{"paper_id":"b0a7e7a4-9e11-43ea-a488-9934334a6488","body_text":"Uterine embryo implantation after transfer is a key step in assisted reproduction\ntreatments. For instance, in Europe in 2019, the mean pregnancy rate per embryo\ntransfer was 34.6% after IVF, 32.1% after ICSI, 35.9% after frozen embryo transfer\nand 50.5% after egg donation ( European IVF\nMonitoring Consortium (EIM) for the European Society of Human Reproduction and\nEmbryology (ESHRE)  et al. , 2023 ). The mechanisms that\nexplain successes and failures are still poorly understood. However, many\nalternatives were proposed to potentially increase the number of successful\nimplantations, and Hyaluronic acid (HA) is frequently used to possibly improve the\nreproductive results ( Holt-Kentwell  et\nal. , 2022 ). HA is a high molecular weight glycosaminoglycan\nof the extracellular matrix without antigenic properties, and it is essential for\ncreating and filling extracellular spaces, being naturally present in almost all\nbody tissues, including the genital tract ( Rodriguez-Martinez  et al. , 2016 ). The early  in\nvitro  studies using a mouse model proposed that HA promotes cell to\ncell and cell to matrix adhesions via its receptor CD44, which is expressed on the\npreimplantation embryo and also on the endometrial stroma in mammals ( Cowman  et al. , 2015 ;  Adeniyi  et al. , 2021 ). HA is\nused in embryo transfer medium because it increases its viscosity without imposing\nany harm to the embryo ( Cowman  et\nal. , 2015 ). Additionally, it is thought that HA can\nfacilitate implantation by promoting cellular interactions ( Adeniyi  et al. , 2021 ), and the HA-enriched\nmedium for embryo transfer is a commercially available add-on for assisted\nreproductive technology (ART) ( Lensen  et\nal. , 2021 ).\nIn the last two decades, several studies have evaluated the effectiveness of\nHA-enriched medium for embryo transfer ( Valojerdi\n et al. , 2006 ;  Urman\n et al. , 2008 ;  Fancsovits  et al. , 2015 ), and there are also some\nrecent systematic reviews on this topic ( Heymann\n et al. , 2020 ; 2022;  Tyler  et al. , 2022 ). All these three systematic reviews\nconcluded that HA-enriched medium might improve reproductive outcomes, and the\nquality of the evidence was judged to be moderate ( Heymann  et al. , 2022 ). However, since the publication\nof these reviews, new RCTs with considerable sample sizes have been published,\nshowing no beneficial effect of this intervention ( Yung  et al. , 2021 ), and there is also evidence that\ninfusion of the uterine cavity with HA might actually worsen the endometrial\nreceptivity ( Marei  et al. ,\n2017 ).\nThe objective of this systematic review is to identify, assess, and summarize\nevidence on the effectiveness and safety of HA-enriched medium compared with the\nstandard medium for embryo transfer.\n\nThe protocol of this review was registered at PROSPERO under CRD42024579125,\navailable at  https://www.crd.york.ac.uk/prospero/display_record.php?ID=CRD42024579125\nPublished and unpublished studies were searched, without language restriction, in\nPubMed, Scopus, Web of Science and Clinicaltrials.gov, until December 2024.\nAdditionally, we hand-searched the reference list of the previously published\nsystematic reviews on this topic.\nThe following terms were used for the searches: (Hyaluronic OR hyaluronan OR\nEmbryoglue OR Glue) AND (embryo OR IVF) AND (Random* OR Trial).\nRegarding the study design, only randomized controlled trials were considered\neligible. Crossover trials were considered eligible, but only data from the\nfirst phase were included in the quantitative analysis, as the crossover is not\na valid design in this context. Quasi-randomized trials (allocation based on\ndate of birth, day of the week, alternated) were not considered eligible. The\nparticipants were all women undergoing embryo transfer. The intervention in\nquestion used an enriched medium in comparison to a standard medium (that might\ncontain lower concentrations of HA) for embryo transfer.\nThe primary outcome is live birth per allocated woman. Ongoing pregnancy would be\nused as a surrogate for live birth when only the latter was available ( Braakhekke  et al. ,\n2014 ).