Article Towards a single embryo transfer

Transcription

1 RBMOnline - Vol 6. No Reproductive BioMedicine Online; on web 17 February 2003 Article Towards a single embryo transfer Dr David Gardner Dr David K Gardner was awarded his D.Phil. from the University of York in 1987 under the supervision of Professor Henry Leese. Following a postdoctoral fellowship with Professor John Biggers at Harvard Medical School, he moved to Australia in 1989 to work with Professor Alan Trounson at Monash University where he founded the Embryo Physiology Laboratory. In 1997 Dr Gardner moved to Denver, USA, to become the Scientific Director of the Colorado Centre for Reproductive Medicine. Dr Gardner is also the Scientific Director of Houston IVF, an Adjunct Professor at Colorado State University, and Visiting Professor at the University of Zhongshan, China. His research has focused on the physiology, metabolism and culture of the mammalian preimplantation embryo. Dr Gardner has edited six books and authored over 100 papers and book chapters. David has four fully developed blastocysts with his wife Jackie in Colorado. David K Gardner 1, Michelle Lane Colorado Center for Reproductive Medicine, 799 East Hampden Avenue, Suite 520, Englewood, Colorado 80110, USA 1 Correspondence: dgardner@colocrm.com Abstract The delivery of a single, healthy child is the desired outcome of human assisted reproduction techniques. To attain this goal, there is an increasing movement toward single embryo transfer. The question is, therefore, at what stage to transfer the human embryo back to the uterus? Maximal implantation rates reported to date have come from the transfer of blastocysts (70% fetal heart rate). In any given cycle of treatment the probability of conceiving a child will be further increased by the ability to cryopreserve those embryos not transferred. It is therefore proposed that the transfer of a single blastocyst is the best treatment for most patients, given the high implantation rates of fresh transfers, and that it is now possible to cryopreserve supernumerary blastocysts effectively. The next decision is how to culture the human embryo to the blastocyst stage. The use of sequential culture media, designed not only to allow for changes in nutrient requirements and metabolism as development proceeds, but also to minimize intracellular trauma, can facilitate the development of highly viable blastocysts. Sequential culture media have been evaluated against a single-step culture system. It has been shown that sequential media (G1/G2) produce more viable blastocysts than those embryos cultured in a single medium formulation (simplex optimized medium with elevated potassium and with amino acids, KSOM AA ) throughout the preimplantation period. Furthermore, even if KSOM AA is used for embryo culture, it is essential that the medium be renewed after 48 h to alleviate the toxicity associated with ammonium build-up. Of great significance, embryos cultured in sequential media G1 and G2 have the same rate of development as embryos developed in vivo. Keywords: single embryo transfer, blastocyst transfer 470 Introduction The objective of IVF should be to provide couples with a single, healthy baby per transfer and more than one healthy baby per treatment cycle. To ensure couples have a singleton pregnancy, there is an obvious requirement to move to single embryo transfers. For more than a single child to be conceived per treatment cycle there is a need to be able to cryopreserve the embryos not transferred without significantly compromising their viability. Which stage is optimal for the transfer of a single embryo back to the uterus? Figure 1 represents the highest published implantation rates obtained for transfer on days 1, 3 or 5 following oocyte retrieval and insemination. Although such data are from selected groups of patients, they indicate that the maximal implantation rate possible will be obtained when embryos are transferred on day 5. This paper presents the evidence for single blastocyst transfer. It is concluded that this is a viable option for most patients because of the high implantation rate of fresh blastocysts, and their ability to be cryopreserved with minimal loss of viability. For the following article, the authors declare that they have no financial interest in the sale of the commercial products listed. Why do blastocysts have a higher implantation rate than pronucleate and cleavage-stage embryos? Not all oocytes or spermatozoa are destined to give rise to a viable embryo. This is not only due to chromosomal anomalies, but also cytoplasmic deficiencies and chromatin damage (Tesarik, 1994; Van Blerkom, 1994; Seli et al., 2002). Furthermore, before blastocyst formation, one is really

2 monitoring a cleaving oocyte, as the maternal embryonic genome transition is not complete (Braude et al., 1988; Taylor et al., 1997). To assess true viability (post-embryonic genome activation), the embryo must be cultured to the blastocyst stage. This point does not detract from studies on pronucleate embryo polarity (Scott and Smith, 1998), as this clearly reflects the inherent quality of the oocyte, and a good embryo cannot be made from a poor quality oocyte (see below). Clearly such data are useful in indicating which embryos have the highest potential early on (Scott and Smith, 1998), but implantation rates greater than 28% have not been reported following the transfer of pronucleate embryos. Similarly the assessment of embryos on day 3 using strict criteria has resulted in increases in implantation rates (48%) in patients aged under 34 on their first treatment cycle (Gerris et al., 1999; Van Royen et al., 1999), but again maximal implantation rates are below those that can be obtained on day 5 (70%) (Gardner et al., 2000a) for patients under 46 with 10 follicles (Figure 1). Figure 1. Maximal implantation rate following the transfer of human embryos to the uterus on either day 1 (Scott and Smith, 1998), 3 (Gerris et al., 1999) or 5 (Gardner et al., 2000a) of development. Why is this the case? The answer may stem from a lack of synchronization of the embryo with the female reproductive tract. In animal models it is evident that the premature replacement of the pronucleate or cleavage-stage embryo is not consistent with high rates of embryo implantation (Marston et al., 1977). Rather the highest rates of implantation in all mammalian species studied to date have come from the transfer of embryos to the uterus at the morula and blastocyst stages. Figure 2. Effect of a 6 h exposure to the uterine environment on the subsequent development of in-vivo developed 8-cell mouse embryos. Control (solid bars); in-vivo developed 8-cell embryos were collected, then transferred to the uteri of naturally cycling pseudo-pregnant recipients and then immediately flushed from the tract and placed in medium G2 for 24 h. The test embryos (open bars) came from the same cohort of in-vivo developed 8-cell embryos but, following transfer to the uterus, were left there for 6 h before uterine flushing and culture in G2 for 18 h. Inner cell mass (ICM) development is expressed as both cell number and as a percentage of the total cells. Blastocyst formation was equivalent in the two groups but ICM development in the treatment group was significantly impaired. **P < Significantly, even a transient exposure of a cleavage-stage embryo to the uterus of a pseudo-pregnant recipient can result in impaired inner cell mass development within the resultant blastocysts (Figure 2). One of the main reasons for this is that the oviduct and uterus provide different nutritional environments for the developing embryo (Gardner et al., 1996) (Table 1) which in turn mimic the changing requirements of the embryo. It has been well documented that nutritional stress, such as that placed on the embryo when transferred to the wrong part of the reproductive tract will cause metabolic perturbations (Gardner, 1998). Such perturbations in metabolism are associated with loss of viability (Lane and Gardner, 1998). Furthermore, the uterus to which an embryo is transferred following IVF/intracytoplasmic sperm injection (ICSI) has been exposed to supraphysiological concentrations of Table 1. Concentration of carbohydrates in the human oviduct and uterus. Pyruvate Lactate Glucose (mmol/l) (mmol/l) (mmol/l) Oviduct (midcycle) Uterus Data from Gardner et al., (1996). Lactate measured as the biologically active L-isoform. 471

