Culture media for human pre-implantation embryos in assisted reproductive technology cycles(review)

Transcription

1 Cochrane Database of Systematic Reviews Culture media for human pre-implantation embryos in assisted reproductive technology cycles(review) YoussefMMA,MantikouE,vanWelyM,VanderVeenF,Al-InanyHG,ReppingS,MastenbroekS YoussefMMA,MantikouE,vanWelyM,VanderVeenF,Al-InanyHG,ReppingS,MastenbroekS. Culture media for human pre-implantation embryos in assisted reproductive technology cycles. Cochrane Database of Systematic Reviews 2015, Issue 11. Art. No.: CD DOI: / CD pub2. Culture media for human pre-implantation embryos in assisted reproductive technology cycles(review) Copyright 2015 The Cochrane Collaboration. Published by John Wiley& Sons, Ltd.

2 T A B L E O F C O N T E N T S HEADER ABSTRACT PLAIN LANGUAGE SUMMARY SUMMARY OF FINDINGS FOR THE MAIN COMPARISON BACKGROUND OBJECTIVES METHODS Figure RESULTS Figure Figure Figure Figure Figure Figure Figure Figure Figure Figure DISCUSSION AUTHORS CONCLUSIONS ACKNOWLEDGEMENTS REFERENCES CHARACTERISTICS OF STUDIES DATA AND ANALYSES Analysis 1.1. Comparison 1 Live birth or ongoing pregnancy per woman randomised, Outcome 1 Cleavage-stage embryo transfer Analysis 1.2. Comparison 1 Live birth or ongoing pregnancy per woman randomised, Outcome 2 Blastocyst-stage embryo transfer Analysis 2.1. Comparison 2 Clinical pregnancy rate per woman randomised, Outcome 1 Cleavage-stage embryo transfer. 81 Analysis 2.2. Comparison 2 Clinical pregnancy rate per woman randomised, Outcome 2 Blastocyst-stage embryo transfer. 82 Analysis 3.1. Comparison 3 Multiple pregnancy rate per woman randomised, Outcome 1 Cleavage-stage embryo transfer. 83 Analysis 3.2. Comparison 3 Multiple pregnancy rate per woman randomised, Outcome 2 Blastocyst-stage embryo transfer Analysis 4.1. Comparison 4 Miscarriage rate per woman randomised, Outcome 1 Cleavage-stage embryo transfer.. 85 Analysis 4.2. Comparison 4 Miscarriage rate per woman randomised, Outcome 2 Blastocyst-stage embryo transfer.. 86 ADDITIONAL TABLES APPENDICES HISTORY CONTRIBUTIONS OF AUTHORS DECLARATIONS OF INTEREST SOURCES OF SUPPORT DIFFERENCES BETWEEN PROTOCOL AND REVIEW INDEX TERMS i

3 [Intervention Review] Culture media for human pre-implantation embryos in assisted reproductive technology cycles Mohamed MA Youssef 1, Eleni Mantikou 2, Madelon van Wely 2, Fulco Van der Veen 2, Hesham G Al-Inany 1, Sjoerd Repping 2, Sebastiaan Mastenbroek 2 1 Department of Obstetrics & Gynaecology, Faculty of Medicine, Cairo University, Cairo, Egypt. 2 Center for Reproductive Medicine, Academic Medical Center, University of Amsterdam, Amsterdam, Netherlands Contact address: Sebastiaan Mastenbroek, Center for Reproductive Medicine, Academic Medical Center, University of Amsterdam, Meibergdreef 9, Amsterdam, 1105 AZ, Netherlands. S.Mastenbroek@amc.uva.nl. Editorial group: Cochrane Gynaecology and Fertility Group. Publication status and date: New, published in Issue 11, Citation: Youssef MMA, Mantikou E, van Wely M, Van der Veen F, Al-Inany HG, Repping S, Mastenbroek S. Culture media for human pre-implantation embryos in assisted reproductive technology cycles. Cochrane Database of Systematic Reviews 2015, Issue 11. Art. No.: CD DOI: / CD pub2. Background A B S T R A C T Many media are commercially available for culturing pre-implantation human embryos in assisted reproductive technology (ART) cycles. It is unknown which culture medium leads to the best success rates after ART. Objectives To evaluate the safety and effectiveness of different human pre-implantation embryo culture media in used for in vitro fertilisation (IVF) and intracytoplasmic sperm injection (ICSI) cycles. Search methods We searched the Cochrane Menstrual Disorders and Subfertility Group s Trials Register, Cochrane Central Register of Controlled Trials (CENTRAL), MEDLINE, EMBASE, the National Research Register, the Medical Research Council s Clinical Trials Register and the NHS Center for Reviews and Dissemination databases from January 1985 to March We also examined the reference lists of all known primary studies, review articles, citation lists of relevant publications and abstracts of major scientific meetings. Selection criteria We included all randomised controlled trials which randomised women, oocytes or embryos and compared any two commercially available culture media for human pre-implantation embryos in an IVF or ICSI programme. Data collection and analysis Two review authors independently selected the studies, assessed their risk of bias and extracted data. We sought additional information from the authors if necessary. We assessed the quality of the evidence using Grades of Recommendation, Assessment, Development and Evaluation (GRADE) methods. The primary review outcome was live birth or ongoing pregnancy. 1

