Discussion

Although the majority of ART-related multiple pregnancies are thought to be dizygotic (derived from two or more fertilized eggs), numerous reports suggest an increase in MZT pregnancies (Alikani et al., Reference Alikani, Cekleniak, Walters and Cohen2003; Aston et al., Reference Aston, Peterson and Carrell2008; Blickenstein et al., Reference Blickenstein, Verhoeven and Keith1999; Derom et al., Reference Derom, Vlietinck, Derom, Van Den Berghe and Thiery1987; Edwards et al., Reference Edwards, Mettler and Walters1986; Sills et al., Reference Sills, Moomjy, Naninovic, Veeck, McGee, Palermo and Rosenwaks2000). Currently, approximately 1.5–2% of all clinical pregnancies that occur after ovulation induction using exogenous gonadotropin therapy (with or without IVF and embryo transfer) are reported as MZT pregnancies, and the incidence of those is four times higher than that of spontaneously conceived MZT pregnancies which eventuated in live births (Alikani et al., Reference Alikani, Cekleniak, Walters and Cohen2003; Derom et al., Reference Derom, Vlietinck, Derom, Van Den Berghe and Thiery1987; Edwards et al., Reference Edwards, Mettler and Walters1986; Schachter et al., Reference Schachter, Raziel, Friedler, Strassburger, Bern and Ron-El2001; Wenstrom et al., Reference Wenstrom, Syrop, Hammitt and Van Voorhis1993).

Some authors have identified ovarian hyperstimulation (Derom et al., Reference Derom, Vlietinck, Derom, Van Den Berghe and Thiery1987, Reference Derom, Leroy, Vlietinck, Fryns and Derom2006; Schachter et al., Reference Schachter, Raziel, Friedler, Strassburger, Bern and Ron-El2001; Sills et al., Reference Sills, Moomjy, Naninovic, Veeck, McGee, Palermo and Rosenwaks2000), assisted hatching (Abusheikha et al., Reference Abusheikha, Salha, Sharma and Brinsden2000; Alikani et al., Reference Alikani, Noyes, Cohen and Rosenwaks1994; Hershlag et al., Reference Hershlag, Paine, Cooper, Scholl, Rawlinson and Kvapil1999; Saito et al., Reference Saito, Tsutsumi, Noda, Ibuki and Hiroi2000; Schieve et al., Reference Schieve, Meikle, Peterson, Jeng, Burnett and Wilcox2000; Skiadas et al., Reference Skiadas, Missmer, Benson, Gee and Racowsky2008; Slotnik & Ortega, Reference Slotnik and Ortega1996; Tarlatzis et al., Reference Tarlatzis, Qublan, Sanopoulou, Zepiridis, Brimbizis and Bontis2002), and/or extended culture (Behr et al., Reference Behr, Fisch, Racowsky, Miller, Pool and Milki2000; Da Costa et al., Reference Da Costa, Abdelmassih, De Oliveira, Abdelmassih, Abdelmassih and Nagy2001; Jain et al., Reference Jain, Boostanfar, Slater, Francis and Paulson2004; Kyono et al., Reference Kyono, Shiotani, Watanabe, Oka, Takahashi and Hayashi2011; Milki et al., Reference Milki, Jun, Behr, Giudice and Westphal2003; Rijinders et al., Reference Rijinders, van Os and Hansen1998; Sheiner et al., Reference Sheiner, Har-Vardi and Potashnik2001; Vitthala et al., Reference Vitthala, Gelbaya, Brison, Fitzgerald and Nardo2009; Wright et al., Reference Wright, Schieve, Vahgratian and Reynold2004) as possible reasons for monozygotic twinning.

Our aim is to clarify the precise timing of embryo splitting resulting in monozygotic DD twins. First, there have been some reports which suggested that embryo splitting occurred in the blastocyst stages.

