Review Chromosome abnormalities and their relationship to morphology and development of human embryos

Transcription

1 RBMOnline - Vol 12. No Reproductive BioMedicine Online; on web 14 December 2005 Review Chromosome abnormalities and their relationship to morphology and development of human embryos Santiago Munné has been director of PGD at Reprogenetics since This company, which he founded, offers PGD services to over 150 IVF centers in the US, and also has labs in Spain and Japan. Originally from Barcelona, Spain, Dr Munné gained his PhD in genetics from the University of Pittsburgh and joined Dr Jacques Cohen at Cornell University Medical College, New York in There he developed the first PGD test to detect embryonic numerical chromosome abnormalities. His work has been recognized by several prizes: in 1994, 1995, 1998 and 2005 from the Society for Assisted Reproductive Technology, and in 1996 from the American Society for Reproductive Medicine. Recently the PGD team has shown higher pregnancy rates and lower spontaneous rates in women of advanced age undergoing PGD. This team has performed more than 700 PGD cycles for translocations and over 5500 PGD cycles for chromosome abnormalities related to advanced maternal age or with recurrent Dr Santiago Munné pregnancy loss. Dr Munné has more than 150 publications to his name, and is a frequent lecturer, both nationally and internationally, on his team s work and the field of preimplantation genetics. He was recently blessed with his first daughter, Mar. Santiago Munné Reprogenetics LLC, West Orange, NJ 07052, USA Correspondence: munne@embryos.net Abstract This review covers the relationship between chromosome abnormalities, morphological abnormalities and embryonic development. The baseline of chromosome abnormalities in human embryos produced by assisted reproduction is higher than 50%, regardless of maternal age. While aneuploidy increases with maternal age, abnormalities arising post-meiotically, such as mosaicism, chaoticism, polyploidy and haploidy, have similar incidence in all age groups (about 33%). Post-meiotic abnormalities do increase with dysmorphism. The most common dysmorphisms found in cleavage-stage embryos are multinucleation, fragmentation and uneven cells, among others. All dysmorphisms are associated with an increase in postmeiotic chromosome abnormalities and a decreased implantation potential. Similarly, embryos developing slowly or with arrested development have higher incidence of post-meiotic abnormalities than normally developing ones. Chromosome studies in blastocysts indicate that mosaicism is the most common abnormality but that the load of abnormal cells decreases with increasing blastocyst quality. Regardless of blastocyst quality, more than 40% of mosaics are still chromosomally abnormal and will not implant or will spontaneously abort. Because aneuploidy is not related to cleavage stage dysmorphism and trisomies can reach blastocyst stage and beyond, morphological analysis is not enough to select against chromosome abnormalities, and thus preimplantation genetic diagnosis should be recommended in patients 35 and older. Keywords: aneuploidy, chaotic embryos, chromosome abnormalities, dysmorphisms, mosaicism, preimplantation genetic diagnosis 234 Introduction Preliminary studies using conventional karyotyping showed great disparity of chromosome abnormality rates, probably because of the few cells analysed per embryo and the small sample size (reviewed by Pellestor, 1995; Almeida and Bolton, 1996). Comparing the data obtained from these studies up to, 1995 (Pellestor, 1995), in which most embryos had a single cell analysed, with the study of Almeida and Bolton (1996) in which at least two cells were analysed, shows that a large proportion of embryos classified as aneuploid by earlier studies were in fact mosaic embryos. Nevertheless, the karyotype literature indicates that morphologically abnormal monospermic embryos (dysmorphic embryos) have a higher rate of chromosome abnormalities than morphologically normal embryos (reviewed by Pellestor, 1995; Almeida and Bolton, 1996). However, chromosomally normal and abnormal human embryos cannot easily be distinguished morphologically (Zenzes and Casper, 1992). This is probably because chromosome abnormalities detected at the 2- to 8-cell stage cannot induce dysmorphisms, since embryonic translation has not yet commenced (Braude et al., 1988; Tesarik et al., 1988). Unlike other methods, large numbers of embryos with various morphological abnormalities have been studied using fluorescence in-situ hybridization (FISH). Thus the following

2 sections will include data mostly obtained by FISH and will cover only those embryos developing from di-pronucleated zygotes (2PN), so zygote abnormalities will not be covered. FISH as the technique of choice Several requirements must be met to study numerical chromosome abnormalities in preimplantation human embryos. Firstly, interphase analysis is necessary to evaluate cleavage arrested embryos because their cells do not divide and consequently cannot be found at metaphase stage. Secondly, analysis of individual chromosomes is important to distinguish differences in aneuploidy rates among different chromosomes. Thirdly, all blastomeres of non-replaced embryos should be analysed to distinguish mosaicism from other abnormalities. Although classical cytogenetic techniques have been applied in the study of preimplantation embryos, classical karyotyping has disadvantages: (i) it requires metaphase stage cells. However, only 24 36% of the embryos produce metaphases of sufficient quality for accurate quantification of chromosomes (Jamieson et al., 1994; Santaló et al., 1995), and in fewer than 25% of embryos with results can all the cells be analysed (Jamieson et al., 1994; Pellestor et al., 1994). This implies that mosaicism is underestimated and that arrested embryos (20% of all embryos) cannot be analysed, since they do not produce metaphases; (ii) optimal banding of blastomere metaphase chromosomes is difficult to obtain so that individual chromosomes are hard to identify; (iii) cells must remain alive after biopsy otherwise only interphase nuclei are obtained; and (iv) the typical loss of chromosomes caused by the fixation of metaphase nuclei and experienced in most karyotyping studies does not allow the differentiation of aneuploidy from mosaicism, much less to determine mechanisms of mosaicism. These deficiencies make karyotyping unusable for PGD in a clinical setting. FISH can be used with efficiencies of over 90% per cell to study the chromosome constitution of cleavage stage human embryos, including interphase cells (Griffin et al., 1991; Munné et al., 1993a; Harper et al., 1994a). Simultaneous analysis of more than three chromosome pairs permits differentiation of polyploidy, haploidy and aneuploidy in arrested human embryos (Munné et al., 1994b) (Figure 1a g). When most or all cells of an embryo are analysed, mosaicism can be differentiated from FISH or fixation failure, as well from aneuploidy, and their mechanism of formation ascertained (Munné et al., 1994c). However, FISH has disadvantages; FISH supplies information only on the chromosomes for which specific probes are used, and only five fluorochromes of the visible spectrum can be used at the same time. Thus, sequential FISH is needed, and even then only a maximum of three rounds of FISH can be used reliably, reducing the potential number of chromosomes studied to 15. Nevertheless, studies using comparative genome hybridization (CGH) (Voullaire et al., 2002) indicate that the 8-chromosome probe FISH test currently used in PGD (Munné et al., 1998) does detect about 85% of chromosome abnormalities detected by CGH, both because these eight chromosomes are the most involved in aneuploidy, and also the other chromosome aneuploidies tend to occur simultaneously with those involving the eight chromosomes (Voullaire et al., 2002; Abdelhadi et al., 2003). Three other techniques have been proposed, but FISH is still the most suitable for diagnosis of chromosome abnormalities in preimplantation embryos. The first is called primed insitu labelling (PRINS), which entails the use of specific primers for each chromosome that bind to a fixed nucleus, and their amplification with Taq polymerase and labelled nucleotides (Gosden et al., 1991; Speel et al., 1995). Up to four chromosomes per blastomere can be detected in about 2 to 3 h, and amplification is found in % of them (see review by Pellestor, 1996). The limitations are that only one colour can be used per reaction, so multiple reactions on the same nucleus have to be performed, each one less efficient that the previous one, for a maximum of four chromosomes analysed in total. The second technique, CGH, is based on amplifying all the DNA of a cell, labelling this DNA with a fluorochrome, and mixing this DNA in a 1:1 ratio with a reference DNA labelled in another colour. The probe is hybridized to normal metaphases. If the test DNA has missing or extra material, the ratio of fluorescence will shift away from 1:1 (1:2 for monosomies and 3:2 for trisomies) due to hybridized DNA (Kallioniemi et al., 1992). CGH can provide information from interphase cells, and produce a full karyotype with information on numerical and structural abnormalities on single cells (Voullaire et al., 1999, 2000; Wells and Delhanty, 2000; Wilton et al., 2001, 2003; Wells et al., 2002); but the utility of this technique within the clinical setting is limited, mainly owing to its time requirements. Firstly, it requires about 1 h per cell of analysis; this explains why large numbers of embryos have not yet been analysed with CGH. Secondly, the analysis takes 3 5 days altogether, thus allowing only either polar body analysis (Wells et al., 2002), or requiring cryopreservation of embryos while these are being analysed (Wilton et al., 2003). Cryopreservation of biopsied embryos can result in an unacceptable loss of implantation potential. The third technique is called quantitative fluorescent polymerase chain reaction (Sherlock et al., 1998; Cirigliano et al., 2002) and has the potential to analyse multiple chromosome types, but so far has not been able to work at the single cell level except for very few chromosomes. Of these three techniques, only CGH has been sparingly used in a clinical setting (Wells et al., 2002; Wilton et al., 2003), thus FISH remains the most suitable method for preimplantation genetic analysis of human embryos in a clinical setting. Morphological abnormalities in cleavage stage The following sections will review different dysmorphisms and the association of each with chromosome abnormalities (Table 1) updating previous reviews by us (Munné et al., 2004a). Fragmentation Clinical outcome Alikani et al. (2000) reported that embryos with >15% fragmentation formed morphologically normal blastocysts with lower frequency than embryos with <15% fragmentation (16.5 versus 33.3%, P < 0.001). Using the day 3 implantation rate for 235