4 Main results We included 32 studies in this review. Seventeen studies randomised women (total 3666), three randomised cycles (total 1018) and twelve randomised oocytes (over 15,230). It was not possible to pool any of the data because each study compared different culture media. Only seven studies reported live birth or ongoing pregnancy. Four of these studies found no evidence of a difference between the media compared, for either day three or day five embryo transfer. The data from the fifth study did not appear reliable. Six studies reported clinical pregnancy rate. One of these found a difference between the media compared, suggesting that for cleavagestage embryo transfer, Quinn s Advantage was associated with higher clinical pregnancy rates than G5 (odds ratio (OR) 1.56; 95% confidence interval (CI) 1.12 to 2.16; 692 women). This study was available only as an abstract and the quality of the evidence was low. With regards to adverse effects, three studies reported multiple pregnancies and six studies reported miscarriage. None of them found any evidence of a difference between the culture media used. None of the studies reported on the health of offspring. Most studies (22/32) failed to report their source of funding and none described their methodology in adequate detail. The overall quality of the evidence was rated as very low for nearly all comparisons, the main limitations being imprecision and poor reporting of study methods. Authors conclusions An optimal embryo culture medium is important for embryonic development and subsequently the success of IVF or ICSI treatment. There has been much controversy about the most appropriate embryo culture medium. Numerous studies have been performed, but no two studies compared the same culture media and none of them found any evidence of a difference between the culture media used. We conclude that there is insufficient evidence to support or refute the use of any specific culture medium. Properly designed and executed randomised trials are necessary. P L A I N L A N G U A G E S U M M A R Y Culture media for human pre-implantation embryos for in assisted reproductive technology cycles Review question: What is the impact of different human embryo culture media in women undergoing in vitro fertilisation (IVF) or intracytoplasmic sperm injection (ICSI) cycles, with regard to live birth or ongoing pregnancy and adverse events? Background: Embryo culture refers to the culture of human gametes and embryos during in vitro fertilisation treatments. The procedure is initiated when eggs and sperm are combined in a culture dish and it ends with the transfer of the resulting embryo to the uterus. The culture period varies between one and six days, and the embryo culture process is vital to the success of any IVF or ICSI procedure. The type of culture medium used may affect embryo quality and hence the success rate of the treatment. Despite its importance, it is unknown which culture medium is most effective and safest. Study characteristics: The evidence is current to March We included 32 randomised controlled trials of a wide variety of different commercially available culture media in women undergoing IVF or ICSI. Sixteen studies randomised women (total 3666), three randomised cycles (total 1018) and twelve randomised oocytes (over 15,230). Most studies (22/32) failed to report their source of funding. Key results: No two studies compared the same culture media. Only seven studies reported our primary outcome of live birth or ongoing pregnancy, and they found no good evidence of a difference between the media compared. A single study found low-quality evidence that for day three embryo transfer, Quinn s Advantage may be associated with higher rates than G5, but this study was available only as an abstract and the methods used were not clearly reported. With regard to adverse effects, three studies reported multiple pregnancies and six studies reported miscarriage. None of them found any evidence of a difference between the culture media used. None of the studies 2

5 reported on the health of offspring. We conclude that there is insufficient evidence to support or refute the use of any specific culture medium. Properly designed and executed randomised trials are necessary. Quality of the evidence: There was very low-quality evidence for nearly all comparisons, the main limitations being imprecision and poor reporting of study methods. 3

6 S U M M A R Y O F F I N D I N G S F O R T H E M A I N C O M P A R I S O N [Explanation] Live birth or ongoing pregnancy in women receiving human pre- implantation embryos using different culture media Population: Women in assisted reproductive technology cycles (ART) receiving human pre-implantation embryos in either early (cleavage stage transfer) or late (blastocyst stage transfer) stages Intervention: specific embryo culture media Comparison: other embryo culture media Comparisons** Illustrative comparative risks*(95% CI) Relative effect (95% CI) Early embryo transfer- GM 501 (specific) versusismi(other) Early embryo transfer- GM 501 (specific) versus Sydney IVF(other) Early embryo transfer - G2 (specific) versus Universal IVF(other) Assumed risk Other embryo culture media 264per per per1000 Corresponding risk Specific embryo culture media 270per1000 (160 to 422) 167per1000 (81 to 311) 269per1000 (139 to 457) OR 1.03 (0.53 to 2.03) OR 0.87 (0.38 to 1.96) OR 1.53 (0.67 to 3.51) No of Participants (studies) 172 (1 study) 158 (1 study) 129 (1 study) Quality of the evidence (GRADE) verylow 1,2 verylow 1,2 verylow 1,2 Comments No evidence of a difference between the two media in live birth rate No evidence of a difference between the two media in live birth rate No evidence of a difference between the two media in live birth rate Early embryo transfer: Cook (K-SIFM or K- SICM) (specific) versus Vitrolife(IVF or G3) (other) 220per per1000 (151 to 302) OR 0.98 (0.63 to 1.54) 449 (1 study) low 3,4 No evidence of a difference between the two media in live birth rate 4

