The vanishing twin phenomenon, first identified by Stoeckel in 1945, describes the spontaneous reduction of one fetus in a twin pregnancy.1 Recent studies have estimated that a vanishing twin occurs in 14.8–36% of twin pregnancies resulting from in vitro fertilization (IVF).2–4 As a result of the frequent and early gestation ultrasonographic evaluation of IVF pregnancies, this population affords an excellent opportunity to evaluate outcomes after a vanishing twin.
Clinical outcomes of surviving singletons in vanishing twin pregnancies have become an area of debate within the literature. Some studies demonstrate an increased risk of low birth weight and preterm birth when compared with singletons,5–8 whereas others report no difference in perinatal outcomes.3,9,10 These discrepancies may be the result of differences in exclusion criteria regarding gestational age of the vanishing twin at demise and of the remaining singleton at birth. Moreover, findings may be complicated by the influence of IVF techniques on perinatal outcomes, such as the association of cryopreserved (as compared with fresh) embryos with large-for-gestational-age (LGA) birth weight and of day 5 (as compared with day 3) embryos with preterm birth.11–14
Peripartum complications are infrequently evaluated in studies of pregnancies with a vanishing twin, and consideration of maternal morbidity is vital to elucidating the full peripartum and perinatal risks after a vanishing twin.6,7 The present study was performed to help address these knowledge gaps. We assessed vanishing twin, singleton, and twin pregnancies resulting from IVF in our program to determine whether the perinatal and peripartum outcomes more closely resemble those of singleton or twin pregnancies.
MATERIALS AND METHODS
This study was approved by the Partners Human Research Committee at the Brigham and Women's Hospital (Protocol #2016P000545).
This was a retrospective cohort study of women undergoing their first IVF cycle at the Brigham and Women's Hospital from January 1, 2007, to December 31, 2015. Data were collected from our prospectively maintained departmental database and the hospital electronic medical record system. In vitro fertilization cycle and embryology data are entered into the database prospectively by clinicians and embryologists. This database is routinely audited twice yearly. The peripartum database is updated prospectively by obstetric clinicians and nurses directly involved in patient care and serves as the official delivery record for the hospital. Key data points and missing data, including patient demographics and pregnancy data and outcomes, were verified in the electronic medical record.
Only those women who had fresh or cryopreserved autologous day 3 or day 5 embryo(s) transferred during the study period and who delivered a liveborn neonate at our institution at or beyond 24 weeks of gestation were included. Our infertility center has a large referral area and many of our patients choose to deliver at their local hospital. To ensure data validity, we limited our study population to patients with directly verifiable outcomes in our electronic medical record. As a result of the variability in research and clinical definitions of viability before 24 weeks of gestation, the resultant variability in clinical management in this range, and the overall poor prognosis for deliveries before 24 weeks of gestation, we chose to define live birth as deliveries occurring at or beyond the broadly accepted threshold of viability (24 weeks of gestation or greater). Gestational age was calculated as the interval from embryo transfer plus 2 weeks and the duration of embryo culture (3 or 5 days). Additional exclusion criteria were use of preimplantation genetic diagnosis or screening, a mixed source of fresh and cryopreserved embryos or both day 3 and 5 embryos, gestational carrier or donor oocyte cycles, and gestations with monochorionic twins or selective fetal reduction. Women who delivered a singleton from either a vanishing twin gestation or a singleton gestation and those who delivered dichorionic–diamniotic twins were identified and analyzed in three groups: vanishing twin, singleton, and twin.
A vanishing twin pregnancy was defined as having two gestational sacs recorded on ultrasonography with the demise of one twin at or before 14 weeks of gestation or the absence of identification of either a yolk sac or embryonic pole in one of the gestational sacs with a resulting singleton live birth at or beyond 24 weeks of gestation. At the time of vanishing twin diagnosis, the gestational age in days was recorded as was the furthest documented developmental stage (gestational sac only, gestational sac with yolk sac, presence of a fetal pole without cardiac activity, or presence of a fetal pole with cardiac activity). For subanalyses, vanishing twin pregnancies were further stratified into two groups according to whether they were diagnosed before compared with after the identification of fetal cardiac activity. A singleton pregnancy was defined as having only one gestational sac ever documented by early ultrasonography (at 5–8 weeks of gestation) and which resulted in a singleton live birth at or beyond 24 weeks of gestation. A twin pregnancy was defined as having only two gestational sacs documented by early ultrasonography (5–8 weeks of gestation) and which resulted in a twin live birth at or beyond 24 weeks of gestation.
