Gonadotropin levels at the start of ovarian stimulation predict normal fertilization after hCG re‐trigger in GnRH antagonist cycles

Abstract Purpose To assess the appropriateness of human chorionic gonadotropin (hCG) re‐trigger in poor responders to gonadotropin‐releasing hormone agonist (GnRHa) trigger in controlled ovarian stimulation (COS) cycles. Methods The 2251 cycles in 2251 patients triggered with GnRHa for oocyte stimulation, with or without requiring hCG re‐trigger between 2013 and 2018, were retrospectively analyzed to compare gonadotropin levels at the start of COS and the rate of normal fertilization between the re‐trigger and non–re‐trigger group. Furthermore, patients in the re‐trigger group were stratified by the rate of normal fertilization (good: ≥60% or poor: <60%) to compare patient demographics, hormone profiles, and clinical outcome between the subgroups. Results In the re‐trigger group, FSH and LH levels at the start of COS were significantly lower in the good fertilization group than in the poor fertilization group (P < .01). Receiver operating characteristic curves identified cutoff values of the FSH and LH levels of 1.30 and 0.35 mIU/mL, respectively, for predicting ≥60% normal fertilization. Conclusion Gonadotropin levels at the start of COS are predictors of response to GnRHa trigger and hCG re‐trigger necessity, and may serve as indicators to help clinicians appropriately choose hCG re‐trigger rather than abandoning the cycles or continuing the first oocyte aspiration attempt.


| INTRODUC TI ON
Ovarian hyperstimulation syndrome (OHSS) is an iatrogenic complication of controlled ovarian stimulation (COS) protocols. The most common symptoms are decreased circulating blood volume 1,2 and pleural/peritoneal fluid accumulation, 3 and in serious cases, resulting in fatal cardiorespiratory impairment. The most possible pathogenesis of OHSS is the increased release of vascular endothelial growth factor (VEGF) by human chorionic gonadotropin (hCG) administered as an ovulation trigger in COS protocols, which induces capillary hyperpermeability. [4][5][6][7] Although the reported incidence of OHSS in in vitro fertilization (IVF) cycles varies widely, mild OHSS and moderate-to-severe OHSS are expected to occur in 20%-33% and 3%-8%, respectively, of IVF cycles. 8 In order to prevent OHSS, GnRH antagonist protocols using a GnRH agonist (GnRHa) as an ovulation trigger have been introduced in IVF cycles. 9 GnRH antagonist protocols, as compared with conventional GnRHa long protocols, require a lower dose of gonadotropin medications 10 and reduce the risk of serious OHSS. 11 However, GnRHa trigger fails to achieve sufficient oocyte yields in some patients. [12][13][14] Failures to retrieve adequate oocytes range from far fewer oocytes retrieved than expected from the number of mature follicles visible on the day of ovulation 15 to empty follicle syndrome. 16,17 We have experienced such oocyte retrieval failures in patients undergoing GnRH antagonist protocols using GnRHa trigger. Some of the patients were rescued by re-triggering oocyte maturation with hCG, leading to successful oocyte retrieval and resulting in live births. 18,19 However, the other patients were found to have morphologically abnormal cumulus-oocyte complexes (COCs) and lower rates of normal fertilization after hCG re-trigger. These experiences have led us to believe that hCG re-trigger may not be appropriate for all poor responders to GnRHa trigger.
This study retrospectively assessed the appropriateness of conducting hCG re-trigger based on fertilization outcomes and attempted to identify patient characteristics or hormone profiles, which might be predictive of normal fertilization after hCG re-trigger.

| Patients
This study reviewed 2251 cycles of GnRH antagonist-based COS using GnRHa trigger for final oocyte maturation conducted in patients who received fertility treatment from 2013 to 2018 at our clinic. The patients who underwent oocyte retrieval more than once during the study period were not included. The opt-out method (potential participants were informed about research, information was available on our Web site, and patients were included unless they objected) approved by the IRB of Asada Ladies Clinic was used for this study to recruit the participants.
Among these, 36 cycles, in which adequate oocytes could not be collected at the first oocyte aspiration attempt probably because of suboptimal oocyte maturation, required hCG re-trigger; these cycles constituted the re-trigger group. The remaining 2215 cycles requiring no hCG re-trigger constituted the non-re-trigger group. Patients in the re-trigger group were matched for age with those in the nonre-trigger group to minimize age bias. The GnRHa trigger was scheduled for 21:00 to 22:00 and oocyte retrieval was conducted 36-40 hours after the GnRHa trigger.

