We performed a Cochrane review to assess which of the treatments for twin–twin transfusion syndrome (TTTS) improves fetal, childhood and maternal outcomes. This article represents a version of the review which includes additional data to the published version.
We searched the Cochrane Pregnancy and Childbirth Group's Trials Register (April 2007) and the Cochrane Central Register of Controlled Trials (The Cochrane Library, October 2007) for randomized and quasi-randomized studies of amnioreduction, laser coagulation and septostomy and compared their outcomes. We also searched conference proceedings and contacted the authors of published trials for clarification and addtional data. No trials compared intervention with no intervention. Three studies (253 women) were included.
Laser coagulation resulted in less overall death (48% vs. 59%; relative risk (RR), 0.81; 95% CI, 0.65–1.01 adjusted for clustering; two trials, 364 fetuses), perinatal death (26% vs. 44%; RR, 0.59; 95% CI, 0.40–0.87 adjusted for clustering; one trial, 284 fetuses) and neonatal death (8% vs. 26%; RR, 0.29; 95% CI, 0.14–0.61 adjusted for clustering; one trial, 284 fetuses) when compared with amnioreduction. There was no difference in perinatal outcome between amnioreduction and septostomy. More babies were alive without neurological abnormality at the age of 6 months in the laser group than in the amnioreduction group (52% vs. 31%; RR, 1.66; 95% CI, 1.17–2.35 adjusted for clustering; one trial). There was no difference in the proportion of babies alive at 6 months that had undergone treatment for major neurological abnormality between the laser coagulation and the amnioreduction groups (4% vs. 7%; RR, 0.58; 95% CI, 0.18–1.86 adjusted for clustering; one trial).
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
Historically, twin–twin transfusion syndrome (TTTS) was associated with the death of one or both fetuses in more than 80% of untreated pregnancies, particularly if problems developed before 28 weeks' gestation1, 2. Sudden deteriorations could occur, leading to death of the cotwin and neurological handicap in the survivor3, 4. The poor outcome of untreated TTTS led to the introduction of a number of interventions, including repeated serial amnioreduction, endoscopic laser photocoagulation of vascular anastomoses, amniotic septostomy and, in rare cases, in which the demise of the cotwin was certain, selective feticide. Data from observational studies suggested survival rates following serial amnioreduction of 37–60%2, 5, 6, with a risk of neurological damage of between 17% and 33%2, 5, 6. Published series on laser photocoagulation showed a 55–73% survival rate, with a 4.2% neurological handicap rate7–9. Septostomy, the deliberate creation of a puncture in the intertwin membrane, was described by Saade et al.1. Survival rates as high as 83% were achieved in observational studies but no figures for neurological outcome have been published. These early studies suggested that laser photocoagulation might result in lower neurological handicap rates compared with other treatments. Randomized controlled trials were therefore designed to evaluate the treatment options further.
We performed a Cochrane review to assess which of the treatments for TTTS improves fetal, childhood and maternal outcomes. This article represents a version of the review10, which can be accessed via The Cochrane Library, available online through Wiley InterScience (http://www.mrw.interscience.wiley.com/cochrane/). This version includes the results from a trial not included in the Cochrane review as it was published after the latter was submitted for publication. The data discussed here do not alter the conclusions of the original review.
We searched the Cochrane Pregnancy and Childbirth Group's Trials Register by contacting the Trials Search Co-coordinator (April 2007), who maintains the register. It contains trials identified from: quarterly searches of the Cochrane Central Register of Controlled Trials (CENTRAL); monthly searches of MEDLINE; hand searches of 30 journals and the proceedings of major conferences; and weekly current awareness searches of a further 36 journals plus monthly BioMed Central email alerts. Details of these searches can be found in the ‘Search strategies for identification of studies’ section within the editorial information about the Cochrane Pregnancy and Childbirth Group11. Trials identified through these searches are given a code (or codes) depending on the topic. The codes are linked to review topics, and the Trials Search Co-ordinator searches the register for each review using these codes rather than keywords.