7 Early embryo transfer - G2 (specific) versus Ham s F10(other) Late embryo transfer- GM 501 (specific) versus Sydney IVF(other) Late embryo transfer - G2 (specific) versus Universal IVF(other) Late embryo transfer -ECM/Multiblast(specific) versus Global (other) 579per per per per per1000 (444 to 846) 143per1000 (7 to 778) 250per1000 (71 to 591) 538per1000 (318 to 746) OR 1.52 (0.58 to 3.99) OR 0.67 (0.03 to 14.03) OR 1.92 (0.44 to 8.31) OR 0.50 (0.20 to 1.26) 72 (1 study) 12 (1 study) 47 (1 study) 79 (1 study) verylow 2,3 verylow 1,2 verylow 1,2 verylow 1,2 No evidence of a difference between the two media in ongoing pregnancy rate No evidence of a difference between the two media in live birth rate No evidence of a difference between the two media in live birth rate No evidence of a difference between the two media in ongoing pregnancy rate *Theassumedrisk is the risk in the study control group. Thecorrespondingrisk (and its 95%confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: Confidence interval; OR: Odds ratio GRADE Working Group grades of evidence High quality: Further research is very unlikely to change our confidence in the estimate of effect. M oderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate. Lowquality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate. Very low quality: We are very uncertain about the estimate. ** In addition to the studies in Table 1, one study compared HTF/CCM versus ISM-1/ISM-2 and reported that blastocyst transfer HTF/CCM was associated with higher birth rates. The data were not entered in data tables as they did not appear reliable; we have been unable to obtain clarification from the study authors. 1 Serious risk of bias. Poor reporting of study methods. 2 Very serious imprecision. Only one small study. 95% confidence interval around the estimate of effect includes both no effect and appreciable benefit or appreciable harm 3 Serious risk of bias. Method of allocation concealment not reported. 4 Serious imprecision: 95%confidence interval around the estimate of effect includes both no effect and appreciable benefit or appreciable harm 5

8 B A C K G R O U N D Description of the condition Human pre-implantation embryos are exposed to culture media for several days during their development in in vitro fertilisation (IVF) or intracytoplasmic sperm injection (ICSI) treatment. A wide variety of culture media is available on the market with variable formulations, ranging from simple salt solutions to complex tissue culture media with synthetically derived amino acids and growth factors. Despite the well recognized need for an optimal in vitro environment for human pre-implantation embryos, there are few data on the success rates of specific culture media. Description of the intervention Embryo culture media used in early human IVF had successfully been used in animal embryo culture. In 1985, an embryo culture medium called Human Tubal Fluid (HTF) was designed, based on the analysis of human tubal fluids (Quinn 1985). Since the development of HTF, many modifications have been made in the composition of human embryo culture media, with the addition of different components such as amino acids, vitamins, and nucleic acid precursors (Gardner 2002 a; Menezo 2004). Culture media can be stratified into two types according to the changing metabolic needs of the rapidly growing and changing embryo. Cleavage media are formulated specifically to support the growth requirements of the early cleavage-stage embryo until day three of pre-implantation development. Embryos that are to be cultured until day five or six are placed into another medium, referred to as blastocyst medium, on day three. Blastocyst medium has different and additional components that are thought to be required by the embryo for its transition from the cleavage stage to the blastocyst stage. For the blastocyst culture, two media can be used sequentially (with cleavage-stage specific medium replaced by blastocyst-specific medium at day three of development), or one medium can be used continuously (possibly replacing the medium with the same but fresh medium at day three of development). How the intervention might work The accumulated knowledge of mammalian embryo physiology and biochemistry in culture media has led to the production of theoretically more stable and optimised environments to which human embryos are exposed (Gardner 2002 b). However, differences between media and their effect on embryonic, prenatal and postnatal development, blastulation rates, embryo implantation rates, live birth rates, and birth weight of the offspring have been reported (Dumoulin 2010; Mantikou 2013). Earlier studies have shown that the environment of gametes and pre-implantation embryos has an effect on the expression and imprinting of a number of key genes that affect the development of the embryos (Blake 2007; Doherty 2000; Fauque 2007; Khosla 2001; Zander 2006). The differences in success rates of IVF treatments could be explained by differences in the composition and balance of the elements present in the culture medium. Currently, there is limited information concerning the composition of the culture media, which restricts the understanding of the mechanism of action. Why it is important to do this review Culture media are generally considered to play an important role in the success rates of an assisted reproductive technology (ART) programme (Mantikou 2013). They could have an effect on the health of the newborns (Dumoulin 2010). However, it is not known which specific culture medium leads to the best outcomes in terms of safety and effectiveness. We performed this systematic review to try to answer this question. O B J E C T I V E S To evaluate the safety and effectiveness of different human preimplantation embryo culture media used in IVF or ICSI cycles. M E T H O D S Criteria for considering studies for this review Types of studies Included studies: We included all randomised controlled trials (RCT) that randomised women, oocytes, or embryos, and that compared different culture media used in IVF or ICSI cycles. Excluded studies: We excluded quasi-randomised trials. Types of participants Women undergoing IVF or ICSI cycles. 6

