Best Practice & Research Clinical Obstetrics and GynaecologyVol. 21, No. 1, pp. 21e40, 2007doi:10.1016/j.bpobgyn.2006.09.008

available online at http://www.sciencedirect.com

2

Blastocyst or cleavage-stage

embryo transfer?

Neil Johnson* CREI, FRANZCOG

Associate Professor

Department of Obstetrics & Gynaecology, University of Auckland, Auckland City Hospital,

Grafton, Auckland, New Zealand

Debbie Blake PhD

Embroyologist

Biotechnology Research Institute, Auckland University of Technology, 19 Mount Street, Auckland, New Zealand

Cindy Farquhar MD, FRANZCOG

Professor

Department of Obstetrics & Gynaecology, University of Auckland, Auckland City Hospital,

Grafton, Auckland, New Zealand

Key issue� Does blastocyst embryo transfer offer benefit over cleavage-stage embryo transfer in assisted

conception treatment?What do we know?� Following natural conception, the embryo is traversing the fallopian tube at cleavage stage; it

is in the uterus at blastocyst stage.� Embryo development to cleavage stage occurs under maternal genomic control.� To reach blastocyst stage, embryos need to develop under their own genomic control.� Selection of embryos, in order to replace the embryo or embryos with maximal viability, has

presented a challenge.What do we think we know?� Implantation rates of blastocysts tend to be higher than those for cleavage-stage embryos.

Should not the chance of success with a blastocyst culture and transfer policy be higherthan that with cleavage-stage embryo transfer?

� It might be possible to select couples who would benefit from blastocyst culture on the basisof milestones at certain stages of an in-vitro fertilization (IVF) cycle.

* Corresponding author. Tel.: þ64 9 3670000; Fax: þ64 9 3035969.

E-mail address: n.johnson@auckland.ac.nz (N. Johnson).

1521-6934/$ - see front matter ª 2006 Elsevier Ltd. All rights reserved.

eg-

heple

ghal

e-rs

es,ns,ityuitof

es.utalsx-

erinofityreorasia-

erreel

e7,u-o-l.8

riab-cyns,s-

edofge

22 N. Johnson et al

� It might it be possible for clinics to reduce the multiple pregnancy rate and maintain the prnancy rate by employing blastocyst culture.

What do we not know?� Whether a policy of blastocyst culture offers genuine advantages by either increasing t

chance of success of IVF/intracytoplasmic sperm injection or reducing the chance of multipregnancy.

� Which couples, if any, benefit from blastocyst culture.

Key words: blastocyst; cleavage stage; embryo transfer; ICSI; IVF.

Improving the outcome of assisted reproductive technology (ART) treatments throuin-vitro fertilization (IVF) and intracytoplasmic sperm injection (ICSI) is a continuchallenge for fertility services.

The fledgling era of ART ended in the mid 1990s when success rates, that had rmained relatively static until that time, began to rise substantially. Numerous factoprobably contributed to the improving success rates, including new techniquhigh-quality consumables and equipment, sophisticated ovarian stimulation regimeand a wealth of clinical experience that had been accumulated by this time. Fertilspecialists are continually inspired to trial newly reported innovations in the pursof improved livebirth outcomes for their patients. However, unravelling the hypenew techniques from the reality of their effectiveness is essential for global advancAs the rate of human implantation is intrinsically low, it is difficult to establish small bsignificant improvements. Systematic meta-analysis of all randomized controlled tri(RCTs) is therefore the only way to establish definitively if an innovation, such as etended embryo culture to the blastocyst stage, offers a true advance.1

Traditionally, embryos have been transferred at the cleavage stage, 2e3 days aftegg collection, although allowing human embryos to develop to the blastocyst stageIVF programmes is not novel. What is new, however, is the accessibility and rangereportedly successful media products, resulting in a widespread rise in the acceptabiland use of this approach.2 Initial reports of blastocyst culture involved a single cultumedium consisting of a mixture of a complex and simple media formulation3

co-culture.4 More recently, the development of stage-specific sequential media hbeen claimed to allow 36e66% of embryos to develop to blastocysts with a high vbility of up to 50% implantation rate.5,6

There are two central reasons why an alternative to cleavage-stage embryo transfwas proposed. Firstly, it has long been recognized that it is physiologically prematuto expose early-stage embryos to the uterine environment. In vivo, embryos travthrough the fallopian tubes and do not reach the uterus before the morula stagwhich equates to at least day 4 of in-vitro culture. The uterus provides a different ntritional milieu from the fallopian tube; it has been postulated that this may cause hmeostatic stress on the embryo, resulting in a reduced implantation potentiaSecondly, there are widely acknowledged shortcomings of the morphological criteused for selection of cleavage-stage embryos for transfer on day 2 or 3. There is sustantial debate over the correlation of embryo morphological features with pregnanrates.9,10 Prior to day 3 of culture, when genomic activation and compaction begiembryonic development is primarily controlled by transcripts and stored RNA mesages of maternal origin.11 Only after this transitional stage does development proceunder the control of an activated embryonic genome, resulting in the expressionnumerous growth factors and receptors. Furthermore, it is suspected that a lar

Blastocyst or cleavage-stage embryo transfer? 23

proportion of morphologically normal day 3 embryos are chromosomally abnormal,thus contributing to the 80e90% rate of implantation failure observed in cleavage-stage embryos.12 Extending embryo culture until the blastocyst stage might provide ad-vantages over traditional protocols by allowing transfer of embryos into a synchronizeduterine environment, and selection of only those embryos that have demonstrated thepotential for continued development under embryonic genomic control.

Until recently, it was widely accepted that in order to achieve acceptable pregnancyrates, two or more embryos had to be replaced.13 However, pressure on the ART in-dustry to reduce the multiple birth rate over the past decade has seen a steady declinein the number of embryos transferred. Single embryo transfers for selected patientgroups are now considered to be standard practice in many clinics throughout theworld. The importance of selecting the single most viable embryo for transfer has in-tensified the search for improving the assessment of the quality of embryos.

