GnRH antagonists in poor responders

Abstract

Gonadotropin releasing hormone (GnRH) antagonists allow endogenous gonadotropins to enhance stimulation in the first part of the follicular phase until there is a need to inhibit the premature luteinizing hormone (LH) surge and this might represent a stimulation option for poor responders 1-3). We hypothesized that an increase in the dose of gonadotropins in combination with the use of GnRH antagonists would improve the stimulation and pregnancy outcome in poor responders previously treated unsuccessfully with GnRH agonists for intracytoplasmic sperm injection (ICSI). Patients should have first undergone an ICSI cycle with a long GnRH agonist protocol in which a poor ovarian response was present, no pregnancy was achieved and subsequently start an ICSI cycle in which GnRH antagonist was used. Poor ovarian response was characterized either by cancellation of the cycle due to poor follicular development after at least 10 days of gonadotropin stimulation (n = 4), or by retrieval of no more than five cumulus oocyte complexes (COCs) (n = 30) at oocyte pick-up. In the agonist cycle the long protocol was used with intranasal Buserelin (Suprefact®, Hoechst, Frankfurt, Germany) administered at a dose of 6 × 100 μg/day, starting in the midluteal phase of the menstrual cycle preceding the ovarian stimulation. Human menopausal gonadotropins (HMGs, Humegon, Organon, Oss the Netherlands) were used for ovarian stimulation at a dose determined by the patient's age and/or basal serum follicle stimulating hormone (FSH) level and/or ovarian response to previous treatment cycles (mean ± SEM of the starting dose: 230 ± 13.9 IU). Recombinant FSH (Puregon, NV Organon) was used in the antagonist cycle and the dose was increased by 100 IU (mean ± SEM of the starting dose: 330 ± 14.5 IU). Stimulation started on day 2 of the menstrual cycle and continued with the same dose until a follicle of ≥ 15 mm was detected by ultrasound. At that point, recombinant FSH was increased by 100 IU, and a daily subcutaneous injection of 0.25 mg GnRH antagonist (Orgalubran, Organon) was started and continued up to and including the day of human chorionic gonadotropin (hCG) administration. Final oocyte maturation was induced with 10 000 IU of hCG (Pregnyl®, Organon) and all patients received luteal-phase support by natural micronized progesterone (Utrogestan®, Piette, Belgium) administered intravaginally. Stimulation characteristics during the agonist and the antagonist cycles in poor responders who underwent oocyte retrieval in both cycles are shown in Table I. No difference in the quality of the embryos transferred was observed between the two groups of patients who had embryo transfer (ET) in both cycles (n = 17). In addition, no LH surge occurred in either the agonist or the antagonist cycle. The outcome for all patients in both the agonist and the antagonist cycles is shown in Table I. This study has shown that an improved stimulation and pregnancy outcome is observed in poor responders previously treated unsuccessfully with GnRH agonists for ICSI by using GnRH antagonists and increasing the starting dose of gonadotropins. A different type of gonadotropin was used in the antagonist cycle as compared to the agonist cycle (recombinant FSH vs. HMG, respectively). However, recombinant FSH has not been shown to be superior to HMG in assisted reproductive technologies (ART) (4) and thus its contribution to the improvement of reproductive outcome is probably unlikely. Moreover, it could be argued that the occurrence of pregnancies in the GnRH antagonist cycle is not due to the modified protocol applied but rather could be attributed to regression to the mean. However, clinical experience for many years in poor responders does not support such an argument, as these patients show a consistently adverse outcome under any modified protocol applied so far. A randomized controlled trial (RCT) would provide a more reliable answer to the usefulness of a GnRH antagonist protocol in these patients. We thank Julie Deconick for correcting the manuscript. This work is supported by grants from the Fund for Scientific Research, Flanders.