9 Types of interventions We included studies that compared any commercially available culture media for pre-implantation embryo in an IVF or ICSI programme. The comparison groups were: 1. Studies comparing different culture media by randomising women for day two or three embryo transfer. 2. Studies comparing different culture media by randomising women for day five or six embryo transfer. 3. Studies which compared different culture media by randomising oocytes or embryos for day two or three embryo transfer. 4. Studies which compared different culture media by randomising oocytes or embryos for day five or six embryo transfer. We excluded studies involving the addition of supplements (such as globulin or human serum albumin) to commercially available media. 8. Fertilisation rate: defined as the number of fertilised oocytes per total number of oocytes injected or inseminated Studies that compared different commercially available media but did not report any of the above outcomes were included for completeness, but none of their data were used in the analysis. We excluded pregnancy outcomes from studies that randomised oocytes or embryos. Search methods for identification of studies We searched for all published and unpublished RCTs that examined the use of different embryo culture media, using a search strategy designed in consultation with the Cochrane Menstrual Disorders and Subfertility Group s (MDSG) Trials Search Co-ordinator, without language restriction. See Appendix 1; Appendix 2; Appendix 3; Appendix 4; Appendix 5. Types of outcome measures Primary outcomes 1. Live birth or ongoing pregnancy. Live birth was defined as the number of live births after twenty completed weeks of gestational age per randomised woman. Ongoing pregnancy was defined as the number of viable gestations with a positive fetal heart beat at 12 weeks per randomised woman. Where studies reported both live births and ongoing pregnancy, we used the data on live births. Secondary outcomes 1. Health of babies born: determined by birth weight and presence of congenital anomalies per baby 2. Clinical pregnancy rate: defined as the number of clinical pregnancies (demonstrated by the presence of a gestational sac on ultrasound scan at six weeks of pregnancy) per randomised woman 3. Multiple pregnancy rate: defined as the number of multiple pregnancies per randomised woman (adverse events) 4. Miscarriage rate: defined as the number of miscarriages per randomised woman (adverse events) 5. Implantation rate: defined as the number of women with one or more fetal sacs detected ultrasonographically at six weeks of gestation per total number of embryos transferred 6. Cryopreservation rate: defined as the number of embryos cryopreserved per total number of embryos 7. Embryo quality: defined as the number of good quality embryos per total oocytes or embryos randomised on day three. Assessment of embryo quality was based on the heterogeneous definitions used by the original investigators. Electronic searches The following electronic databases, trials registers and websites were searched up to March 2015: Cochrane MDSG Trials Register Appendix 1 Cochrane Central Register of Controlled Trials (CENTRAL) Appendix 2 MEDLINE Appendix 3 EMBASE Appendix 4 PsycINFO Appendix 5 Trials registers for ongoing and registered trials: Current Controlled Trials; the US National Institutes of Health: Clinical Trials Gov; The World Health Organisation International Trials Registry Platform search portal: WHO DARE (Database of Abstracts of Reviews of Effects), on the Cochrane Library (DARE); reference lists from relevant non- Cochrane reviews LILACS database, providing a source of trials from the Portuguese- and Spanish-speaking world (LILACS) PubMed (PubMed) Searching other resources In addition, we searched: The reference lists of all known primary studies, review articles and citation lists of relevant publications Abstracts of major scientific meetings (e.g. the European Society of Human Reproduction and Embryology (ESHRE), the American Society for Reproductive Medicine (ASRM)) in the ISI Web of Knowledge (ISI) Known experts and personal contacts were contacted regarding any unpublished materials 7

10 Data collection and analysis Selection of studies Two review authors (MY, EM) independently selected the trials to be included, using forms designed according to our inclusion criteria and Cochrane guidelines. We sought additional information on trial methodology from the authors of trials which appeared to meet the eligibility criteria but had unclear methodology. Differences of opinion were resolved after discussion with the other authors (MvW, FvdV, SR, SM, and HG). Subsequently, a detailed Characteristics of excluded studies table was constructed for trials that did not satisfy the inclusion criteria. A Characteristics of included studies table was constructed for those considered suitable. The search and selection process was documented using a PRISMA flow diagram (Figure 1). 8

11 Figure 1. Study flow diagram. 9

12 Data extraction and management A standardised data extraction form was developed and piloted for consistency and completeness. Data extraction was performed independently by two review authors (MY, EM). The two sets of extracted data were compared and discrepancies were resolved by discussion with the other authors. Where studies had multiple publications, the main trial was used as a reference, and additional information was derived from the secondary papers. We sought additional information from the authors of trials that reported data in a form unsuitable for meta-analysis. Studies that included such data (e.g. means, percentages) were still included for completeness. Assessment of risk of bias in included studies Two review authors (MY, EM) independently assessed the risk of bias in the included studies, using the Cochrane risk of bias assessment tool to assess randomisation, allocation concealment, blinding of participants and personnel, blinding of outcome assessor, completeness of outcome data, selective outcome reporting, and other biases. Each trial characteristic was judged and given a risk of bias rating as low, unclear, or high. We entered the information into Risk of bias tables. 1. Random sequence generation: risk of bias was considered low if an adequate method of sequence generation was described e.g. reference to a random number table or a computer random number generator. 2. Allocation concealment: risk of bias was considered low if a third-party system, serially-numbered sealed opaque envelopes, or a similar system was described. Concealment was stated as uncertain if sealed envelopes were mentioned without further description or if no information was available pertaining to allocation concealment. 3. Blinding: examined with regard to who was blinded in the trials. All levels were sought and categorized as follows: single or double blinding; no blinding (doctors, embryologist and participant knew the allocated treatment); unclear. 4. Completeness of outcome data: risk of bias was considered low if there were no missing outcome data or the missing outcome data balanced in numbers across intervention groups, with similar reasons for missing data across groups. 5. Selective outcome reporting: risk of bias was considered low if the study protocol was available and all of the study s prespecified (primary and secondary) outcomes that were of interest in the review had been reported as pre-specified. Measures of treatment effect For dichotomous data (e.g. live birth rates), we used the number of events in the control and intervention groups of each study to calculate the odds ratio (OR) with 95% confidence interval (CI) using the Mantel-Haenszel method with a fixed-effect model. If continuous data had been reported (e.g. birth weight), we had planned to calculate mean differences between the groups, with 95% CIs. Unit of analysis issues Our analysis was based on an intention-to-treat protocol; all clinical outcomes were analysed per randomised woman, even if this definition was not used by the original investigators. For studies that randomised cycles, we only included studies that had one cycle per woman in the analysis. For studies that randomised women, we developed additional tables for all outcomes reported per oocyte or embryo (implantation rate, embryo quality, fertilisation rate). We also developed additional tables for all studies that randomised oocytes or embryos. We counted multiple live births as one live birth event. When the studies did not use the correct unit of analysis, we attempted to contact authors to obtain data per randomised woman. Dealing with missing data We analysed the data on an intention-to-treat basis as far as possible; losses to follow-up were included as failure for pregnancy and live birth. We attempted to obtain missing data from the original investigators. We calculated miscarriage rates when possible, based on differences between clinical and ongoing pregnancies. Assessment of heterogeneity We had planned to assess heterogeneity by visual inspection of the forest plots (Higgins 2011). We had planned to calculate the I² statistic when appropriate, and to quantify any apparent inconsistency. We defined I 2 of 50% or more as indicating substantial heterogeneity. Assessment of reporting biases In view of the difficulty of detecting and correcting for publication bias and other reporting biases, the review authors minimised their potential impact by performing a comprehensive search for eligible studies, and by being alert for duplication of data. If there were ten or more studies in an analysis, we had planned to use a funnel plot to explore the possibility of small study effects (a tendency for estimates of the intervention effect to be more beneficial in smaller studies). 10

