In humans, the vast majority of births are singleton; however, multiple births, chiefly dizygotic twins, occur with low but steady frequency. The estimated rate of twin birth in the United States is 32.2 per 1,000 births.1
In the early days of ultrasonographic imaging arose the idea that many pregnancies actually started with a double implantation, although one of the embryos died and was reabsorbed. One study from the mid-1970s estimated that over 70% of all twin gestations ended in single births.2 Subsequent reviews, however, indicated that the occurrence of twin pregnancy was heavily overdiagnosed as a result of misinterpretation of embryonic structures.3 The introduction of transvaginal ultrasonographic scanning considerably improved pregnancy monitoring, and soon the concept of high embryo loss in twin pregnancies was dismissed.4
The booming of assisted reproduction provided abundant data to study multiple pregnancies. An early study found a higher “take-home baby” rate in pregnancies with multiple implantations,5 and further studies firmly established that, indeed, the incidence of embryo loss is lower in twin than in singleton pregnancies.6 – 13
We studied the basis for the higher live birth rate per implanted embryo, observed in twin pregnancies, and examined the effect of age, presumably the most determinant factor in human reproductive success.14 – 16 We considered two forms of embryo interaction leading to a higher live birth rate: “collaboration” and “assistance.” In collaboration, two embryos that would fail to survive on their own, when implanting together, help each other out, surviving both to birth. In assistance, an embryo that would fail on its own, when implanted along with a sibling fully fit to survive, receives support from it and survives too.
MATERIALS AND METHODS
This is a retrospective study based, entirely, on records from the Assisted Reproduction Unit database at the Hospital Clinic of Barcelona, University of Barcelona, Spain. It includes data from patients who underwent treatment for assisted conception at the center through either in vitro fertilization or intracytoplasmic sperm injection from January 1991 to January 2007. The study received approval from the scientific and ethical committees responsible for supervising all research activities at the hospital.
The study is based, solely, on cases in which the patient's own fresh oocytes were used in the treatment. Clinical implantation was diagnosed at 6–7 weeks of gestation by transvaginal ultrasonographic imaging as the visualization of gestational sacs containing an embryo. Cases reported here are gestations with a single chorionic sac with an embryo (singleton pregnancy) and two chorionic sacs with an embryo each (twin pregnancy). Ectopic pregnancies, anembryonic sacs, cases of monozygotic twin pregnancies (identified as monochorionic twins), and induced embryonic reductions were all excluded from the study. Only cases in which complete and accurate records existed were regarded. The participants involved in this study were, in the vast majority (over 95%) of white origin.
Of 1,747 pregnancies originally regarded (1,213 singleton and 534 twin), several cases were excluded from the study. These include 12 monochorionic monozygotic twin pregnancies subsequently identified (seven in singleton and five in twins). In singleton pregnancies, 47 other cases were also excluded: five induced abortions, 21 anembryonic sacs, and 21 ectopic pregnancies. In twin pregnancies, six further cases were also excluded: three induced embryonic reductions, one anembryonic sac, and two ectopic pregnancies.
We assume that the intrinsic fitness of an embryo is independent of the number of implanted embryos in a pregnancy. Likewise, there is no evidence that potentially deleterious factors, either from maternal or environmental origin, affect differentially singleton and twin pregnancies. We, thus, hypothesize that the higher live birth rate observed in twin pregnancies results from some form of direct, or indirect, interaction between the cohabiting embryos. In this respect, we consider embryo collaboration and assistance.
To test the likelihood of collaboration and assistance, we compare the observed results in twin pregnancies for each possible outcome (double birth, single birth, and double loss) with those expected regarding as the null hypothesis that the probability of survival for either embryo is independent from the presence of the other (ie, there is no interaction between embryos). The expected values, assuming no interaction, are calculated based on the results obtained in singleton pregnancies. Accordingly, if in singleton pregnancies the probabilities of “live birth” and “pregnancy loss” are “p” and “q,” respectively, the expected values for each possible outcome in twin pregnancy are: “p2” for double birth, “2pq” for single birth, and “q2” for double loss.
Either hypothesis is consistent with specific predictions. According to collaboration, the number of observed double losses should be lower than expected. The hypothesis of assistance predicts a higher number of observed double births at the expense of single births. In contrast to collaboration, assistance predicts no difference between observed and expected double losses.
To study the effect of maternal age, the study population was divided into two groups according to the median age value, which produced two subsets of different age and identical size. Comparisons between any two groups were carried out using odds ratio (OR) with 95% confidence interval (CI) and χ2 with continuity correction (Yates correction). Comparisons between observed and expected values in twin pregnancies were performed using the χ2 goodness-of-fit test. Comparison of age between groups was carried out using the two-tailed Wilcoxon rank-sum test and values were expressed as mean±standard deviation.
