Comparisons of GnRH antagonist protocol versus GnRH agonist long protocol in patients with normal ovarian reserve: A systematic review and meta-analysis.

Since gonadotropin-releasing hormone agonist (GnRH-a) was developed in the 1980s [ 1 ], it has played an important role in controlled ovarian hyperstimulation (COH) among patients who are undergoing assisted reproductive technology (ART). The advantage of GnRH agonist is to prevent premature luteinizing hormone (LH) surge, thereby increasing the number of retrieved oocytes and pregnancy rates and decreasing the number of cycle cancellations [ 2 , 3 ]. These advantages, however, may lead to ovarian hyperstimulation syndrome (OHSS) or other side effects [ 4 ]. GnRH antagonist (GnRH-ant), which was discovered in the 1990s, can competitively block GnRH receptors and cause rapid suppression of Gn release [ 5 ]. This protocol has fewer complications and is more convenient for patients because of the shorter treatment time and fewer injections [ 6 ]. However, its effectiveness is still debated. Multiple studies, including meta-analyses and randomized controlled trials (RCTs), of the GnRH-a protocol and GnRH-ant protocol on pregnancy rate and live birth rate have yielded controversial findings [ 6 – 8 ]. A 2006 Cochrane systematic review of 27 RCTs showed that GnRH-ant protocol has a significantly lower clinical pregnancy rate and live birth rates than those in GnRH-a long protocol, while the incidence of OHSS is significantly lower in GnRH-ant protocol [ 9 ]. However, a 2011 Cochrane systematic review of 45 RCTs found that there was no significant difference in the live birth rates between the GnRH-a and GnRH-ant groups [ 10 ]. A recent Cochrane systematic review of 73 RCTs in 2016 also concluded that these two protocols have equivalent live birth rates, and GnRH-ant protocol has a lower incidence of OHSS [ 11 ]. The finding that GnRH-ant protocol reduces the pregnancy rate may result from the fact that some centers only choose GnRH-ant protocol as their second treatment option in COH, or use it to treat the patients with an unfavorable prognosis, such as repeated implantation failures, older patients, and low responders [ 12 ]. This study’s purpose is to determine the effectiveness and safety of GnRH-a long protocol and GnRH-ant protocol among patients with normal ovarian reserve to unify the influencing factors.

