Stillbirth is a devastating obstetric problem whose causal factors remain complex and incompletely understood. Multiples are at higher risk for in utero fetal death than singletons. Much of the risk of in utero fetal death in twins has been attributed to complications found more frequently in multiples, such as growth abnormalities or congenital anomalies, and problems specific to twin gestations, such as twin-twin transfusion syndrome. Prior studies on multiples and the risk of fetal death have focused on the number of fetuses or the gestational age-specific death rates.1–3 The impact of chorionicity on the gestational age-specific risk of stillbirth is not known. Although the death of a fetus is a tragic outcome, the event can be particularly problematic in monochorionic placentations where the death can be associated with a greater than 20% risk for adverse neurologic sequelae in a co-twin survivor.4 In addition, concern has grown surrounding an unexpectedly high rate of fetal death in uncomplicated monochorionic-diamniotic twins, despite reassuring antenatal surveillance.2,5 Our study of 1,000 consecutive twin pairs calculated the risk of fetal death with advancing gestational age in ongoing pregnancies by chorionicity. The risk of stillbirth was also investigated in the subgroup of “apparently normal” twins unaffected by growth abnormalities, anomalies, or twin-twin transfusion syndrome. The purpose of this study was to compare in utero survival between monochorionic-diamniotic and dichorionic-diamniotic twins.
MATERIALS AND METHODS
For this retrospective cohort study, all consecutive twin gestations delivered at a single tertiary care center from December 1, 2000, to May 11, 2007, were identified from a departmental perinatal database. The Columbia University Medical Center Institutional Review Board approved this study. Inclusion criteria were twins with two viable fetuses at 23+6/7 weeks of gestation and delivery at 24 weeks or later. Monoamniotic twins and twin sets within triplets or higher order multiples were excluded. Computerized and written medical charts were reviewed. Sonographic data were retrieved from computerized archiving systems R2 (R2 Technology, Inc., Sunnyvale, CA) and Observer (Vision Chips, Inc., Laguna Hills, CA).
The primary exposure was chorionicity as determined by placental histopathologic examination or, if unavailable, by standard sonographic criteria.6 Major structural anomalies were defined as those prenatally diagnosed and described in the RADIUS trial.7 Fetuses were classified as growth restricted if the estimated fetal weight calculated using the Hadlock formula fell at or less than the 10th percentile for the gestational age on a recent United States national growth chart.8,9 Twin-twin transfusion syndrome was diagnosed using the sonographic criteria described by Quintero et al10 or if a therapy such as laser ablation or amnioreduction was performed. Intertwin discordance was calculated in the standard fashion as the difference between the estimated fetal weights expressed as a percentage of the heavier twin estimated fetal weight, with significant intertwin discordance defined as 20% or greater. Twins were classified as “apparently normal” if they were not diagnosed antenatally with intrauterine growth restriction (IUGR), significant intertwin discordance, major anomaly, or twin-twin transfusion syndrome. Maternal age was the mother's age on the delivery date. For all variables except gender, data were analyzed per pregnancy rather than per fetus.
Monochorionic-diamniotic twins were compared with dichorionic-diamniotic twins across a number of maternal characteristics and pregnancy complications. Continuous variables were summarized as means and standard deviations and groups compared using the Student t test. Categorical data were compared using the χ2 test. All hypothesis tests were two-sided and significance set a priori at P<.05.
Overall in utero survival was analyzed using a time-to-event (Kaplan-Meier) analysis in which week of gestation was used as the time scale, stillbirth per pregnancy defined as the event, and all other births treated as censored. A stillbirth event was considered to have occurred if one or both twins died in utero and, for purposes of this analysis, was enumerated only once for a given pregnancy. For example, if an initially surviving co-twin subsequently suffered an in utero death, only the initial stillbirth was counted. Survival data for monochorionic-diamniotic and dichorionic-diamniotic twins were plotted as cumulative percentage without event and compared by the log-rank test. To assess the degree of difference in survival, hazard ratios and 95% confidence intervals were estimated using a Cox proportional hazards model and adjusted for covariates that differed at baseline.