Stimulation protocols for fresh and cryopreserved embryo transfer cycles are listed in Appendix 1, available online at http://links.lww.com/AOG/B87.15–18 Approximately 2 weeks after embryo transfer, patients had at least two serum human chorionic gonadotropin (hCG) levels checked at 2-day intervals. If at any point the hCG level rise was below 66%, an early ultrasonogram was obtained at approximately 5 weeks of gestation, with further monitoring and management dictated by clinical findings. From January 1, 2007, to November 1, 2012, a pelvic ultrasonogram was obtained at both 6 and 8 weeks of gestation if the serial serum hCG levels rose normally (greater than 66% in 2 days). After November 2012, patients with such normally rising hCG levels underwent a single pelvic ultrasonogram at 7–8 weeks of gestation as a result of a change in practice.
Preterm birth was defined as delivery before 37 weeks of gestation. Birth weight (grams) was recorded as both a continuous variable and as a categorical variable as follows: low birth weight (less than 2,500 g), appropriate birth weight (2,500–4,000 g), and high birth weight (greater than 4,000 g). Furthermore, gestational age-specific categorization of birth weight (Z-scores) were calculated using published data derived from U.S. birth certificates and corrected for gestational age and neonatal sex.19 These scores were used to qualify a neonate as small for gestational age (SGA), appropriate for gestational age, or LGA. In twin pregnancies, the birth weight and growth category of each twin were recorded to identify whether either twin was LGA or SGA. A twin pregnancy was categorized as appropriate birth weight or appropriate for gestational age only if both twins were either appropriate birth weight or appropriate for gestational age, respectively. To calculate the mean birth weight of the entire cohort of twin pregnancies, only the larger twin in each pregnancy was included. This method was chosen to identify how even the best grown twins would compare with singleton and vanishing twin birth weights. Had the smaller twin in each pregnancy also been included in the birth weight calculation, the mean birth weight of the twin group would be smaller. Good-quality cleavage-stage embryos were defined as having seven cells or more, less than 10% fragmentation, and perfect or moderate asymmetry of blastomeres. Good-quality blastocysts were defined as expanded, hatching, or hatched blastocysts with fair- or good-quality inner cell mass or trophectoderm.
Maternal records were evaluated for peripartum complications including gestational hypertensive diseases (gestational hypertension or preeclampsia), gestational diabetes mellitus, estimated blood loss, postpartum hemorrhage, abruption, primary cesarean delivery, abnormal placentation (placenta previa with the placental edge coming within 2.0 cm of the internal os on final ultrasonogram; or placenta accreta, increta, or percreta diagnosed clinically at the time of delivery or on pathology review), and hysterectomy. The placentas of all deliveries with a suspected abruption were evaluated by a pathologist, and an abruption was only included in the analyses if noted on the pathology report as a pathologic diagnosis. Postpartum hemorrhage was defined as an estimated blood loss greater than 1,500 mL or if the patient received a red blood cell transfusion.
Logistic regression analysis was used to estimate the odds ratio (OR) and 95% CI of patient demographics, IVF cycle characteristics, perinatal outcomes, and peripartum outcomes associated with vanishing twin live births (referent) as compared with singleton live births and compared with twin live births separately. The primary outcomes were gestational age and birth weight at delivery; secondary outcomes included peripartum morbidities. Patient age at oocyte retrieval was included a priori in all analyses. Further variables were retained in the regression model if their addition to the base model changed the OR from the crude model by 10% or more.20 Accordingly, logistic regression analysis of patient and cycle characteristics was adjusted for the number of embryos transferred to estimate the adjusted OR and 95% CI of a vanishing twin. Additional covariates that were tested as confounders of the relationship between vanishing twin and perinatal outcomes included gestational hypertensive diseases, gestational diabetes mellitus, body mass index (calculated as weight (kg)/[height (m)]2), assisted hatching, day of embryo transfer, and cryopreserved or fresh transfer; these did not meet criteria for inclusion in the final adjusted model. Birth weight analyses were adjusted a priori for gestational age at birth and neonatal sex given the known effect of these variables on birth weight.21 Statistical analyses were performed using SAS 9.3.