In addition, we searched the Cochrane Central Register of Controlled Trials (The Cochrane Library) using the following search strategy: (1) TWIN-TWIN; (2) (TWIN near TRANSFUSION); (3) SEPTOSTOMY; (4) ((LASER next COAGULATION) and TWIN-TWIN); (5) (FETO-FETAL next TRANSFUSION); (6) (FOETO-FOETAL next TRANSFUSION); (7) AMNIOREDUCTION; (8) AMNIODRAINAGE; (9) ((LASER next ABLATION) and TWIN-TWIN); (10) any combination of these nine search terms. The last search was on 13th October 2007.
We also searched the following conference proceedings (in February 2007): British Maternal and Fetal Medicine Society, Annual Clinical Meeting of the Society for Maternal and Fetal Medicine, International Society of Ultrasound in Obstetrics & Gynecology and World Congress of Obstetrics and Gynaecology. We did not apply language restrictions for any of the searches.
Due to an anticipated lack of randomized controlled trials, we considered both quasi-randomized (allocation to different arms by a method not truly random, e.g. by medical record number or date of birth) and randomized studies of one treatment versus another or no intervention. We considered women with a twin pregnancy with TTTS diagnosed on ultrasound using the minimum criteria of: confirmation of monochorionicity; oligohydramnios in one sac and polyhydramnios in the other; normal anatomy of both fetuses. We considered any intervention performed as a therapy for TTTS with a view to improving maternal symptoms, improving fetal, neonatal and childhood outcomes and prolonging pregnancy. Comparisons of amnioreduction versus laser coagulation, septostomy versus laser coagulation or septostomy versus amnioreduction were suitable for inclusion.
The outcome measures we considered included pregnancy, perinatal and maternal outcomes, as well as procedure-related details. Pregnancy outcome included: fetal death; death of at least one infant per pregnancy; death of both infants per pregnancy, preterm labor within 48 h of the procedure; preterm prelabor rupture of membranes within 48 h of the procedure; preterm labor/prelabor rupture of membranes prior to 32 weeks' gestation. Perinatal outcome included: gestational age at delivery; use of mechanical ventilation; fetal hemoglobin discordance at delivery; need for blood transfusion within 48 h of delivery; weight/head circumference discordance at delivery; ventriculomegaly; intraventricular hemorrhage (any grade); intraventricular hemorrhage (Grades III/IV); cystic periventricular leukomalacia (PVL); seizures within 28 days of delivery/anticonvulsant therapy; admission to neonatal intensive care unit; length of stay on neonatal intensive care unit; perinatal death; early/late neonatal death; overall death; developmental delay at less than 2 years; developmental delay at 2 years. Maternal outcome included: tocolysis; maternal death; amniotic fluid embolism; placental abruption; chorioamnionitis; intraperitoneal bleeding; admission to intensive care unit for procedure-related reasons; relief of symptoms; maternal satisfaction with the procedure. Procedure-related details included: number of interventions per pregnancy; first intervention to delivery time; need for a combination of therapies; type of anesthesia.
Three studies were identified for inclusion. We assessed the trials for methodological quality and appropriateness for inclusion without consideration of results. The trials were not assessed blind; the review authors knew the authors' names and institutions and the sources of publication. We resolved disagreement by discussion until a consensus was reached. One author (D.R.) extracted the data and a second author (M.K.) double-checked them for discrepancies. Authors of included studies were contacted for clarification and additional data. We processed the data as described by Clarke and Oxman12. Allocation concealment was assessed using the criteria described in section six of the Cochrane Handbook (Clarke 2000b)13: adequate (A), unclear (B), inadequate (C) or not used (D). Studies rated D were not used. Information about blinding was collected. Completeness of follow-up for the outcomes outlined was categorized as follows: fewer than 5% participants excluded (A), 5–9.9% participants excluded (B), 10–19.9% participants excluded (C), ≥ 20% participants excluded (D), unclear (E). We excluded studies rated D. We analyzed outcomes on an intention-to-treat basis.
