Prevalence and risk factors of zygotic splitting after 937 848 single embryo transfer cycles

Abstract

STUDY QUESTION

What is the prevalence of multiple pregnancy with zygotic splitting after single embryo transfer (SET)?

SUMMARY ANSWER

The prevalence of multiple pregnancy with zygotic splitting after SET was 1.36%.

WHAT IS KNOWN ALREADY

In 2008, the Japan Society of Obstetrics and Gynaecology (JSOG) recommended the adoption of SET to reduce multiple births. Since then, to improve the clinical pregnancy rate, elective SET using blastocyst transfer and frozen-warmed ET has increased. Blastocyst culture and zona pellucida manipulation, including ICSI and AH, have been widely reported as risk factors for monozygotic twinning. However, all these studies may have included cases with dizygotic pregnancies produced by a transferred embryo and a spontaneous conception.

STUDY DESIGN, SIZE, DURATION

A retrospective observational study was performed, based on 937 848 SET cycles in registered ART data from the JSOG between 2007 and 2014. The study was approved by the Registration and Research Subcommittee of the JSOG and Juntendo University Ethics Committee.

PARTICIPANTS/MATERIALS, SETTING, METHODS

To identify possible factors affecting the prevalence of zygotic splitting, we identified pregnancies, in which the number of foetuses exceeded the number of gestational sacs (GSs), to restrict our analysis to ‘true’ zygotic splitting. Multiple logistic regression analysis was performed using singleton pregnancy after SET, as control. P < 0.05 was considered statistically significant.

MAIN RESULTS AND THE ROLE OF CHANCE

Fresh and frozen-warmed SET produced 276 934 clinical pregnancies (29.5%/SET), including 4310 twins (1.56% of pregnancies) and 109 triplets (0.04% of pregnancies). Based on sex analysis of dichorionic twins after SET, the prevalence of multiple pregnancy with zygotic splitting was 1.36%. Statistical analysis revealed that compared to singleton pregnancies zygotic splitting pregnancies were associated with frozen-warmed ET cycles (odds ratio [OR] = 1.34; 95% CI: 1.16–1.55), blastocyst culture (OR = 1.79; 95% CI: 1.54–2.09) or AH (OR = 1.21; 95% CI: 1.08–1.35). In fresh ET cycles, the prevalence rate of zygotic splitting pregnancy after single blastocyst transfer was significantly higher than that after SET cycles with cleavage embryos (OR = 2.20; 95% CI: 1.83–2.66). However, no significant difference in ovarian stimulation and fertilization methods was recognized.

LIMITATIONS, REASONS FOR CAUTION

In the current Japanese ART registry system, data regarding frozen-warmed ET do not include information about ovarian stimulation and fertilization methods. Registration for AH only began in 2010. There is no way of validating if data submitted by clinics is correct

WIDER IMPLICATIONS OF THE FINDINGS

Clinicians should consider whether to counsel couples about the small increase in the risk of zygotic splitting associated with some embryo manipulations.

STUDY FUNDING/COMPETING INTEREST(S)

No external funds were used for the study. The authors have no competing interests to declare.

TRIAL REGISTRATION NUMBER

None.

assisted reproductive technology, single embryo transfer, zygotic splitting, multiple pregnancy, monozygotic twinning, monozygotic triplet, blastocyst culture, assisted hatching, frozen-warmed embryo transfer