\nThe secondary outcomes were clinical pregnancy per allocated woman and pregnancy\nloss per clinical pregnancy. Multiple pregnancies and the birth of twins were\ncounted as single events.\nPregnancy loss was evaluated using clinical pregnancy as the denominator to avoid\nconfusion when interpreting the results. For example, assume these theoretical\nresults: Group 1=100 women, 60 clinical pregnancies, 12 pregnancy losses, 48\nlive births; Group 2=100 women, 30 clinical pregnancies, six pregnancy losses,\n24 live births. By using the number of randomized women as the denominator, one\nshould conclude that the risk of pregnancy loss is higher in group 1 (12%\n vs . 6%, groups 1 and 2, respectively). However, by using\nclinical pregnancy as the denominator, one would conclude that the pregnancy\nloss by clinical pregnancy is similar between groups (20%  vs .\n20%). We believe the latter makes more sense, avoiding the conclusion that the\nrisk of pregnancy loss is greater in group 1 while the pregnancy loss by\nclinical pregnancy was precisely the same, and the observed difference only\noccurred because there were more clinical pregnancies in group 1.\nTwo reviewers (RMR and MCRA) read the title/abstracts of all records after\nexcluding duplicates. All records that were considered potentially eligible by\nat least one of the reviewers were selected for evaluating the full text. The\nnext step was performed by the same reviewers, reading the entire manuscript of\nthe records that were selected in the first phase to evaluate whether the study\nwas eligible or not. Disagreements were solved by consulting another reviewer\n(WPM).\nTwo reviewers (RMR and MCRA) independently extracted data from the eligible\nstudies. Additionally, other reviewer (WPM) compared the extract data with those\nreported in previously published systematic reviews. Disagreements were solved\nby discussion.\nThe following outcomes and data items were assessed: Live birth, ongoing\npregnancy, clinical pregnancy and pregnancy loss.\nWe also evaluate the mean age (differences greater than 1.0 years were considered\nto be relevant), ovarian antral follicle count (differences greater than 2.0\nfollicles were considered to be relevant), anti-mullerian hormone AMH\n(differences greater than 0.5 ng/mL were supposed to be appropriate), and the\nnumber of transferred embryos (differences greater than 0.2 embryos were\nconsidered to be relevant). If age or the number of embryos transferred were not\nreported, the study was considered to be at high risk of bias.\nAdditionally, we assessed whether the embryo transfer medium in the control group\nhad HA on its composition, and we performed a subgroup analysis separating the\nstudies by this criterion.\nTwo reviewers (RMR and MCRA) independently evaluated the risk of bias of the\nincluded studies by using the Cochrane risk-of-bias tool for randomized trials\nversion 2 (RoB 2) ( Higgins  et\nal. , 2024 ). Disagreements were solved by discussion with\na third author (WPM).\nWe assessed the risk ratios and their respective 95% confidence interval\n(CI).\nData were combined for meta-analysis using Review Manager 5.4 using the\nMantel-Haenszel method and a random-effects model. Heterogeneity was assessed by\nI 2  statistics. Forest-plots were produced to summarize the\nanalyses, and a sensitivity analysis was performed excluding the studies deemed\nat high-risk of bias.\nWhen the one of the evaluated outcomes could not be extracted from the full\ntexts, we evaluated whether the outcomes were reported in the published protocol\nclinicaltrials.gov. When the results were not available, we tried to contact the\nstudy authors to provide additional information. Additionally, we assessed the\nfunnel-plot to evaluate the risk of reporting bias.\nWe evaluate the quality/certainty of the evidence as suggested GRADE working\ngroup ( https://www.gradeworkinggroup.org/ ), evaluating the limitations\nof the included studies, inconsistency, indirectness, imprecision, and\npublication bias. An evidence table was created reporting the judgements about\nevidence quality (high, moderate, low or very-low) with the justifications\n( Schünemann  et al. ,\n2023 ).