{"paper_id":"840f6850-3570-426f-9876-bd34ed5d553b","body_text":"Adenomyosis is a benign gynecological disease characterized by the invasion of\nendometrial glands and/or stroma within the myometrium, leading to hypertrophy and\nhyperplasia of the surrounding smooth muscle cells ( Ferenczy, 1998 ). According to a population-based cohort study with\n333,693 women, the overall incidence was reported to be 1.03% or 28.9 per 10.000\nwomen-years ( Yu  et al. ,\n2020 ).\nFor decades, adenomyosis was considered a disease that affected multiparous women\nover the age of 40 years. Nowadays, with the widespread use of TVUS (transvaginal\nultrasonography) and the implementation of the ultrasonographic features to detect\nadenomyosis, clinicians are able to diagnose the disease at earlier stages,\nespecially among nulliparous women ( Pinzauti\n et al. , 2015 ;  Harada\n et al. , 2016 ;  Tellum\n et al. , 2020 ).\nAdenomyosis is linked to infertility. Even though the rationale behind this\nrelationship remains not clearly established, several authors have reported a\nnegative impact on reproductive outcomes among these women. In a meta-analysis by\n Vercellini  et al. \n(2014) , a 28% reduction in the likelihood of clinical pregnancy and an\nincreased risk of miscarriage was reported among women with adenomyosis undergoing\nIVF compared with women without the disease.\nLater  Younes & Tulandi (2017)  confirmed in\nanother meta-analysis a 41% decrease in live pregnancy rates and an increased risk\nof miscarriage among women with adenomyosis. Consistently, in two other systematic\nreview and meta-analysis, both  Horton  et\nal.  (2019)  and  Huang\n et al.  (2020)  also described an increased risk of\nmiscarriage among these women.\nIn this regard, the use of long and ultra-long GnRH analogue protocols as\npretreatment before COS has been considered of choice by many clinicians in an\nattempt to mitigate the hyperestrogenism that results from the increased expression\nof estrogen receptors, and the functional link between estradiol action and\nadenomyotic cell proliferation ( Sztachelska\n et al. , 2022 ). However, conflicting results have\nbeen reported in terms of reproductive outcomes.\nIn the meta-analysis by ( Nirgianakis  et\nal. , 2021 ), women with adenomyosis were associated with a\nsignificant reduction in clinical pregnancy are and higher miscarriage rate after\nIVF, especially when following short GnRH agonist or antagonist protocol for\ncontrolled ovarian stimulation (COS). In contrast,  Cozzolino  et al.  (2022) , found that the use of GnRHa\ndownregulation did not show significant benefits in reproductive outcomes.\nThe aims of this systematic review and meta-analysis are to (i) evaluate the effect\nof GnRH agonist protocols for controlled ovarian stimulation, and (ii) determine\nwhether the pretreatment with GnRH agonist before frozen-thawed embryo transfer\nimprove reproductive outcomes among women with adenomyosis.\n\nThe study was exempt from Institutional Review Board approval because this was a\nsystematic review and meta-analysis. The review was made in accordance with the\nEQUATOR Reporting Guideline ( EQUATOR, 2004 ).\nThe review protocol was registered in International Prospective Register of\nSystematic Reviews (PROSPERO) (ID: CRD42024555052).\nThe systematic review spanned MEDLINE and Embase databases up to April 2024. The\nsearch combined terms and descriptors related to adenomyosis, GnRH agonist,\npituitary down regulation and embryo transfer. The search strategy was modified\nto comply with the requirements of each database consulted. The reference lists\nof all the relevant articles and overviews was screened to identify additional\nrelevant citations. No language limits were used. The complete search strategy\nis available upon request from the authors.