A total of 1,159 singleton and 523 twin pregnancies were included in the analyses. Table 1 shows data on age and sterility causes of all the participants. Overall mean age was 33.88±3.5 years (median age 34 years, range 21–46 years).
Table 2 shows the overall results for all pregnancies. Altogether, there were 1,372 births, 996 single (881 from singleton and 115 from twin pregnancies) and 376 twin births. The overall rate of embryo survival to birth per implanted embryo was higher in twin pregnancies: 83% (867 of 1,046) compared with 76% (881 of 1,159) in singleton (OR 1.53, 95% CI 1.24–1.88, χ2=15.39, P<.001).
Because 76% of the singleton pregnancies ended in births, the expected probability values for each possible outcome in twin pregnancies are: .76×.76=.577 for double birth, 2×(.76×.24)=.365 for single birth, and .24×.24=.058 for double loss.
Figure 1 compares the observed and expected values for each outcome in twin pregnancies. The expected number of double losses is 30 (5.8%), whereas the observed value is 32 (6%); the expected number of single births is 191 (36.5%) and the observed 115 (22%); finally, the expected number of double births is 302 (57.7%) and the observed 376 (72%). Goodness-of-fit test reveals significant differences between expected and observed values (χ2=48.5, P<.001). There was, however, no significant difference between expected and observed cases of double loss (OR 0.95, 95% CI 0.56–1.56, χ2=0.02, P=.887).
These results support the existence of a positive interaction between embryos in twin pregnancies, leading to a higher live birth rate. The results, too, are consistent with assistance but not collaboration.
The study population was divided into two subsets according to the median age: younger than 34 years (younger group) and 34 years or older (older group). Results for each age subset are shown in Table 2. Figure 2 compares the live birth rate in singleton and twin pregnancies within and between each age subset.
The mean age in the younger subset was 30.75±2.1 years (range 21–33 years): 30.74±2.02 for singleton pregnancies and 30.77±2.28 years for twins (P=.812). The mean age in the older subset was 36.37±2.1 years (range 34–46 years): 36.49±2.14 years for singletons and 36.07±2.01 years for twins (P=.006).
In singleton pregnancies, of 479 implanted embryos, there were 391 live births and 88 losses (survival rate 81.6%). In twin pregnancies (261 cases), of 522 implanted embryos, 445 ended in live births and 77 in loss (survival rate 85.2%). This difference is not statistically significant (OR 1.3, 95% CI 0.93–1.82; χ2=2.12, P=.145).
In twin pregnancies, the expected number of double losses, assuming no interaction, is nine (3.4%) and the observed value 14 (5%). Expected single births are 78 (30%) and the observed 49 (19%). Expected double births are 174 (66.6%) and the observed 198 (76%). The goodness-of-fit test reveals significant differences between expected and observed values (χ2=16.87, P<.001). There is no significant difference between observed and expected double losses (OR 0.63, 95% CI 0.27–1.48; χ2=0.73, P=.392).
In singleton pregnancies, of 680 implanted embryos, there were 490 live births and 190 losses (survival rate 72%). In twin pregnancies (262 cases), of 524 implanted embryos, 422 ended in live births and 102 in loss (survival rate 80.5%). This difference is statistically significant (OR 1.6, 95% CI 1.2–2.1, χ2=11.12, P<.001).
In twin pregnancies, the expected number of double losses is 20 (8%) and the observed value 18 (7%). Expected single births are 106 (40%) and the observed 66 (25%). Expected double births are 136 (52%) and the observed 178 (68%). Goodness-of-fit test reveals significant differences between expected and observed values (χ2=28.26, P<.001). There is no significant difference between observed and expected double losses (OR 1.12, 95% CI 0.58–2.17, χ2=0.03, P=.862).
Live birth rate in singleton pregnancies, per implanted embryo, is considerably higher in the younger group (81.6% compared with 72%; OR 1.72, 95% CI 1.29–2.29, χ2=13.6, P<.001). For twin pregnancies, the rate of embryo survival to birth is also higher in the younger group, although to a lesser extent (85.2% compared with 80.5%; OR 1.4, 95% CI 1.00–1.93, χ2=3.77, P=.052) (Fig. 2). Although the live birth rate is superior in twin pregnancies in both age groups, the difference was higher for women 34 years and older (80.5% compared with 72%) than for women for women younger than 34 years (85.2% compared with 81.6%).
The rate of double loss in twin pregnancies is similar in younger and older women (5% compared with 7%; OR 0.77, 95% CI 0.37–1.58, χ2=0.29, P=.590). Single births are slightly higher in the older group (25% compared with 19%, OR 0.69, 95% CI 0.45–1.04, χ2=2.78, P=.095). Double births are slightly higher in the younger group (76% compared with 68%, OR 1.48, 95% CI 1.00–2.18, χ2=3.68, P=.055).