Advocates of blastocyst culture have also emphasized:

� the increased implantation rate compared with cleavage-stage embryos; and� the need to transfer fewer embryos lowers the costly multiple birth rate.14

Whilst critics of blastocyst culture have countered, claiming an increase in:

� having no embryos to transfer15, although the day of patient recruitment into theblastocyst programme is crucial to this argument. It is important to be awarethat clinic policies may differ, not only on the day on which this decision is made,but also on the minimum criteria for blastocyst culture (e.g. the numbers of folliclesprior to egg collection, fertilized eggs collected, eight-cell embryos on day 316;� failure to have extra embryos that can be freeze-stored for future use17;� monozygotic (MZ) twinning;� altered sex ratio in births18; and� sensitivity of the system to suboptimal conditions.

They have also questioned whether there are patient groups for whom blastocystculture is disadvantageous.

The multiple birth rate in ART has increased since the 1970s, with the increase inthe rate of triplets being greater than the rise in twins.19 In addition to the trend ofincreased multiple births relating to multiple embryo transfer, there is also a trend to-wards an increase in the rate of MZ twinning, resulting from the early splitting of anembryo. About one in 330 of all spontaneous births is an MZ twin birth, thus onein 165 babies born is an MZ twin. MZ twinning is associated with increased morbidityabove that of dizygotic twinning; miscarriage, structural congenital anomaly (includingconjoined twins, acardiac state, microsomia, limb reductions, club feet, hip disloca-tion), growth discordancy and twin-to-twin transfusion syndrome occur more com-monly in MZ twins.20 Mouse studies show a possible increase in epigeneticdysfunction in MZ twins, possibly increasing the chance of imprinting disorders suchas Beckwith-Weideman syndrome. MZ twinning can also be produced in animals bytoxic exposure and by so-called ‘zona tampering’.21

An increased frequency of MZ twins has been reported in association with concep-tions following ART for over 20 years.22 Ovulation induction, ICSI, blastocyst cultureand assisted hatching have all been implicated in MZ twinning. In ART cycles, MZ twin-ning rates above 1% are reported frequently and it is estimated that MZ twinning isincreased two to five times in babies born following IVF.21 MZ twinning is a result

24 N. Johnson et al

of a hatching event, caused by a breech in the integrity of the zona pellucida, herniationof the blastomeres and splitting of the embryo. Such a breech in the zona pellucidamay be caused by ART procedures such as ICSI and assisted hatching. An alternativetheory is that developmental discordance between adjacent cells may occur thatresults in repulsion and early splitting of the embryo. MZ twins produced by IVFare also more likely to be dichorionic, diamniotic twins, which suggests that the MZtwinning process and subsequent separation of twins seen with IVF happens at avery early stage.

There are pitfalls in estimating the rate of MZ twinning.23 The most frequently usedestimation of monozygosity counts the number of monochorionic twins, either by an-tenatal ultrasound or by postpartum placental examination, in a population of childrenconceived by ART. The problem of detection bias arises as this assessment is unlikelyto be as rigorous amongst spontaneously conceived multiple pregnancies. There is alsothe problem that the diagnosis of zygosity cannot be determined without genetic test-ing in as many as 43% of cases, and therefore studies based on clinical estimations maymiss as many as one-third of MZ twins. Generally, it is agreed that many of the studiesin ART are thought to underestimate the rate of MZ twins. The most reliable studiesshould include both ultrasound determination and postpartum zygosity estimation byDNA ‘fingerprinting’ using cord blood, saliva or buccal smears.

OBJECTIVES

� To determine if blastocyst embryo transfer results in higher livebirth and pregnancyrates, and lower multiple pregnancy rates than cleavage-stage embryo transfer.� To assess the factors that contribute to the primary outcome (including failure to

transfer any embryos, rate of embryo freezing, implantation and miscarriage). Sub-group analysis will assess the effect of patient prognosis and differences in the num-ber of embryos transferred.� To determine if blastocyst culture and embryo transfer is associated with an

increase in MZ twinning.

METHODOLOGY

A systematic review of randomized trials was undertaken according to Cochrane Men-strual Disorders and Subfertility Group guidelines.1

Criteria for study inclusion

� Only RCTs were eligible for inclusion in the meta-analysis. Non-randomized studieswere only considered where MZ twinning rates were considered, if they specificallyconsidered MZ twins separately from dizygotic twins in an ART population.� Participants: couples undergoing IVF, ICSI or oocyte donation treatment.� Interventions: all known culture methods for IVF/ICSI where embryos were grown

for 2e6 days in vitro prior to embryo transfer, and where cleavage-stage (day 2or 3) embryo transfers were compared with blastocyst (day 5 or 6) embryotransfers.� Outcome measures: primary outcome was live birth per couple; secondary out-

comes (per couple) were clinical pregnancy, multiple pregnancy, high-order multiple

Blastocyst or cleavage-stage embryo transfer? 25

pregnancy, MZ twins, miscarriage rate, embryo freezing rate, and failure to have anyembryos transferred.

Search strategy and data extraction

Standard Cochrane Menstrual Disorders and Subfertility Group search and data ex-traction strategies were employed for RCTs.1 The last search date was in May2005. Non-randomized studies were also identified using the search term ‘monozy-gotic twin’ for this outcome.

DESCRIPTION OF STUDIES

Data from randomized controlled trials

Seventy-two and 199 publications were retrieved by the MEDLINE search and theEMBASE search, respectively, of which 44 publications were selected as primarystudies. Fifteen studies were selected for inclusion. Twenty-nine primary studieswere excluded [this includes all the ‘quasi-randomized’ trials that were included inthe original Cochrane Review (2000), but which have now been excluded24 fromthis update3,25e29]. Other trials were excluded as the Vero co-culture is no longerused30, and in one trial, more than 50% of the group allocated to blastocyst transferactually underwent transfer on day 3.31 All other excluded trials failed to use a randomdesign.