\nThe review included randomized controlled clinical trials and cohort studies of\nwomen with adenomyosis diagnosed by ultrasound or magnetic resonance imaging\n(MRI). The types of interventions evaluated were (i) long compared to ultra-long\npituitary downregulation with GnRHa prior to controlled ovarian stimulation\n(COS), (ii) short, long or ultra-long GnRHa compared to GnRH antagonist protocol\nbefore COS, (iii) long or ultra-long GnRHa compared to no GnRHa pretreatment\nprior to FET regardless of the COS protocol followed.\nLong GnRHa protocol was defined as the use of GnRH analogues starting at the\nmid-luteal phase of the cycle prior to COS, while the ultra-long protocol was\ndefined as the use of GnRHa at least one month prior to COS.\nStudies that mixed women undergoing different GnRH agonist protocols within the\nsame group, or that included combined treatments for adenomyosis such as surgery\nor radiofrequency were also excluded.\nThe data were collected using standard forms in which the characteristics of the\nstudy design, participants, interventions, comparisons, and main outcomes were\nrecorded ( Tables 1  and  2 ). Where data were missing, the original\nreview authors were contacted for assistance. Two independent authors (E.I.A.O.\nand M.G.C.) judged study eligibility, assessed the risk of bias and extracted\ndata solving discrepancies by agreement, and if needed, reaching consensus with\na third author (M.A.C). The agreement between reviewers was analyzed using the\nweighted kappa for each inclusion criterion (Fleiss, 1993).\nDescription of the interventions by treatment group of the included\nstudies.\nNo data regarding the protocol for COS is provided the authors.\nDescription of the population included by type of intervention and\ntreatment group.\nOvarian reserve: AMH expressed in ng/ml or basal FSH in italics and\nexpressed in mlU/ml when AMH was not available; PCOS = Polycystic\nOvarian Syndrome.\nWe assessed the risk of bias in the included studies following the guidance\nsuggested by the Cochrane Collaboration for non-randomized studies, using\nROBINS-E tool ( Higgins  et al. ,\n2024 ). We addressed seven specific domains, which include i)\nconfounding, ii) measurement of exposure, iii) selection of participants into\nthe study, iv) Post-exposure interventions, v) Missing data, vi) measurement of\nthe outcome, and vii) selection of the reported results. A judgment of “Low” for\nall domains indicates a low risk of bias, a judgment of “High” for one or more\ndomains indicates a high risk of bias. We interpreted the risk of bias in the\nspecific domains as “Some concerns” when the information was unclear. The risk\nof bias for the included trials is detailed in  Fig. 1 .\nFigure 1 Assessment of the quality of the included studies (ROBINS-E).\nAssessment of the quality of the included studies (ROBINS-E).\nStatistical analyses were performed using Review Manager software (version 5.3;\nCopenhagen: Nordic Cochrane Centre, Cochrane Collaboration, 2014). The odds\nratio (OR) and 95 confidence interval (CI) was used for the outcomes described.\nThe degree of heterogeneity was evaluated through I2 statistic, and tests for\nsubgroup differences were based on random-effects models in order to minimize\nthe false-positive results ( Higgins  et\nal. , 2003 ). Thresholds for the interpretation of\nI 2  is as follows: 0-40% might not be important, 30-60% may\nrepresent moderate heterogeneity, 50-70% may represent substantial\nheterogeneity, and 75% to 100% considerable heterogeneity. From 30% to 100%, the\nimportance of the I 2  value will depend on the magnitude and direction\nof the effect as well as the strength of evidence ( p  value\n<0.10 shows statistical significance) ( Deeks\n et al. , 2024 ).\n\nA total of 50 articles were identified in the initial electronic search, but only 24\nwere considered for eligibility. During the second phase of the inclusion process,\n15 studies were excluded because they had no control group for comparison (n=5), the\ncontrol group was comprised of women without endometriosis (n=3), the target\npopulation of the study was women with endometriosis and concomitant adenomyosis\n(n=8), combined multiple agonist protocols within the same group (n=1) or combined\nsurgery with the agonist protocol (n=1). Two studies were excluded because of i) a\npotential duplicate ( Li  et al. ,\n2023 ) since it included women possibly evaluated in a separate cohort\nstudy ( Ge  et al. , 2023 ), or\nii) were unable to extract data (even though the authors had been contacted directly\n( Dian  et al. , 2022 ).