13 Data synthesis If the studies were sufficiently similar, we had planned to combine the data using a fixed-effect model to calculate pooled odds ratios or mean differences, with 95% CIs. We conducted separate analyses for the following groups: studies that compared culture media for days two or three embryo transfer (cleavage-stage transfer) studies that compared culture media for days four to six embryo transfer (blastocyst-stage transfer) studies that randomised oocytes or embryos instead of women for the following outcomes: fertilisation rate and embryo quality We displayed the results of the included studies graphically in analysis tables. Subgroup analysis and investigation of heterogeneity We planned to conduct subgroup analyses if there was obvious clinical heterogeneity, to examine whether the effects varied by clinical setting (e.g. in different populations). If substantial statistical heterogeneity had been detected, we planned to explore whether effects varied by differences in the methodological characteristics of the studies (e.g. with differing risks of bias.) Sensitivity analysis We planned to conduct sensitivity analyses for the primary outcome to determine whether the conclusions were robust for the decisions we made regarding study eligibility and analyses. These analyses would have included consideration of whether the review conclusions would have differed if: 1. Analysis had been restricted to studies without high risk of bias; 2. A random-effects model had been adopted; 3. Alternative imputation strategies had been implemented; 4. The summary effect measure had been relative risk rather than odds ratio; 5. The primary outcome had been live birth only. Overall quality of the body of evidence: Summary of findings table We generated a Summary of findings table using GRADEpro software. This table evaluated the overall quality of the body of evidence for the primary review outcome, using Grades of Recommendation, Assessment, Development and Evaluation (GRADE) criteria (study limitations (i.e. risk of, consistency of effect, imprecision, indirectness and publication. We justified, documented, and incorporated our judgements about the quality of the evidence (high, moderate, low or very low) into the reporting of results for each outcome. Due to the large number of comparisons in the review, we chose to focus on our primary outcome and report findings for all relevant comparisons in a single table. R E S U L T S Description of studies The diversity of media used and outcome measures assessed can be seen in the Characteristics of included studies table. Each of the trials studied the effects of different culture media on different outcome measures, precluding a meta-analysis. Results of the search We identified 1242 potentially relevant abstracts by our search strategy. After reading the abstracts, we discarded 1088 because they compared cleavage- stage versus blastocyst-stage embryo transfer, media supplementation, or because they studied homemade media or media for non-human embryos. We retrieved the full papers of the remaining 154 abstracts that appeared to meet the inclusion criteria.thirty-two studies (38 articles) met our inclusion criteria (Arenas 2007; Artini 2004; Barak 1998; Baum 2004; Bird 2012; Bungum 2003; Campo 2010; Cano 2001; Carrasco 2013; Chatziioannou 2010; Chen 2009; Fechtali 2004; Findikli 2004; Gimeno 2006; Hazlett 2003; Jamieson 1997; Khan 2004; Kyono 2000; Lambert 2005; Mauri 2001; Mayer 2003; Mendoza 2003; Paternot 2010; Rubino 2004; Sepulveda 2009; Shih 2014; Sifer 2009; Stevens 2000; Tedesco 1990; Von During 2004; Yamamoto 2006; Zollner 2004), We excluded 100 studies (112 articles), and four studies are ongoing (See Figure 1). See Characteristics of excluded studies table for full exclusion details. Included studies Thirty two studies met the inclusion criteria (See Characteristics of included studies table). Study design and setting All included studies were randomised controlled trials. All were performed at a single private, university-based or medical centre with the exception of three studies, where this was unclear (Arenas 2007; Baum 2004; Bird 2012). Participants Seventeen studies randomised 3666 women (Arenas 2007; Bungum 2003; Campo 2010; Carrasco 2013; Chatziioannou 2010; Gimeno 2006; Hazlett 2003; Jamieson 1997; Khan 2004; Mauri 2001; Mendoza 2003; Paternot 2010; Rubino 2004; 11