After excluding 1,698 patients because of delivery at an outside institution and 51 patients as a result of a miscarriage occurring between 14 and 24 weeks of gestation (two were vanishing twin pregnancies), this study consisted of 100 vanishing twin pregnancies, 798 singleton pregnancies, and 291 twin pregnancies (Fig. 1). Demographic characteristics for patients in each of the three exposure groups are shown in Table 1. Vanishing twins occurred in patients who were older at oocyte retrieval (37.2±3.7 years) compared with patients with singletons (35.2±3.8 years) or twins (35.0±3.8 years). Patients with vanishing twin pregnancies had more embryos transferred (2.8±1.4) than those with singleton pregnancies (2.1±1.2). The proportion of day 3 embryo transfers was 85% in vanishing twin pregnancies, 74% in singleton pregnancies, and 77% in twin pregnancies. Additionally, the proportion of fresh embryo transfers was 81% in vanishing twin pregnancies, 79% in singleton pregnancies, and 82% in twin pregnancies.
Among all vanishing twin pregnancies, the gestational age by IVF dating at demise ranged from 41 to 85 days with a mean of 59.2±10.9 days. The distributions of patients by the furthest developmental stage documented before identification of a vanishing twin were: 15 patients (15%) had only a gestational sac; 21 (21%) had a gestational sac with a yolk sac; nine (9%) had a fetal pole with no documented cardiac activity; and 55 (55%) had documented fetal cardiac activity.
Perinatal outcomes in the three exposure groups of patients are shown in Table 2. The mean gestational age at birth and birth weights were 38.6±2.3 weeks and 3,207±644 g in singleton pregnancies, 35.5±2.7 weeks of gestation and 2,539±610 g in twin pregnancies, and 38.5±1.8 weeks of gestation and 3,175±599 g in vanishing twin pregnancies, respectively. Comparison of vanishing twin and singleton pregnancies revealed a similar incidence of preterm birth (17.0% vs 14.8%, respectively; OR 1.18, 95% CI 0.67–2.06; adjusted OR 1.18, 95% CI 0.67–2.06), a similar incidence of SGA (14.0% vs 9.7%; OR 1.53, 95% CI 0.83–2.28; adjusted OR 1.67, 95% CI 0.88–3.15), and an identical incidence of LGA (8.0% vs 8.0%; OR 1.05, 95% CI 0.83–2.84; adjusted OR 1.11, 95% CI 0.50–2.44) neonates. Conversely, gestational age and birth weight outcomes were all clinically and statistically significantly different between vanishing twin and twin pregnancies. Preterm births occurred in 17% of vanishing twin pregnancies and 61% of twin pregnancies (OR 0.13, 95% CI 0.07–0.23; adjusted OR 0.12, 95% CI 0.07–0.22); and neonates were SGA in 14% of vanishing twin pregnancies vs 42% of twin pregnancies (OR 0.24, 95% CI 0.13–0.45; adjusted OR 0.14, 95% CI 0.07–0.28) (Fig. 2).
Peripartum complications in the three exposure groups are shown in Table 2. When comparing vanishing twin with singleton pregnancies, no difference was statistically significant; however, the study was not powered to detect uncommon complications. In contrast, when comparing vanishing twin and twin pregnancies, gestational hypertensive diseases, postpartum hemorrhage, and primary cesarean deliveries were more common in twin pregnancies. Gestational hypertensive diseases occurred in 25.8% of twin pregnancies as compared with 9% of vanishing twin pregnancies (OR 0.29, 95% CI 0.14–0.59; adjusted OR 0.30, 95% CI 0.14–0.64); postpartum hemorrhage occurred in 5.8% of twin pregnancies and 1.0% of vanishing twin pregnancies (OR 0.16, 95% CI 0.02–1.24; adjusted OR 0.15, 95% CI 0.02–1.14); primary cesarean deliveries occurred in 73.2% of twin pregnancies and 29% of vanishing twin pregnancies (OR 0.15, 95% CI 0.09–0.25; adjusted OR 0.17, 95% CI 0.10–0.28). The odds of placental abnormalities (placenta previa or accreta), abruption, and hysterectomy were similar among all groups.
