Initial studies in in vitro fertilization (IVF)/intracytoplasmic sperm injection (ICSI) patients have demonstrated that the use of a GnRH agonist (GnRHa) trigger, followed by fresh transfer and a standard luteal phase support (LPS) was associated with unacceptably high rates of loss in early pregnancy compared to hCG trigger, particularly in normal responder (NR) patients (1, 2). It has been subsequently concluded that this early pregnancy loss was caused by luteal phase (LP) insufficiency, despite the use of a standard LPS package of progesterone (P) and estradiol (E2). The LP defect was primarily caused by reduced early-mid-luteal luteinizing hormone (LH) activity, resulting in a significant reduction in progesterone output by the corpora lutea (CL) as no adverse effects were seen with respect to the maturity rate of oocytes, fertilization rates, embryo quality, and reproductive outcomes during the subsequent replacement of frozen embryos derived from women who had received a GnRHa trigger (3–7). Following these first disappointing results several studies have shown a LP rescue with the use of a modified LPS, resulting in reproductive outcomes comparable to those observed after hCG triggering (3, 4, 8–10). Up until now, the method of modified LPS reported by Humaidan et al. (3) appears to be the most frequently used regimen in fresh embryo transfer cycles (11). However, further fine-tuning of LPS after the GnRHa trigger could still possibly be warranted (12–14). Significantly increased implantation rates (IRs) were previously reported in oocyte recipients as well as in patients who were triggered with hCG, if they received a single mid-luteal bolus of GnRHa in addition to standard LPS (15–17). However, others have failed to support this finding (18, 19). Importantly, there are significant differences in the early-mid-LP endocrine pattern when GnRHa triggers and hCG triggers are compared, especially in terms of LH levels. From this, it could be hypothesized that the GnRHa-triggered IVF cycle could benefit more from the addition of a bolus of GnRHa to boost the circulating endogenous LH and thus, progesterone levels around the time of implantation than the hCG triggered cycle. No studies previously investigated this issue. Therefore, the aim here was to explore a possible fine-tuning of the LPS of GnRHa-triggered IVF/ICSI cycles, using the previously suggested protocol of Humaidan et al. (3). Here, we have combined a GnRHa trigger and the 1,500 IU HCG dose administered on the day of oocyte pick-up (OPU) (3, 4) with a single dose of GnRHa administered 6 days after OPU, i.e., 2 days before the expected day of implantation in an attempt to mimic the natural cycle peak of progesterone during the mid-LP. In addition, a standard LPS consisting of vaginal progesterone (P) and oral estrogen (E2) was used. By possibly optimizing the early and mid-LP in terms of steroids, it was also hypothesized that we would subsequently optimize IRs, which was the primary outcome measure of the study.

The trial was performed according to CONSORT guidelines. The flow of participants is shown in Figure 1. A total of 341 patients met the study criteria and were recruited. However, 13 patients were subsequently excluded, 6 patients due to hyper-responsiveness to stimulation (>30 follicles), 4 patients due to fertilization failure or failed embryo development, and 3 patients for other reasons. Therefore, a total of 328 patients were randomized into the study group (n = 165) or the control group (n = 163) on the day of ET. None of these patients were lost at follow-up; therefore, a total of 165 patients in the study group and 163 patients in the control group were included in the main analyses.

This randomized controlled study explored the possible effect of a mid-luteal GnRHa bolus on IRs in IVF/ICSI patients. Overall, no significant differences were seen in implantation and clinical pregnancy rates. In contrast, significantly higher mid-luteal steroid levels were seen in the group of patients who received a mid-luteal bolus of GnRHa (Table 2). Two cases of moderate-late onset OHSS occurred; one case in the study group and one in the control group. Both of these cases were associated with twin pregnancies (Table 3).

