{"paper_id":"89fcdc65-ec0b-4329-b66d-e2ccedcd3845","body_text":"Freeze-all protocol along with frozen-thawed embryo transfer (FET) strategies is expanding\nsubstantially because of effective cryopreservation techniques and convenience. This\ntechnique reduces the risk of ovarian hyperstimulation syndrome and increases endometrial\nreceptivity for embryos. Moreover, the duration of the preimplantation genetic testing\nprocess is allowed before the transfer ( Singh  et\nal.,  2020 ). Several studies revealed that live birth and clinical\npregnancy rates were significantly increased in the FET compared with fresh cycle transfer\n( Shapiro  et al.,  2011 ;  2013 ;  2014 ;  Özgür  et al.,  2015 ).\nConsequently, there is a tendency toward using embryo freezing with FET rather than fresh\ntransfer. One metaanalysis suggests that FET was shown to lower the risk of poor neonatal\noutcomes such as preterm birth and low birth weight ( Sha\n et al.,  2018 ).\nFET should be done during an appropriate endometrial window, because synchronization\nbetween the developing embryo and receptive endometrium is required to achieve implantation\n( Fazleabas & Strakova, 2002 ). Before the FET,\nendometrial preparation methods which consist of natural cycle, modified natural cycle, and\nartificial cycle were opted and applied. It is still undetermined which endometrial\npreparation protocol is more beneficial to yield a high live birth rate ( Groenewoud  et al.,  2013 ). However, the\nartificial cycle is scheduled and less monitored.\nCurrently, evidence has suggested gonadotrophin-releasing hormone (GnRH) activity in the\nconceptus. It was reported that identified GnRH receptors in embryos at the morula and\nblastocyst stages reflect a role for GnRH in embryonic development and probable implantation\n( Casañ  et al.,  1999 ;  Kawamura  et al.,  2005 ). Additionally,\nGnRH and its receptor expression have been detected in endometrium, with greater expression\nin epithelial cells during the luteal phase than in the proliferative phase ( Raga  et al.,  1998 ).\nSeveral studies have discovered that a single dose of GnRH agonist (GnRHa) during the\nluteal phase has a positive influence on pregnancy outcomes such as implantation, clinical\npregnancy, and live birth ( Tesarik  et al., \n2004 ;  2006 ;  Isik  et al.,  2009 ;  Razieh\n et al.,  2009 ). Recent meta-analyses have shown some inconsistent\nresults regarding the benefit of GnRHa as a result of different embryo transfer protocols\nand various regimens of GnRH administrations ( Oliveira\n et al.,  2010 ;  Martins  et\nal.,  2016 ;  Li & Li, 2018 ;\n Chau  et al.,  2019 ;  Song  et al.,  2020 ). Furthermore, no\nprior meta-analyses have intended to evaluate the effect of GnRHa in the FET cycle and\nemphasize randomized controlled trials (RCTs) with single-dose regimens.\nThe purpose of this systematic review and metaanalysis was to summarize the data from RCTs\nto assess the impact of GnRHa given once for the FET cycle on pregnancy outcomes.\n\nThis systematic review protocol was registered at the International Prospective Register\nof Systematic Reviews (PROSPERO) and accepted with registration number CRD42021291651.\nSince this is a systematic review, this protocol is exempt from review by the Research\nEthics Committee of the Faculty of Medicine, Chiang Mai University.\nAll published or abstract reports of RCTs, including parallel group and cross-over\nstudies, were considered eligible for review. When cross-over trials were included, all\ndata from all treatment protocols for each participant were analyzed. All RCTs that assess\npregnancy outcomes after receiving an additional single dose of GnRHa injection compared\nwith practical luteal phase support during FET cycles regardless of endometrial\npreparation technique. Any RCTs including fresh embryos or more than one dose of GnRHa\ninjection were excluded.\nThe primary pregnancy outcome was clinical pregnancy rate. Secondary outcomes consisted\nof positive pregnancy rates (or other similar terms used in the studies), miscarriage\nrates, implantation rates, ongoing pregnancy rates, live birth rates, and extrauterine\npregnancy rates (or ectopic pregnancy).