Introduction

In Japan, since the first IVF baby was born in 1983, the number of ART cycles has increased each year and in 2015 more than 420 000 cycles were done resulting in 51 000 neonates, so that 1 in 19.7 of all babies born in Japan were conceived by ART (Irahara et al., 2017; Saito et al., 2018). Behind the spread and development of this technique, the Japan Society of Obstetrics and Gynaecology (JSOG) has promoted the establishment of safe pregnancy and delivery notably a decrease in the number of multiple conceptions after ART. JSOG recommended single embryo transfer (SET) in 2008 ahead of other countries (Hazekamp et al., 2000). Subsequently, the proportion of cycles using SET increased from 52.2% in 2007 to 80.1% in 2015 accompanied by a decrease in the prevalence of multiple pregnancy caused by ART from 10.7 to 3.2% (Takeshima et al., 2016; Saito et al., 2018). Perinatal outcomes improved with a decrease in the critical obstetric risks due to multiple conception, including preterm birth, low birth weight and small size for gestational age (Takeshima et al., 2016). However, there was concern that use of SET would produce low pregnancy rates, leading to low cost efficiency. To improve the clinical pregnancy rate using a single embryo, elective SET with selection of the most viable embryo from a larger number of embryos has expanded (Takeshima et al., 2016).

Yet, even use of SET produced multiple pregnancies. Nakasuji et al. (2014) reported a 1.4% prevalence of monozygotic twinning after SET in 2010 in Japan. Many reports showed that the 1.01–2.24% prevalence rate of monozygotic twinning after SET is higher compared to 0.40–0.45% of live births after spontaneous conception (MacGillivray, 1986; Derom et al. 1987; Kawachiya et al., 2011; Nakasuji et al., 2014; Kanter et al., 2015).

The mechanism of zygotic splitting has been debated, based on numerous studies. Potential risk factors include manipulations of the zona pellucida, such as assisted hatching (AH), ICSI and embryo biopsy, which can lead to herniation of blastomeres through the zona and embryo splitting during blastocyst expansion (Alikani et al., 1994; da Costa et al., 2001; Skiadas et al., 2008; Saravelos et al., 2016). The most consistent risk factor is blastocyst culture. Extended exposure of the blastocyst weakens the intracellular bonding in the inner cell mass, leading to zygote splitting (Steinman and Valderrama, 2001; Menezo and Sakkas, 2002; Cassuto et al., 2003). Some investigations of pregnancy outcomes after SET, indicate that blastocyst culture increased the risk of zygotic splitting compared to SET, using cleavage stage embryos; however, other reports observed no relationship between monozygotic splitting and blastocyst culture (Moayeri et al., 2007; Papanikolaou et al., 2010; Osianlis et al., 2014; Vega et al., 2018). Further studies have shown that genetic factors, ovarian stimulation, embryo quality and maternal age to be potential risk factors, and that multiple influences may affect embryo division (Schachter et al., 2001; Kallen et al., 2002; Steinman, 2003; Knopman et al., 2014; Otsuki et al., 2016).

The ‘true’ zygotic splitting rate remains unknown, because multiple pregnancy may include multi-zygotic pregnancy after SET combined with sexual intercourse. Same-sex dizygotic twins cannot be distinguished from dichorionic monozygotic twins in the pregnancy and even postpartum (Blickstein et al., 2003). Furthermore, when detecting a gestational sac (GS)—like structure using ultrasound, we cannot prove it as a GS. Therefore, the ‘true’ risk factors of zygotic splitting remain unclear. We examined the ‘true’ prevalence and risk factors for zygotic splitting after SET.

Materials and Methods

Japanese ART national registry data

In Japan, JSOG authorises the facilities that can provide ART treatment, and imposes a duty to report treatment procedures and outcomes. More than 99% of ART treatment cycles were entered in this registry since 2007. The JSOG implemented an online cycle-based registry system (individual surveys) and reported on the analysed ART data annually (Irahara et al., 2017; Saito et al., 2018). The cycle-based registry form has been described previously (Irahara et al., 2017). To analyse the prevalence and risk factors of zygotic splitting rate in Japan, we extracted the data below from the registered ART database between 2007 and 2014; patient age at initiation of ART cycle; pregnancy history; causes of infertility; methods of controlled ovarian stimulation; transferred embryos, such as frozen-warmed and fresh embryos; fertilization procedures, including conventional IVF and ICSI; number and grade of transferred embryos; AH, number of GSs and foetuses, courses of pregnancy and number and findings of live born infants. Registration of AH began in 2010; thus, the AH data were analysed between 2010 and 2014. Regarding controlled ovarian stimulation and fertilization procedures, no data on frozen-warmed embryo transfer (ET) cycles are available. Therefore, we analysed the data in fresh ET cycles. Controlled ovarian stimulation includes ‘natural’, ‘clomiphene citrate’, ‘clomiphene citrate + hMG or FSH’, ‘hMG or FSH’, ‘GnRH agonist + hMG or FSH’, ‘GnRH antagonist + hMG or FSH’, ‘others’ and ‘hormone replacement cycle’. Therefore, for statistical analysis, we regrouped data into two groups as follows: natural or only oral medicine group, including clomiphene citrate, and gonadotropin injection group. ‘Others’ was excluded in this study.