Gestational age-specific stillbirth risks were then calculated. The risk of stillbirth was defined as the number of stillbirths, delineated per pregnancy, during or after a given 2-week gestational period divided by the total number of ongoing pregnancies at the start of the time period. This statistic has been previously described as the “prospective risk of stillbirth.”1 Ongoing pregnancies were defined as those that remained undelivered and that had not experienced a stillbirth event. All analyses were repeated comparing “apparently normal” monochorionic-diamniotic twins with “apparently normal” dichorionic-diamniotic twins. Statistical analysis was performed using SAS 9.1 for the personal computer (SAS Institute, Inc., Cary, NC).
A total of 1,024 twin pairs delivered at Columbia University Medical Center during the 6.5-year study time period. Chorionicity was confirmed by placental histopathologic examination in 689 (68.9%) twin pairs and diagnosed by standard sonographic criteria alone in the remaining pregnancies. The accuracy of ultrasonographic prediction of chorionicity at our institution has been previously described.6 After excluding 19 monoamniotic twins and five monochorionic-diamniotic pairs within triplets and higher order multiples, 1,000 consecutive twins fit our inclusion criteria. There were no cases of twin reversed arterial perfusion sequence or conjoined twins. The overall characteristics of the twin pregnancies are shown in Table 1. Mothers of monochorionic-diamniotic twins were older and delivered earlier than dichorionic-diamniotic parturients. Monochorionic-diamniotic twins were, on average, smaller at delivery and had a higher frequency of significant intertwin discordance or major congenital anomalies than dichorionic-diamniotic twins.
Overall in utero survival was lower for monochorionic-diamniotic twins compared with their dichorionic-diamniotic counterparts (log-rank P=.004) (Fig. 1). The risk of in utero death for monochorionic-diamniotic twins was approximately fourfold higher (age-adjusted hazards ratio 4.04, 95% confidence interval 1.48–11.03). Even among those that were “apparently normal,” overall in utero survival was lower for monochorionic-diamniotic twins (log-rank P=.039) (Fig. 2). The magnitude of the increased risk among “apparently normal” twins was similar to the entire group (age-adjusted hazards ratio 3.95, 95% confidence interval 0.86–18.15), but this did not reach statistical significance in this smaller group. The assumption of constant proportionality was confirmed graphically. Adjustments were not made for pregnancy complications used to distinguish “apparently normal” from complicated pregnancies because these groups were analyzed separately. Models were adjusted for age alone since no other remaining antenatal characteristics differed at baseline.
The number of stillbirths at each week of gestation and the risk of stillbirth with advancing gestation for monochorionic-diamniotic and dichorionic-diamniotic twins are shown in Table 2. Data for “apparently normal” twins are detailed in Table 3. An excess risk for stillbirth was associated with monochorionicity at all gestational ages and was apparent as early as 24 weeks of gestation even among “apparently normal” twins.
All stillbirths were single in utero fetal deaths except a double in utero fetal death of dichorionic-diamniotic twins in a patient who presented with acute fatty liver disease at 35 weeks. Of the 17 dead fetuses, eight were the presenting twin. The time interval between stillbirth diagnosis to delivery ranged from 0 to 59 days. There were no intrapartum deaths in either group. Autopsy and placental pathology findings for each stillbirth are described in Table 4. Of the three in utero fetal deaths within the “apparently normal” monochorionic-diamniotic group, placental pathology in one case demonstrated arterial-arterial vessel anastomoses, placental infarcts, and cytomegalovirus inclusions. Additionally, a 1-mm ventricular septal defect was found on autopsy in the dead twin, and the co-twin survivor subsequently demonstrated a cytomegalovirus infection. Stillbirth remained unexplained in two of the three cases, both of which had a normal sonogram within a week of the diagnosis. Autopsy was declined in one case, but in the other the autopsy was unremarkable, with very few threadlike vascular anastomoses found on placental examination.