The initial literature search yielded 1,304 studies. Screening of the titles and abstracts resulted in 68 published articles that could possibly compare the GnRH-a long protocol and GnRH-ant protocol in patients with normal ovarian reserve. There is no official definition of normal ovarian reserve. Therefore, the studies included in our article mainly depended on a consensus, which met any two of the following criteria: Age < 40 years, normal menstrual cycle, basal FSH ≤ 10 IU/L, basal E2 5, previous IVF attempts < 3, no previous poor response/OHSS history, no polycystic ovary syndrome (PCOS) / severe endometriosis. After reading the full papers, only 29 studies (6399 patients) finally met our inclusion criteria [ 8 , 14 – 41 ]. Although the cited studies used different drugs and doses for COH, they all belong to either GnRH-ant protocol or GnRH-a protocol, so we included these studies in a general way. The literature screening process and the results are shown in Fig 1 , and the basic characteristics of the included papers are shown in Table 1 . The quality assessment results are shown in S1 Table . From : Moher D, Liberati A, Tetzlaff J, Altman DG, The PRISMA Group (2009). P referred R eporting I tems for S ystemetic Reviews and M eta- A nalyses: The PRISMA Statement. PLoS Med 6(7): e1000097. doi: 10.1371/journal.pmed1000097 . For more information, visit www.prisma-statement.org . A total of 25 studies [ 8 , 14 – 23 , 25 – 29 , 32 – 39 , 41 ](5,814 cases) were included in the clinical pregnancy rate meta-analysis. There was no heterogeneity (P = 0.91, I 2 = 0%) among the trials; therefore, the fixed-effects model was used for the meta-analysis. The results indicated that there was no statistically significant difference in the clinical pregnancy rate between the GnRH-ant group and the GnRH-a long-protocol group (OR [95% CI] = 0.90 [0.80, 1.01], P = 0.08, Fig 2 ). A total of 18 studies [ 8 , 14 – 20 , 23 , 28 , 30 , 31 , 34 , 36 – 39 , 41 ](5,119 cases) were included in the ongoing pregnancy rate meta-analysis. There was no heterogeneity (P = 0.96, I 2 = 0%) among the trials; therefore, the fixed-effects model was used for the meta-analysis. The results indicated that there was no statistically significant difference in the ongoing pregnancy rate between the GnRH-ant group and the GnRH-a long-protocol group (OR [95% CI] = 0.88 [0.77, 1.00], P = 0.05, Fig 3 ). A total of 6 studies [ 8 , 14 , 31 , 35 , 37 , 41 ](2,237 cases) were included in the live birth rate meta-analysis. There was no heterogeneity (P = 0.88, I 2 = 0%) among the trials; therefore, the fixed-effects model was used for the meta-analysis. The results indicated that there was no statistically significant difference in the live birth rate between the GnRH-ant group and the GnRH-a long-protocol group (OR [95% CI] = 0.95 [0.74, 1.09], P = 0.27, Fig 4 ). A total of 17 studies [ 8 , 14 , 16 , 21 – 27 , 31 , 33 – 35 , 39 – 41 ](3,171 cases) were included in the number of stimulation days meta-analysis. There was heterogeneity (P<0.00001, I 2 = 96%) among the trials; therefore, the random-effects model was used for the meta-analysis. The results indicated that the stimulation days were statistically significantly fewer in the GnRH-ant group than in the GnRH-a long-protocol group (MD [95% CI] = -0.8 [-1.36, -0.23], P = 0.006, Fig 5 ). A total of 18 studies [ 8 , 14 , 16 , 21 – 27 , 31 – 36 , 40 , 41 ](3,424 cases) were included in the Gn dosage meta-analysis. There was heterogeneity (P<0.00001, I 2 = 94%) among the trials; therefore, the random-effects model was used for the meta-analysis. The results indicated that the Gn dosage was statistically significantly less in the GnRH-ant group than in the GnRH-a long-protocol group (MD [95% CI] = -3.52 [-5.56, -1.48], P = 0.0007, Fig 6 ). A total of 6 studies [ 23 , 24 , 27 , 35 , 36 , 40 ](698 cases) were included in the meta-analysis of endometrial thickness on the day of HCG administration. There was no heterogeneity (P = 0.68, I 2 = 0%) among the trials; therefore, the fixed-effects model was used for the meta-analysis. The results indicated that there was no statistically significant difference in the endometrial thickness on the day of HCG administration between the GnRH-ant group and the GnRH-a long-protocol group (MD [95% CI] = -0.06 [-0.22, 0.11], P = 0.5, Fig 7 ). A total of 15 studies [ 8 , 14 , 16 , 22 – 27 , 32 – 34 , 36 , 39 , 40 ](1,770 cases) were included in the E2 level on the day of HCG administration meta-analysis. There was heterogeneity (P<0.00001, I 2 = 91%) among the trials; therefore, the random-effects model was used for the meta-analysis. The results indicated that the E2 level on the day of HCG administration was statistically significantly lower in the GnRH-ant group than in the GnRH-a long-protocol group (MD [95% CI] = -365.49 [-532.93, -198.05], P<0.0001, Fig 8 ). A total of 22 studies [ 14 – 17 , 20 , 21 , 23 – 27 , 30 – 36 , 38 – 41 ](4,919 cases) were included in the meta-analysis of the number of oocytes retrieved. There was heterogeneity (P = 0.01, I 2 = 45%) among the trials; therefore, the random-effects model was used for the meta-analysis. The results indicated that the retrieved oocytes were statistically significantly fewer in the GnRH-ant group than in the GnRH-a long-protocol group (MD [95% CI] = -1.41 [-1.84, -0.99], P<0.00001, Fig 9 ). A total of 11 studies [ 16 , 17 , 22 – 24 , 27 , 30 , 32 , 33 , 35 , 38 ](1,588 cases) were included in the meta-analysis of the number of embryos obtained. There was no heterogeneity (P = 0.68, I 2 = 0%) among the trials; therefore, the fixed-effects model was used for the meta-analysis. The results indicated that the obtained embryos were statistically significantly fewer in the GnRH-ant group than in the GnRH-a long-protocol group (MD [95% CI] = -0.99 [-1.38, -0.59], P<0.00001, Fig 10 ). A total of 21 studies [ 8 , 14 – 19 , 23 – 25 , 28 , 29 , 31 – 33 , 35 – 39 , 41 ](5,763 cases) were included in the incidence of OHSS meta-analysis. There was no heterogeneity (P = 0.27, I 2 = 15%) among the trials; therefore, the fixed-effects model was used for the meta-analysis. The results indicated that the incidence of OHSS was statistically significantly lower in the GnRH-ant group than in the GnRH-a long-protocol group (OR [95% CI] = 0.69 [0.57, 0.83], P<0.0001, Fig 11 ). A total of 14 studies [ 8 , 14 – 20 , 23 , 32 , 34 – 37 ](3,198 cases) were included in the miscarriage rate meta-analysis. There was no heterogeneity (P = 0.94, I 2 = 0%) among the trials; therefore, the fixed-effects model was used for the meta-analysis. The results indicated that there was no statistically significant difference in the miscarriage rate between the GnRH-ant group and the GnRH-a long-protocol group (OR [95% CI] = 0.98 [0.69, 1.40], P = 0.93, Fig 12 ). A total of 19 studies [ 8 , 14 – 20 , 23 , 25 , 29 – 31 , 33 – 36 , 38 , 41 ](5,209 cases) were included in the cycle cancellation rate meta-analysis. There was heterogeneity (P = 0.004, I 2 = 53%) among the trials; therefore, the random-effects model was used for the meta-analysis. The results indicated that there was no statistically significant difference in the cycle cancellation rate between the GnRH-ant group and the GnRH-a long-protocol group (OR [95% CI] = 0.86 [0.52, 1.44], P = 0.57, Fig 13 ). The sensitivity analysis was performed by excluding the maximum weight studies [ 8 , 17 , 31 , 36 , 37 , 41 ] in each outcome. The results showed that there was no influence on the pooled OR value; therefore, the outcomes are stable. Begg’s funnel plots were symmetrical, and Begg’s tests had no significant publication bias (Begg’s test P>0.05, S1 – S12 Figs).

