1 Division of Reproductive Health, National Center for Chronic Disease Prevention and Health Promotion, Centers for Disease Control and Prevention, Atlanta, GA, USA
2 To whom correspondence should be addressed. Email: vwright{at}cdc.gov
![]() |
Abstract |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Key words: blastocyst/extended culture/IVF/monozygotic/twinning
![]() |
Introduction |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Multiple embryo transfer is still the norm in ART treatment because there is concern that limiting the number of embryos transferred might reduce the likelihood of achieving pregnancy. Many advocate developing strategies to improve selection of high-quality embryos, thus maximizing the potential for implantation while limiting the number of embryos transferred (Racowsky et al., 2000). With the development of commercially available culture media, an increasingly feasible and popular treatment option is to extend embryo culture and transfer embryos at the blastocyst stage. This method allows the evaluation of embryo quality and the selection of fewer, high-quality embryos for transfer which, in turn, should reduce the number of multiple births.
Several randomized studies have demonstrated high efficacy with blastocyst transfer in comparison wirh the transfer of cleavage stage embryos (Gardner et al., 1998; Van der Auwera et al., 2002
; Frattarelli et al., 2003
). However, a recent meta-analysis suggests no difference between blastocyst and cleavage stage transfer in implantation and pregnancy rates (Blake et al., 2004). Moreover there is an emerging concern that extending embryo culture and transfer to the blastocyst stage may increase the risk for embryo splitting and subsequent monozygotic (MZ) multiple-gestation pregnancies. Several case studies, case series and small comparative studies suggest an association between blastocyst transfer and MZ twinning (Rijnders et al., 1998
; Peramo et al., 1999
; Behr et al., 2000
; Van Langendonckt et al., 2000
; da Costa et al., 2001
; Sheiner et al., 2001a
,b
; Milki et al., 2003
). MZ multiple gestations are of particular concern because morbidity and mortality rates are even higher for MZ twins than for dizygotic (DZ) twins (Deromet al., 1991). We used a population-based data set of IVF/embryo transfer procedures to perform a casecontrol analysis of the association between day of embryo transfer and MZ twinning.
![]() |
Materials and methods |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
We limited our analysis to IVF/embryo transfer procedures performed in 1999 and 2000 in which the embryos were freshly fertilized with the patient's oocytes (i.e. non-frozen, non-donor). Of 110 381 such transfer procedures, we excluded small proportions of procedures for which the number of embryos transferred was either missing (n=134) or implausible (n=6); the data necessary to calculate the day of embryo transfer were missing (n=32) or resulted in implausible values [i.e. <2 (n=884) or >6 (n=257)]; and where the patient age was aged either <20 years (n=17) or >44 years (n=715). Among the remaining 108 336 transfer procedures, there were 41 534 clinical pregnancies. We excluded those pregnancies missing data on the number of fetal hearts on ultrasound (n=2336). Thus, our final sample consisted of 39 198 pregnancies.
Definitions of cases and controls
Cases were defined as pregnancies in which the number of fetal hearts on ultrasound exceeded the number of embryos transferred. We considered these pregnancies to contain at least one set of MZ twin fetuses. We acknowledge that such pregnancies might not entirely be MZ multiple gestations; our definition allows for the inclusion of pregnancies with a mix of MZ and DZ fetuses. We identified 226 cases.
We selected two control groups: (i) IVF/embryo transfer singleton pregnancies (one fetal heart); and (ii) other IVF/embryo transfer multiple-gestation pregnancies (2 fetal hearts on ultrasound, but the number of fetal hearts was less than or equal to the number of embryos transferred). These control groups included 15 092 and 23 880 pregnancies, respectively. Having two control groups was desirable because each group had a possible limitation. First, comparison of our cases of MZ pregnancies with the control group of other multiple-gestation pregnancies could be biased toward the null hypothesis of no association between presumed blastocyst culture and MZ twinning. The other multiples control group could contain some MZ and DZ multiple pregnancies, a misclassification that would be independent of day of embryo transfer, i.e. a non-differential misclassification. On average, non-differential misclassification tends to attenuate associations between two variables (Schlesselmann, 1982
). In contrast, comparison of MZ multiples with a control group of singleton pregnancies might produce a spurious association. Previous studies suggest that blastocyst culture may increase implantation and pregnancy rates; if true, such a comparison might overstate an association with blastocyst culture if the MZ case pregnancies were actually a mix of MZ and DZ fetuses. This is likely to be the case with many case pregnancies in this study.