\nAfter screening, 8 cohort studies were finally included in the metaanalysis. The\narticle search and screening process is shown in  Fig.\n2 . The two reviewers achieved good agreement on the selection of the\ntrials (weighted kappa 0.75).\nFigure 2 Flow chart for the trial identification and selection process.\nFlow chart for the trial identification and selection process.\nEight retrospective cohort studies were included. They were grouped according to\nthe comparison in i) ultralong versus long GnRHa protocol before COS and fresh\nET ( Chen  et al. , 2020 ;\n Hou  et al. , 2020 ;\n Lan  et al. , 2021 ;\n Wu  et al. , 2022 ;\n Ge  et al. , 2023 ),\nii) ultra-long GnRHa versus antagonist protocol before COS and fresh ET ( Park  et al. , 2016 ;  Ge  et al. , 2023 ), iii)\nlong GnRHa versus antagonist protocol before COS and fresh ET ( Ge  et al. , 2023 ), iv)\nshort GnRHa versus antagonist protocol before COS and fresh ET ( Ge  et al. , 2023 ), v) Long\nGnRH agonist pretreatment versus no treatment before FET regardless of the\nprotocol followed for COS ( Niu  et\nal. , 2013 ;  Li  et\nal. , 2022 ;  Zhang\n et al. , 2022 ).\nThe mean age across the studies was comparable. The inclusion criteria were\nrelatively homogeneous. Most studies included women under the age of 42-45\nyears, although the studies by  Hou  et\nal.  (2020)  and  Niu\n et al.  (2013) only considered for inclusion women\nunder 38 years.  Park  et al. \n(2016)  did not consider age an inclusion criteria. The\nultrasonographic criteria for diagnosing adenomyosis was also relatively\nhomogeneous across the studies, although the group of Zhang failed to provide\ninformation regarding this aspect ( Zhang\n et al. , 2022 ).\nFive retrospective cohort studies were identified that reported on reproductive\noutcomes after ultralong and long GnRH agonist protocol before COS and fresh ET\n( Chen  et al. , 2020 ;\n Hou  et al. , 2020 ;\n Lan  et al. , 2021 ;\n Wu  et al. , 2022 ;\n Ge  et al. , 2023 ). A\ntotal of 1018 cycles were included in the ultralong GnRH agonist protocol and\n594 in the long GnRH agonist protocol.\nThe inclusion and exclusion criteria were relatively homogeneous across the\nstudies ( table 2 ). Only women under 42\nyears were included, without uterine malformations, submucosal myomas or\nuntreated hydrosalpinx ( Hou  et\nal. , 2020 ;  Lan\n et al. , 2021 ;  Wu\n et al. , 2022 ;  Ge\n et al. , 2023 ), although the study by  Hou  et al.  (2020)  had more\nrestrictive criteria and only considered women under 38 and with good ovarian\nreserve (basal FSH<10 mU/mL, and AFC above 10). In the study by  Chen  et al.  (2020) \nexclusion criteria were not reported.\nConcerning the basal characteristics, women who received ultralong GnRHa protocol\nbefore COS had overall larger uterine size compared to long GnRHa protocol\n( Lan  et al. , 2021 ;\n Wu  et al. , 2022 ;\n Ge  et al. , 2023 ),\nalthough in the study by  Chen  et\nal.  (2020)  no differences were registered in this\nregard, and ( Hou  et al. ,\n2020 ) did not report on this variable. However the studies that\nregistered the type of adenomyosis of women included (focal versus diffuse)\nreported no differences between treatment groups ( Lan  et al. , 2021 ;  Wu  et al. , 2022 ).\nRegarding ovarian reserve status of women included, measured by AMH (or AFC and\nbasal FSH when AMH was no available), 4 studies reported no differences between\ngroups ( Chen  et al. ,\n2020 ;  Hou  et al. ,\n2020 ;  Lan  et al. ,\n2021 ;  Wu  et al. ,\n2022 ). In the study by  Ge  et\nal.  (2023) , a significantly lower AMH was registered\namong women in the ultralong GnRHa group ( p <0.001).\nAdditionally, in terms of coexisting of endometriosis, a significantly higher\npercentage of endometriosis ( Lan  et\nal. , 2021 ) or moderate/severe dysmenorrhea ( Ge  et al. , 2023 ) was\nreported among women in the ultralong GnRHa protocol compared to the long GnRH\nprotocol. In the study by  Wu  et\nal.  (2022) , both groups were comparable, although  Chen  et al.  (2020)  did not\nreport on this variable. The group of  Hou\n et al.  (2020)  considered endometriosis as an\nexclusion criteria ( Table 2 ).\nWith regard to the protocols reported before COS, most authors reported\nultra-long treatment with GnRH analogues over 2 to 4 months ( Chen  et al. , 2020 ;  Hou  et al. , 2020 ;  Lan  et al. , 2021 ;  Wu  et al. , 2022 ), although\nGe included women who received only one month but could extend the treatment up\nto 6 months depending on the uterine diameter ( Ge  et al. , 2023 ). The long GnRHa protocol was more\ncomparable across studies, starting 10 to 14 days before the beginning of the\nCOS ( Table 1 ).\nTwo retrospective studies evaluated reproductive outcomes between women who\nreceived ultra-long GnRHa protocol compared antagonist protocol before COS and\nfresh ET ( Park  et al. ,\n2016 ;  Ge  et al. ,\n2023 ). A total of 267 cycles were evaluated in the group of\nultra-long GnRHa protocol and 181 cycles in the antagonist protocol ( Table 2 ).\nRegarding basal characteristics, both groups of treatment across the studies were\ncomparable in terms of ovarian reserve. However, ( Ge  et al. , 2023 ) registered a\nsignificantly higher history of severe dysmenorrhea in the ultralong GnRHa\nprotocol (37.96% versus 11.76%,  p <0.001).\nNone of the studies registered uterine volumes or type of adenomyosis between\ngroups, and  Park  et al. \n(2016)  did not report on previous history of either dysmenorrhea or\ncoexisting endometriosis.\nOnly one retrospective study evaluated reproductive outcomes between women who\nreceived long GnRHa protocol before COS and fresh ET compared antagonist\nprotocol ( Ge  et al. ,\n2023 ). A total of 56 cycles of women who received long GnRHa protocol\ncompared to 34 with antagonist protocol for COS and fresh ET were included\n( Table 2 ).\nRegarding basal characteristics, women who received long GnRHa had significantly\nsmaller uterine size and better ovarian reserve compared to women in the\nantagonist protocol, and no differences in the previous history of dysmenorrhea\nwe registered.\nOnly one retrospective study evaluated reproductive outcomes between women\nundergoing short GnRHa compared to antagonist protocol before COS and fresh ET,\nwith a total of 59 and 34 cycles included, respectively ( Ge  et al. , 2023 ).\nNo differences were registered regarding ovarian reserve or previous history of\ndysmenorrhea. However, no information was given regarding the uterine volume or\nthe type of adenomyosis.\nThree retrospective studies evaluated the use of the pretreatment with GnRHa\ncompared with no pretreatment before FET ( Niu\n et al. , 2013 ;  Li\n et al. , 2022 ;  Zhang  et al. , 2022 ). In all, 399 cycles of women\nwith adenomyosis who received at least 28 days of GnRH pretreatment before FET\nwere included, while 544 underwent FET without pretreatment ( Table 2 ).\nRegarding the basal characteristics, ovarian reserve was comparable between\ngroups of treatment in all studies. Regarding the inclusion of women with\nendometriosis, only  Niu  et al. \n(2013)  reported on this variable, and described no differences\nbetween groups ( p =0.18).\nRegarding the sonographic features of adenomyosis,  Zhang  et al.  (2022)  Zhang only considered\nfor inclusion women with uterine volumes between 56 and 100cm 3 . In\nthis study, both groups were balanced regarding the size of the uterus\n(p=0.220), and the presence of focal versus diffuse adenomyosis\n( p =0.145). However, neither the group of Niu nor Li\nreported on these variables ( Niu  et\nal. , 2013 ;  Li  et\nal. , 2022 ).\nThe three studies failed to provide data regarding the stimulation protocol used\nfor COS before embryo cryopreservation. With regard to the pretreatment received\nbefore FET,  Niu  et al. \n(2013)  and  Li  et al. \n(2022)  used GnRHa during 28 days, while  Zhang  et al.  (2022)  extended the treatment\nup to 3 months. Regarding the group who received no pretreatment,  Niu  et al.  (2013)  and\n Li  et al.  (2022) \nused hormonal replacement therapy before FET, while  Zhang  et al.  (2022)  used the natural cycle\nfor endometrial preparation ( Table\n1 ).\nIn general, studies provided data regarding methodological aspects particularly\nwith reference to the inclusion and exclusion criteria, the mode of diagnosis of\nthe disease and the protocols used for either controlled ovarian stimulation or\nthe treatment received before FET, depending on the study.\nAs shown in  Figure 1 , and according to the\nROBINS-E tool ( Higgins  et al. ,\n2024 ), studies showed in general low risk of bias in terms of bias\ndue to confounding, bias arising from measurement of the exposure, bias due to\npost-exposure interventions, due to missing data and bias in measurement of the\noutcome.\nHowever, in four of the included studies, the authors ( Niu  et al. , 2013 ;  Chen  et al. , 2020 ;  Li  et al. , 2021 ;  Zhang  et al. , 2022 ) failed to provide in\ndetail some characteristics regarding the population included that would allow\ncomparison across studies, especially regarding the co-existence of\nendometriosis and the type of adenomyosis, raising some concerns regarding the\nrisk of bias in selection of participants into the study.\nIn addition, there were some concerns regarding the selection of the reported\nresults, and some authors failed to provide data for some of the main outcomes\nof interest ( Niu  et al. ,\n2013 ;  Park  et al. ,\n2016 ;  Chen  et al. ,\n2020 ;  Zhang  et al. ,\n2022 ).\nThe pooled analysis of the data from the 5 studies ( Figure 3 ) showed no significant differences in live\nbirth (OR 1.1, 95% CI 0.69-1.77), clinical pregnancy (OR 0.77, 95% CI\n0.46-1.31) or miscarriage rates (OR 1.18, 95% CI 0.81-1.73) when comparing\nthe use of ultra-long versus long GnRHa protocols for COS and fresh ET,\nHowever, a significant increase in implantation rate was detected (OR 1.24,\n95% CI, 1.03-1.,50), although only 4 of the studies reported on this outcome\n( Hou  et al. ,\n2020 ;  Lan  et\nal. , 2021 ;  Wu  et\nal. , 2022 ;  Ge\n et al. , 2023 ). The test for heterogeneity\nacross subgroup showed low heterogeneity (I 2 =0%).\nFigure 3 Reproductive outcomes after Ultra-long compared to Long GnRHa\nprotocol for COS and fresh ET.\nReproductive outcomes after Ultra-long compared to Long GnRHa\nprotocol for COS and fresh ET.\nWith regard to the use of ultra-long GnRHa versus antagonist protocols for\nCOS ( Figure 4 ), the pooled analysis of\nthe data showed no differences in live birth (OR 1.84, 95% CI 0.78-4.32),\nclinical pregnancy (OR 1.56, 95% CI 0.96-2-51) or miscarriage rates (OR\n1.37, 95% CI 0.64-2.92). In contrast, a significant increase in implantation\nrate was observed (OR 1.98, 95% CI 1.04-3.75). However, only the study by\n Ge  et al.  (2023) \nreported on both implantation and live birth rates.\nFigure 4 Reproductive outcomes after Ultra-long GnRHa compared to\nAntagonist protocol for COS and fresh ET.\nReproductive outcomes after Ultra-long GnRHa compared to\nAntagonist protocol for COS and fresh ET.\nThe test for heterogeneity across subgroup showed low heterogeneity\n(I 2 =0%).\nThe use of long GnRHa compared with the antagonist protocol, only one study\nreported this comparison ( Ge  et\nal. , 2023 ). A significant increase in live birth (OR\n2.59, 95% CI 1.03-6.52), clinical pregnancy (OR 2.91, 95% CI 1.20-7.02) and\nimplantation rates (OR 2.18, 95% CI 1.09-4.37) were observed ( Figure 5 ). However, no differences we\nobserved in the miscarriage rate (OR 0.75, 95% CI 0.19-3.04).\nFigure 5 Reproductive outcomes after Long GnRHa compared to Antagonist\nprotocol for COS and fresh ET.\nReproductive outcomes after Long GnRHa compared to Antagonist\nprotocol for COS and fresh ET.\nThe test for heterogeneity across subgroup showed low heterogeneity\n(I 2 =0%).\nOnly one study compared the use of short GnRH versus antagonist protocol for\nCOS ( Ge  et al. ,\n2023 ). The analysis showed no significant differences in terms of\nlive birth, clinical pregnancy, miscarriage or implantation rates ( Figure 6 ).