3 a b c d e f g 236 Figure 1. Single nuclei analysed by fluorescence in-situ hybridization using probes for the first hybridization of the two usually used for testing of 9 10 chromosomes. In this case, the probes tested were for chromosome 13 (red), 16 (aqua), 18 (turquoise), 21 (green), and 22 (yellow). (a) Normal nucleus; (b) nucleus with a trisomy 22; (c) monosomy 16; (d) haploid; (e) polyploid;(f) trisomy 21; and (g) two nuclei from a mosaic embryo, one being monosomic for 22 and the other trisomic for 22. Original magnification 1000.

4 Table 1. Summary of morphological abnormalities and their relationship to chromosomal abnormalities. Embryo morphology FISH analysis Reference Normal morphology years old 16% abnormal years old 37% abnormal years old 53% abnormal 1 Dysmorphic 2PN embryos Uneven 2PN 73 87% abnormal 2, 3 Abnormal PB distribution 71 81% abnormal 3, 4 Giant embryos (>220 mm) Triploid 5, 6 Dominant single blastomere Polyploid 5 >35% fragments 70 90% abnormal 8, 9 Multinucleated embryos % abnormal 7, 8 Asymmetric blastomeres 67% 9 References: 1 = Munné et al. (1995), 2 = Sadowy et al. (1998), 3 = Gamiz et al. (2003), 4 = Coskun et al. (2003), 5 = Munné et al., 1994c, 6 = Balakier et al. (2002), 7 = Kligman et al. (1996), 8 = Magli et al. (2001), 9 = Munné et al. (2004a). FISH = fluorescence in-situ hybridization; PB = polar body. embryos with >15% fragmentation (18%) and considering the low frequency with which the same embryos formed apparently normal blastocysts on day 5, they proposed that the implantation rate of embryos with moderate fragmentation would be lower if replaced on day 5 (8.2%). By comparison, non-dysmorphic and normally developing embryos had much higher implantation rates (40.1 and 49% respectively). Other investigators have also reported low implantation rates, around 5%, for day 3 fragmented embryos transferred on day 5, although they did not use sequential media for blastocyst culture (Racowsky et al., 2000; Terriou et al., 2001). In another important study, Racowsky et al. (2003) studied cycles with replacement of embryos with either homogeneous morphology, or for which the outcome of each embryo was known (0 or 100% implantation). In those cycles, when replacement was on day 3, they found that embryos with <10% fragmentation produced the highest rate of viable offspring at 23% compared with 11% with 10 25% fragmentation, and 0.8% with >25% fragmentation. However, if the embryo had 8 cells by day 3 and reached expanding/expanded blastocyst on day 5, fragmentation did not have an effect on implantation. In addition, different patterns of fragmentation have been observed (Alikani et al., 1999; Antczak and Van Blerkom, 1999). One of the fragment types (type IV), which is characterized by having large fragments, produces notably lower pregnancy rates than other types of fragments, as well as lower blastocyst formation rates (Alikani et al., 2000). Figure 2 shows two fragmented embryos with different patterns. Chromosome abnormalities Fragmentation rate has been associated with chromosome abnormalities (Plachot et al., 1987; Pellestor and Sele, 1988; Munné and Cohen, 1998; Magli et al., 2001; Munné et al., 2001). Chromosome abnormalities increase from 50 60% in non-fragmented embryos to 70 90% in embryos with >35% fragmentation, but while fragmentation is strongly related to mosaicism and other post-zygotic abnormalities, aneuploidy does not appear to increase with fragmentation (Munné and Cohen, 1998; Magli et al., 2001). Table 2 shows a significant increase in mosaicism (P < 0.001), polyploidy and haploidy with increasing fragmentation, in particular when fragmentation is 35% or higher, but a similar trend is not observed in the rate of aneuploidy. In fact, while aneuploidy increases with maternal age, there is no link to an increase in embryo fragmentation (Alikani, 2001). There is no obvious correlation between chromosome abnormalities and fragmentation type (Munné et al., 2004a). However, in view of the recent findings regarding the cytoskeletal origin of fragmentation in the mouse (Alikani et al., 2005), a revision of the fragmentation classification may be necessary (M. Alikani, personal communication) and this issue may have to be revisited. Mechanism by which fragmentation affects implantation Even though fragmentation seems to affect blastocyst formation and implantation potential when it exceeds 15%, these embryos are not necessarily chromosomally abnormal. Thus, their reduced viability reflects a broader problem (Alikani et al., 2005). As suggested by Antczak and Van Blerkom (1999), the depletion of cortically positioned regulatory proteins from polarized domains in certain types of fragmented embryos may limit their potential for implantation. It has recently been shown that the distribution of the vital cell adhesion protein, E-cadherin, is perturbed in fragmented embryos; this could be one pathway to abortive blastulation (Alikani, 2005). In some cases, microsurgical removal of fragments may alleviate downstream effects of fragmentation (at least up to 35%), perhaps by restoration of the spatial relationship of cells within the embryo, facilitating cell-to-cell contact, compaction, cavitation and blastocyst formation (Alikani, 2001). 237