3 472 gonadotrophins. This raises a question regarding the normalcy of such a uterine environment. Data from animal models have shown that uterine receptivity is significantly compromised if the recipient female has undergone ovulation induction (Van der Auwera et al., 1999; Ertzeid and Storeng, 2001). It would seem prudent to minimize the embryo s exposure to such an environment, and this can be achieved through blastocyst transfer. Another plausible reason for the high implantation rates following blastocyst transfer has been provided by the work of Fanchin and colleagues (1998), who have shown that uterine contractions are inversely related to pregnancy rates. Subsequent studies have revealed that such uterine contractions decrease significantly over time following human chorionic gonadotrophin (HCG) administration. It is therefore probable that by transferring human embryos at the blastocyst stage, there is significantly less chance of the embryo being expelled from the uterus (Fanchin et al., 2001). To summarize, it is fortuitous for human medicine that among all the mammalian species, the human cleavage-stage embryo can tolerate the uterus during development. However, this does not necessarily make the transfer of the human embryo to the uterus before compaction an optimized procedure. How can viable blastocysts be cultured? The most important issue is viability, as blastocysts can be cultured in a wide variety of conditions but the resultant viability varies enormously (Gardner and Lane, 1997; Gardner and Lane, 1998). This fact continues to cause tremendous confusion in the medical literature, which often groups papers on blastocyst culture together and yet disregards the conditions used (see below). The authors approach has been to learn from the environment to which embryos are exposed in vivo (Gardner et al., 1996), while also studying the physiology and metabolism of the embryo in culture to determine what causes intracellular stress to the embryo (Gardner and Lane, 1993a; Edwards et al., 1998; Gardner, 1998; Gardner et al., 2000b; Lane and Gardner, 2000). By being able to identify and monitor such stress, the authors have developed stage-specific culture media that substantially reduce culture-induced trauma. The development and characterization of such sequential media has been published in detail elsewhere (Gardner, 1994; Gardner and Lane, 1997; Gardner and Lane, 1999; Pool, 2002). This approach is fundamentally different to that taken by Biggers et al. (2000), who developed mouse embryo culture media using the simplex optimization procedure. This method used a computer program to generate culture media formulations based on the response of mouse embryos in culture (Lawitts and Biggers, 1991, 1993). Once a specific medium was formulated, tested and blastocyst development analysed, the software would then generate several more media formulations for use in the next series of cultures. This procedure was performed several times to generate media that supported high rates of blastocyst development of embryos derived from the oocytes of outbred mice (CF1) crossed with the spermatozoa of an F1 hybrid male. Such a medium (KSOM simplex optimized medium with elevated potassium) were subsequently modified by another laboratory to include amino acids (KSOM AA ) (Ho et al., 1995). This last phase of medium development was based on previous studies on the mouse embryo (Gardner and Lane, 1993b) and did not involve the simplex procedure. Recently this single medium formulation has been used to produce human blastocysts in culture (Biggers and Racowsky, 2002; Wiemer et al., 2002). It was observed that equivalent rates of human blastocyst development could be obtained in culture compared with the use of media P1 and CCM in sequence (Biggers and Racowsky, 2002). It is important to note that the media P1 and CCM were not designed for use together. Although five babies have been born following the transfer of blastocysts cultured in KSOM AA, the implantation and pregnancy rates were not reported by Biggers and Racowsky (2002). In a study by Wiemer and colleagues (2002) on a selected group of patients who had at least five embryos with 6 12 blastomeres on day 3 with <20% fragmentation after culture in human tubal fluid medium, it was determined that KSOM AA could support 44% blastocyst development from the pronucleate stage (62% from 6- to 12-cell embryos on day 3 with <20% fragmentation). When such blastocysts were transferred, the resultant implantation rate was 37%. These values are somewhat lower than can be attained using modern sequential media, in which blastocyst development from the pronuclear stage of 64% and implantation rates of 58% in IVF patients can be attained (Gardner et al., 2002a, 2002b). Indeed, a 37% implantation rate has been reported by many groups for day 3 transfers. As published extensively (Gardner and Lane, 1998, 1999, 2002; Gardner et al., 2000b), it is of no surprise that a single culture medium can support human blastocyst development, as several studies have used a single medium formulation throughout development. For example, although Bolton and colleagues (1991) obtained a very respectable 40% blastocyst development from human pronuclear embryos using Earle s medium supplemented with pyruvate and 10% maternal serum, the resultant implantation and pregnancy rate was only 7%. The question remains, are blastocysts cultured in a singlemedium formulation more viable than those obtained through the use of sequential culture media? For humans, this has yet to be answered in prospective, randomized clinical trials. However, there are data from two different animal models in which KSOM AA has been compared directly with sequential media (Gardner and Lane, 2002). In these studies the efficacy of KSOM AA was compared with the sequential media G1/G2 in their ability to support the development of the cow and mouse embryo. In the cow, blastocyst numbers were significantly higher when embryos were cultured in sequential media (Gardner and Lane, 2002). The mouse model also supported these findings. In both models, the blastocysts cultured in sequential media had significantly more cells than those cultured in KSOM AA ; and