\nFigure 6 Reproductive outcomes after Short GnRHa compared to Antagonist\nprotocol for COS and fresh ET.\nReproductive outcomes after Short GnRHa compared to Antagonist\nprotocol for COS and fresh ET.\nThe test for heterogeneity across subgroup showed low heterogeneity\n(I 2 =0%).\nWith regarding to the use of GnRH analogues as a pretreatment before FET\ncompared to no treatment, no significant differences were registered in\nterms of live birth, clinical pregnancy, miscarriage or implantation rates\nin the pooled analysis of the data ( Figure\n7 ).\nFigure 7 Reproductive outcomes after pretreatment with GnRHa before FET\nversus no pretreatment.\nReproductive outcomes after pretreatment with GnRHa before FET\nversus no pretreatment.\nOnly two of the three studies provided data regarding implantation ( Niu  et al. , 2013 ;\n Li  et al. , 2022 )\nand live birth rates ( Li  et\nal. , 2022 ;  Zhang\n et al. , 2022 ).\nThe test for heterogeneity across subgroup showed moderate heterogeneity\n(I 2 =47.6%), value considered non significant\n( p =0.13).\n\nThis systematic review and meta-analysis show that pituitary downregulation before\ncontrolled ovarian stimulation with either long or ultra-long protocols among women\nwith adenomyosis does not improve reproductive outcomes. In fact, the use of\nultra-long GnRHa protocols compared to either long GnRHa or antagonist protocol seem\nto significantly lower implantation rate, although results in terms of live birth\nrate were comparable. However, the use of long GnRHa protocol before COS seem to\nnegative affect reproductive outcomes compared to antagonist protocols, including\nimplantation, clinical pregnancy, live birth rates.In addition, the pretreatment\nwith GnRHa before FET compared to no treatment does not offer any significant\nbenefit in reproductive outcomes among these women.\nDuring the past decade, several authors have described how adenomyosis is associated\nwith poorer reproductive outcomes compared to women without the disease ( Vercellini  et al. , 2014 ;  Younes and Tulandi, 2017 ;  Horton  et al. , 2019 ;  Bourdon  et al. , 2021 ;  Nirgianakis  et al. , 2021 ;  Cozzolino  et al. , 2022 ).\nHowever there has been high heterogeneity within and across groups that limits the\ninternal and external validity of the results observed. Studies that evaluated IVF\nresults among women with and without adenomyosis were based on women with colorectal\nendometriosis ( Ballester  et al. ,\n2012 ), used oocyte donated cycles ( Martínez-Conejero  et al. , 2011 ), or used the\njunctional zone thickness ( Maubon  et\nal. , 2010 ) or even the myometrial thickness ( Youm  et al. , 2011 ) as an\nindirect marker to define women with adenomyosis.\nIn this regard, the use of GnRHa have been extensively used for the treatment of\nadenomyosis, offering a reduction in uterine volume and also a relief in pain\nsymptoms ( Grow & Filer, 1991 ;  Nelson & Corson, 1993 ;  Akira  et al. , 2009 ). However,\ntheir potential beneficial effect of IVF treatments is yet to be established. So\nfar, only two meta-analysis included studies that evaluated the different IVF\ntreatments among women with adenomyosis ( Younes and\nTulandi, 2017 ;  Cozzolino  et\nal. , 2022 ). Yet, a comparison between treatments could not\nbe made, since different protocols were used. The meta-analysis by  Younes & Tulandi (2017)  reported that the\nuse of GnRHa protocols for COS could be beneficial to improve IVF outcomes in\nadenomyosis, although they only included the study by  Niu  et al.  (2013) , that assessed GnRHa\npre-treatment compared with no treatment before FET, and the study by  Park  et al.  (2016)  that\ncompared long versus antagonist GnRH protocols for COS. The meta-analysis by  Cozzolino  et al.  (2022) \nadditionally included a third study ( Hou  et\nal. , 2020 ) that assessed the use of ultra-long versus long\nGnRHa before COS and fresh ET, and authors concluded that with available evidence,\nthe use of long-term GnRHa did not seem beneficial.