5 a b Figure 2. Laser scanning confocal images of human embryos with extensive fragmentation on day 3 of development. These embryos developed from normally fertilized eggs. In (a), a number of anucleate fragments (F) and two equally sized nuclei (arrowheads) in an interphase cell are visible. In (b), extensive micronucleation is visible in two cells (arrowheads) and roughly a quarter of the embryo has been lost to fragmentation. Images are courtesy of M Alikani (this work has been described in Alikani, 2005). Table 2. Chromosomal abnormalities detected by fluorescence in-situ hybridization and fragmentation rate. Data from Munné et al. (2004a). Fragments Analysed Aneuploid Mosaic and Normal (%) (%) other (%) (%) > P < 0.01 a P < a P < a a All values within column are significantly different from one another. 238 Multinucleation Frequency Multinucleated blastomeres (MNB) may occur at any time between the first cleavage division and the blastocyst stage, but are found more often in 2-cell embryos. The latter may be an artefact since visualization of nuclei may be more difficult when embryos have more cells and fragments. The frequency of cycles with multinucleated embryos has been reported to range between 14 and 79%, and the frequency of multinucleated embryos per patient ranges from 15 to 33.6% (Balakier and Cadesky, 1997; Jackson et al., 1998; Van Royen et al., 2003; Meriano et al., 2004). These differences between centres are significant, and could be attributed to differences in hormonal stimulation and culture conditions (Munné et al., 1997). A much larger study comprising 55,612 2PN embryos found that 12.5% were multinucleated on day 2 of development, and 5% on day 3 of development, so in total 17.6% exhibited some degree of multinucleation (Walmsley, 2003). More MNB were also seen in day 2 (27%) than day 3 (15%) by Van Royen et al. (2003). Figures 2 and 3 show embryos with multinucleated blastomeres. Clinical outcome Alikani et al. (2000) reported that only 16% of embryos with one or more multinucleated cells on day 2 or day 3 reached blastocyst stage compared with non-multinucleated embryos (32%, P < 0.001). However, embryos with binucleated cells formed blastocysts more frequently than those with multinucleated cells (38 versus 9%, P < 0.001; Meriano et al., 2004).

6 a b Figure 3. Laser scanning confocal images of cleaved human embryos on days 5 (a) and 6 (b) of development. These embryos developed from eggs in which pronuclei were never seen. Arrows in (a) point to micronuclei within one cell. In (b), one normally sized nucleus is visible along with micronuclei. The remaining cells are mono-nucleated. Green stain is tubulin and red stain is DNA. Images courtesy of M Alikani (unpublished). Embryos with MNB can implant (Mohr et al., 1983) and healthy babies have been born from exclusive transfer of such embryos (Balakier and Cadesky, 1997; Jackson et al., 1998; Pelinck et al., 1998), but clinical pregnancy from mixed transfers was significantly reduced, from 29% to 17% (P < 0.01) (Pelinck et al., 1998), and from 29 to 8% (Jackson et al., 1998). In pure replacements, the reduction was from 29 to 4% (Jackson et al., 1998) (P < 0.05) and from 53 to 19% (P < 0.005) (Walmsley, 2003), and the implantation rate from 25 to 6% (Van Royen et al., 2003) (P < 0.05), and from 35 to 13.6% (P < 0.001) (Walmsley, 2003). As with fragmented embryos, the implantation potential of MNB embryos was higher if replaced on day 3 (13%) or day 5 (7%), compared with non-dysmorphic and normally-developing embryos (40.1 and 49% respectively) (Walmsley et al., 2003). As with fragmented embryos, the implantation potential of MNB embryos was higher if replaced on day 3 (13%) or day 5 (7%), compared with non-dysmorphic and normally developing embryos (40.1 and 49% respectively) (Walmsley et al., 2003). The type of multinucleation was also taken into account in several studies. Transfer of pure binucleated (BN) embryos resulted in higher pregnancy rates than pure multinucleated (MN) ones (48 versus 15%; Meriano et al., 2004; 36 versus 26%; Walmsley et al., 2003). In addition, according to Jackson et al. (1998) implantation sites from cycles containing MNB embryos are six times more likely to abort (19 versus 3%, P = 0.006). Chromosome abnormalities FISH studies on MNB showed that the chromosomal content of each MNB nucleus was not always the same as the chromosomal content of the nuclei of the sibling blastomere MNB (Munné and Cohen, 1993; Munné et al., 1994a). Several studies have analysed MNB with multinucleation observed at the 2-cell stage, and all have detected high rates of abnormalities, ranging from 55 to 100%, the differences between studies depending mostly on the number of chromosomes being analysed by FISH (Kligman et al., 1996; Laverge et al., 1997; Staessen and Van Steirteghem, 1998; Magli et al., 2001). In the largest study so far, involving 558 MNB embryos compared with 1952 non-mnb embryos, the overall rate of abnormalities was 68% in non-mnb and 75% in MNB, but if aneuploidy is excluded (since it is related to maternal age), 53% of non-mnb were extensive mosaics, polyploids, or haploids compared with 68% of MNB (P < 0.001). In all these studies, when the multinucleated cell was abnormal the rest of the cells were also usually abnormal. Regarding the time at which MNB are observed in the past, transfer of bi-nucleated embryos on day 3 over those binucleated on day 2 was recommended when no other embryos were available for transfer (Kligman et al., 1996; Munné and Cohen, 1998; Munné et al., 2004a). In contrast, the most recent study has found little difference between day 2 (75% abnormal) and day 3 4 multinucleation (73% abnormal) (Magli et al., 2001; Walmsley, 2003), compared with non-mnb (60% abnormal). 239

