Embryo Culture Media Influence on Live Birth Rate and Birthweight after IVF/ICSI: A Systematic Review Comparing Vitrolife G5 Media to Other Common Culture Media

Previous studies have indicated that culture media vary in efficiency and outcomes, such as live birth rate, birthweight and embryo quality. Does Vitrolife G5 series culture media result in higher live birth rates and birthweight compared to other common culture media? This study is a systematic review based on the PRISMA criteria. Relevant search terms, mesh terms (PubMed and Cochrane) and Emtree terms (Embase) were identified. We searched the literature using PubMed, Embase and Cochrane, on November 10, 2019. The inclusion criteria involved published articles in English comparing Vitrolife G5 to other common culture media. We included randomized controlled trials (RCTs) and cohort studies. The quality of the studies was assessed using the Cochrane Risk of Bias tool 2.0 and the Newcastle-Ottawa Scale. Primary outcomes were live birth rate and birthweight. Secondary outcomes were fertilization rate, implantation rate, biochemical pregnancy rate, clinical pregnancy rate, miscarriage rate, multiple pregnancies and congenital malformations. Of 187 articles screened, 11 studies fulfilled the inclusion criteria: Five RCTs and six retrospective cohort studies. Only one study reported live birth rate, showing a non-significantly higher live birth rate for Vitrolife G5 media. Birthweight had equivocal results with three of six studies, showing significantly lower (2)/higher (1) birthweights, whereas the others were non-significant. Overall, there were no significant differences concerning secondary outcomes. The results are equivocal, and we need more studies to evaluate culture media and their effect on short- and long-term health.


INTRODUCTION
In in vitro fertilization (IVF) and intracytoplasmic sperm injection (ICSI), the fertilized embryos are cultivated in culture media to choose the best embryo to transfer to the uterus either at cleavage stage or as a blastocyst. To make this possible, the oocytes and embryos are transferred to one or several culture media that support the early development of the embryos. These media have evolved from simple culture media based on blood serum to complex media containing a variety of different substances such as amino acids, human albumin, vitamins, antibiotics and growth factors (Chronopoulou & Harper, 2015).
While the culture media of the early years were homemade in fertilization clinics, fewer, but more specialized companies now commercially produce them. This has added economic interests, resulting in lack of transparency regarding media composition, but it has also led to increased quality and more quality control (Chronopoulou & Harper, 2015). Culture media can be divided into sequential media such as the G5 series (Vitrolife), where different culture media are used throughout the embryo development; or single media, such as GL BLAST sole medium (Ingamed), where only one single medium is used for the whole period, until the blastocyst stage.
Previous studies have indicated that different culture media vary in their efficiency and outcomes, such as live birth rate, birthweight and embryo quality (Youssef et al., 2015;Mantikou et al., 2013). Studies suggest that culture media influence gene expression and epigenetics in animals and humans, which might affect the long-term health of the children (Schwarzer et al., 2012;Kleijkers et al., 2015).
The number of infertile women submitted to IVF is increasing. Therefore, we undertook this study to compare the common culture media G5 series (Vitrolife, Sweden) to other common culture media, with the prime focus on live birth rates and birthweight.

MATERIALS AND METHODS
We used the PRISMA criteria in this review. The study is registered in Prospero (CRD42020153820). Two of the review team members (L Bick and A S Nielsen) did data collection, data extraction and the assessment of the studies independently. Discussion or a third person (U B Knudsen) solved disagreements.

Outcomes
The primary outcomes were live birth rate and birthweight. Live birth rate was defined as the proportion of women giving birth to at least one child born alive, independent of gestational age. Birthweight was defined as the mean birthweight of the babies measured in grams.
Secondary outcomes were fertilization rate, implantation rate, biochemical pregnancy rate, clinical pregnancy rate, miscarriage rate, multiple pregnancy rate and congenital malformations. Most definitions were based on Kleijkers et al. (2016), but may vary slightly among the different studies. The fertilization rate was defined as the percentage of fertilized oocytes (containing two pronuclei) among the number of mature oocytes (metaphase II) inseminated or injected. The implantation rate was defined as the number of gestational sacs identified by transvaginal ultrasound after six to eight weeks of gestation, divided by the number of embryos transferred. The biochemical pregnancy rate was defined as the percentage of women having at least one serum beta-hCG test of at least 50 UI/l two weeks after embryo transfer. The clinical pregnancy rate was defined as the percentage of women with a gestational sac and a fetal heartbeat, identified by transvaginal ultrasound examination at six to eight weeks of gestation. A miscarriage was determined as a biochemical pregnancy not resulting in a live birth. The multiple pregnancy rate was defined as the percentage of live births resulting in more than one child. Congenital malformations were divided into minor and major malformations. Major JBRA Assist. Reprod. | v.25 | n o 3| July-Aug-Sept/ 2021 malformations were defined as malformations causing functional impairment or requiring surgical correction, and the remaining malformations were considered minor.

