Twin–twin transfusion syndrome complicates approximately 15% of monochorionic twin pregnancies 1 and despite contemporary obstetric and neonatal management strategies is associated with a 30–50% perinatal mortality rate. 2 Additionally, there is significant morbidity in surviving neonates due to complications of the underlying pathology and the high preterm birth rate that invariably accompanies this condition.
The hallmark diagnostic feature of twin–twin transfusion syndrome in the antenatal period is a marked discordance in amniotic fluid volumes between the fetuses. It is this polyhydramnios–oligohydramnios sequence that typifies the second-trimester presentation of this disorder. It has become apparent, however, that twin–twin transfusion syndrome is a heterogeneous condition and may manifest a variety of concomitant features including fetal hydrops and abnormal fetoplacental vascular Doppler studies.
The precise etiology of twin–twin transfusion syndrome remains uncertain, although functioning placental vascular anastomoses with secondary alterations in fetal intravascular volume are central to the pathophysiology. The fetal responses to this volume alteration in addition to amniotic fluid volume alterations include growth discordance and varying degrees of cardiac and renal dysfunction.
As a consequence of improvements in sonographic diagnosis and monitoring of twin–twin transfusion syndrome, several therapeutic strategies have been proposed in an attempt to improve perinatal outcomes. Amnioreduction, 3–5 photocoagulation of placental vascular anastomoses, 6–8 amniotic septostomy, 9 and selective feticide by cord occlusion 10 have been advocated as treatment options. These treatment modalities currently appear to produce similar overall perinatal survival rates, in the order of 50–70%. Prolongation of gestation appears central in management strategies, as the complications of preterm birth are dominant causal factors in fetal and neonatal losses. 11,12 As twin–twin transfusion syndrome is a heterogeneous condition, stratification of treatment options to individual case severity may further improve reported outcomes while minimizing maternal and fetal complication rates.
In 1992 a prospective longitudinal cohort of pregnancies complicated with twin–twin transfusion syndrome in Western Australia was established. Because of the existence of a single tertiary referral center for perinatal medicine in Western Australia, a unique opportunity to review all cases of prenatally diagnosed twin–twin transfusion syndrome has arisen. The aim of this study was to assess the perinatal outcomes of all cases of monochorionic twin pregnancies complicated by twin–twin transfusion syndrome in a contemporary obstetric environment.
MATERIALS AND METHODS
In 1992 a prospective cohort encompassing all cases of prenatally diagnosed twin–twin transfusion syndrome pregnancies was established at the Women and Infants Research Foundation, Perth, Western Australia. The study population for this article includes all pregnancies from this cohort over the 10-year time period from May 1992 through May 2002.
The study design involved the prospective collection and assessment of obstetric and neonatal information using a purpose-designed data collection sheet. Any missing data were obtained by review of the individual maternal or neonatal medical record chart. The Institutional Ethics Committee of King Edward Memorial Hospital for Women provided permission for the conduct of this study.
The pregnancies were retrospectively staged at initial diagnosis as per Quintero 13 before 2000 and prospectively since then. This staging classification for twin–twin transfusion syndrome consists of five discrete stages reflecting disease severity. Stage I describes the isolated discrepancy in amniotic fluid volumes between fetuses, for stage II there is absence of a urine-filled bladder in the donor fetus, absent end-diastolic flow velocity waveform pattern in one or both fetuses identifies stage III, and in stage IV fetal hydrops is present. In stage V there is the demise of one or both fetuses.
The primary diagnostic criterion used in our institution for twin–twin transfusion syndrome is based on the presence of discordant amniotic fluid volumes in a monochorionic twin pregnancy (one twin with a maximum vertical pocket of at least 8 cm and the other with a maximum vertical pocket of up to 2 cm). If placental chorionicity had not been clearly established sonographically in the first trimester, single placentation with a thin dividing membrane in concordant sex fetuses was used as presumptive evidence of monochorionicity. Placentae were examined postdelivery for histopathologic confirmation of monochorionicity.