This review consists of a total of 1944 couples in 15 RCTs from 16 publications.32e47

Two conference abstracts presented different aspects of data from the sametrial.32,33 A detailed description of studies is found in the published Cochranereview.1

The size of trials ranged from 2334 to 46035 couples. The majority of trialswere carried out in less than 6 months except for the two largest studies. All studieswere reported to have been performed at single private or university-based clinics.Nine countries were represented in the included studies, with Belgium being themost prolific (five studies). Countries represented were Brazil32,33, Belgium34e37, Aus-tralia38, Israel39e41, Jordan42, Italy43,44, Denmark45 and the USA/Hawaii.46,47

Patient selection criteria comprised three main groups: unselectedpatients32,33,35e37,42,44; good prognostic factors where those expected to do wellwith blastocyst culture were positively selected38e40,43,45e47; and poor prognosticfactors where couples were selected who had experienced multiple failures with con-ventional treatment or had poor response to ovulation induction.34,41

Most studies recruited women aged less than 40 years; the mean age across all thestudies varied from 29 to 34 years. Trials that provided details on the ovarian stimulationregimen reported using a similar gonadotrophin-releasing hormone pituitary downre-gulation protocol prior to human menopausal gonadotrophin/follicle-stimulatinghormone administration. Eight trials used sequential media, five did not state themedia used, and two used single media for day 2/3 culture and sequential mediafor day 5/6 culture. Cryopreservation of embryos in both experimental groupswas reported in eight of the 15 RCTs included in this review.32,33,35,36,40,42,43,45,46

Other interventions, such as assisted hatching, were either not provided or notreported for the majority of trials. One RCT had a co-intervention of assisted

26 N. Johnson et al

hatching for the day 3 embryo transfer group and not for the blastocyst group.46

For the day 2/3 transfer groups, most transfers were on day 3, with a minorityon day 232e34,36,37; one RCT performed transfers on day 2 or day 3.41

Data from non-randomized controlled trials

Nineteen studies were identified that reported the rate of MZ twins following ARTcycles: six studies reported on blastocyst transfers48e53; six reported on assistedhatching53e58; one reported on ovulation induction22; and the remainder were eithermixed techniques or did not specify the ART technique used.23,59e64 Two studies re-ported on MZ twins in spontaneous pregnancies.55,65

METHODOLOGICAL QUALITY OF INCLUDED STUDIES

Data from randomized controlled trials

Allocation concealment

Only seven RCTs had enough information to confirm adequacy of allocation conceal-ment; sealed envelopes in each case.36,38e42,45

Randomization method

Six RCTs used computer-generated randomization35,38,41,43,46,47, one used a list37, andthe remainder did not state their randomization method.

Blinding

Duration of culture and the day of embryo transfer was different for each of the ex-perimental groups, making it impossible to blind which group a patient was in from thedoctor, scientist, nurse or patient. There was no evidence to suggest that the assessorof the follow-up outcomes or the statistician in any trial was blinded to the assignmentstatus.

Intention to treat, withdrawals and dropouts

No published trial stated that an intention-to-treat analysis was performed. Identifica-tion of patients failing to have an embryo transfer was not stated or unclear in sometrials. One RCT implied that a 100% embryo transfer rate was achieved in both the day2/3 and the day 5/6 groups, which is unexpectedly high.39 Where the number ofcouples and the number of embryo transfers were different, the number of coupleswas used as the denominator, even when exclusions took place after randomization(with an assumption that no pregnancies occurred), this imputation making a post-hoc intention-to-treat analysis possible for a number of RCTs.36e38,47 The timing ofrandomization (before egg collection, on the day of fertilization check, or on day 2)had an effect on the number of withdrawals in each trial.1

Power analysis

Six trials reported having performed a power analysis35e38,45,47, but all were unable toachieve the level of statistical significance required with the number of patients

Blastocyst or cleavage-stage embryo transfer? 27

recruited. Attempts were made to obtain additional information regarding all aspectsof randomization, blinding, power analysis and intention to treat from all trial authors.

Data from non-randomized trials for monozygotic twinning

Only one study used postpartum DNA testing to establish zygosity; three used post-partum examination of the placenta and membranes; six did not specify how the diag-nosis was made; and the remainder defined MZ twins on the basis of the number ofembryos transferred being exceeded by the number of gestational sacs.

RESULTS

Livebirth rate

Figure 1 demonstrates that there was no evidence of a difference in livebirth rate percouple between the two treatment groups [seven RCTs; odds ratio (OR) 1.03, 95%confidence interval (CI) 0.74e1.44]. No heterogeneity was detected (I2¼ 37.9%).

There was no evidence of any differences in livebirth rate when RCTs were sub-grouped, according to a priori specified criteria, by:

� number of embryos transferred: in four RCTs, more cleavage-stage embryos thanblastocyst embryos were transferred (OR 1.11, 95% CI 0.60e2.07); in threeRCTs, equal numbers of embryos were transferred (OR 1.00, 95% CI 0.67e1.49);� prognosis: three RCTs had good-prognosis patients (OR 0.94, 95% CI 0.56e1.57);

two RCTs had poor-prognosis patients (OR 2.05, 95% CI 0.53e7.96); and twoRCTs did not select for prognosis (OR 1.02, 95% CI 0.64e1.63); and� timing of randomization: the day of randomization was at the start of the cycle in

three RCTs (OR 1.05, 95% 0.67e1.64); where randomization was on the day ofovum pick-up or the following day, there was no significant difference (OR 0.94,95% 0.56e1.57).

A priori specified sensitivity analysis, excluding the studies that did not report ad-equate concealment of allocation, did not affect the significance of livebirth rates (OR1.40, 95% CI 0.88e2.23). Exclusion of the study41 where different media were used ineach arm of the study did not affect the significance of the livebirth rates (OR 1.15,95% CI 0.76e1.74).

Clinical pregnancy

Figure 2 shows no evidence of a difference in pregnancy rate per couple between thetwo treatment groups (15 RCTs, OR 1.05, 95% CI 0.88e1.26). No heterogeneity wasdetected (I2¼ 26.3%).

There was no evidence of any differences in pregnancy where RCTs were sub-grouped by number of embryos transferred or prognosis.