Data Collection
We ran a systematic search on PubMed, Embase and the Cochrane Library on November 10, 2019.
The inclusion criteria were published articles in English on clinical trials containing well-defined data on at least one of the primary and/or secondary outcomes comparing Vitrolife G5 series culture media with other common culture media in humans. Both randomized controlled trials (RCTs) and cohort studies were included. Initially, there was no time limit on the search, but since Vitrolife G5 series was introduced in 2007, all articles from before 2007 were later excluded.
The research keywords was set up using the PICO model and divided into four search blocks. We used relevant search terms, mesh terms (PubMed and Cochrane) and Emtree terms (Embase). The four search blocks were used to run a combined search. The PICO table, search terms and examples of search queries can be found in the Supplements section of this review.
The data collection is illustrated on the PRISMA flow diagram ( Figure 1). The search in the three databases resulted in 44 results in PubMed, 63 results in Cochrane and 112 results in Embase. This yielded 219 results.
We removed the duplicates, resulting in 186 hits. We read the abstracts from the 186 results. When information was missing in the abstracts, for instance, whether the culture media was part of the Vitrolife G5 series, we searched for information in the full article. Of the 186 studies, 176 were excluded because either they did not contain any data comparing between Vitrolife G5 series and other culture media, they were animal studies, reviews, conference abstracts, or they were published before the G5 series was introduced in 2007. The participants in one study (Kleijkers et al., 2015) were enclosed in a larger study (Kleijkers et al., 2016), and therefore, only Kleijkers et al. (2016) was included. The reviews and their references were searched to find any data comparing G5 series media to other culture media.
We ran a Scopus citation search on the 10 remaining studies. The titles and abstracts of articles citing the 10 studies were searched to identify other relevant studies in which the systematic search might have been missing. One additional study was found, resulting in 11 studies to be included in this review.

Data extraction and assessment of included studies
We read the included articles and extracted the data regarding primary and secondary outcomes. When available, we collected additional data, such as the type of G5 product, whether the study was an IVF/ICSI study, and whether the study used fresh or frozen embryos. RCTs were assessed by Version 2 of the Cochrane riskof-bias tool for randomized trials (RoB 2), shown in Table  1. The Newcastle-Ottawa Scale (NOS), shown in Table 3, assessed cohort studies. After the individual assessment was completed, a final assessment was found, and disagreements were solved by discussion or by a third person.

RESULTS
The data collection is illustrated on the flow diagram ( Figure 1). Of the 219 articles, only 11 studies qualified to be included in this review.
In Table 1, you find the RoB 2 assessment of the RCTs, and Table 2 shows additional information about the studies. Table 3 shows the NOS assessment of the retrospective cohort studies; and Table 4 shows additional information about the studies. Table 5 shows an overview of the 11 studies regarding the type of culture media, whether the study included IVF or ICSI or both, and which of the outcomes each study included. Table 6 shows the results of the primary outcomes, and Table 7 shows the results of the secondary outcomes.

Primary outcomes
Live birth rate Kleijkers et al. (2016) found in an RCT that G5 culture media tended to have a slightly higher live birth rate than the HTF culture media, but the difference was nonsignificant ( Table 6). None of the other studies report on live birth rate.

Birthweight
Two studies out of six found a significantly lower birthweight for G5 media, whereas one study found a significant higher birthweight for G5 (Table 6). Kleijkers et al. (2016) found in an RCT that G5 had a 158g lower birthweight compared to the HTF culture media; and Hassani et al. (2013) found in an RCT a 370g lower birthweight comparing G5 to ISM1. Eskild et al. (2013) found in a retrospective study a significant higher birthweight comparing G5 to Universal IVF medium and ISM1, where G5 was found to have a 92.4 g higher birthweight compared to ISM1.
Three of the retrospective cohort studies did not find any differences (Gu et al., 2016-Quinn's media, De Vos et al., 2015-Medicult and Lin et al., 2015, even though all three studies included more than one thousand embryos. Kleijkers et al. (2016) included both fresh and frozen embryos in their analysis with total numbers only. De Vos et al., 2015 included both fresh and frozen embryos, and had separate results. The other studies included fresh embryos only.