The demographic characteristics of vanishing twin pregnancies further stratified by whether diagnosis occurred before (n=45 [45%]) or after (n=55 [55%]) documentation of fetal cardiac activity are shown in Table 3. No patient characteristic or IVF cycle characteristic was statistically significantly associated with absence or presence of fetal cardiac activity before a vanishing twin diagnosis. The day of embryo transfer was similar for both groups. Day 3 transfers occurred in 80% of pregnancies with vanishing before fetal cardiac activity and 85% of pregnancies with vanishing after fetal cardiac activity (OR 1.47, 95% CI 0.52–4.18; adjusted OR 1.76, 95% CI 0.60–5.23). Likewise, use of fresh embryo transfer was similar between the two groups: 80% and 82% of patients, respectively (OR 1.13, 95% CI 0.41–3.06; adjusted OR 1.21, 95% CI 0.43–3.38). Furthermore, no differences in presence or absence of fetal cardiac activity were observed regarding odds of adverse perinatal outcome or peripartum complications (Table 4). Of note, however, there was a threefold increase in odds of primary cesarean delivery in those pregnancies in which the vanishing twin was identified after the presence of fetal cardiac activity (38% vs 18%; OR 2.86, 95% CI 1.12–7.30; adjusted OR 3.10, 95% CI 1.19–8.09).
The most notable findings of this study are 1) vanishing twin pregnancies had comparable perinatal and peripartum outcomes with singleton pregnancies and improved outcomes compared with twin pregnancies and 2) occurrence of twin demise after documentation of fetal cardiac activity was not associated with increased risk of adverse outcomes other than primary cesarean delivery rate.
Our rate of 25.6% vanishing of all twin pregnancies is consistent with previous studies.2–4 Our demographic predictors of a vanishing twin pregnancy are in agreement with risk factors for multiple pregnancy and miscarriage. Additionally, that our vanishing twin patient population was older than either singletons or twins is likely explained by the age-related risk of miscarriage, largely attributable to aneuploidy.22
The increased perinatal and peripartum morbidity associated with twin pregnancies as compared with vanishing twin pregnancies is consistent with known risks of multifetal gestations as compared with singleton births.23 Of note, our perinatal and peripartum findings indicate that demise of one twin in the first trimester returns the patient and the surviving twin to the risk profile of a singleton pregnancy.
Despite the theoretically increased vascularity and tissue mass of a twin demised at a more advanced developmental stage, timing of demise was not associated with increased risk for any adverse perinatal or peripartum outcome, apart from primary cesarean delivery rate. The etiology underlying the association between increased use of cesarean delivery when vanishing occurred after the presence of fetal cardiac activity is unclear and warrants a more detailed obstetric investigation. In the absence of any biological plausibility to explain this association, we conclude that perinatal and peripartum outcomes do not appear adversely affected by timing of vanishing during the first trimester.
Prior publications have reported inconsistent associations between vanishing twin and singleton pregnancies for perinatal outcomes ranging from no adverse outcome in the surviving twin to significantly reduced gestational age and birth weight.3,6,7,9 One explanation for these varied observations is lack of a consistent definition of vanishing twin, which could lead to misclassification of the exposure. The majority of prior research has included demises occurring in the second and third trimesters.6,7,24 It is well established that first-trimester losses are most commonly associated with cytogenetic abnormalities; for losses beyond the first trimester, placental dysfunction, maternal disease, and infection become more common causes of demise.25,26 Although a more liberal definition of a vanishing twin may help increase study power, it can have unintended effects on reported outcomes as a result of the occurrence of varying pathologies at different points in a gestation. We therefore chose to limit the definition of a vanishing twin to demise of one twin occurring within the first trimester. Furthermore, many prior studies do not define a lower limit of gestational age at which the surviving twin is delivered. This, in turn, may lead to bias in perinatal outcome data, because previable deliveries are often related to maternal anatomy or infections and cannot be attributed solely to the vanishing twin. By including these patients, previously reported perinatal and maternal outcomes likely were influenced by multiple pathologic processes unrelated to the vanished twin per se, thereby introducing substantial bias.