Sensitivity analysis showed no effect on the results of excluding studies that did not re-port adequate allocation concealment, and no effect of excluding studies where differentmedia were used in each arm of the study.32,33,39,41,42 The day of randomization did notaffect the results. The outcomes in the four RCTs with the best pregnancy rate in the con-trol arm and the four RCTs with the worst pregnancy rates did not show different results.

adaptation of a meta-analysis graph in Blake et al (2006).1 Copyright

28

N.

Johnso

net

al

Figure 1. Livebirth rate per couple (grouped by number of embryos transferred). This figure is an

Cochrane Library, reproduced with permission.

re is an adaptation of a meta-analysis graph in Blake et al (2006).1

Blasto

cystor

cleavage-stageem

bryo

transfer?

29

Figure 2. Clinical pregnancy rate per couple (grouped by number of embryos transferred). This figu

Copyright Cochrane Library, reproduced with permission.

30 N. Johnson et al

Multiple pregnancy

Figure 3 shows that there was no evidence of a difference in multiple pregnancy rateper couple between the two treatment groups in 11 RCTs (OR 0.86, 95% CI0.64e1.15). No heterogeneity was detected (I2¼ 0%).

A key consideration, seen in Figure 3, regarding multiple pregnancy is subgroupingby the number of embryos transferred; there was no difference in multiple pregnancyrates in the six RCTs where more cleavage-stage embryos than blastocyst embryoswere transfered (OR 0.76, 95% CI 0.48e1.22), or in the five RCTs where equal num-bers of embryos were transferred (OR 0.93, 95% CI 0.63e1.35).

Subgrouping by prognosis had no effect on multiple pregnancy rates. Six RCTs hadwomen with a good prognosis for pregnancy (OR 0.76, 95% CI 0.5e1.14), one RCThad women with a poor prognosis (OR 0.89, 95% CI 0.14e5.81), and four RCTs didnot select for prognosis (OR 0.96, 95% CI 0.62e1.47).

Sensitivity analysis, excluding the studies that did not report adequate concealmentof allocation, did not affect the significance of multiple pregnancy rates (OR 0.95, 95%CI 0.65e1.39). Exclusion of the studies in which different media were used in each armof the study32,33,39,41,42 did not affect the significance of the multiple pregnancy rates(OR 1.02, 95% CI 0.82e1.27).

High-order multiple pregnancy

There was no evidence of a difference in pregnancy rate per couple between the twotreatment groups in nine RCTs (OR 0.44, 95% CI 0.15e1.33). No heterogeneity wasdetected (I2¼ 0%).

There was no evidence of an effect of subgrouping by:

� number of embryos transferred: in five RCTs, more cleavage-stage embryos thanblastocyst embryos were transfered (OR 0.33, 95% CI 0.01e8.28); in four RCTs,equal numbers of embryos were transferred (OR 0.46, 95% CI 0.14e1.49); or� prognosis: in six RCTs, the prognosis for pregnancy was good (OR 0.29, 95% CI

0.08e1.06); in one RCT, the prognosis for pregnancy was poor (OR 4.2, 95% CI0.16e107); and in two RCTs, there was no selection for prognosis and therewere no high-order pregnancies in either group.

Monozygotic twinning

Only six RCTs reported if any of the multiple pregnancies were MZ; there was one MZtwin pregnancy following day 2/3 transfer47 and one following day 5/6 transfer.41

Data from non-randomized studies showed that overall in ART pregnancies, therate of MZ twins ranged from 0.4% to 8.6%. Amongst pregnancies following blastocysttransfers, the rate of MZ twins ranged from 1.16% to 5.7%.

Miscarriage

There was no evidence of a difference in miscarriage rate per couple between the twogroups (nine RCTs, OR 1.33, 95% CI 0.89e2.01). No heterogeneity was detected(I2¼ 0%).

gure is an adaptation of a meta-analysis graph in Blake et al (2006).1

Blasto

cystor

cleavage-stageem

bryo

transfer?

31

Figure 3. Multiple pregnancy rate per couple (grouped by number of embryos transferred). This fi

Copyright Cochrane Library, reproduced with permission.

32 N. Johnson et al

Embryo freezing

Rates of embryo freezing per couple showed a significant difference in favour of day 2/3 transfers (eight RCTs, OR 0.45, 95% CI 0.36e0.57). Significant heterogeneity wasdetected (I2¼ 88%). Subgrouping by number of embryos transferred and prognosisdid not affect the significance of the result.

Failure to transfer any embryos

Failure to transfer any embryos per couple was significantly more common in the day5/6 transfer group than in the day 2/3 transfer group (14 RCTs, OR 3.35, 95% CI2.20e5.10). No heterogeneity was detected (I2¼ 19.8%). Subgrouping by number ofembryos transferred and prognosis did not affect the significance of the results.

Per cycle data

Per cycle data, more often reported in clinical trials, cannot be included in the meta-analysis as they would not generate valid estimates or confidence intervals. Reportednumbers confirm the results for ‘per couple’ data.

Blastulation

Rates of formation of blastocysts (day 5/6 transfer only) ranged from 28%39 to89.9%.37 There was no apparent association between RCTs reporting a high blastula-tion rate and superiority of blastocyst transfer over cleavage-stage embryo transfer inthose RCTs.

Implantation data

For day 2/3 transfer, the implantation rate varied from 3% to 69%, and for day 5/6transfer, the implantation rate varied from 20% to 50% (Figure 4).

Possible bias amongst trials

The funnel plot for clinical pregnancy rate demonstrates that the studies are distrib-uted evenly across the graph, suggesting that publication bias is not present.

DISCUSSION

Higher implantation rates were widely expected with blastocyst culture. This wasgenerally the case, with six of the 15 RCTs reporting implantation rates showinga significantly higher rate for blastocyst embryo transfer compared with cleavage-stage embryo transfer, although one RCT reported a significantly higher implanta-tion rate in women having cleavage-stage transfer (Figure 4). Embryo transfer onday 5 or 6 offers the opportunity to replace embryos into a more synchronizeduterine environment, and extended culture provides an opportunity to select thoseembryos that have proven ability to survive and develop to a more advanced stagein vitro. However, there is no evidence from the meta-analysis that blastocyst

is an adaptation of a meta-analysis graph in Blake et al (2006).1 Copyright

Blasto

cystor

cleavage-stageem

bryo

transfer?