There are various reports of atypical hatching leading to twinning in the literature, in particular those of Van Langendonckt et al. (Reference Van Langendonckt, Wyns, Godin, Toussaint-Demylle and Donnez2000), Meintjes et al. (Reference Meintjes, Guerami, Rodriguez, Crider-Pirkle and Madden2001), and Behr and Milki (Reference Behr and Milki2003), but our case was the first that reported successful birth of same-sex DD twins after a single-blastocyst transfer (Doshida et al., Reference Doshida, Toya, Araki and Kyono2012). Van Langendonckt et al. (Reference Van Langendonckt, Wyns, Godin, Toussaint-Demylle and Donnez2000) reported a case of atypical hatching of a human blastocyst leading to monozygotic twinning. They showed a half-blastocyst in the zona pellucida and a half-blastocyst herniated through the zona pellucida in one blastocyst. After a two-blastocyst transfer, they confirmed one gestational sac and fetus with a fetal heartbeat, and two gestational sacs and fetuses without fetal heartbeats, and it was concluded that this was a dizygotic trichorial triamniotic triplet pregnancy. Meintjes et al. (Reference Meintjes, Guerami, Rodriguez, Crider-Pirkle and Madden2001) reported a trichorionic heterozygotic triplet pregnancy with dichorionic monozygotic twins following transfer of two expanded blastocysts, one of which had two distinct ICMs. We speculate that one blastocyst had two distinct ICMs in one trophectoderm that formed two half-blastocycts with two ICMs and two trophectoderms, which developed into monozygotic DD twins. One of the triplets was ongoing at 33 weeks at the time of the original publication.

Behr and Milki (Reference Behr and Milki2003) reported a case of atypical hatching of a human blastocyst in vitro forming two identical embryos. On day 5, they observed two trophoblast populations sharing an ICM from a single blastocyst. On the afternoon of day 5, the now completely hatched blastocysts of similar size, both of which had a visible trophectoderm and ICM, connected together. On day 6, the two blastocysts separated completely. However, the blastocysts were not cryopreserved or transferred. Tokunaga et al. (Reference Tokunaga, Shibahara, Shimada, Hirano, Suzuki, Takamizawa and Suzuki2010) reported monozygotic DD twins (two healthy girls) following a single warmed blastocyst (Grade 5BC) with assisted hatching.

Knopman et al. (Reference Knopman, Krey, Lee, Fino, Novetsky and Noyes2010) reported 98 MZTs confirmed on transvaginal ultrasonography at 7–12 weeks gestation (2% incidence). Only one DD MZT pregnancy was clearly diagnosed in their cohort. However, their diagnosis of DD gestations is limited, because DNA analysis, the gold standard for diagnosing monozygotic siblings, was not performed on the fetuses or children.

Knopman et al. (Reference Knopman, Krey, Lee, Fino, Novetsky and Noyes2010) also reported that they had not seen an embryo splitting in half spontaneously before the blastocyst stage in over 15 years of laboratory experience. In their one case of DD monozygosity, no embryo splitting was noted in a daily microscopic examination during development from zygote to blastocyst.

Second, there are our two cases in which embryo splitting occurred during blastocyst stages. In our two cases, ultrasonography showed a λ sign between two gestational sacs and fetuses with fetal heartbeats at seven weeks of gestation. The diagnosis was a single zygote with two chorions and two amnions. Carroll et al. (Reference Carroll, Soothill, Abdel-Tattah, Porter, Montague and Kyle2002) reported that the most reliable indicator for dichorionicity was a combination of the λ sign and two separate placentae with a sensitivity and specificity of 97.4% and 100%, respectively. The pathological finding was that the DD twin placentas had fused in the third trimester in the second case. DNA fingerprinting (short tandem repeat) confirmed that the two babies originated from the same zygote in the first case.

In the second reported case of DD twins from an embryo that splits in the blastocyst stage, the embryo had been chosen due to its grade being higher than that of the other cryopreserved embryos. When we found the abnormality (two half-blastocysts) in the warmed embryo, we informed the patients, who decided to proceed with the embryo transfer as they did not want to waste the embryo or wait for another embryo to warm. In our laboratory, we measure the pH of media (7.20–7.40), CO₂ levels (5.5–6.5%), O₂ levels (4.5–5.5%), and temperature (36.5–37.5 °C) every day, and as far as we can ascertain, there were no irregularities. We speculate that this development was neither laboratory-specific nor patient-specific.