We performed statistical analysis using Review Manager software (RevMan 2000)14. For dichotomous data, relative risks and 95% CIs were calculated. The analysis involved multiple pregnancies; therefore, whenever possible and whenever there was evidence of design effect suggesting correlation between babies from the same pregnancy, analyses were adjusted for clustering15. Treating babies from multiple pregnancies as if they were independent, when they are more likely to have similar outcomes than are babies from different pregnancies, would overestimate the sample size and give confidence intervals that were too narrow. Each woman can be considered a cluster in multiple pregnancy, with the number of individuals in the cluster being equal to the number of fetuses in her pregnancy. Analysis using cluster trial methods allows calculation of relative risk and adjustment of confidence intervals. Usually this will mean that the confidence intervals get wider. Although this may make little difference to the conclusion of a trial, it avoids misleading results in those trials where the difference may be substantial. The inverse variance method for adjusted analyses, as described in the Cochrane Handbook, was used16.
We limited primary analysis to prespecified outcomes. Subgroup analysis was planned for maternal and perinatal outcomes in percutaneous versus open procedures, but all three trials included used percutaneous procedures. Differences in unspecified outcomes or subgroups were found and analyzed post hoc; these are identified clearly to avoid unjustified conclusions being drawn.
Description of included studies
Three studies were identified for inclusion: the 2004 Eurofetus trial17–20, the 2005 Moise trial21–23 and the 2007 NIH trial24–27, with data available from 253 women (506 fetuses) (Table 1). All three studies compared two treatment arms; the Moise trial compared serial amnioreduction with septostomy, while the other two compared laser photocoagulation with serial amnioreduction. The Moise and NIH trials were conducted in the USA whilst the Eurofetus trial was conducted in Europe.
Table 1. Characteristics of the three included studies of treatment in twin–twin transfusion syndrome (TTTS)
40 women with TTTS, randomized and treated < 24 weeks
Survival of donor and recipient twins to 30 days after delivery
Three planned interim analyses: after ¼, ½ and ¾ of the planned sample size included
Excluded: cervical length < 2 cm,
Trial stopped after 42 subjects included, at request
bladder filling 12–24 h after a
of investigators due to reluctance of referring
clinicians to refer to centers only offering laser
amnioreduction to 5 cm DVP
as part of trial
TOC also noted trend towards adverse outcome in
recipients in one arm
For all pregnancies in the NIH trial, the only data available for inclusion were for death of at least one infant per pregnancy, overall death and maternal death. While the published NIH study gives data for 30-day survival of fetuses treated in either arm, by donor or recipient status, per pregnancy and overall survival, it is not clear whether these are neonatal deaths alone or the data include fetal deaths as well; we therefore did not use the 30-day data. No data on other death outcomes are available from the published study. Results from the NIH trial were not included in the Cochrane review10, which was published before the results of the trial were published, and further information has been requested from the authors to allow full analysis of the main outcomes. The NIH study allowed crossover between arms of the trial when treatment was deemed to have failed, i.e. in cases of progressive hydrops or cardiac failure with imminent fetal demise (33% of cases in the amnioreduction arm, none in the laser arm).
The Moise trial allowed crossover to the amnioreduction arm if two consecutive septostomies failed to resolve oligohydramnios in the donor sac or if the recipient sac continued to have polyhydramnios (32% of cases). Laser ablation of the anastomotic vessels (one in each arm) and umbilical cord occlusion (two in the amnioreduction arm, one in the septostomy arm) were used in cases in which there was progression of TTTS.
The Eurofetus trial performed amnioreductions in the laser arm after 26 weeks' gestation as the trial protocol did not allow laser coagulation after that gestation (one case). Six women in the amnioreduction arm in this trial underwent laser treatment for progression of TTTS after repeated amnioreductions. Two cases in the laser arm required repeat laser treatment.