Data selection

Of 2 167 159 ART treatment cycles, SET, defined as one ET in one cycle, was chosen in 937 848 cycles (Fig. 1). After SET treatment, 276 934 clinical pregnancies were detected (29.5%/SET). We excluded 2330 ectopic pregnancies and recruited 274 604 intrauterine pregnancies. As shown in Fig. 1, we divided the data into 12 groups based on the number of GSs and foetuses, except for 6138 pregnancies with no information about GS number ‘True’ zygotic splitting is defined as the pregnancies in which the number of foetuses exceeded the number of GSs. To identify potential risk factors of zygotic splitting, we compared the 1872 ‘true’ zygotic splitting pregnancies, including 1808 monozygotic twins, 37 monozygotic triplets and 17 dizygotic triplets, with 232 992 ‘true’ singleton pregnancies. There was no information about ovarian stimulation and fertilization methods in frozen-warmed ET cycles or about AH before 2009. Therefore, to analyse risk factors of ‘true’ zygotic splitting, we selected two groups; fresh ET cycles from 2007 to 2014 and all ET cycles from 2010 to 2014.

Figure 1

Prevalence of multiple pregnancy after single embryo transfer (SET) cycles. Of 2 167 159 treatment cycles from 2007 to 2014, 937 848 cycles of SET had been performed. Of 276 934 clinical pregnancies, one, two and three gestational sacs were detected after SET in 265 892 (96.0% of pregnancies), 2519 (0.9% of pregnancies) and 55 (0.02% of pregnancies) pregnancies, respectively.

Statistical analyses

All statistical analyses were performed using Statistical Analysis System ver.9.4 (SAS Institute, Cary, NC, USA). To identify possible factors affecting the prevalence of zygotic splitting, multiple logistic regression analysis was performed using singleton pregnancy after SET as control. Potential clinical risk factors for zygotic splitting pregnancy were examined in the multivariate model, including age, ovarian stimulation (reference: natural or only oral medicine), fertilization method (reference: conventional IVF) and embryo stage (reference: cleavage embryo) in fresh ET cycles (Table II) and age, fresh or frozen embryo (reference: fresh embryo), embryo stage (reference: cleavage embryo) and AH (reference: no) in all ET cycles (Table III). The level of significance was defined as P < 0.05. Each pregnancy was included in the analyses as an independent event. Odds ratios (ORs) and 95% CI were determined. The differences in the characteristics between the true singleton and the true zygotic splitting were evaluated using the non-paired t test for continuous variables and the Chi squared test for categorical variables.

Ethical approval

This study was approved by the local ethical committee of Juntendo University, Faculty of Medicine (No. 16-159) and the Registration and Research Subcommittee of the JSOG Ethics Committee (2016-48), and all data analysed in this study were provided by the JSOG.