4 Figure 3. Blastocyst cell number and differentiation of mouse (CF1 x CF1) embryos cultured in sequential media (solid bars) or KSOM AA (open bars). Inner cell mass (ICM) development is expressed as both cell number and as a percentage of the total cells. To ensure the two media were treated in a similar way, embryos in medium KSOM AA were transferred to fresh medium after 48 h of culture at the time when embryos cultured in medium G1 were moved to medium G2. *P < 0.05; **P < From Gardner and Lane (2002) with permission. of greatest importance, they also had significantly better development of their inner cell mass, which in turn reflected their increased viability (Table 2; Figure 3). The morphology of embryos cultured in both types of systems is shown in Figure 4, and the advanced development of embryos in sequential media can be seen clearly. Other examples of a single-medium formulation being used to support human blastocyst development comes from Huisman and co-workers (2000) who, in a retrospective analysis, compared day 2, day 3 and day 4 transfers. Overall implantation and pregnancy rates were not significantly different. However, when cavitating morula were available on day 4, an implantation rate of 41% was attained. Without cavitating morula, implantation rates on day 2, day 3 or day 4 range between 12.7% and 14.4%. In 1996 Scholtes and Zeilmaker (1996) performed a randomized trial comparing day 3 versus day 5 embryo transfer. The pregnancy rates were 40% for day 5 transfers compared with 28% for day 3. The implantation rates for day 5 were 23% compared with 14% for day 3. Both of the above studies used a mixture of Earle s and Ham s F10 media for culture, which cannot be described as sequential even though technically two media were used. Therefore care must be taken when comparing data where different culture media were used. It is evident that, although different culture conditions can support similar rates of blastocyst development, subsequent implantation rates can be dramatically different (Gardner and Lane, 1998). Figure 4. Photomicrographs of CF1 x CF1 mouse embryos cultured from the pronucleate stage for 96 h in (a) sequential media G1 and G2 or (b) KSOM AA. Note the increased expansion and hatching of blastocysts cultured in sequential media. To ensure the two media were treated in a similar way, embryos in medium KSOM AA were transferred to fresh medium after 48 h of culture at the time when embryos cultured in medium G1 were moved to medium G2. Scale bar represents 100 µm. From Gardner and Lane (2002) with permission. Stresses other than media formulation that can adversely affect embryo development and viability As indicated above, one of the keys to the successful culture of all mammalian embryos is to minimize intracellular stress. Stress can come in many forms other than inappropriate media formulations. An example of this is the release of ammonium ions into the culture medium when amino acids are present. Amino acids are among the most important regulators of mammalian embryo development (Gardner et al., 2000b) and their inclusion in embryo media formulations is a prerequisite to optimizing embryo development. However, owing to both their metabolism by the embryo and their spontaneous deamination, amino acids release ammonium ions into the culture system (Gardner and Lane, 1993b). Significantly, it is the latter that is responsible for the vast majority of ammonium ions produced, with most ammonium ions coming from glutamine, which is highly labile at 37 C. 473

5 Table 2. Efficacy of sequential media G1 and G2 versus KSOM AA in supporting CF1 CF1 mouse zygotes in culture and their subsequent viability. n 8-cell on Compaction at Blastocyst at Hatching at Implantation Fetal day 3 (%) 72 h 117 h 117 h (%) development post-hcg post-hcg post-hcg (%) (%) (%) (% of total) Sequential media KSOM AA c 7.3 c 45.5 a,c 16.8 c 50 b 36.1 From Gardner and Lane (2002). All media were supplemented with 5 mg/ml human serum albumin (HSA); all media were renewed after 48 h; number of embryos transferred = 36 per treatment. Implantation and fetal development were determined on day 15 of pregnancy, day 1 being the day of copulation plug. a Blastocyst development in KSOM AA (45.5%) is lower than that reported by Biggers et al. (2000) who obtained 82% blastocyst development when CF1 females were mated to hybrid males. In the authors laboratory when CF1 females are mated to hybrid males, embryo development is increased by around 30% over embryos derived from a CF1 CF1 mating. Using this correction factor, the data obtained in KSOM AA are very similar to that previously reported (Biggers et al., 2000). b P < 0.05; c P < 0.01, significantly lower than sequential media. Ammonium has been shown to induce embryo retardation in culture, specifically the development of the inner cell mass, and to induce retarded fetal development (Lane and Gardner, 1994). In the worse case, ammonium has also been shown to induce neural tube birth defects (Lane and Gardner, 1994; Sinawat, 2001). To alleviate any problems associated with ammonium, the culture medium should be renewed at least every 48 h (Gardner and Lane, 1993b; Lane and Gardner, 1994) or the ammonium transaminated in situ to alanine using the appropriate enzyme system (Lane and Gardner, 1995). Even if a monoculture system is used (one medium formulation throughout development), if the medium contains amino acids it should be renewed every 48 h. As glutamine is the source of most of the ammonium from deamination, it should be replaced with alanyl-glutamine (Gardner et al., 1998), which is stable at 37 C and is equally as effective as glutamine in promoting embryo development (Gardner and Lane, unpublished observations). The use of a medium containing L-glutamine for human embryo culture should be considered carefully. presence of 300 µmol/l ammonium being significantly reduced from 20% of the total cell population (control) to 9.7%. It can be seen from Figure 5 that the blastocysts cultured by Biggers and Racowsky (2002) were exposed to very high concentrations of ammonium ions during their development because the medium KSOM AA was not renewed during culture. The long-term clinical consequences of exposure to high concentrations of ammonium ions for offspring are not known. Another component of the culture system that can have a highly detrimental impact on embryo physiology, metabolism, gene expression, ultrastructure and development, is the inclusion of serum in the culture system. The negative effects of serum on embryo development have been extensively characterized (Gardner, 1994; Thompson et al., 1995; Gardner and Lane, 1998, 2002; Khosla et al., 2001). Given the published data on the adverse effects of serum it is morally 474 The significance of ammonium production must be taken very seriously. Figure 5 shows the release of ammonium ions into the medium by KSOM AA as used by Biggers and Racowsky (2002), who did not renew the medium, compared with G1/G2 (Gardner and Lane, 2002). The differences in ammonium production by the two culture systems can be attributed primarily to the 1 mmol/l glutamine in the KSOM AA, whereas the sequential media contain alanyl-glutamine, as discussed above. The negative impact of ammonium ions on mouse embryo development was first reported 10 years ago by Gardner and Lane (1993b), who later showed that, at a concentration of 150 µmol/l, ammonium ions induced exencephaly in 20% of all fetuses (Lane and Gardner, 1994). More recent work has revealed that those fetuses not exhibiting a neural tube defect after exposure to 75 µmol/l ammonium during culture were retarded in fetal growth by 24 h on day 15 of development (Lane and Gardner, 2002). Interestingly, the inner cell mass of the blastocyst is most susceptible to ammonium ions, its development in the Figure 5. Production of ammonium into the culture medium (lacking embryos) by the spontaneous breakdown of amino acids. Solid circles, KSOM AA ; open circles, G1/G2.