Numerous independent studies consistently sustain that live birth rate, per implanted embryo, is higher in twin than in singleton pregnancies.5 – 13 Our results provide further support of this conclusion and offer a possible explanation for it that is consistent with a model of assistance, whereby embryos that would fail as singletons survive when implanted along with a competent sibling.
Despite the robust evidence for embryo assistance, a considerable number of twin pregnancies still ends in single births (22% in our study) indicating that not all embryos can benefit from assistance. It is well established that some embryos have major flaws, chiefly aneuploidies that render them intrinsically nonviable.17,18 Other embryos, however, might have minor impeditive factors that, although precluding their survival on their own, can be overcome by cohabiting with a fully fit one.
Our findings, thus, involve the existence of embryos that are capable of reaching full development and birth but relying on the presence of a fully fit sibling and, thus, are contingent embryos. Comparative pathological studies on miscarried embryos from singleton and twin pregnancies should shed light on the differences between these contingent embryos and those intrinsically nonviable.
Embryo quality diminishes with advancing maternal age19,20 and is reflected in a higher rate of miscarriages.15 As expected, in our study, the live birth rate (for singleton and twin pregnancies) was higher in younger women. However, the difference in live birth rate between twin and singleton pregnancies was 8.5 points in women aged 34 years and older (80.5% compared with 72%), whereas in younger women, it was only 3.6 (85.2% compared with 81.6%). Accordingly, the rate of contingent embryos is higher in older women, suggesting that the contingent embryos are somehow defective. Because more than 20% of all births from assisted reproduction in Europe and the United States are multiple,21,22 establishing the nature and extent of this potential deficiency seems a major issue.
The most extensive study to date on neonatal outcome in dizygotic twin births born from assisted reproduction, based on 3,393 cases, reported major weight discordances between the newborns in 20.6% of the cases.23 In our study, because 76% of the singleton pregnancies ended in live births, we can extrapolate that, of the 376 twin births considered, 302 would be pairs of fully fit embryos and 74 pairs of fully fit and contingent embryos (accounting for 19.7% of the cases). This figure is very close to that found for the proportion of cases with major weight discordance in the study regarded above.23 Because the mean maternal ages of both studies are comparable (33.4 and 33.1 years), a possible link between both features should not be ignored.
Understanding the process involved in embryo assistance is a key issue. One study reported that, although the overall risk of spontaneous abortion is higher in singleton than in twin pregnancies, the difference lessens as gestation advances, becoming comparable after 13 weeks.6 This would involve that embryo assistance is restricted to the first trimester of development.
Some potential caveats in our study deserve consideration. The mean maternal age of the singleton group is slightly, although statistically significant, higher (34.1±3.5 compared with 33.4±3.4 years); however, both groups have the same median age (34 years). According to the latest US assisted reproduction technology report (year 2007), in autologous pregnancies, live birth rate per transfer was 43.3% for age 33 years and 43% for age 34 years.22 In our study, the live birth rate for twin pregnancies is seven points higher than in singleton (83% compared with 76%). Thus, the slight difference in mean age between the groups is unlikely to have significant relevance in the overall conclusions.
Our diagnosis of dizygotic pregnancy is based on the visualization of dichorionic embryos. Although exceptional, monozygotic twins can split, forming dichorionic embryos.24 Also, a recent review concludes that monozygotic twin pregnancies are more frequent in assisted reproduction with an incidence of 0.9% compared with 0.4% in natural pregnancies.25 In our original sample, there were 12 monochorionic monozygotic twins (0.68% of the total). Assuming that 25–30% of all monozygotic pregnancies become dichorionic,25 we would expect, in our study, approximately five dichorionic monozygotic cases at most, a figure too low to affect our overall conclusions.
Altogether, this study supports that the live birth rate, per implanted embryo, is higher in twin than singleton pregnancies and indicates that the underlying mechanism is embryo assistance. This involves the existence of contingent embryos that rely on the presence of a fully fit one to survive to birth. Assistance is more pronounced in older mothers, which strongly suggests that the contingent embryos present some deficiency. Understanding the nature of this assistance and the potential deficiency of the contingent embryos is paramount. Finally, it is required to establish whether these conclusions fully apply to spontaneous pregnancies too.
1. Martin JA, Hamilton BE, Sutton PD, Ventura SJ, Mathews TJ, Kirmeyer S, et al.. Births: final data for 2007. Natl Vital Stat Rep 2010;58:1–85.
2. Levi S. Ultrasonic assessment of the high rate of human multiple pregnancy in the first trimester. J Clin Ultrasound 1976;4:3–5.