Once an antenatal diagnosis of twin–twin transfusion syndrome was made a uniform management protocol was adhered to, with all cases managed by the Maternal–Fetal Medicine Service of our institution. Amnioreduction was the principal therapeutic modality used for twin–twin transfusion syndrome and instituted when the maximum vertical pocket of amniotic fluid exceeded 8 cm on sonographic assessment. Amniotic fluid was aspirated from the sac of the putative recipient fetus using an 18-gauge spinal needle until the residual maximum vertical pocket was 4 cm or less. Three operators performed all of the amniocenteses in this series. Antibiotic prophylaxis was not used, and 0.25-mg terbutaline was subcutaneously employed for prophylactic tocolysis. In the latter part of this study our institution participated in the controlled clinical Multicenter Randomized Trial for the Evaluation of Septostomy Versus Serial Amnioreduction for Therapy. 14 A total of 12 pregnant women from our center were recruited to this trial, with the septostomy randomization allocated to six cases. Septostomy was not offered or performed outside the confines of this clinical trial. The septostomy procedure was performed as dictated by the study design, using a 22-gauge spinal needle to pierce the dividing membrane between the recipient and donor fetuses.
In all pregnancies in which amnioreduction and/or septostomy were performed, fetal karyotype analysis and microbiologic culture for infection from the polyhydramniotic sac at first procedure were undertaken. Ultrasound assessment was performed at least weekly and included amniotic fluid volume measurement by maximum vertical pocket technique, biometric measures, and fetoplacental vascular Doppler studies on both fetuses. Antenatal corticosteroid administration was routine for all pregnancies before delivery if the gestation was at or beyond 24 weeks and less than 34 weeks. The timing and mode of delivery was individualized for each case.
The data analyses were conducted using SAS 8 (SAS Institute, Cary, NC). Numerical data, which were normally distributed according to the Shapiro–Wilk statistic, are expressed as means and standard deviations, and otherwise as medians and interquartile ranges (25th percentile to 75th percentile). Categoric variables are expressed as n (%). Comparisons among groups were made using χ2 tests or the Fisher exact test for categoric variables, F tests for normally distributed variables, and Kruskal–Wallis χ2 tests for nonparametric distributions. Comparison of survival with stage was made using logistic regression with a cumulative logit model (PROC LOGISTIC SAS 8). Birth weight discordance was determined as the difference in birth weights as a percentage of the average weight of the twin pair. The same method was used for discordance in hemoglobin values. Tests for the median of these differences to be equal to zero were made using the Wilcoxon signed rank test. For all tests, a P value of .05 was considered to be statistically significant.
A cohort of 69 pregnancies has 80% power to construct a 95% confidence interval for survival of both fetuses within 11.7% of the basal rate. A one-sided log rank test with an overall size of 69 pregnancies (36 in stages I and II and 33 in stages III–V) achieves 77% power at a 5% significance level to detect a difference of 25% survival between 86% (stages I and II) and 60% (stages III–V). This corresponds to a hazard ratio of 3.14.
Since May 1992, 69 twin pregnancies have met the diagnostic criteria for inclusion in this study (Table 1). The median gestation at diagnosis was 22.1 weeks (interquartile range 19.7–25.4). Of all study cases, 59 were diagnosed at 26 weeks' gestation or less (85.5%). The median maximal vertical pocket was 11 cm (interquartile range 9–15) for the putative recipient and 1 cm (interquartile range 0–2) for the putative donor at the time of diagnosis. Critically abnormal umbilical artery Doppler flow velocity waveform patterns (either absent or reverse flow in diastole) were present at the time of diagnosis in 15.1% (ten of 66) of recipient fetuses and 32.8% (21 of 64) of donor fetuses. In five cases there was concordance of critically abnormal umbilical artery Doppler studies at diagnosis, and this was associated with intrauterine death of both fetuses in four cases (80%).
Therapeutic amniocentesis and/or septostomy were performed in 75.4% of cases (52 of 69 pregnancies). The median number of amniocenteses per pregnancy was three. The mean volume of amniotic fluid removed per procedure was 1605 mL, giving a mean total volume removed per pregnancy of 5585 mL. In two cases, isolated septostomy, without concomitant amnioreduction, was performed. The reasons for nonuse of therapeutic amniocentesis were death of one or both fetuses at or soon after the time of diagnosis (n = 9 [53%]), termination of pregnancy (n = 2 [11.8%]), urgent delivery for fetal compromise at the time of diagnosis required (n = 4 [23.5%]), or amniotic fluid volume not felt to warrant amnioreduction (n = 2 [11.8%]). The complication rate from amnioreduction procedures was 4.6% (eight of 174), with preterm membrane rupture (four of eight) and preterm labor leading to delivery (two of eight) the most common complications. In two cases intrauterine fetal death occurred within 72 hours of amnioreduction. There were no cases of chorioamnionitis or placental abruption in this series. In no case of amnioreduction or septostomy was there evidence of intrauterine bacterial, parasitic, or viral infection on culture of the amniotic fluid aspirated at the initial procedure (n = 52). The median interval from the diagnosis of twin–twin transfusion syndrome until delivery was 6.9 weeks (interquartile range 1–11.1). In 79.7% (55 of 69) of all cases antenatal corticosteroids were administered before delivery.