| Treatment
On the day of oocyte retrieval, 5-10 follicles of ≥20 mm in diameter were initially aspirated from one or both ovaries. Patients with an oocyte recovery rate (the ratio of the number of collected oocytes to the number of aspirated follicles ≥20 mm in diameter) of >50% were deemed to require no re-trigger and subsequent aspirations were continued (non-re-trigger group). Patients with an oocyte recovery rate of ≤50%, and no or only a few oocytes were recovered, despite 5-10 follicles measuring ≥20 mm in diameter aspirated, the first oocyte stimulation with GnRHa trigger was presumed to have been suboptimal and the patient was given an hCG injection for re-triggering in the same cycle at the discretion of the clinicians (re-trigger group). The other follicles were kept in ovaries for the second oocyte retrieval. The second oocyte retrieval was scheduled 36 hours after administration of 5000 IU (or 3333 IU if the risk of OHSS was expected to be increased) of hCG (human chorionic gonadotropin for injection; Fuji Pharma).
Retrieved oocytes were fertilized using intracytoplasmic sperm injection (ICSI), and the embryos were cultured for up to 20 hours in a time-lapse imaging incubator (EmbryoScope; Vitrolife; or CCM-iBIS NEXT; Astec) that acquired images at 15-minute intervals.
The appearance of two pronuclear (2PN) signified normal fertilization and that of ≥3 PN abnormal fertilization. All of the resultant embryos were cryopreserved at the pronuclear or blastocyst stage to prevent OHSS, until thawed for embryo transfer which was carried out in hormonally primed cycles using sequentially adminis-

| Patient characteristics and hormone profiles
The following patient characteristics and hormone profiles were compared between the re-trigger group and the non-re-trigger group: age, anti-Mullerian hormone (AMH) level, body mass index (BMI), total number of COS cycles, FSH and LH levels at the start of COS (day 3 of the menstrual cycle), total dose of GnRH antagonist required, total dose of gonadotropin required, serum estradiol (E2) and progesterone (P4) levels on the day of GnRHa trigger, and number of oocytes retrieved. We defined a 60% normal fertilization as a threshold for "good fertilization," based on previously reported rate of normal fertilization in ICSI cycles (68.6%-77.1%). [21][22][23] The retrigger group consisting of 36 patients was subdivided into a group of a ≥60% normal fertilization ("the good fertilization group") and a group of a <60% normal fertilization ("the poor fertilization group") to assess the following patient characteristics and hormone profiles: age, AMH level, FSH and LH levels at the start of COS (day 3 of the menstrual cycle), serum E2 and P4 levels on the day of GnRHa trigger, P4 level on the day of hCG re-trigger, and dose of hCG used for re-trigger. In addition, clinical outcomes (clinical pregnancy rate, clinical loss rate, and live birth rate) were compared between the good fertilization group and the poor fertilization group.

| Male partner's characteristics
To assess the effects of male partner's characteristics on fertilization outcomes, age, LH, FSH, total testosterone, semen volume, sperm concentration, and sperm motility in male partners were compared between the poor fertilization group and the good fertilization group.

| Statistical analyses
Before between-group comparisons, the Kolmogorov-Smirnov test or the Shapiro-Wilk test was conducted to assess data distribution in each group. For normally distributed data, F test was conducted and if the data showed homoscedasticity, t test was used to compare mean values between the groups. For data not normally distributed, the Mann-Whitney U test was used to compare median values between the groups. Rates in the groups were compared with the use of Fisher's exact test. The proportion of patients using each GnRHa medication was compared using chi-squared test. A free software "R" was used for all data analyses. A P-value of <.05 was considered statistically significant.