Exposure of Interest
Day of embryo transfer was calculated using dates of oocyte retrieval and embryo transfer. All embryo transfers that occurred on day 5 were defined as presumed blastocyst transfer. Data were not available to classify specifically whether blastocyst sequential culture medium was used, nor to determine the developmental stage of the embryos transferred.
Embryo transfers that occurred on day 3 (71%) were considered the current standard of care and were thus considered the reference category for our analyses. In addition to examining day 5 transfers, we also considered cycles in which embryo transfer occurred on days 2, 4 or 6.
Data analysis
We calculated the percentage of pregnancies that were multiple gestations and the percentage with at least one MZ multiple gestation for embryo transfers conducted on days 2, 3, 4, 5 and 6. We also calculated odds ratios (ORs) and 95% confidence intervals (CIs) for the association between MZ twinning and day of embryo transfer.
We then considered the possibility of confounding or effect modification by patient and treatment factors for pregnancies resulting from a day 5 or day 3 transfer. We performed stratified analyses and multivariable logistic regression to evaluate and adjust for maternal age, number of prior ART procedures and number of prior births. We calculated adjusted ORs and 95% CIs for comparisons of case pregnancies with pregnancies from each control group. To address concerns that an association between MZ twinning and extended embryo culture was attributable to various treatment subgroups, we repeated these analyses after restricting our sample to procedures with a minimum of five oocytes retrieved; procedures with a minimum of 10 oocytes retrieved; procedures with extra embryos available and cryopreserved; procedures in which ICSI or assisted hatching techniques were not used; and procedures in which three or more embryos were transferred in a cycle. To address concerns that MZ twinning might occur through a natural conception concurrent with ART, we restricted the analyses to women with tubal infertility, a subset least likely to experience such an effect. We used Statistical Analysis Software (SAS), Version 8.2, to perform all analyses.
This study was approved by the Institutional Review Board at the CDC.
![]() |
Results |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
|
|
Women who had MZ multiple-gestation pregnancies were significantly younger than women with a singleton gestation (Table III). When compared with pregnancies in either of the control groups, MZ case pregnancies were less likely to be the result of ART procedures that included assisted hatching and were more likely to be the result of ART procedures with 10 oocytes retrieved, only one or two embryos transferred, and extra embryos cryopreserved for future use. We were unable to determine whether the difference in having embryos cryopreserved was because cases were more likely to have had extra embryos available for future use or whether some other patient or treatment factor was different between the case and control groups. Cases of MZ pregnancies were not significantly different from either control group for prior ART, prior live birth, tubal infertility or use of ICSI.
|
|
|
![]() |
Discussion |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Methodological limitations must be considered when interpreting our results. We did not have specific information on zygosity. We relied on early ultrasound data to classify cases of MZ pregnancies based on the number of fetuses relative to embryos transferred and, thus, may have missed some cases. However, we do not believe this limitation biased our comparisons. Our study compared presumed MZ case pregnancies with pregnancies in two different control groups: other multiple gestations and singleton gestations. Potential biases related to case ascertainment were expected to be in opposite directions for these control groups. Nonetheless, our results were quite comparable across the two control groups. Because we did not have specific zygosity data, the percentages of presumed MZ twinning reported in this study should not be used to assume absolute risk for MZ twining; the strength of this study is in the assessment of relative differences. Although some past investigators have assessed the absolute risk for MZ twinning based on data limited to single embryo transfer (SET), in the USA, <2% of the pregnancies reported were conceived with SET. In addition to the limitation this poses with respect to statistical power, pregnancies from SET in the USA are likely to be a very select subgroup and thus are not representative of the entire ART population.
We also lacked specific data on embryo stage at transfer and culture media. In our study, day 5 embryo transfer procedures were presumed to have been blastocyst. We cannot completely discount the possibility that some embryo transfer dates were reported incorrectly; however, misclassification was likely to have been random, and this would have biased our results toward the null hypothesis. Nonetheless, the lack of specific laboratory data on the embryo and the culture media precluded us from evaluating specific media components that may have been related to MZ twinning.
Our study was based on pregnancies rather than women, so it is possible that the women in our sample with more than one pregnancy during the 2 year period of interest were represented more than once. However, when we limited our sample to a subset of women undergoing their first ART cycle, our results remained the same.
Finally, our study subjects were not randomized. Treatment options, such as the day of embryo transfer, were based on the patient's choice and the provider's assessment. Because blastocyst transfer may be differentially performed depending on the presumed prognosis for the treatment cycle, we performed various subset analyses to determine if our results persisted within more homogeneous treatment groups; we found similar results for all. We also considered that the association documented in this study may have been attributable to high MZ twinning in only a few ART centres. We thus examined the distribution of MZ cases across clinics. MZ cases were identified from 114 ART centres. There was no difference between MZ cases identified from day 5 and day 3 embryo transfer procedures (110 cases from day 5 embryo transfers were from 66 clinics and 98 cases from day 3 transfers were from 58 clinics). Moreover, no clinic contributed an excessive number of cases.