14 Sepulveda 2009; Shih 2014 Stevens 2000; Zollner 2004); three studies randomised 1018 cycles (Cano 2001; Barak 1998; Von During 2004); and twelve studies randomised 15,230 oocytes or embryos (Artini 2004; Baum 2004; Bird 2012; Chen 2009; Fechtali 2004; Findikli 2004; Kyono 2000; Lambert 2005; Mayer 2003; Sifer 2009; Tedesco 1990; Yamamoto 2006). Three of the studies that randomised oocytes did not report their sample size (Baum 2004; Lambert 2005; Yamamoto 2006). Sampling methods varied: two studies randomised all women who visited their centre (Bungum 2003; Von During 2004); fourteen studies had restrictive inclusion criteria (Artini 2004; Bird 2012; Campo 2010; Carrasco 2013; Chatziioannou 2010; Chen 2009; Findikli 2004; Gimeno 2006; Lambert 2005; Mayer 2003; Paternot 2010; Sepulveda 2009; Sifer 2009; Zollner 2004); and it was unclear in the rest of the studies. Maternal age was reported in seventeen of the studies (Artini 2004; Barak 1998; Bird 2012; Campo 2010; Carrasco 2013; Findikli 2004; Gimeno 2006; Hazlett 2003; Lambert 2005; Mauri 2001; Mendoza 2003; Paternot 2010; Rubino 2004; Sepulveda 2009; Sifer 2009; Stevens 2000; Zollner 2004). The mean age for women across these studies was 33.7 years, with a range from 25.6 to 39.3 years. The mean paternal age was 34.9 years, and was reported in two studies (Zollner 2004; Findikli 2004). The mean number of oocytes retrieved was 11.2, reported in eleven studies (Arenas 2007; Artini 2004; Campo 2010; Findikli 2004; Hazlett 2003; Lambert 2005; Mauri 2001; Mendoza 2003; Paternot 2010; Rubino 2004; Zollner 2004). The mean duration of subfertility was five years, with a range of 3.8 to 8.3 years, and was reported in five studies (Mendoza 2003; Zollner 2004; Findikli 2004; Artini 2004; Sifer 2009). Interventions Intracytoplasmic sperm injection (ICSI) was performed in five studies (Arenas 2007; Findikli 2004; Khan 2004; Kyono 2000; Mauri 2001); IVF was performed in five studies (Chen 2009; Hazlett 2003; Mayer 2003; Stevens 2000; Tedesco 1990); five studies did not report the fertilisation method used (Bird 2012; Bungum 2003; Cano 2001; Jamieson 1997; Yamamoto 2006); and both IVF and ICSI were performed in the rest of the studies. Two studies reported on donor oocytes only (Gimeno 2006; Sepulveda 2009); one study used both non-donor and donor oocytes (Cano 2001); and two studies used only non-donor oocytes (Campo 2010; Paternot 2010). For the remaining studies, the use of donor oocytes was unclear. None of the studies commented on frozen-thawed cycles. Fifteen studies performed the transfer during the cleavage-stage embryo development (day two or three; Artini 2004; Barak 1998; Baum 2004; Campo 2010; Carrasco 2013; Chatziioannou 2010; Chen 2009; Hazlett 2003; Jamieson 1997; Khan 2004; Mauri 2001; Rubino 2004; Sifer 2009; Stevens 2000; Von During 2004); five studies performed the transfer during the blastocyst-stage embryo development (day four to six; Bird 2012; Findikli 2004; Kyono 2000; Lambert 2005; Sepulveda 2009); six studies performed transfers during both the cleavage and blastocyst stage (Gimeno 2006; Mayer 2003; Mendoza 2003; Paternot 2010; Yamamoto 2006; Zollner 2004); while the day of transfer was unknown in five studies (Arenas 2007; Bungum 2003; Cano 2001; Fechtali 2004; Tedesco 1990). The data from these studies have been analysed separately, according to the timing of the intervention. All included studies transferred fresh embryos. All studies transferred multiple embryos per treatment cycle. For late embryo transfer, one study used continuous interrupted (renewed) single medium versus sequential media ( Sepulveda 2009); three studies used continuous uninterrupted single medium versus sequential media (Bird 2012; Findikli 2004, Paternot 2010); and seven studies used sequential media (Gimeno 2006; Kyono 2000; Lambert 2005; Mayer 2003; Mendoza 2003; Yamamoto 2006; Zollner 2004). Culture media from twelve commercial companies were represented in the review. The companies represented and their included media were: MediCult - ISM, BlastAssist, IVF (Barak 1998; Campo 2010; Cano 2001; Fechtali 2004; Findikli 2004; Jamieson 1997; Lambert 2005; Mendoza 2003; Sifer 2009; Von During 2004; Yamamoto 2006; Zollner 2004); Irvine Scientific - P1, HTF, ECM, MultiBlast, SSM (Artini 2004; Barak 1998; Baum 2004; Chen 2009; Hazlett 2003; Khan 2004; Mauri 2001; Mayer 2003; Sepulveda 2009; Yamamoto 2006); Vitrolife - IVF, CCM,G2, G3, G5 (Arenas 2007; Bungum 2003; Carrasco 2013; Chatziioannou 2010; Findikli 2004; Lambert 2005; Mayer 2003; Mendoza 2003; Shih 2014; Sifer 2009; Stevens 2000; Zollner 2004); Gynemed - GM501 (Campo 2010; Paternot 2010); Cook - Sydney IVF (Baum 2004; Mendoza 2003; Paternot 2010; Von During 2004); Scandinivian IVF - IVF (Cano 2001; Jamieson 1997; Mauri 2001; Rubino 2004); IVF Online - Global (Bird 2012; Sepulveda 2009); In Vitro Care - IVC (Arenas 2007; Hazlett 2003); Sage - Quinn s advantage (Bird 2012; Chatziioannou 2010; Mayer 2003; Rubino 2004; Shih 2014); Gibco - Ham s F10 (Stevens 2000); FertiPro - Ferticult (Fechtali 2004); and Api-System (Tedesco 1990). SeeTable 1 Outcomes Primary clinical outcomes Seven of the thirty two studies reported our primary outcomes. Five reported live birth rate (Campo 2010; Carrasco 2013; Gimeno 2006; Paternot 2010; Zollner 2004), and four reported ongoing pregnancy rate (Campo 2010; Paternot 2010; Sepulveda 2009; Stevens 2000). One of these studies reported data that did not appear reliable, as reported live birth rates were higher than clinical pregnancy rates (Gimeno 2006). Therefore, these data were not entered into data 12