| Comparisons of patient characteristics and hormone profiles between the re-trigger group and the non-re-trigger group
There were no significant differences between the re-trigger group and the non-re-trigger group in age, AMH level, BMI, the total number of COS cycles, the total dose of GnRH antagonist required, and serum E2 and P4 levels on the day of GnRHa trigger (Table 1). In the re-trigger group, the total dose of gonadotropin required was significantly higher, and FSH and LH levels at the start of COS were significantly lower than those in the non-re-trigger group (all P < .01). Since a decision to re-trigger was made by clinicians on the fact that no or only a few oocytes were retrieved, the number of oocytes retrieved in the re-trigger group (0.5 oocytes) was significantly lower than that in the non-re-trigger group (23 oocytes, P < .01).
The proportions of patients using each GnRHa medication in the non-re-trigger group were 75.3% with injectable leuprolide acetate and 24.7% with intranasal buserelin acetate, whereas those in the re-trigger group were 63.9% and 36.1%, respectively, with no difference in the choice of GnRHa medication between the groups (chisquared test). TA B L E 1 Characteristics and hormone profiles in patients treated with GnRHa trigger for final oocyte maturation (non-re-trigger group versus re-trigger group)

| Comparisons of patient characteristics and hormone profiles after hCG retrigger between the good fertilization group and the poor fertilization group
There were no significant differences in age and AMH level between 17 patients in the good fertilization group and 19 patients in the poor fertilization group ( Table 2). The LH and FSH levels at the start of COS (day 3 of the menstrual cycle) were significantly lower in the good fertilization group than in the poor fertilization group (P < .01). Total dose of gonadotropin required did not differ between the groups, but the good fertilization group required a significantly smaller dose of GnRH antagonist (P < .01). Serum P4 level on the day of hCG re-trigger was significantly lower in the good fertilization group than in the poor fertilization group (P < .01), whereas no significant differences were noted in serum P4 and E2 levels at GnRHa trigger between the groups.

| Comparisons of clinical outcomes after hCG retrigger between the good fertilization group and the poor fertilization group
There were no differences in the numbers of all oocytes and MII oocytes retrieved after hCG re-trigger between the good fertilization group and the poor fertilization group ( Table 3). The good fertilization group had a significantly lower rate of abnormal fertilization, than did the poor fertilization group (P < .01). In the good fertilization group, all the patients (17 patients) underwent embryo transfer in or after the next cycle; however, in the poor fertilization group, 9 of the 19 patients (47.4%) had fertilization abnormalities and did not undergo embryo transfer. There were no significant differences in clinical pregnancy rate, clinical loss rate, and live birth rate between the groups.

| Hormone profiles and semen characteristics in male partners
The age, LH, FSH, and total testosterone levels in male partners did not significantly differ between the poor fertilization group and the good fertilization group (Table 4). Semen volume was significantly higher in the good fertilization group than in the poor fertilization group (P = .012), but no differences were noted in key factors for successful fertilization by ICSI, such as sperm concentration and sperm motility, between the groups.

| Morphological change over time in fertilized oocytes
Time-lapse observation of fertilized oocytes in the good fertilization group revealed no morphological change or abnormality in the ooplasm ( Figure 3A-C), except for poly-pronuclear embryos probably resulting from a failure of polar body extrusion occurring during oocyte maturation ( Figure 3D). In the poor fertilization group, on the other hand, marked changes were seen in ooplasmic morphology. Figure 4A shows reverse cleavage, a phenomenon where oocyte cytoplasm was divided into two before pronuclear formation, and subsequently fused back to one. Figure 4B shows normal fertilization, despite debris being evident in the perivitelline space. Figure 4C shows abnormal fertilization resulting from an oocyte that appeared to be morphologically normal before fertilization but had "twisted" cytoplasm after fertilization. Figure 4D shows an oocyte with morphological abnormalities where direct cleavage occurred within several hours after fertilization and multiple distorted and shrunken blastomeres appeared 17.5 hours after fertilization.
Thus, abnormal fertilizations occurred through different processes in the good fertilization group and the poor fertilization group.