\nThis latter meta-analysis also included the study by  Zhang  et al.  (2022) , however the goal of this one was\nto evaluate whether the use of surgery among women undergoing IVF with GnRHa\nprotocols could be beneficial among these women. Interestingly, several authors have\nfocused on evaluating the different adenomyosis phenotypes to better understand\ntheir impact on fertility ( Bourdon  et\nal. , 2020 ;  Cozzolino\n et al. , 2024 ).In fact, the inner and outer\nmyometrium are two functionally different entities ( Chapron  et al. , 2017 ;  Khan  et al. , 2019 ;  Bourdon  et al. , 2020 ). In this regard,  Cozzolino  et al.  (2024) \nreported in a prospective study a threefold higher relative risk of miscarriage when\nadenomyosis involved the inner myometrium, in contact with the Junctional zone, and\na higher ongoing pregnancy rate when adenomyosis was exclusively in the outer\nmyometrium.\nNot only the location, but also the type of adenomyosis has been the spotlight among\ninfertile women with adenomyosis.  Han  et\nal.  (2023)  compared retrospectively the use of ultra-long\nGnRHa protocol prior to COS and reported worse reproductive outcomes among women\nwith diffuse adenomyosis in terms of clinical pregnancy, miscarriage and live birth\ncompared with focal adenomyosis, although this latter group still showed worse\nmiscarriage rates compared with women with tubal infertility.\nIn this regard, the study by  Lan  et\nal.  (2021)  included in the present meta-analysis, reported\nno significant differences between the use of ultra-long versus long GnRHa protocols\nbefore COS, except for a reduction in early miscarriage rate in the group who\nreceived ultra-long GnRHa ( p =0.045). Interestingly, when\nstratifying by the type of adenomyosis, women with diffuse adenomyosis showed an\nincrease in clinical pregnancy and live birth rates when using ultra-long GnRHa\ncompared with long protocols, while the group with focal adenomyosis showed\ncomparable results. Other authors have focused on the uterine volume, as an\nindicator of severity, since it reveals the accumulation of adenomyotic tissue\nwithin the myometrium ( Li  et al. ,\n2021 ;  Han  et al. ,\n2023 ;  Cozzolino  et al. ,\n2024 ). In this matter,  Cozzolino\n et al.  (2024)  revealed worse reproductive outcomes\nin severe cases of adenomyosis, while  Zhang\n et al.  (2022)  described turning point of uterine\nvolumes larger than 8 weeks gestation (130cm 3 ) for worse reproductive\noutcomes in terms of miscarriage and live birth rates ( Han  et al. , 2023 ).\nAmong the studies included in this meta-analysis, several authors provided details\nregarding the uterine size of the study population ( Chen  et al. , 2020 ;  Lan\n et al. , 2021 ;  Wu\n et al. , 2022 ;  Zhang\n et al. , 2022 ;  Ge\n et al. , 2023 ). Only in three of them a significantly\nlarger uterine size was registered in the group who received ultra-long GnRHa\ncompared to those in the long GnRHa group. Yet none of them performed subanalysis of\nthe reproductive outcomes according to the uterine volume or even reported the\nuterine size reduction after the treatment with ultra-long GnRHa treatment before\ninitiating COS.\nSo far, even though adenomyosis has been associated to negative IVF outcomes ( Vercellini  et al. , 2014 ;  Younes and Tulandi, 2017 ;  Horton  et al. , 2019 ;  Huang  et al. , 2020 ;  Nirgianakis  et al. , 2021 ;  Cozzolino  et al. , 2022 ;  Moawad  et al. , 2022 ), there is\nlack of evidence regarding the benefits of long or ultra-long GnRHa protocols in\nreproductive outcomes among these women. In addition, even though there is some\nevidence that different types of adenomyosis or even the uterine volume could have\ndifferent negative effect of reproductive outcomes, the potential benefit of these\nprotocols for COS and ET is yet to be addressed.