7 240 The type of multinucleation has also been taken into account. Some studies have found that embryos with binucleated cells on day 3 have fewer abnormalities (68%) than those with multinucleated blastomeres (96%) (Meriano et al., 2004), but others did not find any difference (Staessen and Van Steirteghem, 1998). Mechanisms and contributory factors MNB have been associated with dysmorphism and fragmentation (Munné et al., 1995, Jackson et al., 1998). Most studies show a decreased developmental potential of MNB, and this impairment has been attributed to the arrest of the MNB cells (Hardy et al., 1993), resulting in a significantly lower cell count in MNB embryos (Jackson et al., 1998; Alikani et al., 1999), and the arrest of the whole embryo in 57% of embryos, while only 14% reached expanded blastocyst stage (Balakier and Cadesky, 1997). However, multinucleated cells on occasion must be able to give rise to normal diploid daughter cells and not always arrest, since pregnancies and normal offspring have been reported (Balakier and Cadesky, 1997; Jackson et al., 1998; Pelinck et al., 1998). Furthermore, time-lapse photography has shown that many binucleated and some multinucleated blastomeres do cleave (Meriano et al., 2004). Interestingly, time-lapse photography also showed that the nuclei in MNB disappear at different times, indicating that they are separate entities (Meriano et al., 2004). Two mechanisms for multinucleation have been proposed: the first occurring at the 2-cell stage of development in which the MII spindle is disorganized and results in extensive chromosome abnormalities and multinucleation, and the second occurring at the 4 16 cell stage at which time bi-nucleation occurs as a result of karyokinesis without cytokinesis (Pickering et al., 1995; Kligman et al., 1996; Munné and Cohen, 1998). Contributing factors increasing the odds of MNB occurrence are: Maternal age: Although most studies have found no increase in MNB with advanced maternal age (Balakier and Cadesky, 1997; Van Royen et al., 2003; Meriano et al., 2004) a significant increase in MNB of young patients than in older ones has been found (Jackson et al., 1998). Ovarian function and response to exogenous gonadotrophins: Follicular under-oxygenation has been related to multinucleation seen at the 2-cell stage of development (Van Blerkom et al., 1997). In addition, cycles with large numbers of oocytes retrieved have significantly more MNB embryos (Jackson et al., 1998; Van Royen et al., 2003), and lower implantation rates (Pellicer et al., 1989). Similarly, in another study cycles containing MNB have double the concentration of oestradiol on the day of human chorionic gonadotrophin (HCG), and need fewer ampoules of gonadotrophins than cycles without MNB (Jackson et al., 1998). This however was not confirmed by other studies (Van Royen et al., 2003; Walmsley, 2003). The duration of hormonal stimulation has also been associated with multinucleation; MNB occur more commonly in embryos of patients requiring fewer days of gonadotrophins before HCG (Walmsley, 2003; Van Royen et al., 2003). FSH concentrations: Patients with low (1 miu/ml or less) concentrations of FSH on day 3 tended to have a higher incidence of multinucleation (Walmsley, 2003). According to Van Royen et al. (2003), the observation of higher rates of multinucleation in patients that require a higher dose of hormone as well as patients with shorter cycles suggests that in both cases the reason could be a high number of immature follicles at the moment of ovulation induction. Immature follicles may reach metaphase II, but either their nucleus or cytoplasm is not properly matured and they are unable to cleave properly. For instance, Nogueira et al. (2000) found that embryos developing from in-vitro matured oocytes were mostly multinucleated. Thus, Van Royen et al. (2003) suggested the modification of hormonal stimulation to minimize the production of MNB embryos. Male factor: There was a difference in incidence of multinucleation between intracytoplasmic sperm injection (ICSI, 18.3% MNB) and IVF (17.2%, P < 0.001) in the largest study so far (Walmsley, 2003). In addition, MNB occurred in 22.2% of cases with severe male factor infertility, where ICSI was required. This tendency was not seen in smaller studies (Van Royen et al., 2003). Culture conditions: Sudden changes in temperature and suboptimal culture conditions are proven to damage the oocyte cytoskeleton and result in improper cleavage (Angell et al., 1987; Winston et al., 1991; Pickering et al., 1990). These situations may also play a role in multinucleation during the cleavage stage (Pickering et al., 1995; Hardy et al., 1993). There was a significant difference in incidence of multinucleation, depending on the source of protein supplementation for human tubal fluid culture media. In all, 21.1% of embryos were multinucleated by day 3 in culture media supplemented with maternal serum, whereas only 13.8% of embryos were multinucleated in culture media supplemented with HSA (P P < 0.001) (Walmsley, 2003). Dysmorphism and embryo development: Most dysmorphisms (fragmentation, MNB, asymmetry, etc.) tend to occur in the same embryos. For instance, Hardarson et al. (2001) has found that asymmetric embryos have higher rates of multinucleation, and Van Royen et al. (2003) found that highly fragmented embryos also have more multinucleation. Regarding cleavage rate, embryos with 4 cells on day 2 and 8 on day 3 had less chance of being multinucleated than those at other stages, but 15 17% of the best developing embryos were still multinucleated, suggesting that multinucleation per se may not be necessarily detrimental to embryo development (Van Royen et al., 2003). Cytoplasm immaturity: Metaphase I oocytes matured in vitro and subsequently fertilized showed significantly higher rates of multinucleation than control embryos (Balakier et al., 2004). Embryos with asymmetric blastomeres Blastomere asymmetry has been linked to reduced embryo competence (Giorgietti et al., 1995; Hardarson et al., 2001; Racowsky et al., 2003). Racowsky et al. (2003) found that embryos on day 3 of development with severe asymmetric blastomeres resulted in fewer viable offspring (1.4%) than embryos with moderate asymmetry (13.3%) or no asymmetry (22.4%) (P < ). However, if the embryo had 8 cells by day 3 and reached expanding/expanded blastocyst on day 5, day 3 blastomere asymmetry did not have an effect on implantation.