Results

Prevalence and prognosis of multiple pregnancy after SET

Of 276 934 clinical pregnancies after SET, one, two and three GSs were detected in 265 892 (96.0%), 2519 (0.9%) and 55 (0.02%), respectively (Fig. 1). The prevalence of twin pregnancy was 1.56% (4310 pregnancies), including 1808 pregnancies with one GS and two foetuses and 2502 pregnancies with two GSs and no to two foetuses. The prevalence of triplets was 0.04% (109 pregnancies), including 54 pregnancies with one to two GSs and three foetuses and 55 pregnancies with three GSs and no to three foetuses. Total multiple pregnancy rate after SET was 1.60%. To confirm the prognosis of multiple pregnancy, we reclassified the cases, based on the number of foetuses (Figs 2 and 3). When live birth was defined as at least one baby born, the live birth rate in the twins with one GS (monochorionic twin pregnancies) was significantly lower (70.4%, 1273/1808 twins), compared to twins with two GSs (dichorionic twins, 81.0%, 1398/1726 twins, P < 0.0001, Fig. 2). The live birth rate in the triplets with one GS (monochorionic triplets) was lowest (45.7%, 37/81 triplets) among triplets with one to three GSs; however, there was no significant difference (P = 0.061, Fig. 3). Monozygotic monochorionic twins and triplets were adverse prognoses of pregnancy.

Figure 2

Prognosis of twins after single embryo transfer (SET) cycles. Of 4310 sets of twins after SET, 3534 sets of twins with two foetuses included 1808 twins with one gestational sac (GS) (monochorionic twin) and 1726 twins with two GSs (dichorionic twin).

Figure 3

Prognosis of triplets after single embryo transfer (SET) cycles. Of 109 sets of triplets after SET, 81 sets with three foetuses included 37 triplets with one gestational sac (GS) (monochorionic triplet), 17 with two GSs (dichorionic triplet) and 27 with three GSs (trichorionic triplet).

Prevalence of zygotic splitting

The ‘true’ prevalence of zygotic splitting remains unknown, because multiple pregnancies after SET include not only monozygotic, but also multi-zygotic pregnancies derived from the transferred embryo combined with sexual intercourse. To identify the prevalence of zygotic splitting, we focused on the sex of twin babies from the pregnancies with two GSs (dichorionic twins). Of 1202 sets of babies, same-, different- and unknown-sex were noted in 979, 150 and 73 sets, respectively, and the prevalence of different-sex dizygotic twins was 13.3% (150/1129 sets), suggesting that the prevalence of dizygotic twins after SET, including same-sex dizygotic twins based on Weinberg’s differential rule, was 26.6% (Weinberg, 1901; Fellman and Eriksson, 2006). Therefore, the predictive prevalence of monozygotic dichorionic twins was 73.4%. Assuming that 109 triplets and 1808 twins with one GS resulted from zygotic splitting, the prevalence of zygotic splitting after SET can be calculated as follows:

(109 triplets + 1808 twins with one GS + 2519 twins with two GSs × 73.4%)/276 934 pregnancies = 1.36%; i.e. we estimate zygotic splitting 1.36% of transferred embryos in Japan.

Risk factors of zygotic splitting

The possible risk factors underlying the increased rate of zygotic splitting remains controversial. In our results of sex analysis of dichorionic twins, one-fourth of dichorionic twins after SET were not monozygotic pregnancies. An intrauterine white ring finding without foetus heartbeat by ultrasound sonography cannot be proved as a GS. Therefore, to analyse the risk factors of zygotic splitting, we recruited 232 992 pregnancies with one GS and one foetus as ‘true’ singleton and 1862 pregnancies with one GS and two or three foetuses and two GSs and three foetuses as ‘true’ zygotic splitting pregnancies. Characteristics of the patients in true singleton and zygotic splitting groups from all cycles are shown in Table I. The patients with zygotic splitting pregnancies were younger and less often diagnosed as having unexplained infertility, compared to those with singleton pregnancies (P = 0.003 and 0.046, respectively).

Table I

Patients’ characteristics in true singleton and zygotic splitting groups. Data are n (%) unless stated otherwise.