\nFor the search strategy, an online literature search of databases in EMBASE, PubMed, and\nCochrane Controlled Trials Register (CENTRAL) was conducted, with no language limitation,\nfrom the date of database inception to March 31, 2022. These terms were used:\n((frozen-thawed) OR (frozen) OR (freezing) OR (freeze) OR (cryopreservation) OR\n(cryopreservative)) AND (embryo transfer) AND ((GnRH) OR (gonadotropin-releasing hormone)\nOR (buserelin) OR (goserelin) OR (leuprolide) OR (nafarelin) OR (triptorelin)). In\naddition, the manual-searched method was performed to recruit more studies that had the\npotential to be eligible among references of articles or similar reviews.\nTwo reviewers (P.C. and N.J.) independently screened the titles and abstracts of each\ntrial to retrieve interesting articles. Subsequently, full-text articles were contemplated\nbeing possibly eligible, and then these trials were scrutinized for eligibility. The\nreferences listed in previous meta-analyses were checked and compared with our search,\nensuring that all related studies were found. Consensus with the participation of another\nreviewer (U.S.) was made if there was a disagreement between the reviewers.\nTwo reviewers independently completed data extraction from the included trials. When a\nmultiple records study was identified, the most recent and most detailed published data\nwas chosen. We tried to contact the corresponding authors of each trial by e-mail if more\ninformation from their trials was required.\nThe extracted data included authors, institution or center, country, study period,\nethical approval, source of funding, conflicts of interest, randomization method, study\ndesign, enrollment period, eligibility criteria, exclusion criteria, the number of\nparticipants, mean age, body mass index, the number of embryos transferred per woman,\nendometrial preparation method and regimen of luteal phase support. Clinical pregnancy is\nour primary outcome which is characterized by the detection of an intrauterine gestational\nsac with or without ultrasound-confirmed fetal heart activity. Secondary outcomes included\nchemical pregnancy or other terms that suggest similar meanings such as positive pregnancy\nand positive beta-human chorionic gonadotropin (β-hCG), ongoing pregnancy,\nmiscarriage or abortion, live birth, and extrauterine pregnancy.\nTwo reviewers (P.C. and N.J.) assessed the risk of bias independently by following a\nrevised Cochrane risk of bias tool for randomized trials (RoB 2) ( Sterne  et al.,  2019 ). RoB 2 assessment of five domains\nof bias included the following 1) Risk of bias arising from the randomization process, 2)\nRisk of bias due to deviations from the intended interventions, 3) Risk of bias due to\nmissing outcome data, 4) Risk of bias in the measurement of the outcome and 5) Risk of\nbias in the selection of the reported result. Overall risk-of-bias judgments among studies\nwere classified as “low risk of bias”, “some concerns”, or “high risk of bias”.\nDisagreements between the two reviewers were reconciled by consensus.\nAll data were pooled and analyzed using Review Manager version 5.4.1 software (The\nCochrane Collaboration, United Kingdom, 2020). The effects of the interventions from each\neligible study were reported as the risk ratio (RR) and the 95% CI was used to evaluate\nthe precision of the estimates. The heterogeneity analysis between studies was assessed by\nthe  I 2  test.  I 2  less than 50%\nindicated low heterogeneity.\n\nOf the 1594 publications initially identified. Following the removal of duplicated and\nexcluded studies, six eligible RCTs were included for analysis. A PRISMA flow diagram of\nstudy identification and selection is depicted in  Figure\n1 .\nFigure 1 Prisma flow diagram of study identification and selection.\nPrisma flow diagram of study identification and selection.\nThe characteristics of the six RCTs are illustrated in  Table 1 . All studies reported clinical pregnancy ( Ben-Ami  et al.,  2015 ;  Davar\n et al.,  2015 ;  Seikkula\n et al.,  2016 ; 2018;  Ye\n et al.,  2019 ;  Wang  et\nal.,  2021 ), and five studies reported chemical pregnancy (other terms;\npositive pregnancy rate ( Seikkula  et al., \n2016 ;  2018 ), beta-hCG positive rate (Ye\n et al.  , 2019), and biochemical pregnancy rate ( Wang  et al.  , 2021 ) and abortion ( Davar  et al.  , 2015 ;  Seikkula  et al.  , 2016 ;  2018 ; Ye  et al.  , 2019;  Wang  et al.  , 2021 ). Four studies\nreported implantation ( Ben-Ami  et al.  ,\n2015 ;  Davar  et al.  , 2015 ;\nYe  et al.  , 2019;  Wang  et\nal.  , 2021 ). Two studies and other two studies reported ongoing\npregnancy ( Davar  et al.  , 2015 ; Ye\n et al.  , 2019), and live birth ( Seikkula  et al.  , 2016 ;  2018 ), respectively. Miscarriage and extrauterine pregnancy were reported in\nfive ( Davar  et al.,  2015 ;  Seikkula  et al.,  2016 ;  2018 ; Ye  et al.  , 2019;  Wang  et al.  , 2021 ), and three studies\n( Seikkula  et al.  , 2016 ;  2018 ;  Wang  et\nal.  , 2021 ), respectively.\nCharacteristics of Included Randomized Controlled Trials\nFET, frozen-thawed embryo transfer; GnRHa, gonadotropin releasing hormone agonist;\nBMI, body mass index; IVF,  in vitro  fertilization; ICSI,\nintracytoplasmic insemination; hCG, human chorionic gonadotropin; CPR, clinical\npregnancy rate; Ong. PR, ongoing pregnancy rate; Imp., implantation; Abor.,\nabortion; Chem. PR, chemical pregnancy rate; Biochem. PR, biochemical pregnancy\nrate; LBR, live birth rate; PPR, positive pregnancy rate; Misc., miscarriage; PR,\npregnancy rate; N/A, not available.\nHowever, one study ( Ben-Ami  et al.  ,\n2015 ) could be found only a published abstract online. Unfortunately, we were\nunsuccessful in having their full paper, resulting in the lack of in-depth details.\nAccording to the six studies, the implantation rate was reported in four studies.  Davar  et al.  (2015)  reported the\nimplantation rate in terms of several cycle transfers, whereas others provided a unit of\nthese parameters as the total number of participants. With our efforts, we could not\nreceive more data on the cycle and patient numbers for extraction since no raw data were\navailable. Consequently, only three studies were included to evaluate the effect on\nimplantation rate ( Ben-Ami  et al.  ,\n2015 ; Ye  et al.  , 2019;  Wang\n et al. , 2021 ).\nIn overall bias, five of six RCTs were rated as “low risk” (83.3%), whereas the other\nreported “some concerns” for overall bias and reporting bias and attrition bias (16.7%).\nAll RCTs had a “low risk of bias” for randomization, allocation, and outcome measurement\n( Figure 2 ).\nFigure 2 Results of risk of bias assessment using RoB 2.\nResults of risk of bias assessment using RoB 2.\nA significantly higher rate of pooled clinical pregnancy was observed in the GnRHa\ngroup than the control group (52.05% [609/1170]  vs.  47.29% [507/1072];\np=0.04; RR=1.09; 95% CI = 1.00 -1.18;  I 2  = 36%;  Figure 3 ).\nFigure 3 Forest plot for meta-analysis on clinical pregnancy rate from six studies.\nForest plot for meta-analysis on clinical pregnancy rate from six studies.\nBecause of some concerns about overall bias, recalculated pooled clinical pregnancy\nrate from five studies ( Davar  et al., \n2015 ;  Seikkula  et al., \n2016 ;  Seikkula  et al., \n2018 ; Ye  et al.,  2019;  Wang\n et al.,  2021 ) was done, and it could not show statistical\ndifferences between the groups (52% [585/1123]  vs.  48.1% [495/1028];\np=0.10; RR=1.15; 95% CI=0.97-1.37;  Figure 4 ).\nFigure 4 Forest plot for meta-analysis on clinical pregnancy rate from five studies,\nexcluding  Ben-Ami  et al. \n(2015) .\nForest plot for meta-analysis on clinical pregnancy rate from five studies,\nexcluding  Ben-Ami  et al. \n(2015) .\nFive studies were included in subgroup analyses to determine the endometrial\npreparation protocol. A subgroup analysis for the natural cycle ( Ben-Ami  et al.,  2015 ;  Seikkula  et al.,  2016 ), the clinical pregnancy rate in the\ninterventional group was significantly higher compared with that in the control group.\n(43.75% [49/112]  vs.  27.35% [29/106]; p=0.01; RR=1.6; 95% CI =\n1.10-2.32;  Figure 5 ). Conversely, for the\nartificial cycle ( Davar  et al., \n2015 ;  Seikkula  et al., \n2018 ; Ye  et al.,  2019), the clinical pregnancy in the GnRHa\ngroup 50.33% (301/598) did not show a significant difference to the clinical pregnancy\nin the control group 48.59% (259/533%). The statistical outcome was a RR value of 1.07\n(95% CI=0.96-1.20;  Figure 5 ).\nFigure 5 Forest plot for meta-analysis on clinical pregnancy rate. Subgroup analysis\naccording to endometrial preparation protocol for FET between artificial and\nnatural cycles.\nForest plot for meta-analysis on clinical pregnancy rate. Subgroup analysis\naccording to endometrial preparation protocol for FET between artificial and\nnatural cycles.