In conclusion, our meta-analysis demonstrated that GnRH antagonist protocol substantially decreased the incidence of OHSS without influencing the pregnancy rate and live birth rate compared to GnRH agonist long protocol among patients with normal ovarian reserve. Therefore, for patients with normal ovarian reserve, GnRH-ant protocol should play a more important role in COH when clinicians are making individualizing and optimizing treatment decisions.

“GnRH agonist”, “GnRHa”, “GnRH antagonist”, “GnRH-ant”, “GnRHA”, “randomized controlled trial”, “RCT”, and “Normal ovarian reserve” were used as the keywords for the literature searches in the PubMed (1992–2016), Cochrane Library (1999–2016), Web of Science (1950–2016), Chinese Biomedical Database (CBM,1979–2016), and China National Knowledge Infrastructure (CNKI,1994–2016) databases. The retrieval time was from the first publication of the journal to the end of December 2016. References included in the studies were also searched. Inclusion criteria were RCTs that compared the effectiveness and safety of GnRH-a long protocol and GnRH-ant protocol in patients with normal ovarian reserve. Exclusion criteria included failure to report appropriate randomized procedures, classification of participants as low or high ovarian response or endometriosis, and unclear or inappropriate outcomes. Editorials, letters to the editor, review articles, case reports and animal experimental studies were also excluded. Studies were screened by two reviewers (R.W. and Y.W.) independently, and any disagreement was settled by consensus. First, the title and abstract of each study was read carefully to exclude the studies that clearly did not meet the inclusion criteria. Then, the full text of the remaining studies was read to determine which studies would be included in this study. The quality assessment of RCTs was compiled using Cochrane’s risk of bias tool [ 13 ], which included sequence generation, allocation concealment blinding of participants, personnel and outcome assessors, incomplete outcome data, selective outcome reporting, and other sources of bias. The main efficacy outcome measures included the clinical pregnancy rate (defined as the presence of a gestational sac on ultrasound or gestational sac with fetal heart tones), the ongoing pregnancy rate (determined as pregnancies with over 12 weeks of gestation), and the live birth rate. The secondary efficacy outcome measures were ovarian stimulation outcomes, which included stimulation days, gonadotropin dosage, endometrial thickness and estradiol (E2) level on the day of human chorionic gonadotrophin (HCG) administration, the number of oocytes retrieved, and the embryos obtained. The safety outcome measures included the incidence of OHSS, the miscarriage rate, and the cycle cancellation rate. All statistical analyses were performed using Revman 5.3 software. Dichotomous outcomes were expressed as odds ratios (ORs) with 95% confidence intervals (CIs). Continuous variables were expressed as weighted mean differences (MDs) with 95% CIs. Heterogeneity was evaluated using the Q-test and I 2 -index values, and reported for each outcome as a P-value and percentage, respectively. If heterogeneity was adopted (I 2 50% with P>0.1), meta-analysis used a fixed-effects model. Otherwise (I 2 ≥50% or I 2 >25% with P≤0.1), meta-analysis used a random-effects model. Sensitivity analysis was used to determine the stability of the results. Begg’s funnel plot and Begg’s test were used to assess publication bias by STATA software (version 12.0, Stata Corp).

(DOC) Click here for additional data file. (DOC) Click here for additional data file. (TIF) Click here for additional data file. (TIF) Click here for additional data file. (TIF) Click here for additional data file. (TIF) Click here for additional data file. (TIF) Click here for additional data file. (TIF) Click here for additional data file. (TIF) Click here for additional data file. (TIF) Click here for additional data file. (TIF) Click here for additional data file. (TIF) Click here for additional data file. (TIF) Click here for additional data file. (TIF) Click here for additional data file.