15 tables. Attempts have been made to contact the study authors for clarification. Secondary clinical outcomes One study reported on congenital abnormalities or birth weight of babies born. However, the data could not be used as a different culture medium (IVF or G3) from the same commercial company (Vitrolife) was used for IVF and ICSI, but the results were presented together (Carrasco 2013). Thirteen studies reported clinical pregnancy (Campo 2010; Carrasco 2013; Chatziioannou 2010; Gimeno 2006; Hazlett 2003; Mayer 2003; Paternot 2010; Rubino 2004; Sepulveda 2009; Sifer 2009; Von During 2004; Yamamoto 2006; Zollner 2004), but only seven were included in the analysis. Three studies were excluded since the exact numbers of participants were not reported (Hazlett 2003; Rubino 2004; Von During 2004); three more were excluded because the unit of analysis (women or cycles) was different from the unit of randomisation (oocytes; Mayer 2003; Sifer 2009; Yamamoto 2006). Seven studies reported pregnancy rates without providing a definition, so it was not possible to assess whether the pregnancy was biochemical, clinical, or ongoing (Arenas 2007; Artini 2004; Barak 1998; Cano 2001; Jamieson 1997; Khan 2004; Mauri 2001). Four studies reported multiple pregnancies (Zollner 2004, Paternot 2010; Sifer 2009;Carrasco 2013 ); three were included in the analysis. One study was not included as the unit of analysis (women or cycles) was different than the unit of randomisation (oocytes; Sifer 2009). Seven studies reported miscarriage (Campo 2010; Cano 2001; Carrasco 2013; Mauri 2001; Paternot 2010; Sepulveda 2009; Zollner 2004). One study was not included in the analysis as the time of transfer was not known, precluding subgroup analysis (Cano 2001). Fifteen studies reported implantation, but none were included in the analysis (Arenas 2007; Artini 2004; Campo 2010; Cano 2001; Gimeno 2006; Hazlett 2003; Mauri 2001; Mendoza 2003; Paternot 2010; Rubino 2004; Sepulveda 2009; Sifer 2009; Stevens 2000; Von During 2004; Zollner 2004). Nine of the studies did not report the numbers in sufficient detail for inclusion in this outcome (Arenas 2007; Cano 2001; Gimeno 2006; Hazlett 2003; Mendoza 2003; Rubino 2004; Sifer 2009; Stevens 2000; Von During 2004); in six studies, the units of analysis (oocytes or embryos) were different from the units of randomisation (embryos, women or cycles; Artini 2004; Campo 2010; Mauri 2001; Paternot 2010; Sepulveda 2009; Zollner 2004). Three studies reported cryopreservation rate, but none were included in the analysis. One study was excluded since the exact numbers were not reported (Khan 2004); in two, the unit of analysis (embryos) was different from the unit of randomisation (women or cycles; Campo 2010; Von During 2004). The number of good quality embryos obtained was reported in eighteen studies, but none were included in the analysis (Arenas 2007; Artini 2004; Baum 2004; Bird 2012; Cano 2001; Chatziioannou 2010; Chen 2009; Fechtali 2004; Findikli 2004; Jamieson 1997; Mendoza 2003; Paternot 2010; Rubino 2004; Sepulveda 2009; Sifer 2009; Von During 2004; Yamamoto 2006; Zollner 2004). Twelve studies were excluded because the exact numbers were not reported (Arenas 2007; Baum 2004; Bird 2012; Cano 2001; Chatziioannou 2010; Fechtali 2004; Findikli 2004; Jamieson 1997; Rubino 2004; Sepulveda 2009; Sifer 2009; Zollner 2004); in two, the unit of analysis (embryos) was different from the unit of randomisation (oocytes, women or cycles; Paternot 2010; Von During 2004). Fertilisation rate was reported in twenty-two studies; none were included in the analysis (Arenas 2007; Artini 2004; Barak 1998; Baum 2004; Campo 2010; Chatziioannou 2010; Fechtali 2004; Findikli 2004; Gimeno 2006; Hazlett 2003; Jamieson 1997; Khan 2004; Lambert 2005; Mauri 2001; Mayer 2003; Mendoza 2003; Paternot 2010; Rubino 2004; Sepulveda 2009; Sifer 2009; Tedesco 1990; Zollner 2004). One study was excluded since embryos were cultured in the same medium until assessment of fertilisation (Sepulveda 2009), ten studies were excluded because the exact numbers of embryos or oocytes were not reported (Arenas 2007; Baum 2004; Chatziioannou 2010; Fechtali 2004; Hazlett 2003; Jamieson 1997; Khan 2004; Lambert 2005; Rubino 2004; Sifer 2009), and in six more, the unit of analysis (embryos or oocytes) was different from the unit of randomisation (women or cycles; Barak 1998; Campo 2010; Gimeno 2006; Mauri 2001; Paternot 2010; Zollner 2004). One study compared culture media and aimed to determine the correlation between levels of reactive oxygen species in different media. It was included for completeness, although it did not report any outcomes of interest to this review (Shih 2014). Excluded studies One hundred studies (112 articles) were excluded from the review for the following reasons: 16 were not relevant to the topic under study 15 were retrospective studies 3 were reviews 53 were quasi randomised studies 12 were not comparative studies 1 was conducted in mice See Characteristics of excluded studies table for details. Risk of bias in included studies The risk of bias was unclear for the majority of the studies due to poor reporting in the available articles. We received some missing data from the authors of included studies, but in many cases, the risk of bias still remained unclear (Figure 2; Figure 3). 13

16 Figure 2. Risk of bias graph: review authors judgements about each risk of bias item presented as percentages across all included studies. 14