\nLarge Language Models tools were used to English scientific language correction\nand readability improvement.\n\nThe search results are reported in  Figure 1 .\nThe last electronic search was performed in 2024.12.01 and retrieved a total of\n431 records: PubMed=128; Scopus=125; Web of Science=160; and\nClinicaltrials.gov=18. Nineteen additional records were added by manual search,\nand 107 duplicates were removed. A total of 343 records were screened based on\ntitle/abstracts, and 301 were excluded. A total of 45 records were completely\nassessed for eligibility: 13 records were excluded for some reason ( Table 1 ), and we identified three ongoing\ntrials ( Cai  et al. , 2025 ;\n Nogueira, 2025 ;  Warhade  et al. , 2025 ). A total of 24\nstudies (from 30 records) were considered eligible ( Table 2 ). Two records ( Ten\n et al. , 2019 ;  Sellers  et al. , 2022 ) reported the same cohort with\nseparated randomization and different control groups; two studies had three\nrecords each, and the other two studies had two records each ( Table 3 ).\nExcluded studies with reasons.\nIncluded studies\nStudies with more than one record.\nFigure 1 Search results.\nSearch results.\nWe extracted data from a previously published systematic review ( Heymann  et al. , 2022 ) for\nthree studies: all data for 2 studies ( Korosec\n et al. , 2007 ) and only data for live birth for\nthe other study . We were not able to extract data from the outcomes of interest\nfrom 4 studies ( Chen  et al. ,\n2001 ;  Khan  et al. ,\n2004 ;  Drew  et al. ,\n2014 ;  Fasano  et al. ,\n2016 ) therefore, we included 20 studies in the quantitative\nanalysis.\nThe risk of bias assessment of the 24 included studies is reported on  Table 4 . One study ( Yung  et al. , 2021 ) was considered to have\nlow risk of bias and we have some concerns regarding the risk of bias for\nanother study ( Hazlett  et al. ,\n2008 ). The remaining 22 studies were deemed to be at high risk of\nbias.\nRisk of bias of the included studies\nThe results for live births are presented in  Figure\n2a . A total of 11 studies were included in this analysis; the total\nnumber of participants allocated to the HA-enriched medium was 2,026 compared to\n2,037 who were allocated to the standard transfer medium, encompassing 803 and\n697 live births, respectively. The relative risk (RR) was 1.14 (95%\nCI=0.99-1.31),  p =0.07. We observed a substantial heterogeneity,\nwith I 2 =54%. Sensitivity analysis, including only the two studies not\nconsidered to be at high risk of bias ( Figure\n2b ), resulted in a RR=1.06 (95% CI=0.85-1.31),\n p =0.62, with low heterogeneity (I 2 =0%). We did not\nuse ongoing pregnancy as a surrogate outcome.\nFigure 2 Results for live birth considering all the included studies (A) and\nonly those not considered to be at high risk of bias (B).\nResults for live birth considering all the included studies (A) and\nonly those not considered to be at high risk of bias (B).\nThe results for clinical pregnancy are presented in  Figure 3a . Only one study included in the quantitative analysis did\nnot report clinical pregnancy but reported live birth ( Child  et al. , 2021 ). A total of 19 studies\nwere included in this analysis; the total number of participants allocated to\nthe HA-enriched medium was 2,452, compared to 2,450 who were allocated to the\nstandard transfer medium, encompassing 1,149 and 985 clinical pregnancies,\nrespectively. The RR was 1.17 (95% CI=1.05-1.29,  p =0.004. We\nobserved a substantial heterogeneity, with I 2 =50%. Sensitivity\nanalysis, including only the two studies not considered to be at high risk of\nbias ( Figure 3b ), resulted in a RR=1.05\n(95% CI=0.87-1.26),  p =0.53, with low heterogeneity\n(I 2 =0%).\nFigure 3 Results for clinical pregnancy considering all the included studies\n(A) and only those not considered to be at high risk of bias\n(B).\nResults for clinical pregnancy considering all the included studies\n(A) and only those not considered to be at high risk of bias\n(B).