Secondary outcomes Fertilization rate
One RCT study reported that G5 had a significantly lower fertilization rate compared to the HTF culture media (Kleijkers et al., 2016), and one RCT study reported that G5 had a significantly higher fertilization rate compared to Universal IVF Medium (Hambiliki et al., 2011) (Table 7). Two RCTs and a retrospective cohort study reported no differences comparing G5 to Cook Sequential Medium, GV Blast Sole and SAGE 1-STEP (Zhang et al., Ceschin et al., 2016;Lopez-Pelayo et al., 2018).

Implantation rate
One RCT found a significantly higher implantation rate for G5 compared to the HTF culture media (Kleijkers et al., 2016) (Table 7).

Clinical pregnancy rate
In an RCT and in a retrospective cohort study, a significantly higher clinical pregnancy rate was found comparing G5 to HTF and Global (Kleijkers et al., 2016;Lin et al., 2015) (Table 7). Five studies including three RCTs and two retrospective cohort studies found no difference comparing G5 to Cook Sequential Media, ISM1, Universal IVF Medium, SAGE 1-STEP medium, Global, and Quinn's advantage medium (Zhang et al., 2016;Hassani et al., 2013;Hambiliki et al., 2011;Lopez-Pelayo et al., 2018;Lin et al., 2013). Table 1. Assessment of RCTs -Version 2 of the Cochrane risk-of-bias tool for randomized trials (RoB 2).

Multiple pregnancy rate
Two RCTs found no difference in multiple pregnancy rates comparing G5 to HTF and ISM1 (Kleijkers et al., 2016, Hassani et al., 2013 (Table 7). The calculation of the percentages for Kleijkers et al. (2016) can be found in the Supplements section of this review.

Congenital malformations
Only one study reported on congenital malformations. In an RCT, no difference in numbers of congenital malformations was found comparing G5 and HTF media (Kleijkers et al., 2016) (Table 7).

DISCUSSION
Defining the best embryo culture media can be a challenge as there are many different outcomes to assess the quality of the culture media. However, it is commonly accepted that live birth rate is the preferable outcome to assess IVF/ICSI success rates (Mantikou et al., 2013). Kleijkers et al. (2016) is the only study that evaluated live birth rate comparing G5 to another media, and they found a slightly higher live birth rate for G5 compared to HTF media, however not significant. The study was designed to detect a difference of 10%, but even a smaller difference may be of interest if this can be confirmed in more RCTs. The fact that only one of the studies included live birth rates (Kleijkers et al., 2016), which is considered the golden standard, clearly emphasizes the lack of RCTs reporting on live birth rate.
Some of the other studies had outcomes that approached live birth rates. Hambiliki et al. (2011) assessed delivery rate defined as the ratio between deliveries and embryos transferred. However, there are different guidelines for the numbers of embryos transferred per cycle. This makes comparison among centers difficult. Hassani et al. (2013) compared "baby take home rates" but gave no clear definition of the term. Future studies should adhere to the same definitions, and use live birth rate as the main outcome, so studies can be compared.
In this review, six of the eleven studies assessed birthweight with varying results. This is in line with previous studies, where some have shown that the type of culture media could influence birthweight (Dumoulin et al., 2010; JBRA Assist. Reprod. | v.25 | n o 3| July-Aug-Sept/ 2021