All embryo transfer cycles from 2007 to 2014	True singleton n = 232 992	True zygotic splitting n = 1862	P-value
Age (mean ± SD)	35.3 ± 4.0	35.1 ± 4.0	0.003a
Causes of infertility
Tubal factor	43,228 (18.6)	362 (19.4)	0.341b
Endometriosis	16,850 (7.2)	126 (6.8)	0.467b
Anti-sperm antibody	1723 (0.7)	15 (0.8)	0.845b
Male factor	69,770 (29.9)	580 (31.1)	0.269b
Unexplained infertility	100,528 (43.1)	760 (40.8)	0.046b
Others	34,868 (15.0)	260 (14.0)	0.240b

Statistical significant values are highlighted in bold.

^aStudent’s t test.

^bFisher’s extract test.

In fresh SET cycles from 2007 to 2014, patients with zygotic splitting pregnancies had more blastocyst cultures, compared to those with singleton pregnancies (OR = 2.20; 95% CI: 1.83–2.66). However, there were no significant differences in ovarian stimulation and fertilization methods (Table II). Whereas, in all SET cycles from 2010 to 2014, multiple logistic regression analysis revealed that frozen-warmed ET (OR = 1.34; 95% CI: 1.16–1.55), blastocyst culture (OR = 1.79; 95% CI: 1.54–2.09) and AH (OR = 1.21; 95% CI: 1.08–1.35) increased zygotic splitting rate compared to true singleton pregnancies (Table III).

Table II

Multivariable risk factors of zygotic splitting pregnancy in fresh embryo transfer cycles. Data are n (%) unless stated otherwise.

Fresh embryo transfer cycles from 2007 to 2014	True singleton n = 67 830	True zygotic splitting n = 526	Odds ratio (95% CI)b
Age (per year)	35.1 ± 3.9	34.8 ± 4.0	0.99 (0.97–1.02)
Ovarian stimulationa
Natural or only oral medicine	14 380 (23.3)	92 (19.5)	Reference
Gonadotropin injection	47 281 (76.7)	381 (80.5)	0.99 (0.78–1.25)
Fertilization method
Conventional IVF	24 686 (36.4)	196 (37.3)	Reference
ICSI	43 134 (63.6)	330 (62.7)	0.92 (0.77–1.11)
Embryo stage
Cleavage	44 295 (65.3)	239 (45.4)	Reference
Blastocyst	23 535 (34.7)	287 (54.6)	2.20 (1.83–2.66)

In odds ratios, bold indicates significant difference.

^aNumber of patients in true singleton and true zygotic splitting groups is 61 661 and 473, respectively.

^bThe logistic regression model included age, ovarian stimulation, fertilization method and embryo stage.

Table III

Multivariable risk factors of zygotic splitting pregnancy in all embryo transfer cycles. Data are n (%) unless stated otherwise.

All embryo transfer cycles from 2010 to 2014	True singleton n = 191 585	True zygotic splitting n = 1508	Odds ratio (95% CI)b
Age (per year)	35.5 ± 4.0	35.2 ± 4.0	0.98 (0.97–0.99)
Fresh or frozen embryo
Fresh	49 802 (26.0)	367 (24.3)	Reference
Frozen-warmed	141 783 (74.0)	1141 (75.7)	1.34 (1.16–1.55)
Embryo stagea
Cleavage	52 569 (27.4)	284 (18.8)	Reference
Blastocyst	138 951 (72.5)	1224 (81.2)	1.79 (1.54–2.09)
Assisted hatching
No	92 544 (48.3)	645 (42.8)	Reference
Yes	99 041 (51.7)	863 (57.2)	1.21 (1.08–1.35)

In odds ratios, bold indicates significant difference.

^aNumber of patients in true singleton group is 191 520.

^bThe logistic regression model included age, ovarian stimulation, fertilization method and embryo stage.

Transition of prevalence and risk factors of zygotic splitting pregnancy

Since the recommendation in 2008 of SET, to improve pregnancy outcomes with single embryos, elective SET has often been used with embryo manipulation, such as frozen-warmed ET, blastocyst culture and AH. However, our results showed that these procedures were associated with increased risk of zygotic splitting. Therefore, we confirmed transition of performance rates of frozen-warmed ET, blastocyst culture and AH and prevalence of multiple pregnancy after SET (Supplementary Fig. S1). All three embryo manipulations had increased every year; however, prevalence of twin as well as triplet pregnancies after SET had not elevated.