At time of delivery, there were 47 pregnancies with both twins alive and nine pregnancies with one surviving fetus. Both twins in 13 pregnancies did not survive until birth. In three cases resulting in double stillborn fetus, the pregnancy was interrupted at parental request. The median gestation at delivery was 29.4 weeks (interquartile range 26.3–33.8), and the majority of pregnancies were delivered before 37 weeks' gestation. Cesarean delivery was the chosen route for delivery in 47.8% of pregnancies. Gestation and the indication for delivery were the two principal factors influencing delivery mode (Table 1).
The overall perinatal survival in this complete series of twin–twin transfusion syndrome was 64.5% (89 of 138) (Table 1). Intrauterine fetal death was responsible for most perinatal losses, with an odds ratio of 1.47 (95% confidence interval 0.68, 3.18) for intrauterine demise of the donor fetus compared with the recipient. Neonatal death accounted for 28.6% of all perinatal losses. The gestation at delivery had a powerful influence upon perinatal survival in this series. The perinatal survival for those born at less than 28 weeks' gestation was 27.1% (13 of 48), increasing to 84.4% (76 of 90) for those born at 28 weeks' gestation or longer (P = .001).
Staging of the disease severity at diagnosis showed 31.9% of cases to be stage I, 20.3% stage II, 31.9% stage III, 11.6% stage IV, and 4.3% stage V (Table 2). The relationship of stage and perinatal outcome is demonstrated in this table. It is evident that, as stage at diagnosis increases, the subsequent perinatal mortality significantly increases. For lesser disease severity (stages I and II) the perinatal survival rate was 76.4%, falling to 51.5% with increasing disease severity (stages III–V) (P = .004). There was at least one perinatal survivor in 86.1% of cases in stages I to II disease, compared with 60.6% in stages III–V disease (P = .033). There was no significant difference in the gestation of diagnosis with increasing stage (Table 3); however, the diagnosis-to-delivery interval and perinatal survival were significantly reduced as stage at diagnosis increased.
At delivery, significant differences in birth weight and initial hemoglobin values were observed between live-born twin pairs (Table 4). There was a 34% discordance in birth weight between live-born twin pairs and a discordance of 20% in the initial hemoglobin levels. The median Apgar score at 5 minutes was 8 for both the recipient and the donor twin. Umbilical cord arterial blood gas levels were ascertained in 78% of live-born twins. There was no significant difference in any arterial blood gas parameter between donor and recipient twin.
Details of neonatal morbidity are shown in Table 5. Just over 50% of neonates required intubation at or within 30 minutes of delivery, and the median duration of ventilation was 4 days. Exogenous surfactant was used in 55.6% (30 of 54) of recipients and 53.1% (26 of 49) of donor neonates (P = .96). There was a high use of antibiotics, with 78.6% of all neonates receiving these agents, principally on a prophylactic basis. No difference in antibiotic use was observed between recipient and donor fetuses (77.8% versus 79.6%, recipient versus donor, P = .99). An infective agent was cultured in specimens from 15.5% of all neonates (14.8% recipient versus 16.3% donor, P = .95). Blood transfusions were performed in 33% of neonates, with no significant difference in the requirement for transfusion between recipient and donor twins (P = .109) (Table 5). Median nursery stays were 36 days for recipient neonates and 34 days for donors (P = .99).
Chronic lung disease developed in 14.6% of neonates, with no significant difference in occurrence between recipient and donor twin. Necrotizing enterocolitis occurred in 2.9% of neonates and was associated with a perinatal mortality of 33%. Renal failure was restricted to donor twins, occurred in 4.8% of neonates, and was associated with 60% perinatal mortality. Head ultrasonography was performed in 81.6% (84 of 103) of neonates. There were two principal reasons for failure to perform a neonatal cranial ultrasound: early death or late third-trimester delivery of a clinically neurologically normal neonate. Major intracranial hemorrhage (grade 3 or 4) occurred in 7.1% (six of 84) of neonates subject to head ultrasonography. Periventricular leukomalacia was diagnosed in 4.8% of neonates. The only infant who died in this series had antenatally diagnosed microcephaly. Overall, of the 103 born alive, 89 were discharged to home and 88 survived the first year of life.