6 Figure 6. Impact of culture system on human blastocyst development. The x-axis represents individual patients in the order that they attended infertility treatment in program (n = 515). The y-axis represents the percentage blastocyst development attained for each patient. The arrows indicate the times when the clinic made two changes to the culture system. At the first arrow, the clinic moved from culturing embryos in 1ml of medium in a test tube to culturing embryos in 50 µl drops. At the same time the gas environment was changed from 5% CO 2 in air to 6% CO 2, 5% O 2 and 89% N 2. In the period before the changes the mean blastocyst development was 47%, whereas after the changes it increased to 56% (P < 0.01). There were no changes to the culture media during this period. This significant increase in embryo development was therefore facilitated by changes to factors in the culture system other than medium formulation. At the second arrow, the clinic introduced new media formulations (GIII series). The mean blastocyst development after the second change was made was 65%, significantly higher (P < 0.01) than in the previous media formulations. After Gardner et al. (2002b). irresponsible to continue its use in clinical assisted reproduction. Other factors that can affect embryo development are often as innocuous as the number of times an incubator door is opened. Gardner and Lane (1996) demonstrated using the mouse model (CF1xCF1) that blastocyst development and resultant cell numbers were significantly decreased when the incubator door was opened 11 times a day. This has profound implications for clinical embryo culture and enforces the requirement for sufficient incubators (Gardner and Lane, 2001). For around 700 retrievals the authors use 14 incubator chambers (in seven double stacks). The top chamber of each stack is for media equilibration, whereas the bottom chamber is used for inseminations and embryo culture, thereby minimizing the amount of access to incubators containing embryos. This means that one double stack of chambers is used for just two to three retrievals per week. Other factors that impact embryo quality include the medium ph, CO 2 concentration, O 2 concentration, incubation volume, number of embryos per unit volume, types of macromolecules used as supplement and the complete avoidance of serum in all media. These have been reviewed in detail (Gardner and Lane, 2001). Figure 6 shows the effect of making changes other than medium formulation to a culture system in clinical IVF. It is evident that a culture system in human IVF is more than the media alone, and that the laboratory is more than the culture system. Furthermore, the IVF treatment is more than the laboratory. These points must be taken into account when comparing laboratories and when trying to emulate other laboratory s results. Figure 7 highlights the complex interactions that exist within clinical IVF procedures. The manifestations of stress on the embryo in culture are many-fold. The most obvious one is retarded cleavage and, in the extreme, developmental arrest. However, there are others not visible to the naked eye, which have a profound effect on the subsequent viability of the embryo and fetal development. These include perturbed metabolism and gene expression, both of which could have long-term consequences for the conceptus. With regard to embryo metabolism, it has been know for over 30 years that inappropriate culture conditions can lead to aberrant metabolic activity (Menke and McLaren, 1970). By studying the aetilogy of such culture-induced changes in metabolism, it has been possible to develop culture media that greatly reduce metabolic stress on the embryo (Gardner and Lane, 1993a; Gardner, 1998; Lane and Gardner, 1998; Gardner et al., 2000b), thereby maintaining the viability of the embryo (Lane and Gardner, 1998). 475

7 Figure 7. The relationship between patient stimulation, the laboratory and transfer outcome in human IVF. This figure serves to illustrate the complex and interdependent nature of human IVF treatment. For example, the stimulation regimen not only impacts on oocyte quality and hence embryo physiology and viability (Hardy et al., 1995) but can also affect subsequent endometrial receptivity (Simon et al., 1998; Van der Auwera et al., 1999; Ertzeid and Storeng, 2001). Furthermore, the health and dietary status of the patient can have a profound effect on the subsequent developmental capacity of the oocyte and embryo (Kwong et al., 2000). The dietary status of patients attending IVF is typically not considered as a compounding variable, but growing data would indicate otherwise. In the schematic, the laboratory has been broken down into its core components, only one of which is the culture system. The culture system has in turn been broken down to its components, only one of which is the culture media. Therefore, it would appear rather simplistic to assume that by changing only one part of the culture system (i.e. culture media), that one is going to mimic the results of a given laboratory or clinic. One of the biggest impacts on the success of a laboratory and culture system is the level of quality control and quality assurance in place. For example, one should never assume that anything coming into the laboratory that has not been pre-tested with a relevant bioassay (e.g. mouse embryo assay), is safe merely because a previous lot has performed satisfactorily. Only a small percentage of the contact supplies and tissue culture ware used in IVF comes suitably tested. Therefore it is essential to assume that everything entering the IVF laboratory without a suitable pre-test is embryotoxic until proven otherwise. In the authors programme, the 1-cell mouse embryo assay (MEA) is used to pre-screen every lot of tissue culture ware that enters the programme, i.e. plastics that are approved for tissue culture. Around 25% of all such material fails the 1-cell MEA (in a simple medium lacking protein after the first 24 h (Gardner and Lane, 1999)). If quality control to this level is not performed, one in four of all contact supplies used clinically will be inhibitory to embryo development. In reality many programmes cannot allocate the resources required for this level of quality control, and when embryo quality is compromised in the laboratory, it is the media that are held responsible, when in fact the laboratory ware or oil are often the culprits. 