Discussion

Japan performs the most IVF cycles worldwide. To aim at a safe live birth for every baby, the SET rate reached 80% of all ET cycles in 2015 (Saito et al., 2018). Therefore, to our knowledge, this is the largest study to analyse zygotic splitting after SET and the first to investigate in detail the number of GSs and foetuses in multiple pregnancies after SET. Furthermore, this is the first report we know of to analyse potential risk factors of zygotic splitting using the data of ‘true’ singleton and ‘true’ zygotic splitting pregnancies after SET.

The chorionicity of multiple pregnancy supposedly depends on the timing of embryo or zygote division. In spontaneous pregnancy, three types of monozygotic twining were identified: (i) monochorionic monoamniotic twins (splitting of the embryo 1–2 weeks after fertilization, only 1–2% of conceptions); (ii) monochorionic diamniotic twins (division at the blastocyst stage, 70–75% of conceptions); and (iii) dichorionic diamniotic twins (division immediately after fertilization, 20–30% of conceptions) (Aston et al., 2008; Sobek et al., 2016). When focusing on twins after SET in Fig. 2, the number of dichorionic twins (two GSs and two foetuses, 1726 sets; 48.8% of twins) was comparable to that of monochorionic twins (one GS and two foetuses, 1808 sets, 51.2% of twins). If 26.6% of 1726 dichorionic twins are dizygotic based on analysis of sex of dichorionic twins, the predictive monozygotic dichorionic twin rate is calculated at 41.2%, which is still higher compared to that in spontaneous conception at 20–30%. In most of studies of monozygotic twinning after spontaneous pregnancy, the live birth data were analysed, but not early ultrasound findings; therefore, our outcomes must be accurate and important in investigation of chorionicity. Furthermore, monozygotic dichorionic twinning was considered when an embryo divides after fertilization, but a developed blastocyst can split into two according to some reports (Van Langendonckt et al., 2000; Zou et al., 2018). Our SET results also demonstrated that the division of many embryos occurred after blastocyst stage. The procedures in ART treatment may increase the chance of complete blastocyst division as multichorionic pregnancy.

Prevalence of multiple pregnancy after SET was 1.60%, including 1.56% of twins and 0.04% of triplets in our data. This rate is comparable to that of other studies of monozygotic twinning after SET, including 1.4% in the Japanese ART data in 2010 (Blickstein et al., 2003; Kawachiya et al., 2011; Nakasuji et al., 2014; Kanter et al., 2015; Vega et al., 2018). The prevalence of monozygotic twinning after SET is obviously higher compared to the 0.40–0.45% universal prevalence following non-stimulated in vivo conception (MacGillivray, 1986; Derom et al. 1987). This difference in prevalence may result from missing ultrasound findings during early pregnancy after spontaneous conception as well as the impact of ART procedures.

