Differential impact of controlled ovarian hyperstimulation on live birth rate in fresh versus frozen embryo transfer cycles: a Society for Assisted Reproductive Technology Clinic Outcome System study

Objective: To study the impact of both controlled ovarian hyperstimulation (COH) length and total gonadotropin (GN) dose individually and in concert on live birth rates (LBR) in both fresh and freeze-all in vitro fertilization embryo transfer (IVF-ET) cycles. Design: Historical cohort study. Setting: Not applicable. Patient(s): The U.S. national database from the Society of Assisted Reproductive Technology Clinic Outcome Reporting System from 2014 to 2015 was used to identify patients undergoing autologous GN stimulation IVF cycles with the use of GnRH antagonist–based suppression protocols where a single embryo transfer was performed as part of a fresh IVF-ET cycle (fresh, n = 14,866) or the first frozen embryo transfer after a freeze-all cycle (frozen, n = 2,964), and not including preimplantation genetic testing cycles. The patients’ demographic and cycle characteristics, duration of COH, total GN dose, and pregnancy outcomes were extracted. Binomial regression models estimated trend and relative risk of live birth with respect to days of stimulation and total GN dose singularly, and after adjustment for a priori confounders including age, parity, body mass index, diagnosis, and maximum follicle-stimulating hormone in both fresh and frozen embryo transfer cycles. Both days of stimulation and total GN dose were then added to the multivariate model to show whether they were independently associated with LBR. Intervention(s): Not applicable. Main outcome measure(s): Live birth rate. Results: In both fresh and frozen cycles, length of COH was significantly associated with total GN dose. On univariate analysis, LBR decreased significantly with increasing length of stimulation and increasing total GN dose in both fresh and frozen cycles. On multivariable analysis including both days of stimulation and total GN dose, days of stimulation was no longer significantly correlated with LBR, whereas total GN dose remained significantly correlated with LBR in fresh cycles only. When total GN doses ranging from <2,000 IU through 5,000 IU to >5,000 IU were compared, a significant improvement in live birth rate was noted with lower total GN doses. Specifically, GN doses <2,000 IU had a 27% higher rate of live birth compared with GN dose >5,000 IU. For GN dose groups up to 4,000 IU, the estimated effect on LBR was similar. There was a marginal improvement (13%) in LBR with GN doses of 4,000 IU to 5,000 IU compared with >5,000 IU. When the multivariate model was applied to the frozen cycles, neither total GN dose nor days of stimulation was significantly associated with LBR. Conclusions: High total GN dose but not prolonged COH is associated with decreasing LBRs in fresh cycles, whereas neither factor significantly affects LBR in frozen cycles. Consideration should be given to minimizing the total GN dose when possible in fresh autologous cycles, either by decreasing the daily dose or by limiting the length of stimulation to improve LBRs. In freeze-all cycles, the use of higher GN doses does not seem to adversely affect the LBR of the first frozen embryo transfer. High total GN dose likely exerts a negative impact on the endometrium and/or oocyte/embryo unrelated to the length of stimulation. The differential effect of total GN dose on LBR in fresh and frozen cycles may imply a greater impact exerted on the endometrium rather than the oocyte.