476 Furthermore, it has been determined that the embryo precompaction is more sensitive to culture-induced stress than an embryo that has formed a transporting epithelium at compaction (Edwards et al., 1998). With regards to gene expression, there are data in animal models that show that embryo culture in suboptimal conditions is associated with altered gene expression (Ho et al., 1995) and in some cases a loss of imprinting (Doherty et al., 2000). Most studies working on embryo culture and gene expression have shown that the addition of serum can induce major genetic problems in the embryo (Khosla et al., 2001; Wrenzycki et al., 2001). However, the work of Schultz and colleagues has shown that suboptimal culture medium formulations can also induce changes in global gene expression and imprinting. Using Whitten s medium (a simple salt solution lacking amino acids), Ho and colleagues (1995) showed that the relative abundance of specific messenger RNAs was significantly decreased in mouse 8-cell embryos and blastocysts derived from 2-cell embryos in culture. The key observation here is that there were significant changes in gene expression evident by the 8-cell stage, within 24 h of culture. Subsequently, Doherty and co-workers (2000) showed that the culture of 2-cell mouse embryos to the blastocyst stage in Whitten s medium was associated with a loss of imprinting for the gene H19. The question is, therefore, when does this loss of parental imprinting occur during in-vitro culture? Work from Shi and Haaf (2002) indicates that, in the mouse, aberrant methylation patterns induced by either ovulation induction or different invitro culture conditions are evident at the 2-cell stage. Therefore, although altered gene expression and altered imprinting is evident in cultured mouse blastocysts, the time at

8 which the changes were induced appears to reside during the first few cleavage divisions. In support of this hypothesis, the relative levels of gene expression in mouse embryos cultured from the zygote and from the 8-cell stage to the blastocyst in Whitten s medium have been determined (Hewitt et al., 2003). Interestingly, when the embryo is cultured from the zygote stage to the blastocyst, the levels of gene expression are significantly altered, consistent with previously published data (Ho et al., 1995). However, when embryos are cultured from the 8-cell stage to the blastocyst, resultant gene expression appears normal, indicating that the embryo is at greatest sensitivity to its environment during the first three cleavage stages. Clearly inappropriate conditions can adversely affect metabolic activity and gene expression in the embryo. It is evident that the embryo is most sensitive to trauma before compaction and that problems detected in the blastocyst can actually be a result of trauma during the first three cell cycles and not with blastocyst culture itself. The work of Shi and Haaf (2002) also indicate that different ovulation induction regimens can be the cause of subsequent methylation and imprinting abnormalities. Therefore, one must be aware of the multiple possible aetiology of problems for the embryo, some of which reside within the oocyte. This supports the authors basic premise that you cannot make a good embryo from a bad oocyte. In-vitro rates of embryo development comparable with those in vivo Historically, embryos cultured in vitro lag behind their in-vivo developed counterpart (Bowman and McLaren, 1970; Harlow and Quinn, 1982). However, with the development of sequential media based on the premise of meeting the changing requirements of the embryo and minimizing trauma, in-vivo rates can now be attained in vitro in the mouse (Gardner and Lane, 2002; Reed et al., 2003). This is a significant development as there is now a culture system capable of producing blastocysts at the same time and with the same cell number and allocation to the inner cell mass as embryos developed in the female tract. This helps to explain why the implantation rates of human blastocysts developed in the laboratory (Gardner et al., 2000a, 2002a) approximate those blastocysts developed in vivo (Buster et al., 1985). Need to set benchmarks to compare laboratories It is difficult to compare the clinical IVF results of different laboratories because the results are also dependent upon stimulation regimens, transfer technique, luteal support and of course the patient population themselves (Figure 7). Unlike animal models, for which there are extensive in-vivo embryo developmental data to use as the gold standard, there is a paucity of data on the in-vivo developed human embryo. In a study on blastocyst transfer from donor patients to recipient females, in which the donor blastocysts were developed in vivo and then collected by uterine lavage, Buster and colleagues (1985) demonstrated that in-vivo developed blastocysts gave rise to an implantation rate of 60% when transferred. Despite the scarcity of in-vivo data, there are groups of patients that can be used to benchmark the performance of an IVF clinic. It is therefore proposed that patient groups such as oocyte donors or patients under 35 years of age be used to compare the data of different clinics. When blastocyst culture and transfer has been applied to an oocyte donation programme, implantation rates of up to 66% can be obtained (Schoolcraft and Gardner, 2000), quite consistent with those reported by Buster and colleagues (1985). Several groups have reported that blastocyst implantation rates decrease in patients over 40 years of age (Marek et al., 1999; Schoolcraft et al., 1999); considering patients in specific age groups will therefore help in comparisons. For example, in the study by Coskun and colleagues (2000), it was reported that in patients of 30 years or less, or in patients with five or fewer good embryos on day 3, the resultant blastocyst implantation rates were less than 30%. Coskun et al. concluded that blastocyst transfer was no more effective than embryo transfer on day 3. However, the data indicate that