\nA subgroup analysis for the vaginal route progesterone ( Davar  et al.,  2015 ;  Seikkula\n et al.,  2018 ), no significant difference in clinical\npregnancy rate was observed between the groups. (p=0.34; RR=1.17; 95% CI=0.85-1.62;\n Figure 6 ).\nFigure 6 Forest plot for meta-analysis on clinical pregnancy rate from the studies using\nvaginal progesterone for luteal phase support.\nForest plot for meta-analysis on clinical pregnancy rate from the studies using\nvaginal progesterone for luteal phase support.\nThe pooled chemical pregnancy rate was comparable between the groups (40.4% [454/1123]\n vs.  39.5% [406/1028]; p=0.9; RR=0.99; 95% CI = 0.91-1.09;\nI 2 =0%;  Figure 7 ).\nFigure 7 Forest plot for meta-analysis on chemical pregnancy rate.\nForest plot for meta-analysis on chemical pregnancy rate.\nThe pooled ongoing pregnancy rate was comparable between the groups (42.0% [221/526]\nvs. 42.3% [195/461]; p=0.73; RR=0.98; 95% CI = 0.85-1.12;\n I 2 =9%;  Figure 8 ).\nFigure 8 Forest plot for meta-analysis on ongoing pregnancy rate.\nForest plot for meta-analysis on ongoing pregnancy rate.\nDue to unavailable raw data, only three studies were included to evaluate the effect on\nimplantation rate. The pooled implantation rate was comparable between the groups (40.0%\n[682/1702]  vs.  37.8% [579/1529];  p =0.21; RR=1.06; 95%\nCI=0.97-1.15;  I 2 =0%;  Figure\n9 ).\nFigure 9 Forest plot for meta-analysis on implantation rate.\nForest plot for meta-analysis on implantation rate.\nThe pooled live birth rate showed higher live birth rates in the GnRHa group compared\nwith that in the group without GnRHa (29.9% [41/137] vs. 21.6% [29/134];\n p =0.12; RR=1.38; 95% CI=0.92-2.08;  I 2 =0%;\n Figure 10 ).\nFigure 10 Forest plot for meta-analysis on live birth rate.\nForest plot for meta-analysis on live birth rate.\nThe pooled miscarriage rate did not differ between the groups (8.1% [91/1123] vs. 9.1%\n[94/1028];  p =0.43; RR=0.90; 95% CI=0.68-1.18;\n I 2 =0%;  Figure\n11 ).\nFigure 11 Forest plot for meta-analysis on miscarriage rate.\nForest plot for meta-analysis on miscarriage rate.\nThe pooled extrauterine pregnancy rate was not different between the groups (0.6%\n[4/597] vs. 0.7% [4/567];  p =0.94; RR=0.96; 95% CI = 0.28 −3.29;\n I 2 =0%;  Figure\n12 ).\nFigure 12 Forest plot for meta-analysis on extrauterine pregnancy rate.\nForest plot for meta-analysis on extrauterine pregnancy rate.\n\nThe exact mechanisms of the positive effect of GnRHa during the luteal phase on pregnancy\noutcomes remain to be elucidated. GnRHa may directly affect either endometrial stroma and\ncells or preimplantation embryos. It has been shown that GnRH gene expression and messenger\nribonucleic acid (mRNA) of GnRH receptors appear in the human placenta and regulate hCG\nproduction throughout pregnancy ( Lin  et\nal.,  1995 ). Besides, GnRH also plays a role in regulating the balance\nbetween tissue-specific inhibitors of matrix metalloproteinases and matrix\nmetalloproteinases expression in decidual cells ( Chou\n et al.,  2003 ).\nCertainly, GnRHa was described in several studies regarding the improvement of endometrium\nreceptivity and embryo development ( Raga  et\nal.,  1998 ;  Casañ  et\nal. , 1999 ;  Nam  et al. ,\n2005 ;  Li  et al. ,\n2022 ).There are several GnRHa studies, which show pregnancy outcomes. A study showed\na higher implantation rate after a single-dose GnRHa injection, six days following ICSI of\ndonated oocytes but no difference in pregnancy rate ( Tesarik\n et al.,  2004 ). In 2006, another study from previous\ninvestigators conducted subsequent trials with a similar protocol with autologous oocytes\nand then reported a significantly greater implantation and birth rate in the GnRHa group\n( Tesarik  et al. , 2006 ). Ata et al.\nreported no improvement in any pregnancy outcomes after adding a single dose of triptorelin\n0.1 mg six days after ICSI following the long GnRHa protocol of ovarian stimulation ( Ata  et al. , 2008 ).\nOur present study evaluated six RCTs including a total of 2242 participants. The\nmeta-analysis results reveal a marginal benefit on clinical pregnancy rate in participants\nreceiving GnRHa. Nevertheless, regardless of one RCT with limited data, the clinical\npregnancy rate was comparable between patients in the GnRH group and those in the control\ngroup. Lately, there has been a randomized clinical pilot study that recruited 156 patients\nto investigate the effect of the extra single dose of GnRHa ( Liu  et al.,  2023 ). The authors concluded that insignificant\ndifferences in all pregnancy outcomes of artificial cycle frozen embryo transfers were\nobserved. We estimate that our pregnancy outcomes would not improve considerably if data\nfrom the recent RCT is involved.