It is interesting to note that a human blastocyst has been observed to cleave spontaneously into two complete blastocysts, both with ICM and trophectoderm components during the hatching process in vitro, a phenomenon that would also result in a DD pregnancy (Behr & Milki, Reference Behr and Milki2003; Shibuya & Kyono, Reference Shibuya and Kyono2012; Van Langendonckt et al., Reference Van Langendonckt, Wyns, Godin, Toussaint-Demylle and Donnez2000). Massip et al. (Reference Massip, Mulnard, Vanderzwalmen, Hanzen and Ectors1982) confirmed this fact in cows using cinematographic and morphometric analysis.

Monozygotic twinning with blastocyst transfer is typically monochorionic, resulting from the division of the ICM within a single trophectoderm (Chida, Reference Chida1990; Milki et al., Reference Milki, Hinvkley, Grumet and Chitkara2001; Mio, Reference Mio2008). When we diagnose DD twins under ultrasonography at 7–11 weeks gestation, we must discriminate between (1) monozygotic DD twins (two half-blastocysts, both with ICM and trophectoderm components from a SET), (2) dizygotic DD twins (a SET with a natural pregnancy: Kyono et al., Reference Kyono, Nakajo, Doshida, Toya and Araki2009; Sugawara et al., Reference Sugawara, Hukuchi, Maeda, Komaba and Araki2010; Van der Hoorn et al., Reference Van der Hoorn, Helmerhorst, Claas and Scherjon2011), and (3) dizygotic DD twins (a two-embryo transfer). In MD twins, we must discriminate between (1) monozygotic MD twins (a SET) and (2) dizygotic MD twins (dual-embryo transfer: blood chimerism; Hackmon et al., Reference Hackmon, Jormark, Cheng, Green and Divon2009; Kuhl-Burmeister et al., Reference Kuhl-Burmeister, Simeoni, Weber-Matthiesen, Milde, Herwartz, Neppert and Suttorp2000; Machin, Reference Machin2009; Miura & Niikawa, Reference Miura and Niikawa2005; Souter et al., Reference Souter, Kapur, Nyholt, Skogerboe, Myerson, Ton and Glass2003; Williams et al., Reference Williams, Wallace, Drury, Kipersztok, Edwards, Williams and Zori2004). We can discriminate by histochemical analysis of placenta, chromosome analysis (blood, skin biopsy specimen, umbilical cord tissue, buccal smear), blood type, enzyme polymorphisms, and human leukocyte antigen types.

The vascular connections can also be expected to lead to the sharing of hematopoietic stem cells, microchimerism between dizygotic twins, and mosaicism if genetic discordance arises between the monozygotic twins. More fetal–maternal microchimerism takes place with complications of pregnancy, such as pre-eclampsia or bleeding, which typically has been seen in twin pregnancies. However, Jang et al. (Reference Jang, Jung, Kim, Park and Kim2010) reported a very rare case of blood chimerism in dichorionic placentas. It is very important for reproductive doctors and obstetricians to exchange information in cases of twin pregnancy.

One possible reason for the common belief that embryo splitting occurs in the two-cell stage is that it is widely known that we can produce monozygotic DD twins by splitting an embryo artificially at the two-cell stage. For example, monozygotic twins have been produced by separation of the blastomeres of two-cell embryos in sheep (Willadesen, Reference Willadesen1979), cows (Willadesen et al., Reference Willadesen, Lehn-Jensen, Fehilly and Newcomb1981), and mice (Tsunoda & McLaren, Reference Tsunoda and McLaren1983). However, natural splitting in two-cell embryos has never been observed. We have shown that, without this technique, natural splitting of embryos leading to monozygotic DD twins occurs during the blastocyst stage after day 5 post-fertilization.