Despite limitations, our study had many strengths. We used a population-based data set containing a large patient population, which allowed for a more thorough analysis of various patient and treatment factors than in previous studies from clinical populations. We reported 226 cases of presumed MZ twinning. The number of MZ twins reported from previous clinical studies ranged from three to eleven.
The underlying mechanism for the association between blastocyst culture and MZ twinning is unclear. One hypothesis raised is that in vitro culture conditions are related to increased hardening of the zona pellucida (Alikani et al., 1994); this could become even more problematic with extended time in culture. It is possible that spontaneous or assisted hatching of the blastocyst through a less flexible zona may increase the risk for its division. Alternatively, Cassuto and colleagues suggested that embryo culture conditions (i.e. culture media) rather than time in culture increase the risk of MZ twinning (Cassuto et al., 2003
). Other ART treatment factors that have been potentially related to MZ twinning are ovulation induction (Derom et al., 1987
) and zona manipulation such as those that occur during assisted hatching techniques (Hershlag et al., 1999
; Schieve et al., 2000
). These factors, however, would not explain the increased risk with extended culture. In our study population, nearly all case and control pregnancies included ovulation induction. Additionally, our results persisted when we excluded pregnancies from embryos that had been exposed to either assisted hatching or ICSI. Clearly, further work is needed to disentangle the mechanism by which extending culture to blastocyst creates a risk for MZ twinning.
Although our study found the highest pregnancy success rates (45%) among women who had undergone embryo transfer on day 5, it is important to note that we found a large percentage of multiple gestations (41%) among these procedures, despite the fact that they involved the transfer of fewer embryos on average (2.5 embryos). Several other studies have reported similar findings. Gorrill et al. (2001) documented a high multiple-gestation rate (40%) when only two blastocysts were transferred. Levron et al. (2002)
reported a multiple-gestation pregnancy rate of 50% for day 5 embryo transfers with an average of 2.3 embryos transferred. Finally, in a randomized study, Coskun et al. (2000)
reported a multiple-gestation rate of 38% multiples when only 2.2 embryos were transferred on day 5. Thus, our findings suggest that extended embryo culture may be associated with increased DZ multiple gestations in addition to MZ multiple gestations. Although there is great potential to reduce dizygotic multiples by limiting the number of embryos transferred, even two embryos may pose a substantial risk.
Taken altogether, the comparative studies to date and the numerous case reports of MZ twinning in blastocyst transfer pregnancies lend support to the hypothesis of an association between blastocyst transfer and MZ twinning (Rijnders et al., 1998; Peramo et al., 1999
; Behr et al., 2000
; Van Langendonckt et al., 2000
; da Costa et al., 2001
; Sheiner et al., 2001a
,b
; Milki et al., 2003
). While this and other studies also suggest that DZ multiple gestation remains a substantial problem with blastocyst culture, it is important not to minimize the risk for MZ twinning. Prevention of adverse health outcomes among ART pregnancies should not be a sequential approach. While extending embryo culture beyond day 3 holds promise of improved success with fewer embryos transferred, more research is needed to better understand the mechanisms underlying the MZ twin effect.
![]() |
Acknowledgements |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
![]() |
References |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Behr B, Fisch JD, Racowsky C, Miller K, Pool TB and Milki AA (2000) Blastocyst-ET and monozygotic twinning. J Assist Reprod Genet 17, 349351.[CrossRef][Medline]
Blake DA, Proctor M and Johnson NP (2004) The merits of blastocyst versus cleavage stage embryo transfer: a Cochrane review. Hum Reprod 19, 795807.
Cassuto G, Chavrier M and Menezo Y (2003) Culture conditions and not prolonged culture time are responsible for monozygotic twinning in human in vitro fertilization. Fertil Steril 80, 462463.[CrossRef][Medline]
Coskun S, Hollanders J, Al-Hassan S, Al-Sufyan H, Al-Mayman H and Jaroudi K (2000) Day 5 versus day 3 embryo transfer: a controlled randomized trial. Hum Reprod 15, 19471952.
da Costa AL, Abdelmassih S, de Oliveira FG, Abdelmassih V, Abdelmassih R, Naggy ZP and Balmaceda JP (2001) Monozygotic twinning and transfer at the blastocyst stage after ICSI. Hum Reprod 16, 333336.