3. Landy HJ, Keith LG. The vanishing twin: a review. Hum Reprod Update 1998;4:177–83.
4. Tong S, Meagher S, Vollenhoven B. Sonography: dizygotic twin survival in early pregnancy. Nature 2002;416:142.
5. Botchan A, Yaron Y, Lessing JB, Barak Y, Yovel I, David MP, et al.. When multiple gestational sacs are seen on ultrasound, ‘take-home baby’ rate improves with in-vitro fertilization. Hum Reprod 1993;8:710–3.
6. Tummers P, De Sutter P, Dhont M. Risk of spontaneous abortion in singleton and twin pregnancies after IVF/ICSI. Hum Reprod 2003;18:1720–3.
7. Glujovsky D, Shamonki M, Bergh P. Embryonic synergism may reduce pregnancy loss: a multivariate regression analysis. Fertil Steril 2007;87:509–14.
8. Lambers MJ, Mager E, Goutbeek J, McDonnell J, Homburg R, Schats R, et al.. Factors determining early pregnancy loss in singleton and multiple implantations. Hum Reprod 2007;22:275–9.
9. Spandorfer SD, Davis OK, Barmat LI, Chung PH, Rosenwaks Z. Relationship between maternal age and aneuploidy in in vitro fertilization pregnancy loss. Fertil Steril 2004;81:1265–9.
10. La Sala GB, Nucera G, Gallinelli A, Nicoli A, Villani MT, Blickstein I. Spontaneous embryonic loss after in vitro fertilization with and without intracytoplasmic sperm injection. Fertil Steril 2004;82:1536–9.
11. La Sala GB, Nucera G, Gallinelli A, Nicoli A, Villani MT, Blickstein I. Spontaneous embryonic loss following in vitro fertilization: incidence and effect on outcomes. Am J Obstet Gynecol 2004;191:741–6.
12. Zegers-Hochschild F, Bravo M, Fernández E, Fabres C, Balmaceda JP, Mackenna A. Multiple gestation as a marker of reproductive efficacy: learning from assisted reproductive technologies. Reprod Biomed Online 2004;8:125–9.
13. Matias A, La Sala G, Blickstein I. Early loss rates of entire pregnancies after assisted reproduction are lower in twin than in singleton pregnancies. Fertil Steril 2007;88:1452–4.
14. Menken J, Trussell J, Larsen U. Age and infertility. Science 1986;233:1389–94.
15. Nybo Andersen AM, Wohlfahrt J, Christens P, Olsen J, Melbye M. Maternal age and fetal loss: population based register linkage study. BMJ 2000;320:1708–12.
16. Baird DT, Collins J, Egozcue J, Evers LH, Gianaroli L, Leridon H, et al.; ESHRE Capri Workshop Group. Fertility and ageing. Hum Reprod Update 2005;11:261–76.
17. Ljunger E, Cnattingius S, Lundin C, Annerén G. Chromosomal anomalies in first-trimester miscarriages. Acta Obstet Gynecol Scand 2005;84:1103–7.
18. Eiben E, Bartels I, Bähr-Porsch S, Borgmann S, Gatz G, Gellert G, et al.. Cytogenetic analysis of 750 spontaneous abortions with the direct-preparation method of chorionic villi and its implications for studying genetic causes of pregnancy wastage. Am J Hum Genet 1990;47:656–63.
19. Janny L, Menezo YJ. Maternal age effect on early human embryonic development and blastocyst formation. Mol Reprod Dev 1996;45:31–7.
20. Ziebe S, Loft A, Petersen JH, Andersen AG, Lindenberg S, Petersen K, et al.. Embryo quality and developmental potential is compromised by age. Acta Obstet Gynecol Scand 2001;80:169–74.
21. de Mouzon J, Goossens V, Bhattacharya S, Castilla JA, Ferraretti AP, Korsak V, et al.; European IVF-monitoring (EIM) Consortium, for the European Society of Human Reproduction and Embryology (ESHRE). Assisted reproductive technology in Europe, 2006: results generated from European registers by ESHRE. Hum Reprod 2010;25:1851–62.
22. Department of Health and Human Services, Centers for Disease Control and Prevention (2010). 2007 assisted reproductive technology (ART) report. Available at: www.cdc.gov/art/ART2007/index.htm
. Retrieved May 2011.
23. Pinborg A, Loft A, Rasmussen S, Schmidt L, Langhoff-Roos J, Greisen G, et al.. Neonatal outcome in a Danish national cohort of 3438 IVF/ICSI and 10,362 non-IVF/ICSI twins born between 1995 and 2000. Hum Reprod 2004;19:435–41.
24. Hall JG. Twinning. Lancet 2003;362:735–43.
25. Vitthala S, Gelbaya TA, Brison DR, Fitzgerald CT, Nardo LG. The risk of monozygotic twins after assisted reproductive technology: a systematic review and meta-analysis. Hum Reprod Update 2009;15:45–55.