33

Figure 4. Implantation rates for individual trials e not suitable for meta-analysis. This figure

Cochrane Library, reproduced with permission.

34 N. Johnson et al

transfer is generally associated with improved clinically meaningful outcomes(Figures 1e3), including live birth, pregnancy, multiple pregnancy, high-order multi-ple pregnancy and miscarriage. Even the RCT where the participants were womenwith multiple IVF failures41, where any endometrial asynchrony could have ac-counted for these prior failures, did not show improved clinical outcomes withblastocyst transfer. However, blastocyst transfer increases the chance of havingno embryos to transfer (owing to the attrition rate of embryos in vitro betweencleavage and blastocyst stage) and reduces the embryo freezing rate. The increasedrate of failure to transfer for day 5/6 is largely the result of patients whose em-bryos had arrested development prior to the day of embryo transfer. A furtherconfounding factor is the widely variable policy for minimal quality of embryosfor transfer that may have existed amongst the trials.

The number of embryos frozen is an important consideration when assessing theeffectiveness of a treatment because it offers the patient an additional opportunityto achieve a pregnancy. While it is clear that blastocyst culture results in a reducednumber of embryos for freezing, it is not clear if this impacts on the final outcomeof the pregnancy and livebirth rate per started cycle. Advocates of blastocyst cultureargue that patients may prefer to be informed on day 5 if their embryos have lowviability and not to have an embryo transfer, rather than to suffer the disappointmentof failed implantation when pregnancy does not result from embryo transfer. Freezingnumerous poor-quality embryos is unlikely to offer any advantage to the infertile cou-ple as they go through several frozen embryo transfer cycles with no improvement inoutcome, and all the inconvenience and distress associated with unsuccessful embryothaw cycles, and it may increase time to achieve a pregnancy. There has been littleresearch into the feelings of women given such choices.66 Only one RCT has fullyreported data on pregnancies following transfer of frozen embryos.35 Such survivalanalysis data reporting is also unlikely to be forthcoming in the future because ofthe long time span between a woman’s pregnancy cycle and a subsequent frozenembryo cycle.

Freezing protocols for early-cleavage-stage and blastocyst embryos are funda-mentally different and the effectiveness of the latter has yet to be widely accepted,particularly in embryos that have been cultured in sequential media. Furthermore,such confidence in the culture conditions during extended culture may need to betreated with caution for two reasons. Embryo viability based on its morphology onday 5 is by no means certain; indeed, there are widespread reports of pregnanciesfrom developmentally delayed morulas on day 5 (although the same can be said forpoor morphology in cleavage-retarded embryos on day 2 or 3). If blastocyst cul-ture is used strictly to select the most viable embryos, if the slow cleaving embryoon day 3 has a higher chance of pregnancy if replaced into the uterus early than ifsubjected to extended culture, blastocyst transfer will have a detrimental effect.68

Adaptability of an embryo to survive extended culture may come at the price ofviability. Recent studies have explored which key indicators might select a patientgroup that will benefit most from blastocyst culture, such as the number of pro-nuclear embryos on day 1, the pronuclear grading profile19, the number of earlycleaving embryos, and the number of eight-cell embryos on day 3.67 These theorieshave been challenged by the results of this assessment of the best available evi-dence concerning blastocyst embryo transfer, which do not suggest any benefitin terms of pregnancy outcome.

One patient selection policy that has become popular is allowing only those coupleswho have two, three or more high-quality eight-cell embryos on day 3 to continue

Blastocyst or cleavage-stage embryo transfer? 35

with blastocyst culture for day 5/6 embryo transfer.67 This so-called ‘a la carte’approach to blastocyst culture was compared with a control group of women random-ized for day 2/3 embryo transfer in an RCT.31 This RCT was not included in thesystematic review as over 50% of the couples randomized to day 5/6 transfer hadtransfer on day 3, so pooling was considered to be inappropriate. It is unclear onthe basis of this one small study if there is benefit in this policy.

Anecdotally, there are many examples where clinics have used blastocyst culture(with an increased confidence in the implantation potential of blastocysts) to moveto a policy of single embryo transfer. Many clinics claim to have reduced their multiplepregnancy rate and maintained their pregnancy rate by employing blastocyst culture.Only one study in this meta-analysis had a policy for single blastocyst transfer38, buttwo cleavage-stage embryos were transferred in this RCT. One of the difficulties indrawing conclusions from the published blastocyst trials is the common occurrenceof variable embryo transfer policies between the two experimental groups. In themeta-analysis, significantly fewer embryos were transferred in the day 5/6 groupthan in the day 2/3 group. Many clinicians are concerned about the high incidenceof multiple pregnancies with multiple blastocyst transfer, and have a policy to transferno more than two blastocysts. Nonetheless, there were very few high-order multiplepregnancies in either group in the RCTs in this meta-analysis. For many patients un-dergoing blastocyst culture, there is simply a lack of choice as only one (if any) embryoreaches the blastocyst stage. The natural evolution of embryo transfer policies inter-nationally is towards single embryo transfer; it is likely that this will be the norm in-ternationally within a decade. In light of this, perhaps the most compelling RCTdata comes from the most recently published trial of blastocyst versus cleavage-stageembryo transfer, not yet included in this meta-analysis, which examined success rateswith single fresh blastocyst transfer versus single fresh cleavage-stage embryo trans-fer.68 After an interim analysis showed a higher pregnancy rate (P¼ 0.02) in womenundergoing single blastocyst transfer (32.0%) compared with single cleavage-stage em-bryo transfer (21.6%), the study was stopped early after recruiting 351 women (halfthe prior target number of recruits). Whilst compelling, this evidence requires repli-cation in other settings before a firm recommendation can be made to all clinics toswitch entirely from cleavage-stage to blastocyst embryo transfer in women undergo-ing single embryo transfer.