Most of the fetuses were classified as Quintero stage II or III in the Eurofetus trial and Quintero stage I–III in the Moise one. The NIH study only included fetuses with Quintero stage II–IV disease. It had more Quintero stage IV fetuses than had the Eurofetus study, performed an amnioreduction on all included pregnancies prior to randomization and randomized and treated only prior to 24 weeks' gestation as opposed to 26 weeks' in the Eurofetus trial.
The Eurofetus trial used corrections for clustering. The NIH one did not adjust for clustering in the published data because the data for the donors and the recipients were analyzed separately for most outcomes. The Moise one did not adjust for clustering.
Methodological quality of included studies
All three studies used computer-generated central randomization sequences in order to maintain adequate allocation concealment. These studies were coded A for allocation concealment. Performance bias is unlikely to have occurred in the studies included in this review. All three studies describe prospective sample-size calculations. The Moise study lost two women to follow up in the septostomy arm. The NIH trial lost one woman to follow up in each arm. Thus, all three trials had fewer than 5% participants excluded.
Two trials were stopped early on the basis of interim analysis: in the Moise trial, the recruitment rate was slower than predicted and it was felt that the primary end point would not be achieved; the Eurofetus trial was stopped after the second interim analysis showed a higher rate of survival of at least one twin in the laser arm. The NIH trial was stopped at the request of the investigators, who cited the increasing unwillingness of referring physicians to refer eligible subjects to centers where laser photocoagulation was available only to patients participating in the trial.
The results of the main comparisons are shown in Figures 1 and 2; the results of all outcome measures are available online (Figures S1 and S2). Only data adjusted for clustering are included. Table 2 shows both unadjusted and adjusted data, to show the difference in CIs as a result of cluster adjustment. All the unadjusted data are presented with the number randomized as the denominator. Most clinicians find it counterintuitive to consider neonatal death as a proportion of the number randomized rather than as a proportion of the live births but this method is unbiased as it compares randomized groups. The most unbiased assessment of the results is to use the analyses adjusted for clustering.
Table 2. Comparison of relative risks (RR) and 95% confidence intervals (CI) for outcomes unadjusted and adjusted for clustering in three studies of treatment in twin–twin transfusion syndrome
There were no significant differences between septostomy and amnioreduction for fetal death (13% vs. 12.5%), death of at least one infant per pregnancy (40% vs. 50%), death of both infants (20% vs. 22%), neonatal death (26% vs. 24%), overall death (30% vs. 36%). There was a significantly higher need for a combination of therapies in the septostomy arm of the trial (40% vs. 8%).
Endoscopic laser surgery versus amnioreduction (Eurofetus and NIH trials, 182 pregnancies, 364 fetuses; Figure 2)
There were more terminations of pregnancy (16% vs. none, RR, 0.04; 95% CI, 0.00 to 0.70, one trial, 284 fetuses) in the amnioreduction group compared with the laser photocoagulation group. In infants treated with laser photocoagulation compared with those treated by amnioreduction, overall death was lower (48% vs. 59%, two trials, 364 fetuses), as were perinatal death (26% vs. 44%, one trial, 284 fetuses) and neonatal death (7.6% vs. 26%, one trial, 284 fetuses). There was no significant difference in the death of at least one infant (58% vs. 63%, two trials, 182 pregnancies) but there was in the number of pregnancies in which both infants died (24% vs. 49%, one trial, 142 pregnancies). This effect was significant in pregnancies with Quintero stages I and II disease (14% vs. 42%, one trial, 73 pregnancies) as opposed to Quintero stages III and IV (34% vs. 56%).
There was less cystic PVL in the babies treated with laser than in those treated with amnioreduction (5% vs. 14%, one trial, 284 fetuses). This difference was not significant at 6 months (3% vs. 4%, one trial, 284 fetuses). More babies were alive without neurological (motor) abnormality at the age of 6 months in the laser group than in the amnioreduction group (52% vs. 31%, one trial, 284 fetuses) but there was no difference in the number of babies alive with neurological abnormality at the age of 6 months (4% vs. 7%, one trial, 284 fetuses).