We acknowledge several limitations of our study: 1) the medically complex patients referred to our center for antepartum and intrapartum care may result in selection bias, with overrepresentation of adverse outcomes in our study cohort compared with lower risk populations; 2) we included only IVF pregnancies, which are associated with increased risk of perinatal and peripartum complications compared with spontaneous pregnancies,23,27 and the generalizability of these results to spontaneous twin pregnancies, which may have a higher incidence of a vanishing twin,28 is unknown; 3) by excluding monochorionic pregnancies (as a result of the low incidence and different pathophysiology compared with dichorionic pregnancies), our results cannot be generalized to monochorionic pregnancies; and 4) although our sample size of vanishing twin pregnancies is larger than most previous studies, we likely lacked statistical power to detect differences in uncommon peripartum outcomes and these nonsignificant findings should not be generalized.
In summary, this study demonstrates that twins vanishing in the first trimester after IVF result in live births that resemble singleton pregnancies regarding perinatal and peripartum outcomes. Our results are reassuring to patients diagnosed with a vanishing twin pregnancy after IVF and provide a foundation for counseling in that associated risks of adverse outcomes after 24 weeks of gestation are comparable with those of singleton pregnancies. Our findings show that vanishing twin outcomes are likely improved by the early demise of one twin rather than if they continued as a twin pregnancy. Future areas of investigation may include noninvasive prenatal testing to determine karyotypes of failed twins, which may elucidate the pathophysiology of this pregnancy complication.29 Additionally, validation of the association between cesarean delivery rate and developmental stage at vanishing twin demise should be performed with a more comprehensive obstetric analysis of this outcome.
1. Stoeckel W. Lehbuch der geburstchilfe. Jena (Germany): Gustav Fisher; 1945. p. 258.
2. Dickey RP, Taylor SN, Lu PY, Sartor BM, Storment JM, Rye PH, et al. Spontaneous reduction of multiple pregnancy: incidence and effect on outcome. Am J Obstet Gynecol 2002;186:77–83.
3. 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.
4. Pinborg A, Lidegaard O, la Cour Freiesleben N, Andersen AN. Consequences of vanishing twins in IVF/ICSI pregnancies. Hum Reprod 2005;20:2821–9.
5. Almog B, Levin I, Wagman I, Kapustiansky R, Lessing JB, Amit A, et al. Adverse obstetric outcome for the vanishing twin syndrome. Reprod Biomed Online 2010;20:256–60.
6. Evron E, Sheiner E, Friger M, Sergienko R, Harlev A. Vanishing twin syndrome: is it associated with adverse perinatal outcome? Fertil Steril 2015;103:1209–14.
7. Pinborg A, Lidegaard O, Freiesleben N, Andersen AN. Vanishing twins: a predictor of small-for-gestational age in IVF singletons. Hum Reprod 2007;22:2707–14.
8. Shebl O, Ebner T, Sommergruber M, Sir A, Tews G. Birth weight is lower for survivors of the vanishing twin syndrome: a case-control study. Fertil Steril 2008;90:310–4.
9. Mansour R, Serour G, Aboulghar M, Kamal O, Al-Inany H. The impact of vanishing fetuses on the outcome of ICSI pregnancies. Fertil Steril 2010;94:2430–2.
10. Rodríguez-González M, Serra V, Garcia-Velasco JA, Pellicer A, Remohi J. The ‘vanishing embryo’ phenomenon in an oocyte donation programme. Hum Reprod 2002;17:798–802.
11. Pelkonen S, Koivunen R, Gissler M, Nuojua-Huttunen S, Suikkari AM, Hydén-Granskog C, et al. Perinatal outcome of children born after frozen and fresh embryo transfer: the Finnish cohort study 1995–2006. Hum Reprod 2010;25:914–23.