Single embryo transfer is not expected to impact on the rate of MZ twinning. Ithas been suggested that extended culture may create alterations in the zona pellu-cida that place the embryo at risk of abnormal hatching, resulting in MZ twinning.69

Unfortunately, only six of the trials included in this review reported on the presenceor absence of MZ twinning, but there was only one set of MZ twins in day 2/3 andin day 5/6 embryo transfer. Several published retrospective analyses of blastocysttransfer have reported an increased frequency of MZ twinning.48e50 Overall, the non-randomized studies suggest that the rate of MZ twinning may be increased inART pregnancies (0.4%e8.6% amongst non-randomized studies) compared with nat-urally conceived pregnancies (0.6%), with a rate of 1.16e5.7% following blastocysttransfer in non-randomized studies.

Cost comparisons of treatment have not been investigated in this review but areworthy of mention. From a laboratory perspective, the cost of setting up blastocystculture is not insignificant. Often an additional incubator is required due to the ad-ditional 2e3 days that the embryos remain in culture. Additional media costs, on theother hand, are negligible. Blastocyst culture is more labour intensive and laboratorystaff may be required to perform more weekend work, particularly if embryos from

36 N. Johnson et al

two different stages of development are required to be cryopreserved. For the pa-tient, the higher chance of cancellation of the treatment cycle prior to embryotransfer (due to the more stringent selection process of blastocyst culture) may re-sult in a lower treatment cost. As this review of the best available evidence fromRCTs of blastocyst versus cleavage-stage embryo transfer has not shown significantdifferences in clinically meaningful outcomes, secondary pros and cons, such as costs,efficiency, emotional impact, burden to society and time to pregnancy, should beconsidered.

ACKNOWLEDGEMENT

This review is an edited version of that published as a Cochrane review in the Co-chrane Library (http://www.cochranelibrary.net). Cochrane reviews are regularly up-dated with the emergence of new evidence and in response to comments andcriticisms. The Cochrane Library should be consulted for the most recent versionof the review.

Research agenda

� is there a threshold blastulation rate below which a policy of blastocyst transferis simply not viable?� is there a subgroup of patients for whom blastocyst culture is beneficial?

Practice points

� there is no evidence of a difference in live birth, pregnancy, multiple pregnancy,high-order multiple pregnancy, or miscarriage for blastocyst embryo transfercompared with cleavage-stage embryo transfer; this is true irrespective of pol-icies of selection of patients according to prognosis, timing of recruitment tothe programme, and numbers of embryos transferred� there is no subgroup of women in whom there is clear evidence of benefit of

blastocyst transfer� the higher chance of having no embryos to transfer and the reduced numbers

of embryos available for freezing should be discussed fully with all couples priorto blastocyst culture� all women undergoing ART cycles should be informed about the possible in-

creased risk of MZ twinning and the adverse events, although there is nogood evidence that blastocyst culture increases this risk significantly� many clinics choose to apply minimum criteria before offering blastocyst cul-

ture to couples, e.g. a minimum of two eight-cell embryos on day 3� some clinics have used policies of blastocyst embryo transfer to give them the

confidence to move to single embryo transfer; one RCT has shown a benefit ofblastocyst culture when single fresh embryos are transferred

Blastocyst or cleavage-stage embryo transfer? 37

REFERENCES

*1. Blake D, Proctor M, Johnson N et al. Cleavage stage versus blastocyst stage embryo transfer in assisted

conception. Cochrane Database Syst Rev 2005; 4: CD002118.

2. Alves da Motta EL, Alegretti JR, Baracat EC et al. High implantation and pregnancy rates with transfer of

human blastocysts developed in preimplantation stage one and blastocyst media. Fertil Steril 1998; 70:

659e663.

3. Scholtes MCW & Zeilmaker GH. A prospective, randomized study of embryo transfer results after 3 or

5 days of embryo culture in in vitro fertilization. Fertil Steril 1996; 65: 1245e1248.

4. Yeung WS, Ho PC, Lau EYet al. Improved development of human embryos in vitro by a human oviduc-

tal cell co-culture system. Hum Reprod 1992; 7: 1144e1149.

5. Jones GM, Trounson AO, Lolatgis N et al. Factors affecting the success of human blastocyst development

and pregnancy following in vitro fertilization and embryo transfer. Fertil Steril 1998; 70: 1022e1029.

6. Gardner DK, Vella P, Lane M et al. Culture and transfer of human blastocysts increases implantation

rates and reduces the need for multiple embryo transfers. Fertil Steril 1998; 69: 84e88.

7. Croxatto HB, Fuentaealba B, Diaz S et al. A simple non-surgical technique to obtain unimplanted eggs

from human uteri. Am J Obstet Gynecol 1972; 112: 662e668.

8. Gardner DK, Lane M, Calderon I et al. Environment of the preimplantation human embryo in vivo:

metabolite analysis of oviduct and uterine fluids and metabolism of cumulus cell. Fertil Steril 1996;

65: 349e353.

9. Steer CV, Mills CL, Tan SL et al. The cumulative embryo score: a predictive embryo scoring technique

to select the optimal number of embryos to transfer in an in-vitro fertilization and embryo transfer

programme. Hum Reprod 1992; 7: 117e119.

10. Palmstierna M, Murkes D, Csemizdy G et al. Zona pellucida thickness variation and occurrence of

visible mononucleated blastomeres in pre-embryos are associated with a high pregnancy rate in IVF

treatments. J Assit Reprod Genet 1998; 15: 70e75.

11. Braude P, Bolton V & Moore S. Human gene expression first occurs between the four and eight-cell

stages of preimplantation development. Nature 1988; 332: 459e461.

� does blastocyst culture predispose to MZ twinning or any other harmfuloutcome?� new generation sequential media for blastocyst culture require rigorous eval-

uation compared with existing day 2/3 culture methods and other forms ofblastocyst culture, such as co-culture� improvement of blastocyst cryopreservation techniques will necessitate fur-

ther research if blastocyst culture is to become a widespread, reliable and suc-cessful replacement for day 2/3 culture� further well-designed RCTs are required to answer the real clinical questions

when considering the options of cleavage-stage with blastocyst embryo trans-fer; such trials must:e have an adequate sample size to detect clinically important differences in the

primary outcomes;e have sound methods of randomization and adequate allocation concealment;e use intention-to-treat analyses;e consider the most relevant outcome, that of delivery of a healthy singleton

baby at term; ande follow-up cumulative livebirth rates for longer (including results from em-

bryo thaw cycles), presented as a survival analysis.� can these techniques be used to assist clinics to move to a responsible policy of

single embryo transfer so that costly multiple birth rates will be minimized?