Many studies of multiple pregnancy after SET have shown that possible risk factors underlying the increased rate of zygotic splitting included extended culture of embryos and zona pellucida manipulation, such as AH, ICSI and embryo biopsy (Alikani et al., 1994; da Costa et al., 2001; Skiadas et al., 2008; Luke et al., 2014; Saravelos et al., 2016). However, all past studies involved cases with potential dizygotic pregnancies produced by a transferred embryo and spontaneous conception. We first recruited the pregnancies with ‘true’ zygotic splitting, such as twins or triplets in which the number of foetuses exceeded the number of GSs. Patients in the ‘true’ zygotic splitting group were younger and had far fewer women with unexplained infertility (Table I). Some studies also showed that younger maternal age is associated with monozygotic twinning (Knopman et al., 2014; Franasiak et al., 2015). Multiple logistic regression analysis showed a higher prevalence of zygotic splitting with younger age in all ET cycles, but not in only fresh cycles; thus, the result was controversial (Tables II and III). Multiple logistic regression analysis showed three procedures; blastocyst transfer, AH and frozen-warmed ET (Tables II and III). Blastocyst culture is the highest OR in three potential risk factors. Embryo selection using a computer-automated time-lapse image analysis test and cleavage-stage ET may be one of solutions to decrease the risk of embryo division (Conaghan, et al., 2013). And freezing and warming of an embryo was first detected as a risk factor. The freeze-all policy using frozen-warmed ET has increased as ART strategy with low risk of complications, such as multiple pregnancies and ovarian hyper stimulation syndrome. This strategy can also avoid the need for ET under non-physiological conditions; thus, the rate of frozen-warmed ET has increased for elective SET (Maheshwari et al., 2011; Kato et al., 2012; Roque et al., 2015). Interestingly, ICSI did not increase the zygotic splitting rate, according to our data; thus, a possible risk factor for zygotic splitting may be only embryo manipulation, and not oocyte manipulation (Table II). After elective SET recommendation in 2008, most Japanese reproductive doctors have made an effort to decrease the prevalence of multiple pregnancy and increase the pregnancy rate with a single embryo; therefore, the rates of blastocyst transfer, AH and frozen-warmed ET increased each year (Supplementary Fig. S1A). Monochorionic twins and triplets have a higher risk of foetus mortality (Figs 2 and 3). Elective SET might increase the prevalence of multiple pregnancies with zygotic splitting iatrogenically. To confirm this, we analysed transition of prevalence of multiple pregnancy after SET; however, the prevalence of twins as well as triplets after SET has not been elevated (Supplementary Fig. S1B). The stress of embryos by artificial procedures and culture condition may have decreased due to improvement of the technology and media, leading to decreasing the risk of zygotic splitting. In fact, OR of blastocyst culture in ET cycles from 2010 to 2014 was lower than it in ET cycles from 2007 to 2014 (1.79 and 2.20, respectively, Tables II and III). Yet the reason is unknown.

Monozygotic triplets after ART procedures occur very rarely. A systematic review of the literature in 2016 showed that a total of 22 sets of monozygotic triplet or quadruplet pregnancies after ART treatment worldwide (Saravelos et al., 2016). Surprisingly, the Japanese ART national registry database included 37 monochorionic, 17 dichorionic and 27 trichorionic triplets. Additional investigations are needed to examine the triplets after use of SET.

This study has some limitations. First, in the current Japanese ART registry system, the data regarding frozen-warmed ET do not include information about ovarian stimulation and fertilization methods. Registration for AH began in 2010, so there are no data before this. There are also no data on embryo biopsy, such as pre-implantation genetic testing. Second, JSOG manages the accuracy of the Japanese ART national registry data, however, it is impossible to prove the validity. Third, this is observational study, therefore, the casual relationship between zygotic splitting and ART procedures cannot be scientifically demonstrated.

In conclusion, based the Japanese ART data between 2007 and 2014, embryo manipulations using the procedures of blastocyst transfer, AH and frozen-warmed ET were potential risk factors for zygotic splitting. Elective SET has increased worldwide; however, the prevalence of zygotic splitting pregnancy has not increased. Therefore, there is no need to avoid embryo manipulations, such as blastocyst culture, to select the single most viable embryo.

Acknowledgements

We wish to thank all Japanese reproductive doctors that registered the data in Japanese ART registry and contributed to decrease multiple pregnancies using SET procedure. We also thank the JSOG for providing the ART data in 2007–2014.

Authors’ roles

Y.I.: formal analysis, investigation, visualization, writing. K.K.: conceptualization, formal analysis, investigation, methodology, resources, supervision, visualization, writing. A.O.: investigation, discussion. S.Y.: discussion. S.I.: discussion. S.N.: formal analysis. A.I.: project administration. S.T.: project administration.

Funding

No external funding was used for this study.

Conflict of interest

All authors declare that they have no conflict of interest.