In our study, the risk of fetal death for monochorionic twins was higher than dichorionic twins at all gestational ages at or greater than 24 weeks. The relationship between chorionicity and in utero survival was observed for the entire cohort and for the subgroup of “apparently normal” twins. Thus, the excess rate of stillbirths in the monochorionic-diamniotic group compared with dichorionic-diamniotic gestations was not merely a result of excess risk due to twin complications because the risk of higher in utero mortality persisted after exclusion of disorders unique to monochorionicity. In addition, the risk of in utero fetal death was similar for “apparently normal” monochorionic-diamniotic twins compared with all monochorionic-diamniotic twins. Finally, the nadir of the stillbirth risk occurred around 34 weeks, both for unselected monochorionic-diamniotic twins and for “apparently normal” monochorionic-diamniotic twins, whereas the stillbirth risk for dichorionic-diamniotic twins continued to decrease in the late preterm gestation.
The association between monochorionicity and higher rates of obstetric complications are well known.11 The risk of fetal death at 34 weeks for all monochorionic-diamniotic twins in this study (1.6%) was higher than the 0.7% found by Simões and colleagues.3 This difference may be explained by a higher rate of twin complications, such as twin-twin transfusion syndrome, in our study population. Uncomplicated or “apparently normal” monochorionic-diamniotic twins and their potentially deceivingly reassuring antenatal testing have been the focus of recent concern. Barigye and colleagues2 reported an unexpectedly high rate of late fetal death in a cohort of 151 uncomplicated monochorionic-diamniotic twins despite intensive ultrasound surveillance. The risk of fetal death at 34 weeks for “apparently normal” monochorionic-diamniotic twins was 3.3%, similar to the 2.0% found in our study.2 Direct comparisons with the prior two studies are limited by different methodologies. Slight differences in the magnitude of stillbirth risk may be attributed to distinct study populations or differences in the definitions of pregnancy complications such as IUGR or intertwin discordance. The major advantage of our study is the additional benefit of comparative dichorionic-diamniotic twin data.
The limitations of this study are acknowledged. Due to the retrospective design and lack of consistent documentation, potential residual confounders such as assisted reproductive technology, obesity, smoking, and antiphospholipid antibody syndrome could not be evaluated. However, common conditions such as chronic hypertension are unlikely to be causative in fetal death without IUGR or an acute event such as abruption, obvious findings that would subsequently explain the stillbirth. Antenatal diagnoses were used to characterize growth disturbances, twin-twin transfusion syndrome, and anomalies. The antenatal approach was selected over postnatal diagnoses because it was felt to more accurately assess information that was clinically relevant and available at the time providers make decisions about timing of twin deliveries. Obviously unknown before delivery, birth weight may have been altered by post mortem changes that also limited autopsy and placental evaluations, particularly if delivery was delayed for the surviving co-twin. In our study, disparities between prenatal and postnatal malformations were minimal. Using postnatal criteria to redefine the group of uncomplicated twins did not alter the overall findings (data not shown). Caution should be taken when interpreting the magnitude of the residual risk for “apparently normal” monochorionic-diamniotic twins. Although this represents a relatively large experience with monochorionic-diamniotic twins, the cohort size remains modest, given the rarity of the outcome. Inferences regarding the risk of fetal death before 24 weeks cannot be made because this study was limited to those pregnancies that had two living fetuses at 23+6/7 weeks.
Unknown before birth in the majority of twins, zygosity was not evaluated due to the study design focused on antenatal diagnoses available to the medical practitioner. Although zygosity is intimately related to chorionicity, the two are not synonymous, as emphasized in recent case reports of monochorionic twins with discordant genders and as observed in three otherwise unremarkable cases in our cohort.6,12 Because the overwhelming majority of monochorionic twins are monozygous, chorionicity may be a close proxy for zygosity although a portion of dichorionic twins are monozygous. The role of zygosity in the risk of stillbirth is unclear but the theoretical etiology for an elevated risk of in utero fetal death is more likely related to shared placental perfusion rather than identical genetic composition.