There were no significant differences for fetal death after 24 weeks' gestation (19% vs. 21%, one trial, 284 fetuses), fetal death within 7 days of the procedure (11% vs. 6%, one trial, 284 fetuses) and need for a combination of therapies (1.4% vs. 8.6%, one trial, 284 fetuses). No data were available for the other outcome measures for either comparison.
The findings of this review support the use of endoscopic laser coagulation in the treatment of TTTS to improve perinatal and neonatal outcome. There was no difference in perinatal outcome between amnioreduction and septostomy, although septostomy was associated with needing a significantly greater number of additional therapies, such as laser, cord occlusion and serial amnioreductions, following the initial procedure.
We found no trials comparing intervention with no intervention, presumably because it is well accepted that the perinatal mortality for untreated TTTS is around 80%. However, this figure is the overall figure for all stages of the disease. Recently, Bebbington et al.28 reported on 42 cases in which no intervention in Stage 1 disease led to similar outcomes, in terms of gestational age at delivery, as did treatment with amnioreduction. This was a cohort study, in which the group that had amnioreduction were found to have progression of the disease requiring laser/cord cautery in 4/22 cases (18%), compared with none in the no-amnioreduction group. This would suggest that the two groups were different even though there was no difference in the mean amniotic fluid index between them at first consultation. A cohort study by Quintero et al.29 on stage-based treatment of TTTS suggested that the relationship between perinatal mortality and stage was more apparent in pregnancies treated by serial amnioreduction than in those treated by laser coagulation. Their results showed 100% neonatal survival per pregnancy and 96% perinatal survival in pregnancies treated by amnioreduction compared with 86% and 76%, respectively, in laser-treated ones. This study did, however, have twice the number of patients in the laser arm than in the amnioreduction arm and no explanation is given as to how they were allocated to the chosen intervention.
The case for no intervention in earlier stages of the disease or treatment by amnioreduction alone is not supported by the findings of this review. The trials included enrolled fetuses with largely Quintero stage I–III disease. Only six fetuses with Quintero stage IV disease were enrolled in the laser coagulation studies (two in the Eurofetus trial and four in the NIH trial), and two in the septostomy (Moise) trial, so the numbers are too small to reach any conclusions for this group of fetuses. In the Eurofetus trial, post-hoc subgroup analysis was performed on the number of fetal deaths per pregnancy according to Quintero stage. The results suggest that laser coagulation improves perinatal outcome in Stage I or II disease more so than it does in disease at higher stages. Such conclusions, drawn from post-hoc analysis, should be seen as hypothesis-generating rather than as evidence of effectiveness. Some weight is added to the argument by the NIH trial, which showed a statistically significant increase in fetal recipient mortality in the laser arm particularly within Quintero stages III and IV, suggesting that the benefit of this treatment may be greater in earlier stages of the disease. They also found that more recipients in the amnioreduction arm of the trial had treatment failure followed by neonatal demise. TTTS-related cardiac function changes in the recipient were cited as the most significant reason for this.
Eurofetus is the only study that has published initial follow-up data on surviving children. Short-term neurological complications were described as cystic PVL (diagnosed either in utero or in the postnatal period), intraventricular hemorrhage Grade III/IV, blindness and deafness. Neurological impairment was described as motor impairment. The brain abnormalities were diagnosed by ultrasound or magnetic resonance imaging performed at centers where the infants received care and were described whether diagnosed antenatally or postnatally. There was a significant difference between the arms of the study in the incidence of PVL at diagnosis. This difference was no longer apparent at 6 months of age. Additional information from the authors has shown that, of the eight fetuses in the laser arm with PVL, four died in the neonatal period and four survived. Among the 20 fetuses with PVL in the amnioreduction arm, there were six terminations, eight neonatal deaths and six survivors. This loss of fetuses in the antenatal and early neonatal periods explains the lack of difference at 6 months of age.
