Oral GnRH Antagonist (Linzagolix) Versus Injectable GnRH Antagonist for Ovarian Stimulation in IVF/ICSI

Controlled ovarian stimulation (COS) for in vitro fertilization requires prevention of a premature LH surge. The established approaches are the injectable GnRH antagonist (ganirelix or cetrorelix), the current standard of care, and progestin-primed ovarian stimulation (PPOS), which suppresses the LH surge with an oral progestin but renders the endometrium non-receptive, requiring a mandatory freeze-all strategy. Oral, non-peptide GnRH antagonists (relugolix, elagolix, linzagolix) are approved for uterine fibroids and endometriosis but have not previously been used in ovarian stimulation cycles. This study addresses that gap and represents the first documented use of linzagolix in this indication. Linzagolix is an oral, non-peptide GnRH receptor antagonist with dose-dependent suppression of the pituitary-ovarian axis. At 200 mg/day it fully suppresses endogenous estradiol secretion; however, in COS the ovary is driven by exogenous FSH acting directly on the follicle, independent of the pituitary, so follicular growth and estradiol production are preserved while linzagolix acts specifically to prevent the endogenous LH surge. The study hypothesis is that this mechanism yields oocyte yield non-inferior to the injectable antagonist, with the added benefit of oral administration. The feasibility of using a GnRH agonist as the final maturation trigger under an oral antagonist depends on residual receptor occupancy at the time of trigger. Linzagolix has a half-life of approximately 15-20 hours and reaches steady state in about 7 days, compared with relugolix (associated with reduced oocyte retrieval when combined with agonist trigger), which has a substantially longer half-life (approximately 37-42 hours) and reaches steady state in about 12 days. The faster clearance of linzagolix is expected to leave lower residual receptor occupancy at trigger, allowing an adequate LH/FSH flare from agonist trigger; preliminary center experience is consistent with this. Prospective verification of agonist-trigger feasibility under linzagolix is an explicit secondary objective, with particular relevance for OHSS prevention via agonist trigger combined with freeze-all in high responders. Because a freeze-all strategy temporally and biologically decouples embryo transfer from the stimulation cycle, the only pathway by which the LH-suppression method, gonadotropin type, or trigger type could affect reproductive outcomes is oocyte and embryo competence, captured by the primary endpoint (MII count) and secondary embryology parameters. Reproductive outcomes are therefore collected as descriptive data but are not direct endpoints of the stimulation comparison. Design: This is a single-center study at CENTRO AMBRA comprising two components with distinct inferential status. The comparison between linzagolix and the injectable GnRH antagonist is an open-label, randomized, controlled, non-inferiority trial with parallel groups and 1:1 allocation; only randomized patients are included in the two antagonist arms. A third, non-randomized PPOS group is concurrently enrolled by clinical indication (freeze-all candidates, e.g., for PGT-A) and serves as a comparative cohort. The randomization sequence is computer-generated with permuted blocks of variable size, prepared by staff not involved in enrollment, with allocation concealed via sequentially numbered sealed opaque envelopes or a centralized system and revealed only after eligibility confirmation and informed consent. Stratification is by ovarian-reserve responder category (normal/poor/high, defined by AMH and AFC); given the single-center sample size, Pocock-Simon minimization on responder category and age is an acceptable pre-specified alternative. The study is open-label given the oral versus subcutaneous routes of administration, but laboratory staff assessing oocyte and embryo outcomes are kept blinded to allocation. Eligible patients who are not randomized (refusal, contraindication, clinical or organizational reasons) are logged in a screening registry and reported in the CONSORT flow diagram with reason, without being included in the study. The PPOS cohort comparison is observational and subject to confounding by indication, given that the linzagolix arm is not restricted to patients with a freeze-all indication; it is analyzed with mandatory multivariable adjustment and sensitivity analyses restricted to comparable patients, and is interpreted as exploratory and hypothesis-generating. The unit of analysis is one cycle per patient (the first study cycle), to preserve independence of observations. Sample size: The design is based on preliminary center data showing a within-group standard deviation of approximately 5.8 MII oocytes, with a strong correlation between MII count and ovarian reserve (AMH, AFC) allowing substantial variance reduction through ANCOVA. Assuming a conservative R² of 0.60 for the adjustment model, the expected residual standard deviation is approximately 3.7 oocytes. With a non-inferiority margin of 2.0 MII oocytes, one-sided alpha of 0.025, and 80% power, 53 analyzable patients per arm are required. Accounting for an expected 15% rate of cancellation/dropout, 63 patients per arm will be randomized (126 across the two antagonist arms), with the PPOS cohort sized at approximately 63 patients, for a total of approximately 189 patients. The margin of 2.0 MII oocytes was defined on clinical rather than statistical grounds, corresponding to a difference that would not alter clinical management, insemination strategy, transfer strategy, or expected cumulative prognosis. Sample size is fixed a priori; no interim efficacy or futility analyses or sample size re-estimation are planned. Statistical analysis: The primary analysis compares MII oocyte count between the linzagolix and injectable antagonist arms using ANCOVA, adjusted for ovarian reserve (AMH/AFC or responder category), age, and initial gonadotropin dose, with negative binomial regression as a pre-specified alternative in case of overdispersion. Non-inferiority is concluded if the lower bound of the two-sided 95% confidence interval for the adjusted mean difference exceeds -2.0 MII oocytes. Both intention-to-treat and per-protocol populations are analyzed, with non-inferiority requiring concordant results across both. Randomized patients whose cycle is cancelled before oocyte retrieval are included in the ITT analysis with an MII count of zero; sensitivity analyses excluding cancelled cycles are pre-specified. If non-inferiority is demonstrated, superiority testing on the same endpoint may proceed via a closed hierarchical procedure without alpha inflation. The comparison with the PPOS cohort uses mandatory multivariable adjustment (age, BMI, ovarian reserve, gonadotropin type/dose, PGT-A indication), with sensitivity analyses restricted to comparable patients and, where sample size allows, propensity scoring; results are reported as associations, not causal effects.