Placental pathology of the twin pregnancies could be assessed in 68 cases (98.6%). Monochorionicity was confirmed by pathologic examination of the placenta in all cases submitted. In 1.5% (one of 68) of recipient and 29.4% (20 of 68) of donor twins the placental umbilical cord insertion was observed to be velamentous in nature (P < .001). There was, however, no evidence that the presence of a velamentous placental cord insertion influenced the course or outcome of the disorder. There was no difference in the gestation at diagnosis (22.8 ± 5.1 weeks versus 22.9 ± 3.3 weeks, P = .94), distribution of stage at diagnosis (stage I, 25% versus 28.9% [P = .99]; stage 2, 25% versus 21% [P = .75]; stage 3, 40% versus 31.6% [P = .73]; stage IV, 5% versus 15.8% [P = .4]; and stage V, 5% versus 2.7% [P > .999]) or gestation at delivery (30.9 ± 4.5 weeks versus 29.8 ± 4.0 weeks, P = .32) between cases with a velamentous cord insertion and those without. Birth weights for recipient (1644.8 ± 808.8 g versus 1404.5 ± 679 g, P = .24) and donor twins (1032 g [608–16,280] versus 940 g (625–1555), P = .60) did not differ when analyzed according to placental cord insertion type (velamentous versus nonvelamentous). Similarly, the hemoglobin values of the recipient (182 ± 21 g/L versus 172 ± 29 g/L, P = .25) and donor twin (135 ± 17 g/L versus 152 ± 21 g/L, P = .14) demonstrated no difference based on placental pathology. Thus, there did not appear to be any clinical relevance to the presence of a velamentous cord insertion in the pregnancies with twin–twin transfusion syndrome.
One of the principal challenges to contemporary perinatal medicine is the timely identification and effective management of twin–twin transfusion syndrome. This is a unique entity, specific to monochorionic placentation, and complicates 15% of monochorionic diamniotic twin pregnancies. The diagnosis is made primarily in the second trimester of pregnancy by the sonographic demonstration of markedly discordant amniotic fluid volumes in a monochorionic twin gestation. There is essentially no other pathology that can replicate this appearance, and thus the diagnosis is not difficult. Several years ago a case report 15 suggested that cytomegalovirus infection of the amniotic cavity could imitate the ultrasound appearances of twin–twin transfusion syndrome. It is our experience that this is improbable. In this series of 69 consecutive cases of antenatally identified twin–twin transfusion syndrome there was no instance of viral or bacterial infection of the amniotic cavity before intervention or on pathology assessment of the placenta postdelivery.
This series reflects the natural history of twin–twin transfusion syndrome in contemporary obstetric practice. We have not selected out those cases that were candidates for any particular therapeutic intervention, but have presented all cases with a prenatal diagnosis of twin–twin transfusion syndrome. As is evident from our data, there were many instances where the opportunity for intervention was lost, either from late presentation where immediate delivery was the most appropriate course of action or where it was clear that adverse events had already dictated a management course (eg, demise or incipient demise of one or both fetuses). Previous large series reports focusing on specific therapies 9 have provided an interventional outcome assessment that may not necessarily reflect the true population occurrence and outcome of this disehis disehis disease. In those pregnancies in which amnioreduction was performed the overall perinatal survival rate was 67.6% (69 of 102), compared with a perinatal survival rate of 55.6% (20 of 36) for those cases in which it was not (P = .271).
This series has confirmed the previous observations by our research group 11 and others 12 that severity of disease and gestation at delivery are the principal factors in determining perinatal outcome. The diagnostic staging criteria proposed by Quintero 13 conveniently partition the severity of twin–twin transfusion syndrome at the time of diagnosis and provide some prognostic estimate of outcome. Using this staging system it is evident that amnioreduction offers satisfactory efficacy for early-stage disease, with at least one fetus surviving in more than 85% of cases and two survivors in 66.7% of cases with stage I or II disease. As the disease severity progressed to critically abnormal umbilical artery Doppler studies and fetal hydrops, there was a reduced survival rate despite the interventions employed. For stage III or greater disease at diagnosis there was a 60.6% chance of achieving at least one survivor, and in only 42.5% of cases did two fetuses survive the perinatal period.
Twin–twin transfusion therapy is characterized by a high preterm birth rate, which is associated with prolonged neonatal nursery admissions and multiple neonatal complications. As this is a contemporary clinical series, fetal lung maturation with antenatal corticosteroid administration was used frequently and postnatal surfactant administration was commonplace. The former agents may have impacted on the lower incidence of grades 3 and 4 intraventricular hemorrhage observed in this series relative to earlier series. This may also reflect the policies of a single tertiary-level neonatal intensive care unit. In the international series of Mari et al 12 the incidence of grades 3 and 4 intraventricular hemorrhage was 24.5%, compared with 7.1% in our series. There was no mention of the frequency of use of antenatal corticosteroids in the Mari et al series, 12 and if the actual usage was low, this could account in part for the observed differences in incidence of intraventricular hemorrhage.