other factors in the laboratory or clinic had a negative impact on results as several laboratories have reported much higher implantation rates for blastocysts derived from such patient groups with good prognosis (Marek et al., 1999; Milki et al., 2000). In a more recent study using a mixture of Ham s F-10 and Earle s media to make a single-culture medium (Macklon et al., 2002), it was observed that sequential media did not perform any better than this monoculture medium. Significantly though, implantation rates in either the monoculture or sequential media were only 20%. These data also indicate that there were other issues in the laboratory or clinic that were compromised resulting in such low implantation rates following blastocyst transfer. These examples support the previous proposition that blastocyst transfer is not a panacea for all ills. Before extended culture should be considered, all aspects of clinical and laboratory procedures need to be optimized. Should problems exist either in patient stimulation protocols or within the laboratory, extended culture may only exacerbate the situation. To ensure successful outcomes for blastocyst transfer, extended culture should first be tried on oocyte donors, patients who respond well to gonadotrophins or patients under 35. Blastocyst formation rates of around 50% with implantation rates of 40% or greater should be readily obtained in such patients. Unfortunately, reports such as those of Alper and colleagues (2001) have added to the confusion about the merits of blastocyst transfer by incorrectly stating that in patients under 35 years of age, blastocyst transfer does not affect implantation and pregnancy rates, as reported by Marekand co-workers (1999). The fact is, this is the very group of patients in which blastocyst transfer has a profound beneficial effect (Marek et al., 1999). 477

9 Table 3. Outcome of prospective randomized trials on embryo transfer at the cleavage and blastocysts stages when sequential media have been used for embryo culture. Study Patient Day 3 Day 5 P-value population Mean no. Implantation Mean no. Implantation embryos rate (% embryos rate (% transferred fetal sacs) transferred fetal sacs) Gardner et al. (1998) >10 follicles of <0.01 >12 mm on day of HCG Coskun et al. (2000) 4 2PN NS Karaki et al. (2002) 5 2PN <0.01 Levron et al. (2002) <38 years old <0.01 and >5 2PN Utsunomiya All NS et al. (2002) Rienzi et al. (2002) <38 years old NS and 8 2PN by ICSI Van der Auwera All (day <0.05 et al. (2002) transfers) Frattarelli et al. (2003) <35 years old, <0.05 no previous IVF and 10 follicles of 14 mm on day of HCG HCG = human chorionic gonadotrophin; 2PN = two-pronucleate; ICSI = intracytoplasmic sperm injection. NS = not significant. 478 Thus the literature is apparently unclear regarding the merits of blastocyst culture and transfer in human assisted reproduction. In Table 3 the outcomes of eight prospective randomized trials on blastocyst transfer following the use of sequential media are listed. From this table it can be seen that one trial concluded that day 3 transfer was superior to day 5, three trials concluded that the outcome was the same, whether day 3 or day 5 transfer was used, and finally four trials concluded that day 5 was superior to day 3 transfers. From the available data from prospective randomized studies using the same culture system (at least in terms of using sequential media), it therefore appears that the balance is more in favour of day 5 transfers. Furthermore, there are numerous retrospective studies that have concluded that day 5 transfer exhibits significant benefits for human assisted reproduction (Cruz et al., 1999; Marek et al., 1999; Schoolcraft and Gardner, 2000; Balaban et al., 2001; Langley et al., 2001; Trout et al., 2001; Damario et al., 2002; Shapiro et al., 2002; Wilson et al., 2002). These papers represent both non-selected and specific patient populations. Significance of cryopreservation An initial premise made in this paper was that in order to successfully introduce any new culture system one had to be able to cryopreserve the supernumerary embryos successfully. The introduction of blastocyst culture was met with much speculation as not all laboratories were able to cryopreserve blastocysts that were not transferred (Alper et al., 2001). However, with the development of more suitable cryopreservation procedures it is now possible to obtain implantation and ongoing pregnancy rates of 30 and 60% respectively using frozen thawed blastocysts (Spandorfer et al., 2002; Gardner et al., 2003). Overall efficacy of an IVF treatment cycle To bring this discussion to completion, one must consider the overall treatment cycle, i.e. fresh and cryopreserved embryos. To calculate the overall efficacy of an IVF treatment, the following model was developed: (mean number of embryos transferred implantation rate) + (mean number of embryos cryopreserved implantation rate) (1 the no-transfer rate). This model factors in both fresh and frozen embryo transfers and corrects for the number of retrievals that did not result in an embryo transfer. Using such a model for a group of non-selected patients (n = 513) it was calculated that blastocyst transfer was 19% more efficient than day 3 transfers (n = 463) (Schoolcraft and Gardner, 2001). However, in a more recent analysis on 50 patients having day 5 transfer and 48 patients having day 3 transfer, Rienzi and colleagues (2002) determined that the cumulative pregnancy rate per oocyte retrieval was higher when embryos were transferred and cryopreserved on day 3. Unfortunately this model only included eight frozen