The true risk of stillbirth in monochorionic-diamniotic twins may be higher than observed here. The methodology of this study does not assess the effect of a specific antenatal surveillance protocol, often more rigorously implemented for monochorionic-diamniotic than dichorionic-diamniotic pregnancies. At our institution, we evaluate fetal weight and growth discordance every 3–4 weeks and with increased frequency if growth restriction or significant intertwin discordance is discovered.5,13 Surveillance for “apparently normal” monochorionic twins includes a minimum of weekly testing in the form of biophysical profile with amniotic fluid assessment in the third trimester and, as previously described, more frequent testing if complications arise.13 Secondly, as a referral center, patients with monochorionic-diamniotic twins affected with twin-twin transfusion syndrome, growth discordance, or anomalies presenting before 24 weeks may have opted for termination or multi-fetal reduction. For example, 11% of monochorionic twins in our cohort had twin-twin transfusion syndrome, which compares with estimations of 20% in the general population of monochorionic-diamniotic twins and 35% eliminated in the study by Barigye et al.2 Therefore our cohort may underestimate the true risks of stillbirth among unselected monochorionic-diamniotic twins.
At our institution, we have adopted a low threshold for delivery of monochorionic-diamniotic twins due to a concern for the high rate of stillbirth in “apparently normal” gestations.5 The term “apparently” reflects our concerns about the limitations of sonographic estimated fetal weight and intertwin discordance assessments and that reassuring, rigorous antenatal surveillance does not obviate the risk of in utero fetal death and neurologic sequelae for a co-twin survivor. Therefore, the actual risk of in utero fetal death may be lowered by our policy to deliver for such complications as growth discordance in the absence of IUGR and a further liberal policy of offering elective delivery at 34–35 weeks for uncomplicated monochorionic-diamniotic twins after corticosteroid administration or confirmation of fetal pulmonary maturity.5 This may partially explain the lower gestational age at delivery for monochorionic-diamniotic twins, thus limiting our ability to make meaningful conclusions at the later gestational ages because of smaller numbers. In this group electively delivered, there were no cases of respiratory distress syndrome, and all neonates were discharged without delay.
Stillbirths among “apparently normal” twins are challenging to predict but can be particularly devastating because a single in utero fetal death can expose the surviving co-twin in a monochorionic placentation to a more than 20% risk for multicystic encephalomalacia.14 The etiology for stillbirth in otherwise “apparently normal” monochorionic-diamniotic twins is not clear but may be due to sudden or subtle variations of the twin-twin transfusion syndrome phenomenon and/or uteroplacental insufficiency as an underlying mechanism.
Optimal twin pregnancy management involves balancing the risks of neonatal prematurity with the potential for in utero complications. The risk of fetal death plays a key role in this equipoise because unexplained stillbirth is common and problematic.15,16 American College of Obstetricians and Gynecologists practice bulletins on multiples do not specify a gestational age for elective delivery of twins but allow that some complaints or conditions that would not mandate delivery in singleton pregnancies may be considered indications for delivery in twins.17 When is the best time to deliver apparently uncomplicated monochorionic-diamniotic twins? A deceivingly simple solution is elective delivery, but there is growing concern that increasing obstetric interventions have contributed to the recent rise in late preterm births.18 The risk of fetal death for twins equals the risk of fetal death for postterm singletons by approximately 36–37 weeks, and the risk of fetal death and twin neonatal mortality intersect at 39 weeks.1 Therefore, elective delivery at that time seems reasonable, but this calculation does not consider chorionicity. The major advantage of our study is the addition of comparative dichorionic data suggesting the persistence of a higher risk for in utero mortality for monochorionic-diamniotic twins. Optimal timing of delivery for monochorionic-diamniotic twins nonetheless remains unanswered until information regarding perinatal mortality data specific to monochorionic-diamniotic twins and to “apparently normal” monochorionic-diamniotic twins becomes available. We acknowledge that our practice pattern is not standard of care but remain concerned about the risk of stillbirth in monochorionic-diamniotic twins, the risk of adverse neurologic sequelae in a co-twin survivor, and the uncertainties of sonographic assessment of twins. Until large, prospective observational studies have been conducted, elective delivery of monochorionic-diamniotic twins in the late preterm period may be a reasonable option, obviating the risk of stillbirth after extensive counseling and informed consent.