\nCertainly, there are cases where the GnRHa-induced hypopituarism has failed to reduce\nthe estrogen production within adenomyotic tissue particularly in severe cases\n( Cozzolino  et al. ,\n2023 ), leading to consider adjuvant treatments. The increased aromatase\nactivity in adenomyotic tissue ( Kitawaki  et\nal. , 1997 ) has led to the use of aromatase inhibitors (AI)\nin symptomatic women with adenomyosis undergoing IVF ( Kimura  et al. , 2007 ;  Badawy  et al. , 2012 ;  Sharma  et al. , 2023 ). Interestingly,  Sharma  et al.  (2023)  reported\nin a randomized controlled trial that the use of low-dose letrozole could be an\neffective option for women with symptomatic adenomyosis awaiting IVF. In addition,\nAI adjuvant to FSH treatment seem to offer IVF outcomes comparable to standard IVF,\nreaching lower estradiol concentrations in blood ( Requena  et al. , 2008 ;  Lin  et al. , 2024 ). Unfortunately, there is lack of\nresearch comparing different IVF protocols among women with adenomyosis undergoing\npre-treatment with AI prior to ovarian stimulation, or even comparing different IVF\nprotocols using AI adjuvant to FSH during ovarian stimulation.\nThis meta-analysis has several potential limitations that need to be pointed out.\nFirst, the retrospective nature of all studies included could compromise the\nhomogeneity of women who received different IVF protocols and affect negatively\non reproductive outcomes regardless of the treatment received. In fact, women\nwho received ultra-long GnRH protocol had larger uterine volumes ( Lan  et al. , 2021 ;  Wu  et al. , 2022 ;  Ge  et al. , 2023 ) and in\nsome studies a higher rate of dismenorrea ( Ge\n et al. , 2023 ) or confirmed endometriosis ( Lan  et al. , 2021 ), and\neven lower AMH levels compared with long GnRHa protocol ( Ge  et al. , 2023 ). Unfortunately not all\nstudies provided with this data, limiting the interpretation of the results from\nthis meta-analysis. Additionally, none of the studies that evaluated the use of\npre-treatment with GnRH before FET reported on the type of protocol used for\nCOS.\nAlthough unavoidable, the studies included in our meta-analysis are\nheterogeneous, which could lead to some biases in the results. Particularly, the\nduration of the ultra-long GnRH treatments received, since it varies across\nstudies, but also within each study. In some studies, the length of the\ntreatment depended on the reduction of the each patient’s uterine size ( Park  et al. , 2016 ;  Chen  et al. , 2020 ;  Wu  et al. , 2022 ;  Ge  et al. , 2023 ).\nThis is the first meta-analysis to compare the different IVF protocols among\nwomen with adenomyosis. Studies that included women with endometriosis and\nconcomitant adenomyosis, that combined other strategies to the IVF treatment\nsuch as surgery, or mixed different protocols within the same group were\nexcluded, in order to minimize bias arising from measurement of the exposure or\ndue to post-exposure interventions ( Wang\n et al. , 2009 ;  Mijatovic  et al. , 2010 ;  Al Jama, 2011 ;  Ballester\n et al. , 2012 ). Nonetheless, we need to identify\nwhat types of adenomyosis are associated with poorer reproductive outcomes.\n\nThere is no evidence regarding the benefit of pituitary downregulation before\ncontrolled ovarian stimulation or even as pre-treatment before frozen embryo\ntransfer systematically among women with adenomyosis undergoing IVF. In fact, the\nuse of long GnRHa seem to worsen reproductive outcomes compared to antagonist\nprotocols. Additionally, the pretreatment with GnRHa before FET among women with\nadenomyosis show not beneficial effect.\nSo far, there is lack of evidence regarding the benefits of the pituitary\ndownregulation among women with specific subtypes of adenomyosis that entail worse\nreproductive prognosis, such as adenomyosis of the inner myometrium or even women\nwith larger uterine volumes. Results from this systematic review and meta-analysis\nshould raise awareness of the risks of using pituitary downregulation systematically\namong women with adenomyosis undergoing IVF. There is a need for prospective trials\nthat assess the benefits of GnRH analogues before COS or as pretreatment before FET\namong women with different types of adenomyosis.","source_license":"public-domain-us","license_restricted":false}