8 According to Hardarson et al. (2001), uneven blastomeres were associated with high incidence of aneuploidy. However, examining their data, it appears the abnormalities they detected are consistent with mosaicism, not with aneuploidy; 4/11 uneven embryos were mosaics or polyploid compared with 1/11 in symmetric embryos. Based on these results, the database of PGD embryos analysed has been evaluated, and, considering only one type of morphological abnormality, it has been found that among embryos with 7 9 cells, with <15% fragmentation and no multinucleation, embryos with asymmetric blastomeres had more chromosome abnormalities occurring post-meiotically (mosaicism, polyploidy, and haploidy) (35%) than symmetric ones (21%) (P < 0.001), less normal embryos (32.5 versus 40%, P < 0.05), and similar rates of aneuploidy (see Table 3) (Munné et al., 2004a). Giant eggs and embryos Giant oocytes have an average diameter of 200 microns, including the zona pellucida, and occur at a frequency of 0.3% (Balakier et al., 2002; Rosenbusch et al., 2002). Giant eggs have also been reported in the Chinese hamster with a frequency of 0.47% and were found to be digynic triploid (Funaki and Mikamo, 1980). Embryos developing from giant oocytes were invariably triploid or triploid mosaics, with XXX or XXY gonosome constitutions, which suggested a higher contribution of maternal genomes [Munné et al., 1994c (n = 6); Balalkier et al., 2002 (n = 4)]. Similar oocytes karyotyped at MII by Rosenbusch et al. (2002) (n = 6) and Balakier et al. (2002) (n = 9) were also found to be triploid. These two reports also analysed zygotes developing from giant eggs and found them all triploid or polyploid. Human giant eggs at germinal vesicle (GV) stage show two nuclei, each one with 23 univalent (two chromatid) chromosomes (Munné and Cohen, 1998). This indicates that giant GV originate from cytokinetic failure producing a tetraploid GV, or from fusion of two GV. According to Rosenbusch et al. (2002). during MII the two sets may unite in a single metaphase or the egg may develop two independent metaphase plates. After fertilization in the first situation, the egg will have two pronuclei (diploid female and a haploid male) and two polar bodies, but in the second it will have three pronuclei (all haploid) and four polar bodies. Both situations were equally common in the Rosenbusch et al. (2002) and Balakier et al. (2002) studies. According to Balakier et al. (2002), embryos developing from giant eggs could reach blastocyst stage, indicating that they can be a source of digynic triploid fetuses. Other dysmorphisms Dominant blastomere embryos Embryos (n = 13) with only one large cell surrounded by smaller blastomere-sized extracellular fragments. These embryos were polyploid and frequently polyploid mosaics. Some of these embryos showed ploidies of up to 20N, and the single cell was normally multinucleated (Munné et al., 1994c). These results were later confirmed by Magli et al. (2001) in a study of 20 embryos with a dominant blastomere blocked at the two cell stage, surrounded by smaller cells. They found that 12 were polyploidy and eight were complex abnormal. In 14/20 embryos the dominant blastomere was also multinucleated. Embryos with cytoplasm irregularities According to Magli et al. (2001), the presence of vacuoles or dark inclusions was not associated with an increase of chromosome abnormalities, but embryos with cytoplasm concentrations had 86% chromosome abnormalities compared with embryos without it (63%, P < 0.001). Elongated shape embryos Magli et al. (2001) analysed the chromosome constitution of 18 embryos and they were found to have the same proportion of chromosome abnormalities than non-elongated embryos. Esfandiari et al. (2005) recently produced successful pregnancies from them. Table 3. Chromosome abnormalities in preimplantation genetic diagnosis embryos with good morphology on day 3 (<15% fragmentation, no multinucleated blastomeres, 7 9 cells) with or without asymmetric blastomeres. Data from Munné et al. (2004a). Symmetric Asymmetric P-value (%) (%) Average age (years) Aneuploid 314 (38.6) 76 (32.1) NS Haploid 8 (1.0) 8 (3.4) <0.025 Extensive mosaic a 157 (19.3) 65 (27.4) <0.01 Normal b 326 (40.0) 77 (32.5) <0.05 Polyploid 9 (1.1) 11 (4.6) <0.005 Total Post-meiotic c 174 (21.4) 84 (35.4) <0.001 Values are numbers with percentages in parentheses. a Extensive mosaics have 3/8 abnormal cells. b Included with normals were mosaics with <3/8 abnormal cells. c Polyploidy, haploid and extensive mosaics. NS = not statistically significant. 241

9 242 Chromosomal abnormalities and embryo development Introduction Cleavage patterns Developmental day 3 embryos can be classified depending on their rate of cleavage in four large groups, arrested, slow, normal, or fast developing embryos. Arrested embryos are those that have not cleaved during a 24-h period. Slow embryos are those that have not reached the 7-cell stage on day 3 of development but have cleaved during a 24-h period. Normal embryos are those that reach 7 9 cells on day 3, with <15% fragmentation and without multinucleation ( normally developing embryos with abnormal morphology are described in the previous section and in several other studies, grouped with slow embryos), and that have cleaved in the preceding 24 h. Accelerated embryos are those that have >9 cells by day 3. Clinical outcome Several articles have reported compromised clinical outcome and/or development for embryos with slow or accelerated cleavage rates (Cummins et al., 1986; Giorgetti et al., 1995; Ziebe et al., 1997; Alikani et al., 2000; Racowsky et al., 2003). According to Alikani et al. (2000), 13.8% of embryos with <7 cells on day 3, 41.9% of those with 7 9 cells, and 27.5% with >9 cells formed apparently normal blastocysts(p < 0.001). Similarly, Racowsky et al. (2003) found that if embryos were replaced on day 3, those with 8 cells produced the highest rate of live births (25%), compared with embryos with >8 cells (18%), 7 cells (18%) or <7 cells (3%) (P < ). In contrast to Alikani et al. (2000), other studies have found that the major and almost sole indicator for embryo viability at day 5 of development (embryos resulting in offspring) was blastocyst morphology, and not day 3 morphology (Rijnders and Jansen, 1998; Graham et al., 2000; Shapiro et al., 2000; Balaban et al., 2001; Racowsky et al., 2003). Method of study There are now three studies analysing more than 500 embryos per study (Munné et al., 1995; Márquez et al., 2000; Magli et al., 2001). The first study analysed 524 embryos (Munné et al., 1995), the second analysed 721 (Márquez et al., 2000), and the third study analysed one cell each from 1596 embryos (Magli et al., 2001). Estimate of chromosome abnormalities following IVF The overall rate of chromosome abnormalities following IVF, including all treated patients, could be estimated based on morphology of the embryos and age of the patients. Embryos included in the study of Márquez et al. (2000) were obtained from a 6380 IVF cycles producing 48,765 embryos. Multiplying these frequencies by the frequency of chromosome abnormalities in each particular group, it was estimated that the overall rate of chromosome abnormalities for the patient population was 43.3%. Aneuploidy, maternal age and embryo development The first two studies mentioned above (Munné et al., 1995; Márquez et al., 2000) classified the embryos in three maternal age groups: 20 34, and years old. Pooled results from these two studies, for a total of 1255 embryos, demonstrated a highly significant relationship between maternal age and aneuploidy (P < 0.001). Individual chromosomes involved in non-disjunction were also analysed using pooled results from the two studies, and chromosomes 16, 18 and 21 showed significantly higher frequencies of aneuploidy with increasing maternal age (P < 0.05, P < 0.05 and P < 0.01 respectively) (reviewed by Munné et al., 2004a). The chromosomes most involved in aneuploidy at the cleavage-stage were found to be, in order, chromosome 22, 16, 15 and 21 (Table 4). Figure 1a and 1f show trisomic cells. This was confirmed by Bielanska et al. (2002a) in a smaller series of 216 embryos in which a relationship between maternal age and aneuploidy was also found. Together, these studies show the relationship between increasing maternal age and aneuploidy, which is also known from prenatal and post-natal data (Hassold et al., 1980; Hassold and Chiu, 1985; Warburton et al., 1986). However, the large difference in the number of aneuploid embryos detected at cleavage stage compared with prenatal data indicates a strong selection against aneuploid embryos before or shortly after implantation. For instance, while autosomal monosomies are rarely detected in first trimester pregnancies (only 1/1000 are monosomy 21), cleavage-stage embryos showed more monosomy than trisomy (Munné et al., 2004b). Monosomic embryos are eliminated during or prior to blastocyst formation, as demonstrated in the mouse (Magnuson et al., 1985) and humans (Sandalinas et al., 2001). Similarly, trisomies 18, 13 and 21 were found in about 8% of day 3 and 4 embryos but only in 4% of chorionic villous samples (CVS) from women years old (Hassold and Chiu, 1985; Warburton et al., 1986). Again, this indicates that even for those trisomies compatible with development, there is a strong negative selection during the preimplantation stages and/or shortly after implantation. Thus, it is no surprise that maternal age is also linked to compromised embryonic development. For instance, several authors have detected decreased cleavage rates in day 3 embryos and reduced blastocyst formation with advance maternal age (Hsu et al., 1999; Terriou et al., 2001; Ziebe et al., 2001; Racowsky et al., 2003). Monosomy was more frequent than trisomy for all chromosomes studied (Márquez et al., 2000; Munné et al., 2004b). Since non-disjunction theoretically produces disomic and nullisomic gametes with the same frequency, either