17 Figure 3. Risk of bias summary: review authors judgements about each risk of bias item for each included study. 15

18 Allocation Generation of random sequence Three studies described satisfactory methods of sequence generation and were rated at low risk of bias in this domain (Carrasco 2013; Mauri 2001; Stevens 2000). Twenty-nine studies did not describe sequence generation and were deemed at an unclear risk of this bias. Allocation concealment All studies were rated at an unclear risk of this bias, as they either failed to mention allocation concealment, or did not report sufficient details about the method used. Blinding Two studies reported use of blinding of either study personnel (Paternot 2010), or outcome assessors (Lambert 2005), and were rated at low risk of bias related to blinding. One study did not perform blinding and was rated at a high risk of bias related to blinding (Zollner 2004). The other studies were rated as unclear risks. Blinding is unlikely to influence our primary review outcome (live birth), but may be relevant for subjective outcomes such as assessment of embryo quality. Selective reporting Six studies reported all expected outcomes and were rated at low risk of this bias (Bird 2012; Campo 2010; Mauri 2001; Mendoza 2003; Paternot 2010; Zollner 2004). Two studies were rated at high risk of bias in this domain because results were not reported for all prespecified outcomes (Barak 1998; Carrasco 2013). All other studies were rated as unclear risks of bias because too few details were reported to assess the risk of selective reporting bias. Other potential sources of bias Five studies were rated at low risk of other potential sources of bias (Carrasco 2013; Lambert 2005; Mendoza 2003; Paternot 2010; Shih 2014). One was rated at high risk of other potential bias as reported ongoing pregnancy rates were lower than live birth rates Gimeno Twenty nine studies were rated as unclear risks of bias, due to poor or insufficient reporting of methods. Nineteen of these studies were published only as abstracts. Effects of interventions See: Summary of findings for the main comparison Live birth or ongoing pregnancy per woman Different culture media and various outcome measures were assessed in each study. A meta-analysis was not possible but a forest plot is provided as a visual summary of the different comparisons and their effects. Incomplete outcome data Fourteen studies reported complete or almost complete outcome data, with few or no losses to follow-up (Campo 2010; Chen 2009; Findikli 2004; Jamieson 1997; Kyono 2000; Mauri 2001; Mayer 2003; Paternot 2010; Sepulveda 2009; Shih 2014; Sifer 2009; Stevens 2000; Tedesco 1990; Zollner 2004). Two studies were rated at high risk of attrition bias (Carrasco 2013; Gimeno 2006). The other sixteen studies had an attrition rate of up to 20%, or (more commonly) did not clearly report the denominator for their reported results. These studies were rated as unclear risks of attrition bias. Primary outcome 1 Live birth or ongoing pregnancy per woman randomised Seven studies reported live birth or ongoing pregnancy. Five reported live birth rate (Campo 2010; Carrasco 2013; Gimeno 2006; Paternot 2010; Zollner 2004) and two reported ongoing pregnancy (Sepulveda 2009; Stevens 2000); all compared different media. None of the studies included in our analysis found a difference between the groups (Figure 4; Figure 5) 16

19 Figure 4. Forest plot of comparison: 1 Live birth or ongoing pregnancy per woman randomised, outcome: 1.1 Cleavage-stage embryo transfer. 17

20 Figure 5. Forest plot of comparison: 1 Live birth or ongoing pregnancy per woman randomised, outcome: 1.2 Blastocyst-stage embryo transfer. One study reported that there was no difference between the live birth rate per woman when HTF was compared with ISM-1 in cleavage-stage embryo transfers (27/44 versus 21/47), but that CCM was associated with a higher live birth rate than ISM-2 in blastocyst-stage transfers (11/12 versus 6/15 ). These data were not included in our analysis because they did not appear reliable, as clinical pregnancy rates reported in this study were lower than the live birth rates (Gimeno 2006). We have sought clarification from the study authors. Secondary outcomes 1. Health of babies born No studies reported this outcome. 2. Clinical pregnancy rate (CPR) per randomised woman Six studies were included in the analysis for clinical pregnancy (Campo 2010; Chatziioannou 2010; Gimeno 2006; Paternot 2010; Sepulveda 2009; Zollner 2004). One reported more clinical pregnancies in the group whose embryos were cultured in Quinn s advantage than in those whose embryos were cultured in G5 ( Chatziioannou 2010). The rest of the studies found no evidence of a difference between the compared media for either cleavageor for blastocyst-embryo transfer (Figure 6; Figure 7). 18

21 Figure 6. Forest plot of comparison: 2 Clinical pregnancy rate per woman randomised, outcome: 2.1 Cleavage-stage embryo transfer. 19

22 Figure 7. Forest plot of comparison: 2 Clinical pregnancy rate per woman randomised, outcome: 2.2 Blastocyst-stage embryo transfer. 3. Multiple pregnancy rate per woman randomised Three studies were included in the analysis for multiple pregnancies (Paternot 2010; Zollner 2004;Carrasco 2013 ), none of which found evidence of a difference between the compared media for either day three or day five embryo transfers (Figure 8; Figure 9). 20

23 Figure 8. Forest plot of comparison: 3 Multiple pregnancy rate per woman randomised, outcome: 3.1 Cleavage-stage embryo transfer. Figure 9. Forest plot of comparison: 3 Multiple pregnancy rate per woman randomised, outcome: 3.2 Blastocyst-stage embryo transfer. 21

24 4. Miscarriage rate per woman randomised Five studies were included in the analysis for miscarriage (Campo 2010; Mauri 2001; Paternot 2010; Sepulveda 2009; Zollner 2004). None found evidence of a difference between the compared media for either day three or day five embryo transfers (Figure 10; Figure 11). Figure 10. Forest plot of comparison: 4 Miscarriage rate per woman randomised, outcome: 4.1 Cleavagestage embryo transfer. 22