. The multivariable models in the current analyses predicted the occurrence of high and low ovarian response sufficiently well to be used in clinical practice to individualize treatment. Study funding/competing interest(s) Financial support for this study was provided by Merck, Sharp & Dohme Corp., a subsidiary of Merck & Co. Inc., Whitehouse Station, NJ, USA. Trial registration number NCT01144416