10 technical error or an alternative mechanism for aneuploidy, e.g. mitotic anaphase lag during the first mitotic division (Ford et al., 1988) or loss of a chromosome during meiosis, could explain the latter observation. Technical error is unlikely to be the cause of these differences (Munné et al., 2004b) and therefore up to 20% of aneuploidies may be due to loss of chromosomes before or during the first mitotic division. Aneuploidy does not necessarily lead to developmental arrest in the first 3 days of culture since the embryonic genome is not fully active until day 2 or 3 of development (Braude et al., 1988; Tesarik et al., 1988). In fact, aneuploidy seems to be more common in good quality embryos. There is probably an underestimation of aneuploidy in those embryos that are also chaotic or polyploid mosaics, where detecting aneuploidy is difficult because of the other chromosome abnormalities (Munné et al., 1995). Aneuploidy and response to hormonal stimulation Although aneuploidy originates during gametogenesis (mitotic aneuploidy is classified here as mosaicism), and therefore is not affected by embryo culture conditions, the hormonal stimulation process and patient response may have an effect on aneuploidy. Poor responders While Gianaroli et al. (2001b) reported that patients with a poor response to hormonal stimulation produced 68% chromosomally abnormal embryos, others have found that the lower pregnancy rate seen in poor responders seems to be related only to a reduced number of embryos available for replacement (De Sutter and Dhont, 2003). Reduced ovarian reserve (high FSH) Ovarian reserve is measured according to FSH concentrations on day 3. High concentrations of FSH are indicative of reduced ovarian reserve (Khalifa et al., 1992) with consequent decrease of oocyte production after ovarian stimulation; and are related to lower pregnancy rates. High FSH is not only found in women of advanced maternal but also in women with unilateral oophorectomy and younger women with premature menopause. However, the ovarian reserve of young patients, measured as FSH production, may only be a good predictor of egg production but less so of egg quality (aneuploidy) (Ubaldi et al., 2003). Warburton (1989) suggested that if aneuploidy was caused by oocyte depletion, women who have had a trisomic conception Table 4. Specific chromosome aneuploidy rates. Chromosome Analysed Monosomic Trisomic Total no. embryos (n) (n) (n) aneuploid (n, %) XY (1.2) (2.5) (0.6) (2.5) (2.0) (1.5) (2.9) (1.2) (3.7) a 19 a 56 (2.9) (1.1) (5.0) (5.3) (2.5) (2.3) (4.9) (6.2) Total b 233 b 549 Acrocentrics 191 c 124 c 315 (13, 14, 15, 21, 22) Non-acrocentrics (others) Double aneuploidies counted twice, once for each chromosome. Tetrasomies and nullisomies were counted as two trisomies and two monosomies respectively. Significance: a versus a': P < 0.05, b versus b', c versus c': P < Source: Munné et al. (2004a) based on pooled data from Munné et al. (2004b) and Abdelhadi et al. (2003). 243

11 244 at a young age might exhibit signs of early oocyte depletion, such as premature menopause. Indeed, some later studies found a significant increase in trisomy 21 in babies from young women with reduced ovarian complement, as well as elevated FSH in women with an aneuploid conceptus (Van Montfrans et al., 1999, Freeman et al., 2000, Kline et al., 2000). Therefore, high FSH concentrations may be directly linked to aneuploidy, regardless of the age of the patient. If that is so, FSH could be a better prognosticator of aneuploidy risk and IVF outcome than maternal age alone. High ovarian response The opposite of the above groups are patients with a high ovarian response to hormone stimulation. In two studies on oocyte donors (Reis Soares et al., 2003; Munné et al., 2006), usually heavily stimulated to produce large cohorts of oocytes, had embryos with chromosome abnormality rates higher than fertile controls and similar to infertile controls. In addition, high ovarian response has been associated with multinucleation (Jackson et al., 1998), and uneven pronuclei (Sadowy et al., 1998), both features were related to chromosome abnormalities, as seen in this review. Post-meiotic chromosome abnormalities and embryo development Mosaicism, haploidy and polyploidy occur post-meiotically (mitotically), and mostly after fertilization (Munné et al., 1995). Therefore, post-meiotic chromosome abnormalities account for more than half the abnormalities detected in cleavage stage embryos. Post-meiotic abnormalities significantly decreased with embryonic competence (polyploidy P < 0.001, extensive diploid mosaicism P < 0.01) but were not affected by maternal age (Munné et al., 1995; Márquez et al., 2000). Study III analysed chromosome abnormalities in relation to cell number, and found that the lowest rate of abnormalities (55%) was in normal day 3 embryos with 7 8 cells, while slow embryos with 4 or fewer had 74% (P < 0.001); accelerated embryos with 9 or more cells were similar with 79% abnormal (P < 0.005) (Magli et al., 2001). Polyploidy Figure 1e shows a polyploidy embryo. Polyploidy increased with decreasing embryo competence, mostly occurring in arrested embryos (Munné et al., 1995; Laverge et al., 1998; Márquez et al., 2000; Bielanska et al., 2002a). As previously noted, it is unlikely that this represents polyspermy, since all these embryos were derived from dipronucleated zygotes with two polar bodies (Munné et al., 1994a). The most likely explanation is that their DNA synthesis continued, although cellular division had stopped. In some instances they also continued karyokinesis, producing multinucleation in almost half their cells. That DNA synthesis is not prevented by cleavage arrest has been demonstrated by Artley et al. (1992). According to Winston et al. (1991), even if karyokinesis and gene activation do not fail, impaired cytokinesis may arrest the embryo because there are insufficient cells to produce a functional inner-cell mass. Since most polyploid embryos arrest before the onset of genome activation, which occurs around the 4- to 8-cell stage (Braude et al., 1988; Tesarik et al., 1988), oocyte quality or embryo culture conditions may be the cause of their arrest, instead of their ploidy. The incidence of polyploidy was much higher when embryos were analysed on day 4 (Munné et al., 1995) compared with day 3 (Márquez et al., 2000). The most probable reason is that arrested embryos studied on day 4 had an extra day to replicate their DNA without cleaving thus becoming polyploid at a higher rate than arrested embryos on day 3. As these embryos seldom reach the blastocyst stage, prolonged culture may effectively eliminate them from transfer. Mosaicism Excluding the >40 years age group, extensive ( 3/8 abnormal) diploid mosaicism is the major chromosome abnormality in IVF-generated human embryos. For the age group 35 39, mosaicism was found in 23.3% of the embryos (Márquez et al., 2000), followed by polyploidy (21.8%), aneuploidy (10.2%), and haploidy (3.6%). Even in the group of normally developing embryos, which are closer in quality to those embryos being replaced, aneuploidy (19.3%) contributes to less than half of the chromosome abnormalities detected, with extensive diploid mosaicism (14.7%), polyploidy (4.5%) and haploidy (4%) together contributing more than aneuploidy. Similarly, another study by Bielanska et al. (2002b) detected 2N/mosaicism (extensive and limited) and chaotic mosaics in 55% of spare embryos, followed by 30% normal ones, and the rest aneuploid, polyploidy and haploid. Figure 1g shows all the cells of a mosaic embryo after FISH. Because the embryonic genome is not fully active until day 3 of development (Braude et al., 1988), mosaicism, polyploidy and haploidy cannot produce dysmorphism originating in the first and second meiotic divisions. However, cytoplasmic impairment could produce both mosaicism and polyploidy, through cytoskeletal and spindle malfunction, cellular division block, or other mechanisms. For instance, abnormalities of the centriole in a fertilizing spermatozoon may produce mosaicism or other chromosome abnormalities in the resulting zygote (Palermo et al., 1994; Hewitson et al., 1997; Silber et al., 2003). In addition, a chaotic mosaicism, but not other types of chromosome abnormalities, has been associated with an intrinsically low mitochondrial activity (Wilding et al., 2003). Most chromosome studies on human embryos have focused on meiotic irregularities as the principal source of chromosome abnormalities. However, the present data indicate that other sources of chromosome abnormalities are more important, and further investigation of other factors such as culture conditions, hormonal stimulation, centriole abnormalities, and cytoplasmic factors is justified. Accelerated cleavage embryos Harper et al. (1994b) first described the occurrence of embryos with accelerated cleavage. These embryos were fertilized (Y presence) but most of them were mosaics, suggesting that they could have been polyspermic and may have become