Several studies have previously shown a positive effect of GnRHa administration during the LP. The evidence for its efficacy is, however, of low quality due to the heterogeneity between trials (21–24). In the present study, the IR, OP, and LB rates were slightly higher in the GnRHa group compared to the control group [27 vs. 23% OR 1.2 (95% CI 0.9–1.7), P < 0.27], [37 vs. 31% OR 1.3 (95% CI 0.8–2.1), P < 0.23], and [36 vs. 31% OR 1.3 (95% CI 0.8–2.0), P < 0.27], respectively; however, the differences fell short of reaching statistical significance (Table 3). However, an absolute risk difference of 5% in LB rate - although not significant - is clinically relevant. In fact, if sample size had been calculated taking OP as the main outcome, which could be regarded more relevant in IVF cycles than IR, we would have needed twice the number of patients included in this study and results might have been statistically significant in favor of the intervention group (at least this is the trend).

In 2004, Tesarik et al. demonstrated that a single supplementary bolus of a GnRHa as LPS in oocyte recipient cycles without pituitary desensitization-improved the IR (36.9 vs. 25.1%) and LB (31.1 vs. 21.5%). In 2006, the same investigators evaluated the effects of 0.1 mg triptorelin administration 6 days after ICSI in GnRHa (n = 300) and GnRH antagonist (n = 300) cycles. IRs were significantly increased with both regimens. However, only in GnRH antagonist cycles a significant increase in the OP was observed (16). Subsequent studies demonstrated that LP GnRHa supplementation seemed to benefit the GnRH antagonist co-treated cycle (8, 16, 25–27) more than the long GnRHa co-treated cycle (18, 19, 28). The main reason for this is the down-regulation of the pituitary, which is induced by the long GnRHa protocol (28). In support of this conclusion, Kung et al. (29) recently verified that luteal GnRHa administration significantly increased clinical pregnancy and LB rates in GnRH antagonist co-treated patients, but not in patients undergoing a long GnRHa down-regulation protocol. Interestingly, Kung et al. (29) suggested that patients with higher basal FSH levels (>8 mIU/mL) and reduced numbers of mature oocytes (≤3) might have better outcomes when receiving LP support with GnRHa.

In the present study, although we demonstrated a significant increase in mid-luteal endogenous gonadotropins and steroids in the study group, we failed to reject the null hypothesis in terms of implantation and pregnancy rates. A subsequent question remaining to be answered is if there were any confounding factors impacting the clinical effects of the intervention. To explore this issue, we conducted a multivariate regression analysis, exploring variables that could independently correlate with the occurrence of OP and LB in the luteal GnRHa arm. Implantation was not chosen as an outcome parameter of the model, as the transfer of more than one embryo could partially compensate for early pregnancy losses, moreover, the prediction of OP and LB was more clinically relevant. In the study population (GnRHa group), 37.6% (62/165) of women had an OP and 36.4% (60/165) had a LB. The results of the multivariate regression (Table 4) show that the ovarian response might confound the relationship between the mid-luteal GnRHa dose and the pregnancy outcome. The NR patient was three times more likely to have OP [P < 0.005, OR 3.18 (95% CI 1.41–7.17)] and LB [P < 0.002, OR 3.37 (95% CI 1.56–7.27)], respectively, compared to the HR patient. Indeed, the OP rate increased from 9.70% (16/165) in HR to 27.88% (46/165) in NR and the LB rate increased from 9.70% (16/165) in HR to 26.66% (44/165) in NR. The reason for this discrepancy may be attributed to a possible threshold of mid-luteal steroids compatible with optimal endometrial receptivity (30–35). Moreover, the availability of good embryo quality, a pre-ovulatory, and mid-luteal LH activities were also associated with increased odds for both OP and LB. Finally, as expected, the number of embryos transferred was independently associated with the occurrence of OP (Table 4). The areas under the ROC curve for OP and LB were 0.80 and 0.70, respectively, indicating that both models fit moderately well and are clinically useful, and as results provide a tool with which to inform patients of the prognosis after the mid luteal addition of the GnRHa. Thus, the addition of a mid-luteal bolus GnRHa to NR (<14 follicles) but not to HR (14–30) seems to be the fine-tuning sought to optimize the protocol suggested by Humaidan et al. (3, 4) in terms of reproductive outcome in both categories. Whereas, in the very HR (>30 follicles) category, which was excluded from the current study, the freeze all strategy remains best option in our opinion.