\nOur results displayed a significantly higher clinical pregnancy rate among the GnRH group\nwith the natural cycles. In the case of existing corpus luteum development in the\nnon-artificial endometrial preparation, the additional GnRHa injection may aid progesterone\nsecretion during early pregnancy. GnRHa supplement plausibly serves to restore luteal phase\nfunction, which could affect serum LH levels, thereby maintaining corpora lutea ( Fujii  et al.,  2001 ;  Pirard  et al.  , 2005 ).\nContrary to previous evidence, GnRH acts as a luteolytic factor by promoting apoptosis in\nluteinized granulosa cells, resulting in decreased progesterone release ( Metallinou  et al. , 2007 ). Moreover,\nstudies suggested that desensitization of GnRH receptors could occur by a luteolytic effect\nof GnRHa, and the use of GnRHa led to a decline in the functioning of the corpus luteum\n( Lemay  et al.,  1983 ;  Herman  et al.,  1992 ).\nIn the artificial cycle, hormonal replacement for endometrial preparation may impair\nendometrial receptivity by earlier closure of the implantation window at high estradiol\nlevels ( Wu  et al.,  2021 ). In vaginal\nprogesterone used, the subgroup analysis demonstrated no beneficial effect of additional\nGnRHa administration. Since only two trials with a small number of participants were\nincluded, it is still unclear whether the route of progesterone affects pregnancy\noutcomes.\nThe definition in this study followed the International Glossary on Infertility and\nFertility Care in 2017 was used for diagnosing the clinical pregnancy by ultrasonographic\nvisualization of one or more gestational sacs or definitive clinical signs of pregnancy\nregardless of fetal heart activity ( Zegers-Hochschild\n et al.,  2017 ). This would cover all clinical pregnancy\nterminology in our six RCTs, which had different terminology of clinical pregnancy.\nConsequently, it may have an impact on clinical pregnancy rates by incorporating abnormal\nearly pregnancies such as blighted ova.\nFor secondary outcomes, the intervention group had greater normal pregnancy rates than the\ncontrol group insignificantly. Similarly, the rates of unfavorable pregnancies, including\nextrauterine pregnancy and loss, did not differ between groups.\nThere are two strengths in our study. First, we intend to pool only RCT-designed studies,\nwhich are statistically considered to be the highest quality of evidence among all types of\nclinical studies. Additionally, our meta-analysis can demonstrate a “low risk” of overall\nbias among the studies. Second, despite the small number of studies for analysis, no\nsignificant heterogeneity was observed across the studies. This indicated that the majority\nof studies had no variation in their findings.\nThis study has certain limitations. The number of eligible studies and the total number of\nparticipants were both small. Inevitably, we chose clinical pregnancy as our primary\nendpoint rather than live birth, which is widely regarded as the best performance indicator\nfor ART methods.Differences in progesterone regimens for improving the luteal phase\nreceptive endometrium could theoretically alter pregnancy rates ( Vuong  et al.,  2021 ;  Greenbaum  et al.,  2022 ). Transferred embryo quality and number\nper cycle varied. High-quality or euploid embryos enhance clinical pregnancy rates higher\nthan unqualified embryos ( Jimenez  et al., \n1997 ;  Veleva  et al.,  2013 ).\nThese can also have a significant effect on pregnancy outcomes and have been recognized as a\nweakness of this study.\n\nThis RCT-only meta-analysis is significant in that it analyzes the effect of a single-dose\nGnRHa administration for luteal phase support during the FET cycle. The findings implied\nthat an additional single-dose GnRHa administration benefits clinical pregnancy rates,\nespecially in the natural cycle FET, but did not affect other pregnancy outcomes. According\nto our research, further high-quality randomized controlled trials are required.","source_license":"CC-BY-4.0","license_restricted":false}