\nThe results for pregnancy loss are presented in  Figure 4a . A total of 9 studies were included in this analysis; the\ntotal number of clinical pregnancies in participants allocated to the\nHA-enriched medium was 736, compared to 647 in the standard transfer medium,\nencompassing 103 and 122 pregnancy losses, respectively. The RR was 0.75 (95%\nCI=0.55-1.02,  p =0.06. We observed a low heterogeneity, with\nI 2 =26%. Sensitivity analysis, including only one study not\nconsidered to be at high risk of bias ( Figure\n4b ), resulted in a RR=1.05 (95% CI=0.61-1.80),\n p =0.86.\nFigure 4 Results for pregnancy loss considering all the included studies (A)\nand only those not considered to be at high risk of bias (B).\nResults for pregnancy loss considering all the included studies (A)\nand only those not considered to be at high risk of bias (B).\nFour studies published solely as abstracts were not included in the quantitative\nanalysis: ( Chen  et al. ,\n2001 ), ( Drew  et al. ,\n2014 ),  Fasano  et al. \n(2016) , and  Khan  et\nal.  (2004) .\n( Chen  et al. , 2001 ): this\nstudy randomized 70 women. The observed results showed no significant difference\nfor positive pregnancy test performed 14 days after the embryo transfer\n(8/35=23%  vs . 5/35=14%, HA enriched medium  vs .\nstandard medium, respectively). The authors concluded that there is a trend\ntowards a better pregnancy rate in an HA-enriched medium. Since this study\nreported only positive pregnancy tests, we could not extract data for the\noutcomes of interest.\n( Drew  et al. , 2014 ): this\nstudy randomized 493 embryo transfers. The observed results showed no\nsignificant difference for clinical pregnancy following either single embryo\ntransfer (D3=30.2%  vs . 23.3%, D5=44.1%  vs .\n42.4%, HA enriched medium  vs . standard medium, respectively) or\ndouble embryo transfer (D3=30.4%  vs . 39.7% and D5=44.4%\n vs . 42.4%). The authors concluded that using an HA-enriched\nmedium resulted in no benefit to reproductive outcomes. We were not able to\nextract data for quantitative analysis because the total number of participants\nin each group was not reported.\nFasano  et al.  (2016) :\nthis study randomized 372 warming cycles from 253 women. Pregnancy rates per\ntransfer were comparable between groups (29.4%  vs . 23.9%\n p =0.32). Authors concluded that HA enriched medium is as\neffective as the standard medium. We were not able to extract data for\nquantitative analysis because authors reported the results for embryo transfer\ncycles, allowing the same participant to be included repeated times.\nKhan  et al.  (2004) : this\nstudy randomized 169 women aged below 39 years. The ongoing pregnancy rates were\ncomparable between groups (53%  vs . 50%). Authors concluded that\nHA enriched medium does not show any significant improvement when compared to\nstandard medium. We were not able to extract data for quantitative analysis\nbecause the total number of participants in each group was not reported.\nThree ongoing studies were identified. One is being conducted in Abu Dhabi,\nUnited Arab Emirates ( Nogueira, 2025 ).\nThe study started on Jan. 2024; the authors estimated that the study will enroll\n783 participants and will be completed by Dec. 2025.\nThe other ongoing study is being conducted in Xian, Shaanxi, China ( Cai  et al. , 2025 ), and it\nstarted on Oct. 2024. The authors estimated to enroll 858 participants and plan\nto complete the study by Sep. 2026.\nFinally, there is a study been held in Wardha, India ( Warhade  et al. , 2025 ). There is an\nestimation to enroll 52 subjects, in the context of recurrent implantation\nfailure. The predicted study completion is by Dec. 2026.\nThe certainty of the evidence is reported in  Table 5 . The quality of the evidence for live birth, clinical\npregnancy and pregnancy loss was judged to be of very low quality. Most of the\nstudies were judged to be at high risk of bias, there was inconsistency among\nstudies, and the estimates were not very precise. Moreover, although some of the\npublished studies have found some benefit of the intervention, the only two\nstudies that were not considered to be at high risk of bias did not show any\neffect, as well as the four studies that were not included in the quantitative\nanalysis.\nGrade evidence profile for live birth, clinical pregnancy and pregnancy\nloss comparing hyaluronic acid enriched medium  vs .