1. Kahn B, Lumey LH, Zybert PA, Lorenz JM, Cleary-Goldman J, D'Alton ME, et al. Prospective risk of fetal death in singleton, twin, and triplet gestations: implications for practice. Obstet Gynecol 2003;102:685–92.
2. Barigye O, Pasquini L, Galea P, Chambers H, Chappell L, Fisk NM. High risk of unexpected late fetal death in monochorionic twins despite intensive ultrasound surveillance: a cohort study. PLoS Medicine 2005;2:e172.
3. Simões T, Amaral N, Lerman R, Ribeiro F, Dias E, Blickstein I. Prospective risk of intrauterine death of monochorionic-diamniotic twins. Am J Obstet Gynecol 2006;195:134–9.
4. Weiss JL, Cleary-Goldman J, Tanji K, Budorick N, D'Alton ME. Multicystic encephalomalacia after first-trimester intrauterine fetal death in monochorionic twins. Am J Obstet Gynecol 2004;190:563–5.
5. Cleary-Goldman J, D'Alton ME. Uncomplicated monochorionic diamniotic twins and the timing of delivery. PloS Med. 2005;2:e180. Epub 2005 Jun 28.
6. Lee YM, Cleary-Goldman J, Thaker HM, Simpson LL. Antenatal sonographic prediction of twin chorionicity. Am J Obstet Gynecol 2006;195:863–7.
7. Crane JP, LeFevre ML, Winborn RC, Evans JK, Ewigman BG, Bain RP, et al. A randomized trial of prenatal ultrasonographic screening: impact on the detection, management, and outcome of anomalous fetuses. RADIUS Study Group. Am J Obstet Gynecol 1994;171:392–9.
8. Hadlock FP, Harrist RB, Sharman RS, Deter RL, Park SK. Estimation of fetal weight with the use of head, body, and femur measurements: a prospective study. Am J Obstet Gynecol 1985;151:333–7.
9. Oken E, Kleinman KP, Rich-Edwards J, Gillman M. A nearly continuous measure of birth weight for gestational age using a United States national reference. BMC Pediatr 2003;3:6. Epub 2003 Jul 8.
10. Quintero RA, Morales WJ, Allen MH, Bornick PW, Johnson PK, Kruger M. Staging of twin-twin transfusion syndrome. J Perinatol 1999;19:550–5.
11. Leduc L, Takser L, Rinfret D. Persistence of adverse obstetric and neonatal outcomes in monochorionic twins after exclusion of disorders unique to monochorionic placentation. Am J Obstet Gynecol 2005;193:1670–5.
12. Miura K, Niikawa N. Do monochorionic dizygotic twins increase after pregnancy by assisted reproductive technology? J Hum Genet 2005;50:1–6.
13. Malone FD, D'Alton ME. Multiple gestation. In: Creasy, RK, Resnik R. editors. Maternal-fetal medicine: principles and practice. 5th ed. Philadelphia (PA): Saunders; 2004:513–36.
14. American College of Obstetricians and Gynecologists and American Academy of Pediatrics. Neonatal encephalopathy and cerebral palsy: defining the pathogenesis and pathophysiology. Washington (DC): ACOG; 2003.
15. Silver RM. Fetal death [published erratum appears in Obstet Gynecol 2007;110:191]. Obstet Gynecol 2007;109:153–67.
16. 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.
17. American College of Obstetricians and Gynecologists. Multiple gestation: complicated twin, triplet, and high-order multifetal pregnancy. ACOG Practice Bulletin No. 56. Obstet Gynecol 2004;104:869–83.
18. Raju TN, Higgins RD, Stark AR, Leveno KJ. Optimizing care and outcome for late preterm (near-term) infants: a summary of the workshop sponsored by the NICHD. Pediatrics 2006;118:1207–14.