Derom R, Vlietnck R and Derom C (1991) Perinatal mortality in the East Flanders Prospective Twin Survey. Eur J Obstet Gynaecol Reprod Biol 41, 2526.[CrossRef][Medline]
Derom C, Vlietnck R, Derom R, Van den Berghe H and Thiery M (1987) Increased monozygotic twinning rate after ovulation induction. Lancet 9, 12361238.[CrossRef]
Frattarelli JL, Leondires MP, McKeeby JL, Miller BT and Segars JH (2003) Blastocyst transfer decreases multiple pregnancy rates in in vitro fertilization cycles: a randomized controlled trial. Fertil Steril 79, 228230.[CrossRef][Medline]
Gardner DK, Schoolcraft WB, Wagley L, Schlenker T, Stevens J and Hesla J (1998) A prospective randomized trial of blastocyst culture and transfer in in-vitro fertilization. Hum Reprod 13, 34343440.[Abstract]
Gorrill MJ, Sadler-Fredd K, Patton PE and Burry KA (2001) Multiple gestations in assisted reproductive technology: can they be avoided with blastocyst transfers? Am J Obstet Gynecol 184, 14711475.[CrossRef][Medline]
Hershlag A, Paine T, Cooper GW, Scholl GM, Rawlinson K and Kvapil G (1999) Monozygotic twinning associated with mechanical assisted hatching. Fertil Steril 71, 144146.[CrossRef][Medline]
Levron J, Shulman A, Bider D, Seidman D, Levin T and Dor J (2002) A prospective randomized study comparing day 3 with blastocyst-stage embryo transfer. Fertil Steril 77, 13001301.[CrossRef][Medline]
Milki AA, Jun SH, Hinckley MD, Behr B, Giudice LC and Westphal LM (2003) Incidence of monozygotic twinning with blastocyst transfer compared to cleavage-stage transfer. Fertil Steril 79, 503506.[CrossRef][Medline]
Peramo B, Ricciarelli E, Cuadros-Fernandez JM, Huguet E and Hernandez ER (1999) Blastocyst transfer and monozygotic twinning. Fertil Steril 72, 11161117.[CrossRef][Medline]
Racowsky C, Jackson KV, Cekleniak NA, Fox JH, Hornstein MD and Ginsburg ES (2000) The number of eight-cell embryos is a key determinant for selecting day 3 or day 5 transfer. Fertil Steril 73, 558564.[CrossRef][Medline]
Rijnders PM, van Os HC and Jansen CAM (1998) Increased incidence of monozygotic twinning following the transfer of blastocyst in human IVF/ICSI. Fertil Steril 70 (Suppl 1), S15.[CrossRef]
Schieve LA, Meikle SF, Peterson HB, Jeng G, Burnett NM and Wilcox LS (2000) Does assisted hatching pose a risk for monozygotic twinning in pregnancies conceived through in vitro fertilization? Fertil Steril 74, 288294.[CrossRef][Medline]
Schlesselmann JJ (1982) Case-Control Studies: Design, Conduct, Analysis. Oxford University Press, New York.
Sheiner E, Har-Vardi I and Potashnik G (2001a) The potential association between blastocyst transfer and monozygotic twinning. Fertil Steril 75, 217218.[CrossRef][Medline]
Sheiner E, Kivilevitch Z, Levitas E, Sonin Y, Albotiano S and Har-Vardi I (2001b) Monozygotic twins following blastocyst transfer: a report of two cases. Eur J Obstet Gynecol Reprod Biol 98, 135138.[CrossRef][Medline]
US Department of Health and Human Services. Fertility Clinic Success Rate and Certification Act of 1992 [FCRSCA]. Section 2[a] of P.L. 102493 [42 U.S.C. 263 9(a)-1]. 10-24-1992.
Van der Auwera I, Debrock S, Spiessens C, Afschrift H, Bakelants E, Meuleman C, Meeuwis L and D'Hooghe TM (2002) A prospective randomized study: day 2 versus day 5 embryo transfer. Hum Reprod 17, 15071512.
Van Langendonckt A, Wyns C, Godin PA, Toussaint-Demylle D and Donnez J (2000) A typical hatching of a human blastocyst leading to monozygotic twinning: a case report. Fertil Steril 74, 10471050.[CrossRef][Medline]
Wright VC, Schieve LA, Reynolds MA and Jeng G (2003) Assisted reproductive technology surveillanceUnited States, 2000. MMWR Surveill Summ 52, 116.[Medline]
Submitted on February 24, 2004; accepted on May 5, 2004.