The finding that there was no difference in the number of babies alive with motor neurological impairment at 6 months may be secondary to the plasticity of the developing brain or to the demise of more severely affected fetuses. The Eurofetus group analyzed the ASQ (Ages and Stages Questionnarie) scores in 120 children (73 treated with laser, 47 with amnioreduction) who were alive at the age of 2 years. Normal ASQ scores were found in 72% and 89% of cases in the laser and amnioreduction arms, respectively20. The included septostomy (Moise) trial has no plans to follow up survivors from the trial.
The analyses in this review were adjusted for clustering for those outcomes in which it was possible to do so. The reasons for this have been explained in the methods section. The graphs demonstrate the effect of cluster analysis on the CIs: they become wider with cluster analysis, suggesting that the true effect lies over a wider upper and lower value, i.e. that the true sample size is smaller than one might think because of the inclusion of multiple pregnancies, making the estimate of effect less precise than it would be in unadjusted analysis. The results of the adjusted analysis did not alter the conclusions in this review but may do so if more pregnancies are added in the future.
All three trials can be criticized for stopping early, particularly because chance can lead to surprisingly large differences early in a trial, which disappear or reverse over time30. The estimates of a treatment effect will be biased when a trial is terminated at an early stage and the earlier the trial is stopped, the larger will be this bias. The reasons for stopping the Moise trial at the midway point interim analysis have been discussed in the Methods section. The data safety monitoring officer would have had to take into account various considerations, such as meta-analyses of data from comparable trials, other existing evidence external to the trial and the nature of the condition and its alternative treatments, before reaching the decision that the primary end-point would not be reached31, 32. It is most likely that the trial was stopped because of the slower than projected enrolment.
The Eurofetus trial was stopped at the second interim analysis by a statistician uninvolved in the study design or analysis of the data. There was no formal data monitoring committee. There was a higher rate of survival of at least one twin in the laser arm compared with the amnioreduction arm (P = 0.002), so the trial was stopped according to the O'Brien-Fleming rule33 with adjustment for multiple evaluations of the data (the rule states that at the second analysis the P-value should be < 0.015 in order to consider early termination of a trial). However, as discussed above, whichever method is used, the estimate of treatment effect may still be biased if a trial is stopped early.
The NIH trial was stopped at the request of the investigators after recruiting only a quarter of the sample size (42 pregnancies) in 5 years. Many eligible subjects were not referred because of the reluctance of referring physicians. The trial oversight committee also noted a trend of adverse outcomes affecting the recipient twin in one arm of the study and requested biostatistical analysis of this trend, which coincided with the request to stop from the investigators. The results of this study on its own must therefore be considered to have considerable bias, but future meta-analysis of all pre-specified outcomes using data from this trial will add more information which will be relevant to clinical practice.
Despite stopping early, these trials remain the best evidence for the treatment of TTTS currently available.
Implications for practice
Endoscopic laser coagulation of anastomotic vessels should be considered in the treatment of all stages of TTTS to improve perinatal and neonatal outcome. However, the current level of evidence does not allow us to state whether treatment with laser coagulation increases or reduces the risk of neurodevelopmental delay or intellectual impairment in childhood compared with other therapies. Amnioreduction can be retained as a treatment option for those situations in which the expertise for laser coagulation is not available or pending transfer to a unit where such treatment can be obtained, but there is no evidence from randomized trials to suggest that this is better than no treatment.
Implications for research
Randomized evaluation of interventions such as septostomy, serial amniocentesis and placental laser ablation or no intervention in cases of very mild forms of TTTS (Quintero stage 1) is required. The results of this review suggest that laser coagulation improves perinatal outcome in Stage I and II disease, but the study included in this review with data for Stage I was not powered to test this hypothesis specifically.
Acknowledgments to Kenneth Moise (Moise trial), Yves Ville and Michel Boulvain (Eurofetus trial), Timothy Crombleholme (NIH trial) for additional data supplied. Thanks to Sonja Henderson and the staff of the Pregnancy and Childbirth Group, Cochrane Office, for editorial support.