In conclusion, this series represents a single geographic area cohort of prenatally diagnosed twin–twin transfusion syndrome over a 10-year period. Perinatal outcome was strongly linked to disease severity as assessed by stage at presentation and gestation at delivery. The use of therapeutic amnioreduction was associated with a satisfactory perinatal survival rate for early-stage disease. With progression of disease severity the interventions used in this series resulted in lower perinatal survival rates than observed in early-stage disease. The results of the recently completed Eurofetus randomized controlled trial of amnioreduction versus placental laser ablation are awaited with interest to provide further information on management options for twin–twin transfusion syndrome, particularly for more advanced stage disease.
1. Ville Y. Monochorionic twin pregnancies: “Les liaisons dangereuses.” Ultrasound Obstet Gynecol 1997;10:82–5.
2. Hecher K, Plath H, Bregenzer T, Hansmann M, Hackelöer BJ. Endoscopic laser surgery versus serial amniocentesis in the treatment of severe twin-twin transfusion syndrome. Am J Obstet Gynecol 1999;180:717–24.
3. Mahony BS, Petty CN, Nyberg DA, Luthy DA, Hickok DE, Hirsch JH. The “stuck-twin” phenomenon: Ultrasonographic findings, pregnancy outcome, and management with serial amniocenteses. Am J Obstet Gynecol 199;163:1513–22.
4. Elliott JP, Urig MA, Clewell WH. Aggressive therapeutic amniocentesis for treatment of twin-twin transfusion syndrome. Obstet Gynecol 1991;77:537–40.
5. Saunders NJ, Snijders RJM, Nicolaides KH. Therapeutic amniocentesis in twin-twin transfusion syndrome appearing in the second-trimester of pregnancy. Am J Obstet Gynecol 1992;166:820–4.
6. Ville Y, Hyett J, Hecher K, Nicolaides K. Preliminary experience with endoscopic laser surgery for severe twin-twin transfusion syndrome. N Engl J Med 1995;332:224–7.
7. De Lia JE, Kuhlmann RS, Harstad TW, Cruikshank DP. Fetoscopic laser ablation of placental vessels in severe previable twin-twin transfusion syndrome. Am J Obstet Gynecol 1995;172:1202–11.
8. Ville Y, Hecher K, Gagnon A, Sebire N, Hyett J, Nicolaides K. Endoscopic laser coagulation in the management of severe twin-to-twin transfusion syndrome. Br J Obstet Gynaecol 1998;105:446–53.
9. Saade GR, Belfort MA, Berry DL, Bui TH, Montgomery LD, Johnson A, et al. Amniotic septostomy for the treatment of twin oligohydramnios-polyhydramnios sequence. Fetal Diagn Ther 1998;13:86–93.
10. Taylor MO, Shalev E, Tanawattanachareon S, Jolly M, Kumar S, Weiner E, et al. Ultrasound-guided umbilical cord occlusion using bipolar diathermy for stage III/IV twin-twin transfusion syndrome. Prenat Diagn 2002;22:70–6.
11. Dickinson JE, Evans SF. Obstetric and perinatal outcomes from The Australian and New Zealand Twin-Twin Transfusion Syndrome Registry. Am J Obstet Gynecol 2000; 182:706–12.
12. Mari G, Roberts A, Detti L, Kovanci E, Stefos T, Bahado-Singh R, et al. Perinatal morbidity and mortality rates in severe twin-twin transfusion syndrome: Results of the International Amnioreduction Registry. Am J Obstet Gynecol 2001;185:708–15.
13. Quintero RA, Morales WJ, Allen MH, Bornick PW, Johnson PK, Kruger M. Staging of twin-twin transfusion syndrome. Am J Perinatol 1999;19:550–5.
14. Dorman K, Saade GR, Smith H, Moise KJ Jr. Use of the World Wide Web in research: Randomization in a multicenter clinical trial of treatment for twin-twin transfusion. Obstet Gynecol 2000;96:636–9.
© 2003 The American College of Obstetricians and Gynecologists
15. Baker ER, Eberhardt, Brown ZA. “Stuck twin” syndrome associated with congenital cytomegalovirus infection. Am J Perinatol 1993;10:81–3.