The exact mechanism behind the presumed beneficial effect of LP GnRHa administration remains poorly defined. It has been hypothesized that GnRHa either supports CL function by inducing LH secretion by the pituitary gonadotrophic cells (8, 26) or stimulates the endometrial GnRH receptors (36). Tesarik et al. (15) postulated a direct effect of GnRHa on the embryo, as suggested by an increase in β-HCG secretion. The latter mechanism does not seem to be supported by our study because there were no significant differences in hCG levels between study and control groups (Table 2). Interestingly, the findings of the present study show that a bolus of GnRHa, when administered 6 days after OPU in GnRH antagonist cycles, is able to induce a surge of pituitary gonadotropins (FSH and LH), eliciting an increase in steroid production (E2 and P) by the CL (Table 2). The above-mentioned hormone variations were still detectable within 24 h following the GnRHa administration; 14–16 h in our protocol precisely. Thus, the increase of FSH, LH, P, and E2 was significantly higher in the study group compared to the control group (Table 2), which is in agreement with previous reports (8, 26). However, similar changes in the hormonal luteal profile have not been reported in other studies (16, 37).

In terms of safety, we did not observe any increase in the development of OHSS in GnRHa-treated patients, which is in agreement with Tesarik et al. (16). Thus, this regimen could be considered a safe approach to support the LP with LH activity when comparing with mid-luteal addition of low dose of hCG (38).

The limitations of the present study are, firstly, from a clinical viewpoint, OP and LB are more relevant outcomes in IVF cycles than IR (39, 40). IR will equate with clinical pregnancy rate only when single embryo transfer (SET) is used (41). In the present study, only 5% (19/328) of cycles were SET cycles. If a clinical pregnancy was considered as the main outcome, the number of patients required would be twice that of the present study (more than 600), which would be beyond the capacity of a single fertility center. Secondly, as the secondary data analyzed used small numbers reported from a single center, this limited generalizability and so these data should be interpreted with caution and be considered only as preliminary to a larger future study.

In conclusion, the present study demonstrates that a single dose of GnRHa administered 6 days after OPU in GnRHa-triggered IVF cycles supplemented with a small bolus of hCG on the day of OPU did not improve reproductive outcomes in the study group as compared to controls, despite the fact that significantly higher mid-luteal endogenous gonadotropins and steroid levels were present in the study group. Results of the multivariate regression analysis performed in the luteal GnRHa population support the concept of “individualized” ovarian response-based luteal support, suggesting that the above-mentioned regimen appears to be a promising alternative to provide an optimal level of LH activity throughout the early-mid-LP to maximize each patient’s chance of a pregnancy with minimal safety issues. The findings need further corroboration in a large-scale study.

All patients provided written informed consent to participate in the study, which was conducted in accordance with the Declaration of Helsinki and Good Clinical Practice. The research project was approved by the Ethics Committee of the University hospital Centre Ibn Badis, Constantine. Moreover, the study was registered in ClinTrial.gov, Number: 02053779.

ABenmachiche and PH designed the study, drafted, and edited the manuscript. ABenmachiche performed data collection, handling of data, and statistical analysis. AZ, SB, and ABoularak actively participated in data acquisition, handling of data and manuscript drafting. All co-authors accepted the final draft.

ABenmachiche, SB, AZ, and ABoularak have nothing to disclose. PH reports personal fees from Merck, MSD, and IBSA, grants from Merck, MSD, and Ferring, outside the submitted work.