12 activated much earlier than normal. Magli et al. (1998) later confirmed that these embryos had high rates of chromosome abnormalities. They usually reach the blastocyst stage less frequently (Alikani et al., 2000). Chromosome abnormalities, embryo selection and PGD Embryo selection is critical to in the success of IVF. Careful evaluation of embryo morphology performed under powerful inverted microscopes will detect many abnormalities such as multinucleation and fragmentation. Dysmorphic and arrested embryos, >50% of which are chromosomally abnormal, should not be replaced if better embryos are available. However, this evaluation does not allow selection against chromosome abnormalities, which occur with a frequency of 30% in embryos with apparently normal morphology, in women years old. This rate increases to about 60% in women aged >40 years. In general, PGD for numerical chromosome abnormalities is indicated for patients 35 and older; but it is also being offered to younger patients who are oocyte donors, have recurrent miscarriages, or a history of failed implantation. So far, close to 10,000 PGD cases have been performed either by embryo biopsy at day 3 of development (Munné et al., 1993b, 1999, 2003, 2004b,c,d, 2005, 2006; Gianaroli et al., 1997, 1999, 2001a,b, 2003; Kahraman et al., 2003) or by polar body biopsy (Verlinsky et al., 1995, 1998, 2001, 2005; Verlinsky and Kuliev, 1996, 2003; Kuliev et al., 2003; Taranassi et al., 2005), and these studies have shown an increase in implantation, pregnancy, and take-home baby rate with a decrease in spontaneous abortion (Gianaroli et al., 1999, 2005; Munné et al., 1999, 2003, 2005; Verlinsky et al., 2005). Effect of chromosome abnormalities on cleavage-stage development Large-scale transcriptional activity begins between the 4- and 8-cell stages in the human (Braude et al., 1988; Tesarik et al., 1988). Although low level transcription has been detected at the 2-cell stage, the functional significance of this activity is uncertain (Taylor et al., 2001; Wells et al., 2005). Therefore, even though chromosome abnormalities may affect embryo development at any stage, the effect is accentuated after the transcriptional burst. In cow embryos, it has been found that depending on the chromosome abnormality, embryos analysed at day 5 of development have the following average number of cells: 7.9 cells for haploids, 7.9 for polyploids, 16.8 for aneuploid, 23.4 for mosaics 2N/4N, and 30 for normal diploids (P < 0.001) (Kawarsky et al., 1996). In other species, mouse haploid parthenogenotes develop slower than diploid embryos (Henery and Kaufman, 1992), while triploid and tetraploid bovine embryos have also been seen to have fewer cells than their diploid counterparts (King et al., 1987). Not surprisingly, in cleavage-stage human embryos it was found that arrested embryos have more chromosomal abnormalities than non-arrested ones (Munné et al., 1995). It seems that chromosome abnormalities such as polyploidy, chaotic mosaicism and haploidy will result in arrested cells or slowly cleaving cells, thus having a direct affect on the overall development of the embryo. Chromosome abnormalities in blastocyst The potential advantages of prolonging culture of IVF embryos to the blastocyst stage remains a subject of discussion and debate. The results so far seem promising because those embryos reaching blastocyst stage implant at very high rates (Gardner et al., 1998). Two types of genetic studies in blastocyst stage embryos have been reported. The first includes studies performed in leftover or surplus embryos that developed to blastocyst stage and therefore were preselected by morphology and development (Clouston et al., 1997, 2002; Veiga et al., 1999; Ruangvutilert et al., 2000; Bielanska et al., 2002b, 2005; Hardarson et al., 2003). The second type were studies analysing blastocysts originated from embryos diagnosed as abnormal by PGD (Magli et al., 2000; Sandalinas et al., 2001). Excluded studies Some have been considered but excluded from this review because: (i) only two chromosome pairs were analysed (Derhaag et al., 2003; Coonen et al., 2004); (ii) they did not use sequential media to culture blastocysts (Evsikov and Verlinsky, 1998; Coonen et al., 2004); (iii) they did not specify what type of embryos reached day 5 (arrested cleavage stage, morulas, blastocysts) and had average cells per blastocyst of 50 or lower (Coonen et al., 2004; Daphnis et al., 2004) or (iv) the data on the blastomeres analysed was not clear (Bielanska et al., 2002a). Chromosome analysis of unselected blastocysts The analysis of surplus blastocysts from non-pgd cycles is a good approach to find answers to several questions such as whether chromosome abnormalities occur differentially within inner cell mass and trophectoderm cells. It is also useful for the study of mosaicism. However, they are less suitable for determination of surviving capacity of abnormal embryos, since the majority are normal or low-degree 2N/4P mosaics. For the latter, embryos left over form PGD cycles are more suitable. Many studies have assessed the chromosome composition of surplus blastocysts and are reviewed here (Clouston et al., 1997, 2002; Evsikov and Verlinsky, 1998; Veiga et al., 1999; Ruangvutilert t et al., 2000, Bielanska a et al., 2002a,b, 2005; Hardarson et al., 2003) (Table 5), while other studies have been excluded for reasons mentioned above (Table 6). Common findings In (of course, they are not replaced if they are surplus!) surplus, low-quality blastocysts excluded from transfer, mosaicism was the most common abnormality. In karyotype studies, the number of metaphases obtained ranged from 0 to 37, with 56% 245