10 blastocysts transfers from the possible 18 patients who had blastocyst freezing (44%). This figure is in contrast to the 38 out of 42 patients (90%) included who had embryos frozen on day 3. Had 90% of the patients who had blastocyst cryopreservation returned for an embryo transfer (similar to the day 3 group), the model would predict no difference between the groups. Furthermore, this study used a methodology for blastocyst cryopreservation that has been superseded (Langley et al., 2001; Gardner et al., 2003). Had a more appropriate blastocyst cryopreservation procedure been used, along with an equal percentage of the two groups included in the analysis, then blastocyst transfer could have superseded the day 3 data. Analysis of embryo viability and the move to single blastocyst transfers It has been shown that blastocyst score has an impact on resultant implantation rates (Gardner et al., 2000a). Using the alphanumeric scoring system of Gardner and Schoolcraft (1999), it is possible to identify blastocysts whose implantation potential is 70%. In such cases the transfer of a single blastocyst should certainly be promoted. Added to this is the possibility of non-invasively assessing the metabolism of individual embryos before transfer, leading to further selection (Gardner and Leese, 1999; Gardner et al., 2001; Van den Bergh et al., 2001; Houghton et al., 2002). The work of Sandalinas and colleagues (2001) has confirmed that some chromosomally abnormal human embryos can reach the blastocyst stage in vitro. To augment blastocyst culture and transfer for patients at specific risk of chromosomal abnormalities, i.e. those over 37 years of age, preimplantation genetic diagnosis should be considered in conjunction with extended culture. Conclusions The safe delivery of a singleton should be considered a successful outcome of infertility treatment. There is no question that this can be attained by the transfer of an individual human embryo at any stage of its development. However, the resultant pregnancy and delivery rates will reflect the implantation rates and so, consistent with Figure 1, the highest rates of pregnancies will be attained with blastocyst transfer. Significantly, in a prospective randomized trial on single blastocyst transfer it has been demonstrated that an implantation rate and subsequent pregnancy rate of 55% can be established in patients who respond well to gonadotrophins (Surrey et al., 2002). With improved methods for blastocyst cryopreservation (Gardner et al. 2001b), together with the data that blastocyst culture and transfer can be applied to non-selected patient populations (Marek et al., 1999; Wilson et al., 2002), there is a definite ability to move to single blastocyst transfers for most patients. Finally, Sutcliffe (2002) has indicated that very little, if any, animal data are available on some of the newer techniques in human assisted reproduction, such as intracytoplasmic sperm injection and preimplantation genetic diagnosis. Leese and Whittall (2001) also raised this criticism of blastocyst transfer. However, unlike the technologies questioned by Sutcliffe (2002), transfers of over 5000 animal embryos were performed by the current authors laboratory alone, before the clinical introduction of blastocyst transfer. Transfers included mice and sheep, and examined both fetal development and full-term normality and fecundity of the offspring. To challenge the proposition of Leese and Whittall (2001), the introduction of blastocyst culture in humans has had among the most extensive pre-clinical work of all assisted reproduction procedures. Acknowledgements The authors gratefully acknowledge the assistance of the embryology teams at the Colorado Center for Reproductive Medicine. Specifically the assistance of Dr Levent Karagenc with studies on transient exposure of embryos to the uterus is appreciated. The authors would like to thank Dr William Schoolcraft for his continued support. References Alper MM, Brinsden P, Fischer R, Wikland M 2001 To blastocyst or not to blastocyst? That is the question. Human Reproduction 16, Balaban B, Urman B, Alatas C et al Blastocyst-stage transfer of poor-quality cleavage-stage embryos results in higher implantation rates. Fertility and Sterility 75, Biggers JD, Racowsky C 2002 The development of fertilized human ova to the blastocyst stage in KSOM AA medium: is a two-step protocol necessary? Reproductive BioMedicine Online 5, Biggers JD, McGinnis LK, Raffin M 2000 Amino acids and preimplantation development of the mouse in protein-free potassium simplex optimized medium. Biology of Reproduction 63, Bolton VN, Wren ME, Parsons JH 1991 Pregnancies after in vitro fertilization and transfer of human blastocysts. Fertility and Sterility 55, Bowman P, McLaren A 1970 Cleavage rate of mouse embryos in vivo and in vitro. Journal of Embryology and Experimental Morphology 24, Braude P, Bolton V, Moore S 1988 Human gene expression first occurs between the four- and eight-cell stages of preimplantation development. Nature 332, Buster JE, Bustillo M, Rodi IA et al Biologic and morphologic development of donated human ova recovered by nonsurgical uterine lavage. American Journal of Obstetric Gynecology 153, Coskun S, Hollanders J, Al-Hassan S et al Day 5 versus day 3 embryo transfer: a controlled randomized trial. Human Reproduction 15, Cruz JR, Dubey AK, Patel J et al Is blastocyst transfer useful as an alternative treatment for patients with multiple in vitro fertilization failures? Fertility and Sterility 72, Damario MA, Phy JL, Tummon IS 2002 Successful elective single blastocyst transfer in a patient with prior repetitive high-order multiple gestations. Journal of Assisted Reproduction and Genetics 19, Doherty AS, Mann MR, Tremblay KD et al Differential effects of culture on imprinted H19 expression in the preimplantation mouse embryo. Biology of Reproduction 62, Edwards LJ, Williams DA, Gardner DK 1998 Intracellular ph of the mouse preimplantation embryo: amino acids act as buffers of intracellular ph. Human Reproduction 13, Ertzeid G, Storeng R 2001 The impact of ovarian stimulation on implantation and fetal development in mice. Human 479