38 N. Johnson et al

12. Magli MC, Gianaroli L, Munne S et al. Incidence of chromosomal abnormalities from a morphologically

normal cohort of embryos in poor-prognosis patients. J Assit Reprod Genet 1998; 15: 297e301.

13. Edwards RG & Steptoe PC. Current status of in vitro fertilisation and implantation of human embryos.

Lancet 1983; 2: 1265.

14. Jones GM & Trounson AO. The benefits of extended culture. Hum Reprod 1999; 14: 1405e1408.

15. Marek D, Langely M, Gardner DK et al. Introduction of blastocyst culture and transfer for all patients in

an in vitro fertilization program. Fertil Steril 1999; 72: 1035e1040.

16. Milki AA, Hinckley MD, Fisch JD et al. Comparison of blastocyst transfer with day 3 embryo transfer in

similar patient populations. Fertil Steril 2000; 73: 126e129.

17. Tsirgotis M. Blastocyst stage transfer: pitfalls and benefits e too soon to abandon practice? Hum Reprod

1998; 13: 3285e3295.

18. Menezo YJ, Chouteau J, Torello J et al. Birth weight and sex ratio after transfer at the blastocyst stage in

humans. Fertil Steril 1999; 72: 221e224.

19. Scott L, Alvero R, Leondires M et al. The morphology of human pronuclear embryos is positively re-

lated to blastocyst development and implantation. Hum Reprod 2000; 15: 2394e2403.

20. Machin GA. Why is it important to diagnose chorionicity and how do we do it? Best Pract Res Clin Obstet

Gynaecol 2004; 18: 515e530.

21. Hall JG. Twinning. Lancet 2003; 362: 735e743.

22. Derom C, Vlietinck R, Derom R et al. Increased monozygotic twinning rate after ovulation induction.

Lancet 1987; 1: 1236e1238.

23. Blickstein I. Estimation of iatrogenic monozygotic twinning rate following assisted reproduction: pitfalls

and caveats. Am J Obstet Gynecol 2005; 192: 365e368.

24. Blake D, Proctor M, Johnson N et al. Cleavage stage versus blastocyst stage embryo transfer in assisted

conception. Cochrane Database Syst Rev 2002; 2: CD002118.

25. Gudmundsson J, Lundqvist M, Rova K et al. The outcome of IVF treatment after two or five days of

embryo culture. Hum Reprod 1998; 13 (Abstracts of the 14th Annual Meeting of the ESHRE, Goteborg),

0e011.

26. Levran D, Weissman A, Farhi J et al. The management of patients with repeated implantation failure:

a randomised prospective trial. Fertil Steril 1999; 72: S30.

27. Demylle D, Godin PA, Van Langendonckt A et al. Day 3 versus day 5/6 randomly elected transfers at the

first two IVF attempts. Hum Reprod 2000; 15 (Abstracts of the 16th Annual Meeting of ESHRE,

Bologna), 0e022.

28. Huisman GJ, Fauser BCJM, Eijkemans MJC et al. Implantation rates after in vitro fertilization and trans-

fer of a maximum of two embryos that have undergone three to five days of culture. Fertil Steril 2000;

73: 117e121.

29. Plachot M, Mayenga JM, Chouraqui A et al. Blastocyst stage transfer: the real benefits compared with

early embryo transfer. Hum Reprod 2000; 15: 24e30.

30. Janny L, Vye P, Pouly JL et al. Cocultures: diagnostic and therapeutic contribution in assisted reproduc-

tive technologies. Contracep Fertil Sexual 1993; 21: 391e394.

31. Boyarsky CY, Vasilevskaya SE & Boykov MV. Blastocyst transfer ‘a la carte’ and routine third day transfer:

a randomized controlled study. Abstracts of the 17th Annual Meeting of ESHRE. Hum Reprod 2001; 16: 8e9.

*32. Motta LA, Alegretti JR, Pico M et al. Blastocyst versus cleaving embryo transfer: a prospective random-

ized trial. Fertil Steril 1998; 70(supplement 1): S17.

*33. Motta ELA, Alegretti JR, Pico M et al. Practice of blastocyst transfer: how much does it help in the man-

agement of embryo cyropreservation? Fertil Steril 1998; 70(supplement 1): S110.

*34. Devreker F, Delbaere A, Emiliani S et al. Prospective and randomized comparison between transfer

on day 2 or day 5 for patients with more than four IVF attempts. In ESHRE Bologna Abstract Book.

2000, p. 135.

*35. Kolibianakis EM, Zilopoulos K, Verpoest W et al. Should we advise patients undergoing in-vitro

fertilization to start a cycle leading to a day 3 or day 5 transfer? Hum Reprod 2004; 19: 2550e2554.

*36. Van der Auwera I, Debrock S, Spiessens C et al. A prospective randomized study: day 2 versus day 5

embryo transfer. Hum Reprod 2002; 17: 1507e1512.

*37. Emiliani S, Delbaere A, Vannin A et al. Similar delivery rates in a selected group of patients, for day 2 and

day 5 embryos both cultured in sequential medium: a randomized study. Hum Reprod 2003; 18:

2145e2150.

Blastocyst or cleavage-stage embryo transfer? 39

*38. Livingstone M & Bowman M. Single blastocyst transfer: a prospective randomised trial. In Abstracts of the

17th World Congress on Fertility and Sterility. Melbourne: International Federation of Fertility Societies,

2001, p. 218.

*39. Coskun S, Hollanders J, Al-Hassan S et al. Day 5 versus day 3 embryo transfer: a controlled randomized

trial. Hum Reprod 2000; 15: 1947e1952.