13 246 yielding between two and 10 karyotypable metaphases. The results showed that the majority of blastocysts were diploid/ tetraploid mosaics and other types of mosaics, with a few pure trisomies (Clouston et al., 2002; Bielanska et al., 2005). The rate of mosaicism detected in FISH studies is up to 90% (Ruangvutilert t et al., 2000; Bielanska et al., 2002a,b, 2005), but the percentage of abnormal cells was low compared with cleavage stage mosaic embryos, being no higher, on average, than 30%. In addition, the majority of abnormal cells found in mosaics were tetraploid, 2N/4N mosaics being very common (23 86% of all blastocysts), but triploid, haploid, aneuploid and chaotic cells were also described. Although aneuploid cells seem detrimental for embryo development, high levels of mosaicism and chaotic embryos can still be detected at the blastocyst stage (Magli et al., 2000; Sandalinas et al., 2001). This, compared with lower rates of 2N/4N mosaics on day 3 of development (Munné and Cohen, 1998), indicates that most 2N/4N mosaics arise at the morula or later stages. In studies comparing chromosome abnormalities in blastocysts from good and poor morphology day 3 embryos, some found higher rates of abnormalities, mostly mosaics, in those with poor morphology (15 versus 57%; Hardarson et al., 2003), while others did not (82 versus 86%; Bielanska et al., 2002b). Nevertheless, in all studies there were higher rates of abnormal cells per blastocyst in those developing from poor morphology day 3 embryos. In a study evaluating chromosome abnormalities in relation to blastocyst morphology, they found that 65% of mosaic blastocysts had good morphology (Bielanska et al., 2005). Thus, by itself, morphology is generally not an appropriate tool to screen for chromosome abnormalities. Figure 4 shows a blastocyst with normal morphology, even though it developed from a digynic zygote. Survival of chromosomally abnormal embryos to the blastocyst stage Several studies have analysed the survival of chromosome abnormalities to the blastocyst stage, either comparing rates of chromosome abnormalities in cleavage, morula and blastocyst stages (Márquez et al., 2000; Bielanska et al., 2002a) or by following to blastocyst stage embryos identified by PGD at cleavage stage as chromosomally abnormal (Magli et al., 2000; Sandalinas et al., 2001). Significant differences in the frequency of chromosome abnormalities have been found among arrested, slow and normally developing embryos in cleavage stages (Munné et al., 1995). Janny and Menezo (1994) argued that selection against chromosome abnormalities may occur during extended culture because many embryos arrest during morula stage. Evsikov and Verlinsky (1998) further speculated that cavitation initiates a negative selection against aneuploid cells, and therefore if the aneuploid cells at morula stage reach some threshold level, this would lead to the self-elimination of the whole embryo. Magli et al. (2000) reported that only 22% of chromosomally abnormal embryos reached blastocyst stage compared with 34% of euploid ones (P < 0.001). Of the embryos surviving to blastocyst stage, most showed that the inner cell mass was mosaic, with 2 16 different cell lines. Márquez et al. (2000) compared day 3 and day 4 embryos, finding that embryos analysed on day 4 had much higher rates of polyploidy than those analysed on day 3, indicating that embryos arresting on day 3 become polyploid by day 4. Mosaics Bielanska et al. (2002b) found that chaotic mosaics were more common in arrested day 3 and day 4 embryos than in blastocyst stage embryos. They also found that diploid/ polyploid mosaics increased with developmental competence while there was an overall decrease in the number of abnormal cells per mosaic embryo, from cleavage stage to blastocyst stage. Sandalinas et al. (2001) found that embryos with high frequency of mosaicism can occasionally develop to blastocyst, although they never had more than 60 cells compared with an average of 114 of other blastocysts. In another study, embryos classified by PGD as complex abnormal, usually mosaics, reached blastocyst stage in 32% of cases. Trisomies Thirty-seven and 34% of trisomies reached blastocyst stage compared with 66 and 61% of normal embryos (Sandalinas et al., 2001 and Rubio et al., 2003 respectively). The difference was statistically significant in both studies. In another study, Magli found that only 24% of normal and 18% of trisomies survived to blastocyst. Monosomies and haploidy Only 9% monosomies and 0% haploidy, reached blastocyst stage in the Sandalinas study, and of those only monosomy X and 21, reached that stage (Sandalinas et al., 2001). Magli et al. (2001) and Rubio et al. (2003) also detected a strong selection against haploidy (2 and 10% blastocyst, respectively) and autosomal monosomy (both, 20% blastocyst) but they did not reanalyse the embryos to confirm the abnormality and some of the embryos reaching blastocyst might have been misdiagnosed and be normal. Rubio et al. (2003) found that 54% of monosomy X survived to blastocyst. The fact that mostly monosomy X and 21 were found in blastocyst stage (Sandalinas et al., 2001) agrees with prenatal diagnosis data, where no other monosomies are detected in first trimester abortions (Eiben et al., 1990, Strom et al., 1992). Polyploidy Polyploid embryos clearly reach blastocyst stage because polyploid pregnancies reach first trimester and beyond. Sandalinas et al. (2001) found that 21% reach blastocyst stage; others have also found polyploid embryos reaching blastocyst stage (Rubio et al., 2003). Translocations Menezo et al. (1997) have also reported in-vitro selection of

14 Table 5. Characteristics of studies included in this review, all made at blastocyst stage. Study Average Average Media No. chromosome included age (years) no. cells sequential analyses >3 pairs >50 1 n/a 175 a DMEM:F12 or G2 G-banding karyotype 2 n/a 55.5 G1/G2 5 probes M3 3 probes 4 n/a >70 G1/G2 3 8 probes 5 n/a 63 b /99 a DMEM:F12 or G2 G-banding karyotype rs2, Vitrolife 7 probes G1/G2 3 probes References: 1 = Clouston et al. (1997), 2 = Veiga et al. (1999), 3 = Ruangvutilert t et al. (2000), 4 = Bielanska et al. (2002b), 5 = Clouston et al. (2002), 6 = Hardarson et al. (2003), 7 = Bielanska et al. (2005). a,b Stage: day 6 or later, and day 5, respectively. DMEM = Dulbecco s modified Eagle s medium. Table 6. Chromosome abnormalities in unselected blastocyst studies. Reference No. analysed No. No. 2N/ No. 2N/ No. 2N/ No. No. No. blastocysts normal polyploid aneuploid chaotic polyploidy haploid aneuploid analysed (polyploid) or chaotic , (6) a b (5) References: 1 = Clouston et al. (1997), 2 = Veiga et al. (1999), 3 = Ruangvutilert t et al. (2000), 4 = Bielanska et al. (2002b), 5 = Clouston et al. (2002), 6 = Hardarson et al. (2003), 7 = Bielanska et al. (2005). a,b Analysis: ploidy only, and karyotype, respectively. Figure 4. Laser scanning confocal image of a blastocyst that developed from a digynic intracytoplasmic sperm injection egg. Note the apparently normal morphology and distribution of E-cadherin (Alikani, 2005), despite (presumed) major chromosomal abnormality. Arrow points to the area of the inner cell mass (which is visible in the section presented in the inset. Green stain is E-cadherin and red stain is DNA. Images courtesy of M Alikani (unpublished). 247