\nstandard 4 medium for embryo transfer\n\nThis systematic review aimed to evaluate the effectiveness of using an HA-enriched\nmedium for embryo transfer based on the currently available evidence. Due to the\nlimitations of the included studies, inconsistency and imprecision, our level of\ncertainty in the observed estimates is still very low for live births, clinical\npregnancies, and pregnancy loss. The quantitative analysis considering all the\nincluded studies did not show a significant difference for live birth (RR=1.14, 95%\nCI=0.99-1.31) and pregnancy loss (RR=0.75, 95%CI=0.55-1.02), although there was a\nsmall increase in clinical pregnancies (RR=1.17, 95%CI=1.05-1.29). When analyzing\nonly the two studies not considered to be at high risk of bias, none of the results\nwere very similar for the live birth (RR=1.06, 95%CI=0.85-1.31), clinical pregnancy\n(RR=1.05, 95%CI=0.87-1.26), and pregnancy loss (RR=1.05, 95%CI=0.61-1.80). The four\nstudies that were not included in the quantitative analysis also reported no benefit\nof the intervention.\nAlthough the lists of included studies were very similar, there were some differences\nbetween this review and the previously published Cochrane review ( Heymann  et al. , 2022 ).\nFirstly, two studies ( Ten  et al. ,\n2019 ;  Sellers  et al. ,\n2022 ) were included that were not previously available. Additionally,\ndata of clinical pregnancy were analyzed from  Walker\n et al.  (2005) , that was included in the previous\nreview, but authors reported that they failed to extract data because the study had\nreported the outcomes as percentage alone. However, authors have reported that there\nwere 34 participants in each group, allowing to ultimately extract the data.\nMoreover, the study of  Fancsovits  et\nal.  (2011)  was not included in our review because it\ncomprises only a preliminary report of the complete study ( Fancsovits  et al. , 2015 ): the enrolment period\nreported by  Fancsovits  et al. \n(2011)  was Jan. 2010 to Dec. 2010 while the enrolment period of  Fancsovits  et al.  (2015)  was\nJan. 2010 to Aug. 2012. By including  Fancsovits\n et al.  (2011) , the previous review counted the\nresults for the same participant more than once, which is not appropriate.\nIn addition, we did not include one study where the comparison was not HA-enriched\nmedium  vs . standard medium for embryo transfer ( Kleijkers  et al. , 2016 ). In\nthis study, participants were randomly assigned to have their oocytes and embryos\ncultured in one of the two media: G5 (Vitrolife, Goteborg, Sweden) and HTF (Lonza\nVerviers, Belgium). Although the G5 medium has HA in its composition and the HTF\nmedium does not have HA, we do not believe the comparison is similar to using the\nHA-enriched medium only for embryo transfer. In this study, the authors compare two\ndifferent standard media for the whole process, including embryo culture.\nAdditionally, the concentration of HA in G5 (0.125 mg/mL) is very different from the\nHA-enriched medium for embryo transfer (0.5 mg/mL).\nFinally, there is only one registered trial that will include only euploid\nblastocysts (19). It will be essential to follow up if the use of HA-enriched medium\nwill influence treatment results within the group of euploid embryos.\nAlthough the differences in the included studies were small, there is a crucial\ndifference in the conclusions between the two reviews. In the previously published\nCochrane review ( Heymann  et al. ,\n2022 ), authors have concluded that moderate-quality evidence shows\nimproved clinical pregnancy and live birth rates with the addition of HA as an\nadherence compound in embryo transfer media. In the present review, we observed a\nvery low level of certainty for the three outcomes, which were live birth, clinical\npregnancy, and pregnancy loss. Moreover, based on the results of the only study\njudged to be at low risk of bias, we believe that there is no clinically relevant\ndifference between using an HA-enriched medium or a standard medium for embryo\ntransfer. Therefore, it is also suggested that we should wait for the results of the\nthree extensive ongoing studies before making further recommendations for clinical\npractice and future research.","source_license":"public-domain-us","license_restricted":false}