12. Sazonova A, Källen K, Thurin-Kjellberg A, Wennerholm UB, Bergh C. Obstetric outcome in singletons after in vitro fertilization with cryopreserved/thawed embryos. Hum Reprod 2012;27:1343–50.
13. Dar S, Lazer T, Shah PS, Librach CL. Neonatal outcomes among singleton births after blastocyst versus cleavage stage embryo transfer: a systematic review and meta-analysis. Hum Reprod Update 2014;20:439–48.
14. Källén B, Finnström O, Lindam A, Nilsson E, Nygren KG, Olausson PO. Blastocyst versus cleavage stage transfer in in vitro fertilization: differences in neonatal outcome? Fertil Steril 2010;94:1680–3.
15. Cheung LP, Lam PM, Lok IH, Chiu TT, Yeung SY, Tjer CC, et al. GnRH antagonist versus long GnRH agonist protocol in poor responders undergoing IVF: a randomized controlled trial. Hum Reprod 2005;20:616–21.
16. Dragisic KG, Davis OK, Fasouliotis SJ, Rosenwaks Z. Use of a luteal estradiol patch and a gonadotropin-releasing hormone antagonist suppression protocol before gonadotropin stimulation for in vitro fertilization in poor responders. Fertil Steril 2005;84:1023–6.
17. Surrey ES, Bower J, Hill DM, Ramsey J, Surrey MW. Clinical and endocrine effects of a microdose GnRH agonist flare regimen administered to poor responders who are undergoing in vitro fertilization. Fertil Steril 1998;69:419–24.
18. Tummon IS, Daniel SA, Kaplan BR, Nisker JA, Yuzpe AA. Randomized, prospective comparison of luteal leuprolide acetate and gonadotropins versus clomiphene citrate and gonadotropins in 408 first cycles of in vitro fertilization. Fertil Steril 1992;58:563–8.
19. Oken E, Kleinman KP, Rich-Edwards J, Gillman MW. A nearly continuous measure of birth weight for gestational age using a United States national reference. BMC Pediatr 2003;3:6.
20. Mickey RM, Greenland S. The impact of confounder selection criteria on effect estimation [published erratum appears in Am J Epidemiol 1989;130:1066]. Am J Epidemiol 1989;129:125–37.
21. Gardosi J, Chang A, Kalyan B, Sahota D, Symonds EM. Customised antenatal growth charts. Lancet 1992;339:283–7.
22. 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.
23. Helmerhorst FM, Perquin DA, Donker D, Keirse MJ. Perinatal outcome of singletons and twins after assisted conception: a systematic review of controlled studies. BMJ 2004;328:261.
24. Zhou L, Gao X, Wu Y, Zhang Z. Analysis of pregnancy outcomes for survivors of the vanishing twin syndrome after in vitro fertilization and embryo transfer. Eur J Obstet Gynecol Reprod Biol 2016;203:35–9.
25. Levy B, Sigurjonsson S, Pettersen B, Maisenbacher MK, Hall MP, Demko Z, et al. Genomic imbalance in products of conception: single-nucleotide polymorphism chromosomal microarray analysis. Obstet Gynecol 2014;124:202–9.
26. Silver RM, Varner MW, Reddy U, Goldenberg R, Pinar H, Conway D, et al. Work-up of stillbirth: a review of the evidence. Am J Obstet Gynecol 2007;196:433–44.
27. Jackson RA, Gibson KA, Wu YW, Croughan MS. Perinatal outcomes in singletons following in vitro fertilization: a meta-analysis. Obstet Gynecol 2004;103:551–63.
28. Márton V, Zádori J, Kozinszky Z, Keresztúri A. Prevalences and pregnancy outcome of vanishing twin pregnancies achieved by in vitro fertilization versus natural conception. Fertil Steril 2016;106:1399–406.
29. Curnow KJ, Wilkins-Haug L, Ryan A, Kirkizlar E, Stosic M, Hall MP, et al. Detection of triploid, molar, and vanishing twin pregnancies by a single-nucleotide polymorphism-based noninvasive prenatal test. Am J Obstet Gynecol 2015;212:79.e1–9.