11 480 Reproduction 16, Fanchin R, Righini C, Olivennes F et al Uterine contractions at the time of embryo transfer alter pregnancy rates after in-vitro fertilization. Human Reproduction 13, Fanchin R, Ayoubi JM, Righini C et al Uterine contractility decreases at the time of blastocyst transfers. Human Reproduction 16, Frattarelli JL, Leondires MP, McKeeby JL et al Blastocyst transfer decreases multiple pregnancy rates in in vitro fertilization cycles: a randomized controlled trial. Fertility and Sterility 79, Gardner DK 1994 Culture of mammalian embryos in the absence of serum and somatic cells. Cell Biology International 18, Gardner DK 1998 Changes in requirements and utilization of nutrients during mammalian preimplantation embryo development and their significance in embryo culture. Theriogenology 49, Gardner DK, Lane M 1993a The 2-cell block in CF1 mouse embryos is associated with an increase in glycolysis and a decrease in tricarboxylic acid (TCA) cycle activity: alleviation of the 2-cell block is associated with the restoration of in vivo metabolic pathway activities. Biology of Reproduction 48 (suppl. 1), 152. Gardner DK, Lane M 1993b Amino acids and ammonium regulate mouse embryo development in culture. Biology of Reproduction 48, Gardner DK, Lane M 1996 Alleviation of the 2-cell block and development to the blastocyst of CF1 mouse embryos: role of amino acids, EDTA and physical parameters. Human Reproduction 11, Gardner DK, Lane M 1997 Culture and selection of viable blastocysts: a feasible proposition for human IVF? Human Reproduction Update 3, Gardner DK, Lane M 1998 Culture of viable human blastocysts in defined sequential serum-free media. Human Reproduction 13 (suppl. 3), Gardner DK, Lane M 1999 Embryo culture systems. In: Trounson AO, Gardner DK (eds) Handbook of in vitro fertilization, 2nd edn, CRC Press, Boca Raton, FL, USA, pp Gardner DK, Lane M 2001 Embryo culture. In: Gardner DK, Weissman A, Howles C, Shoham Z (eds) Textbook of assisted reproductive techniques, Martin Dunitz, London, UK, pp Gardner DK, Lane M 2002 Development of viable mammalian embryos in vitro: evolution of sequential media. In: Cibelli J, Lanza RP, Campbell KHS, West MD (eds) Principles of cloning, Academic Press, New York, USA, Gardner DK, Leese HJ 1999 Assessment of embryo metabolism and viability. In: Trounson AO, Gardner DK (eds) Handbook of in vitro fertilization, 2nd edn, CRC Press, Boca Raton, FL, USA, pp Gardner DK, Schoolcraft WB 1999 In-vitro culture of human blastocysts. In: Jansen R, Mortimer D (eds) Towards reproductive certainty: fertility and genetics beyond 1999, Parthenon Press, Carnforth, Lancs, UK, pp Gardner DK, Lane M, Calderon I, Leeton J 1996 Environment of the preimplantation human embryo in vivo: metabolite analysis of oviduct and uterine fluids and metabolism of cumulus cells. Fertility and Sterility 65, Gardner DK, Schoolcraft WB, Wagley L et al A prospective randomized trial of blastocyst culture and transfer in in-vitro fertilization. Human Reproduction 13, Gardner DK, Lane M, Stevens J et al. 2000a Blastocyst score affects implantation and pregnancy outcome: towards a single blastocyst transfer. Fertility and Sterility 73, Gardner DK, Pool TB, Lane M 2000b Embryo nutrition and energy metabolism and its relationship to embryo growth, differentiation, and viability. Seminars in Reproductive Medicine 18, Gardner DK, Lane M, Stevens J, Schoolcraft WB 2001 Noninvasive assessment of human embryo nutrient consumption as a measure of developmental potential. Fertility and Sterility 76, Gardner DK, Lane M, Schoolcraft WB 2002a Physiology and culture of the human blastocyst. Journal of Reproductive Immunology 55, Gardner DK, Lane M, Stevens J, Schoolcraft WB 2002b Ongoing development of a human blastocyst culture system. Fertility and Sterility 78, S8. Gardner DK, Lane M, Stevens J, Schoolcraft WB 2003 Changing the start temperature and cooling rate in a slow freezing protocol increases human blastocyst viability. Fertility and Sterility 79, ). Gerris J, De Neubourg D, Mangelschots K et al Prevention of twin pregnancy after in-vitro fertilization or intracytoplasmic sperm injection based on strict embryo criteria: a prospective randomized clinical trial. Human Reproduction 14, Hardy K, Robinson FM, Paraschos T et al Normal development and metabolic activity of preimplantation embryos in vitro from patients with polycystic ovaries. Human Reproduction 10, Harlow GM, Quinn P 1982 Development of preimplantation mouse embryos in vivo and in vitro. Australian Journal of Biological Science 35, Hewitt E, Lane M, Gardner DK 2003 Culture effects on mouse embryo gene expression are limited to the first three cleavage divisions. Theriogenology 59, 420. Ho Y, Wigglesworth K, Eppig JJ, Schultz RM 1995 Preimplantation development of mouse embryos in KSOM: augmentation by amino acids and analysis of gene expression. Molecular Reproduction and Development 41, Houghton FD, Hawkhead JA, Humpherson PG et al., 2002 Noninvasive amino acid turnover predicts human embryo developmental capacity. Human Reproduction 17, Huisman GJ, Fauser BC, Eijkemans MJ, Pieters MH 2000 Implantation rates after in vitro fertilization and transfer of a maximum of two embryos that have undergone three to five days of culture. Fertility and Sterility 73, Karaki RZ, Samarraie SS, Younis NA et al Blastocyst culture and transfer: a step toward improved in vitro fertilization outcome. Fertility and Sterility 77, Khosla S, Dean W, Brown D et al Culture of preimplantation mouse embryos affects fetal development and the expression of imprinted genes. Biology of Reproduction 64, Kwong WY, Wild AE, Roberts P et al Maternal undernutrition during the preimplantation period of rat development causes blastocysts abnormalities and programming of postnatal hypertension. Development 127, Lane M, Gardner DK 1994 Culture of preimplantation mouse embryos in the presence of amino acids increases post implantation development whilst the concomitant production of ammonium induces birth defects. Journal of Reproduction and Fertility 102, Lane M, Gardner DK 1995 Removal of embryo-toxic ammonium from the culture medium by in situ enzymatic conversion to glutamate. Journal of Experimental Zoology 271, Lane M, Gardner DK 1998 Amino acids and vitamins prevent culture-induced metabolic perturbations and associated loss of viability of mouse blastocysts. Human Reproduction 13, Lane M, Gardner DK 2000 Regulation of ionic homeostasis by mammalian embryos. Seminars in Reproductive Medicine 18, Lane M, Gardner DK 2002 Ammonium affects ICM development, metabolism, intracellular ph, and fetal growth rates. Biology of Reproduction 66, 17. Langley MT, Marek DM, Gardner DK et al Extended embryo culture in human assisted reproduction treatments. Human Reproduction 16, Lawitts JA, Biggers JD 1991 Optimization of mouse embryo culture media using simplex methods. Journal of Reproduction and Fertility 91,