Individualized versus standard FSH dosing in women starting IVF/ICSI: An RCT. Part 2: The predicted hyper responder

STUDY QUESTION: Does a reduced FSH dose in women with a predicted hyper response, apparent from a high antral follicle count (AFC), who are scheduled for IVF/ICSI lead to a different outcome with respect to cumulative live birth rate and safety? SUMMARY ANSWER: Although in women with a predicted hyper response (AFC > 15) undergoing IVF/ICSI a reduced FSH dose (100 IU per day) results in similar cumulative live birth rates and a lower occurrence of any grade of ovarian hyperstimulation syndrome (OHSS) as compared to a standard dose (150 IU/day), a higher first cycle cancellation rate and similar severe OHSS rate were observed. WHAT IS KNOWN ALREADY: Excessive ovarian response to controlled ovarian stimulation (COS) for IVF/ICSI may result in increased rates of cycle cancellation, the occurrence of OHSS and suboptimal live birth rates. In women scheduled for IVF/ICSI, an ovarian reserve test (ORT) can be used to predict response to COS. No consensus has been reached on whether ORT-based FSH dosing improves effectiveness and safety in women with a predicted hyper response. STUDY DESIGN SIZE, DURATION: Between May 2011 and May 2014, we performed an open-label, multicentre RCT in women with regular menstrual cycles and an AFC > 15. Women with polycystic ovary syndrome (Rotterdam criteria) were excluded. The primary outcome was ongoing pregnancy achieved within 18 months after randomization and resulting in a live birth. Secondary outcomes included the occurrence of OHSS and cost-effectiveness. Since this RCT was embedded in a cohort study assessing over 1500 women, we expected to randomize 300 predicted hyper responders. PARTICIPANTS/MATERIALS, SETTING, METHODS: Women with an AFC > 15 were randomized to an FSH dose of 100 IU or 150 IU/day. In both groups, dose adjustment was allowed in subsequent cycles (maximum 25 IU in the reduced and 50 IU in the standard group) based on pre-specified criteria. Both effectiveness and cost-effectiveness were evaluated from an intention-to-treat perspective. MAIN RESULTS AND THE ROLE OF CHANCE: We randomized 255 women to a daily FSH dose of 100 IU and 266 women to a daily FSH dose of 150 IU. The cumulative live birth rate was 66.3% (169/255) in the reduced versus 69.5% (185/266) in the standard group (relative risk (RR) 0.95 [95% CI, 0.85-1.07], P = 0.423). The occurrence of any grade of OHSS was lower after a lower FSH dose (5.2% versus 11.8%, RR 0.44 [95% CI, 0.28-0.71], P = 0.001), but the occurrence of severe OHSS did not differ (1.3% versus 1.1%, RR 1.25 [95% CI, 0.38-4.07], P = 0.728). As dose reduction was not less expensive ((sic)4.622 versus (sic)4.714, delta costs/woman (sic)92 [95% CI, -479-325]), there was no dominant strategy in the economic analysis. LIMITATIONS, REASONS FOR CAUTION: Despite our training programme, the AFC might have suffered from inter-observer variation. Although strict cancellation criteria were provided, selective cancelling in the reduced dose group (for poor response in particular) cannot be excluded as observers were not blinded for the FSH dose and small dose adjustments were allowed in subsequent cycles. However, as first cycle live birth rates did not differ from the cumulative results, the open design probably did not mask a potential benefit for the reduced dosing group. As this RCT was embedded in a larger cohort study, the power in this study was unavoidably lower than it should be. Participants had a relatively low BMI from an international perspective, which may limit generalization of the findings. WIDER IMPLICATIONS OF THE FINDINGS: In women with a predicted hyper response scheduled for IVF/ICSI, a reduced FSH dose does not affect live birth rates. A lower FSH dose did reduce the incidence of mild and moderate OHSS, but had no impact on severe OHSS. Future research into ORT-based dosing in women with a predicted hyper response should compare various safety management strategies and should be powered on a clinically relevant safety outcome while assessing non-inferiority towards live birth rates