S
Live birth rate is measured in percent. Birthweight is measured in grams: mean ± SD. S/NS: significant/non-significant. *Both G5 and G3 culture medium. 401 out of 710 are G5 culture medium. † The first row shows results for singletons and the second row shows results for twins. Nelissen et al., 2012), other studies found no differences (Eaton et al., 2012;Vergouw et al., 2012). Birthweight is a popular outcome, but it is associated with several potentially confounding factors and it is complicated to interpret regarding the health of the child. On the contrary, larger birthweight might result in a higher risk of caesarian section, fetal hypoxia and stillbirth (Berntsen & Pinborg, 2018) and there may be later health risks for the child (Pinborg, 2019). As mentioned, altering epigenetics is believed to be a mechanism that may be influenced by different culture media, and therefore might influence birthweight and future health of the child (Kleijkers et al., 2015).
Some of the included studies assessed fresh embryo transfers only, while other studies assessed both fresh and frozen embryo transfers. Previous studies suggest the use of either fresh or frozen embryos could influence perinatal outcomes, and frozen embryo transfers might result in a higher birthweight than fresh embryo transfers (Wong et al., 2017;Berntsen & Pinborg, 2018). This is supported by the results on birthweight from Gu et al. (2016); and therefore, birthweight should be related to whether the child was the result of fresh or frozen embryo transfer.
The comparison between G5 series and other culture media is complicated by the fact that the Vitrolife G5 series consists of more than ten products according to their brochure (A link to the list of Vitrolife G5 products can be found in the references). Even inside the G5 series, there are different options for embryo culture media: G-1 Plus and G-2 Plus are ready for use, while addition of human serum albumin is needed in the equivalent G-1 and G-2. Previous studies suggest that these two options of protein sources inside the G5 series might result in a difference in birthweight (Zhu et al., 2014).
There were no significant findings in fertilization rate, biochemical pregnancy rates, miscarriage rates, multiple pregnancy rates and congenital malformations. The secondary outcomes must be interpreted with care regarding the quality of embryo culture media. Like birthweight, they become relevant if there is a clear correlation to IVF success rates, such as measured in live birth rates or child's health.
In general, the comparison of the studies is difficult since there are varying definitions of inclusion criteria for women, varying definitions of outcomes and varying laboratory routines. For instance, different guidelines for transferring one or more embryos at a time could influence some of the outcomes and may increase live birth rates. Most of the studies did not report on dropouts. It is unclear whether there were no dropouts or if they did not include them in their analyses and this might cause bias. In one retrospective cohort study (Lin et al., 2013), there were no clear descriptions of when they used one culture media or the other. If the distribution of the culture media is not random, this might cause selection biases. Only one study (Kleijkers et al., 2016) described a proper blinding in their methods. The lack of good description of the randomization between the culture media might be a problem. While some of the outcomes such as the biochemical pregnancy rates are measurable facts, there is a considerable subjectivity in the assessment of the best embryo for transfer. There are some limitations to this review. There were five RCTs and six retrospective cohort studies. Only one RCT had a description of a good design including blinding of both patients and doctors, proper description and handling of dropouts and a power calculation. The other studies had varying data quality due to the description of the population, the randomization, handling of dropouts etc. Most studies randomized women, whereas in two studies the oocytes were randomized. As long as the randomization is done properly (and blinded), and the study has a reasonable size, this will most likely not influence the results. G5 media is compared to different culture media, which means that there are only few results examining some of the same outcomes (Table 6 and 7). The results of this review are based on comparing the results of the individual studies. Since the culture media, the inclusion criteria for the women and the definitions of the outcome vary in the studies; it was not possible to do a metanalysis on the topic.
As mentioned, there are many culture media available and many different outcomes, and so far there is very limited good evidence when comparing different culture media. This review indicates that no culture media is clearly superior or inferior to others, which allows the embryologist to take other factors such as affordability, availability, workload in the laboratory and experience/preference into account when choosing a media. Furthermore, the different JBRA Assist. Reprod. | v.25 | n o 3| July-Aug-Sept/ 2021 outcomes highlight the importance of further research into media effects, both on success rates and on the long-term health issues, where evidence hopefully becomes available during the next years.

CONCLUSION
In conclusion, Vitrolife G5 series culture media was found to have a trend towards higher live birth rates, but not significant compared to other common culture media. This result is comprised of only one trial (RCT).
Birthweight had equivocal results with three out of six studies showing significantly lower (2)/higher (1) birthweights, whereas the others were non-significant. Likewise, overall no significant differences were found concerning the secondary outcomes.

Embase search
First, the search terms were divided into PICO search blocks: P: IVF OR in vitro fertilization OR infertility OR ART OR assisted reproductive technology OR 'in vitro fertilization'/exp OR 'infertility'/exp OR 'infertility therapy'/exp I: Culture media OR culture medium OR culture system OR embryo culture OR 'culture medium'/exp C: Vitrolife OR G5 OR v5 OR G-1 OR G-2