| Cutoff FSH and LH levels at the start of ovarian stimulation (day 3 of the menstrual cycle)
As above, the good fertilization group had significantly lower FSH and LH levels at the start of COS (both P < .01). Based on the results,

| D ISCUSS I ON
There have been sparse reports as to failures of final oocyte maturation with GnRHa trigger, and the mechanism of the failures has remained unclear. 19 This is the first report focusing on the endocrine environment at the start of COS and assessing the appropriateness of conducting hCG re-trigger after failed final oocyte maturation triggered with a GnRHa, based on fertilization outcome.
This study attempted to determine optimal cutoff FSH and LH levels at the start of COS to identify patients requiring re-trigger in advance. There have been several reports that the need for re-trigger may be predictable from the endocrine environment after GnRHa trigger. Kummer et al 13 measured LH and P4 levels at 8 to 12 hours after GnRHa trigger and found that a low LH level of <15 mIU/mL was associated with a suboptimal oocyte yield.
Similarly, Shapiro et al reported that a LH level of <12 mIU/mL at 12 hours after GnRHa trigger was associated with a dramatically reduced oocyte yield. 14 However, measuring hormone levels on the day following  Although the incidence of abnormal fertilization after ICSI has been reported to be 2.4%-6.5%, [33][34][35][36]  been many reports of oocyte aging. In mice, it was reported that abnormal fertilization increased by postovulatory oocyte aging 37 and that in vitro aged postovulatory oocytes exhibited chromosomal misalignment and abnormal spindle formation, and failed to reach full-term development. 38 It was also reported that human oocytes which had failed to fertilize after an 18-hour incubation with spermatozoa and had spent a further 6-8 hours in culture showed an increased incidence of spindle abnormalities and of the proliferation of ooplasmic microtubules, 39 and chromosomes no longer aligned at the spindle equator but scattered all over the spindles. 40 F I G U R E 4 Morphological change over time in oocytes after ICSI in the poor fertilization group. a, Reverse cleavage is shown (oocyte cytoplasm was divided into two before pronuclear formation and subsequently fused back to one); b, normal fertilization is shown, despite debris being evident in the perivitelline space; c, abnormal fertilization is shown resulting from an oocyte that appeared to be morphologically normal before fertilization but had the twisted cytoplasm after fertilization; and d, morphological abnormalities are shown (direct cleavage occurred within several hours after fertilization and multiple distorted and shrunken blastomeres appeared 17.5 h after fertilization.) Failures of final oocyte maturation triggered with a GnRHa requiring hCG re-trigger occur with an extremely low frequency, and the endocrine environment can fluctuate even within the same patient. It can be therefore said that low FSH and LH levels at the start of COS are not direct causes of infertility but may be incidental symptoms. There have been many reports regarding suboptimal oocyte yield after GnRHa trigger 43 ; however, most of the reports describe the possible association of hormone levels with oocyte yield after re-trigger, but not with fertilization or embryo development.
Further investigation will be necessary focusing on the association of hormone levels with fertilization, embryo development, pregnancy, and delivery after re-trigger.
Currently, there are no established criteria or protocols to determine the appropriateness of hCG re-trigger. This study proposes that a FSH level of ≤1.30 mIU/mL and a LH level of ≤0.35 mIU/mL at the start of COS are expected to be optimal predictors of poor pituitary response to GnRHa trigger and need for hCG re-trigger.
Conducting re-trigger solely based on the fact that adequate oocytes were not retrieved by the first oocyte aspiration attempt may cause oocyte overmaturation resulting in the increased incidence of abnormal fertilization. Low levels of FSH and LH at the start of COS, in combination with the fact that adequate number of oocytes were not retrieved, would warrant the conduct of hCG re-trigger to collect normally fertilized oocytes. Gonadotropin levels, that is, a FSH level of ≤1.30 mIU/mL and a LH level of ≤0.35 mIU/mL at the start of COS are predictors of poor pituitary response to GnRHa trigger and the necessity of hCG re-trigger, and may serve as indicators to help clinicians appropriately choose hCG re-trigger rather than abandoning the cycles or continuing the first oocyte aspiration attempt.

ACK N OWLED G M ENTS
The authors would like to thank Tomokuni Yoshimura, Emiko Asano, and Dr Colin M Howles for valuable comments and suggestions on the manuscript.