*40. Levron J, Shulman A, Bider D et al. A prospective randomized study comparing day 3 with blastocyst-

stage embryo transfer. Fertil Steril 2002; 77: 1300e1301.

*41. Levitas E, Lunenfeld E, Har-Vardi I et al. Blastocyst-stage embryo transfer in patients who failed to con-

ceive in three or more day 2e3 embryo transfer cycles: a prospective, randomized study. Fertil Steril

2004; 81: 567e571.

*42. Karaki RZ, Samarraie SS, Younis NA et al. Blastocyst culture and transfer: a step toward improved in

vitro fertilization outcome. Fertil Steril 2002; 77: 114e118.

*43. Rienzi l, Ubaldi F, Iacobelli M et al. Day 3 embryo transfer with combined evaluation at the pronuclear

and cleavage stages compares favourably with day 5 blastocyst transfer. Hum Reprod 2002; 17:

1852e1855.

*44. Schillaci R, Castelli A, Vassiliadis A et al. Blastocyst stage versus day 2 embryo transfer in IVF cycles.

Hum Reprod 2002; 17 (Abstracts of the 18th Annual Meeting of ESHRE, Vienna), p. 418.

*45. Bungum M, Bungum L, Humaidan P et al. Day 3 versus day 5 embryo transfer: a prospective randomized

study. Reprod Bio Med Online 2003; 7: 98e104.

*46. Gardner DK, Schoolcraft WB, Wagley L et al. A prospective randomized trial of blastocyst culture and

transfer in in-vitro fertilization. Hum Reprod 1998; 13: 3434e3440.

*47. Frattarelli JL, Leondires MP, McKeeby JL et al. Blastocyst transfer decreases multiple pregnancy rates in

vitro fertilization cycles: a randomized controlled trial. Fertil Steril 2003; 79: 228e230.

48. Behr B, Fisch JD, Racowsky C et al. Blastocyst-ET and monozygotic twinning. J Assit Reprod Genet 2000;

17: 349e351.

49. da Costa AL, Abdelmassih S, de Oliveira FG et al. Monozygotic twins and transfer at the blastocyst

stage after ICSI. Hum Reprod 2001; 16: 333e336.

50. Jain JK, Boostanfar R, Slater CC et al. Monozygotic twins and triplets in association with blastocyst

transfer. J Assit Reprod Genet 2004; 21: 103e107.

51. Milki AA, Jun SH, Hinckley MD et al. Incidence of monozygotic twinning with blastocyst transfer com-

pared to cleavage-stage transfer. Fertil Steril 2003; 79: 503e506.

52. Wright V, Schieve LA, Vahratian A et al. Monozygotic twinning associated with day 5 embryo transfer in

pregnancies conceived after IVF. Hum Reprod 2004; 19: 1831e1836.

53. Alikani M, Noyes N, Cohen J et al. Monozygotic twinning in the human is associated with the zona

pellucida architecture. Hum Reprod 1994; 9: 1318e1321.

54. Hershlag A, Paine T, Cooper GW et al. Monozygotic twinning associated with mechanical assisted

hatching. Fertil Steril 1999; 71: 144e146.

55. Saito H, Tsutsumi O, Noda Y et al. Do assisted reproductive technologies have effects on the demog-

raphy of monozygotic twinning? Fertil Steril 2000; 74: 178e179.

56. Schieve LA, Meikle SF, Peterson HB et al. Does assisted hatching pose a risk for monozygotic twinning

in pregnancies conceived through in vitro fertilization? Fertil Steril 2000; 74: 288e294.

57. Sills ES, Moomjy M, Zaninovic N et al. Human zona pellucida micromanipulation and monozygotic twin-

ning frequency after IVF. Hum Reprod 2000; 15: 890e895.

58. Slotnick RN & Ortega JE. Monoamniotic twinning and zona manipulation: a survey of U.S. IVF centers

correlating zona manipulation procedures and high-risk twinning frequency. J Assit Reprod Genet 1996;

13: 381e385.

59. Abusheikha N, Salha O, Sharma V et al. Monozygotic twinning and IVF/ICSI treatment: a report of 11

cases and review of literature. Hum Reprod Update 2000; 6: 396e403.

60. Australia and New Zealand Assisted Reproductive Database, http://www.npsu.unsw.edu.au/

art8high.htm; 2002. National Perinatal Statistics Unit, Australia.

61. Alikani M, Cekleniak NA, Walters E et al. Monozygotic twinning following assisted conception: an anal-

ysis of 81 consecutive cases. Hum Reprod 2003; 18: 1937e1943.

62. Elizur SE, Levron J, Shrim A et al. Monozygotic twinning is not associated with zona pellucida micro-

manipulation procedures but increases with high-order multiple pregnancies. Fertil Steril 2004; 82:

500e501.

40 N. Johnson et al

63. Schachter M, Raziel A, Friedler S et al. Monozygotic twinning after assisted reproductive techniques:

a phenomenon independent of micromanipulation. Hum Reprod 2001; 16: 1264e1269.

64. Wenstrom KD, Syrop CH, Hammitt DG et al. Increased risk of monochorionic twinning associated

with assisted reproduction. Fertil Steril 1993; 60: 510e514.

65. Bulmer M. The Biology of Twinning in Man. Oxford: Clarendon Press, 1970.

66. Borg K, Moller A, Hammar M et al. Blastocyst culture e more or less stressful for patients?. In Abstract

Book ESHRE. Bologna, 2000, p. 48.

67. Racowsky C, Jackson KV, Cekleniak NA et al. The number of eight-cell embryos is a key determinant

for selecting day 3 or day 5 transfer. Fertil Steril 2000; 73: 558e564.

*68. Papanikolaou EG, Camus M, Kolibianakis EM et al. In vitro fertilization with single blastocyst-stage ver-

sus single cleavage-stage embryos. N Engl J Med 2006; 354: 1139e1146.

69. Cohen J, Elsner C & Kort HMH. Impairment of hatching process following IVF in the human and im-

provement of implanation by assisted hatching using micromanipulation. Hum Reprod 1990; 5: 7e13.