Correspondence: Dr Y. Ville, Fetal Medicine Unit, St. George's Hospital Medical School, Blackshaw Road, London SW 17 0QT, UK.
Objective To assess the clinical effectiveness of endoscopic laser coagulation of placental vessels in the
Design Prospective study.
Setting Three referral centres for the management of twin-to-twin transfusion syndrome.
Population One hundred and thirty-two pregnancies complicated by severe twin-to-twin transfusion syndrome, reflected by polyhydramnios and enlarged bladder of one twin and oligoanhydramnios and collapsed bladder of the other twin, presenting before 28 weeks of gestation.
Methods Prospective collection of data on pre-procedure assessment, the procedure and the follow up were collected prospectively. Laser coagulation of placental vessels crossing the intertwine membrane on the chorionic surface under sono-endoscopic guidance, followed by amniodrainage.
Main outcome measures Maternal and pregnancy complications, perinatal death and morbidity were assessed over the last five years with follow up of survivors.
Results Endoscopic laser was carried out at a median gestation of 21 weeks. The total number of surviving infants was 144 (55%) and there was at least one survivor in 97 cases (73%). At a minimum age of one year neurological handicap was suspected in six survivors (4.2%).
Conclusions The results of this multicentre study are similar to those in our original report on the first 45 cases. In comparison with serial amniodrainage, the survival rate may be similar, but the handicap rate in survivors appears much lower. This study stresses the need for a prospective study comparing treatment of severe transfusion syndrome threse two techniques.
Monochorionic twin pregnancies presenting during the second trimester with severe twin-to-twin transfusion syndrome is associated with a high risk of miscarriage, perinatal death and subsequent handicap in survivors1–5. Although survival can be improved by serial amniodrainage, there remains a risk of serious handicap in 15% to 50% of survivors1–5. A more recent development in the management of this condition is fetoscopy and laser coagulation to interrupt the placental vascular communications between the twins which may constitute the underlying mechanism of the syndrome9,10. Preliminary data have suggested that survival of at least one twin can be achieved in about 70% of pregnancies, and this improved survival may be associated with a substantially reduced risk of handicap (about 5%)11. We report the results of a multicentre study on endoscopic laser coagulation reports in the management of 132 pregnancies with severe twin-to-twin transfusion syndrome including at least one year follow up.
Sono-endoscopic laser coagulation was carried out in 132 pregnancies complicated by severe twin-to-twin transfusion syndrome with severe polyhydramnios during the second trimester of pregnancy, between 1992 and 1996. The women were managed in one of three centres by members of the original team who developed this technique11. The subsequent obstetric care was undertaken in the referring hospital.
In all cases, ultrasound examination demonstrated the characteristic features of twin-to-twin transfusion syndrome: same sex fetuses, single placental mass, thin intertwin membrane, polyhydramnios surrounding the recipient fetus whose bladder was enlarged, oligoanhydramnios around the smaller donor fetus whose bladder was collapsed.
The women were counselled with regard to the reported outcome in terms of number of survivors and risk of handicap associated with three available options of management: 10% and 30%, respectively, for expectant management4,6, 30% to 80% and 30%, respectively, for amniodrainage4,6,12–15 and 55% and 5% respectively for endoscopic surgery11. Termination of pregnancy was also presented as a possible option before 22 weeks of gestation although never requested as a first therapeutic option. Informed consent was obtained before the procedure. The study was approved by the respective local Ethics Committees. The results of the first 45 patients have already been reported11.
Laser coagulation of the communicating placental vessels
Detailed ultrasound examination was first performed to localise the placenta, the inter-twin amniotic membrane, the placental insertion of the umbilical cords and to ascertain any fetal anomalies. Umbilical artery Doppler was carried out to determine the presence or absence of end-diastolic flow. The appropriate site of entry on the maternal abdomen was chosen to avoid injury to the placenta or the fetuses and to allow access through the recipient sac to the inter-twin membrane, ideally perpendicular to the long axis of the donor fetus (Fig. 1).
Tocolytic therapy (indomethacin or diclofenac per rectum or salbutamol intravenously) and antibiotics (a third generation cephalosporin intravenously) were given for prophylaxis. Local anaesthetic (1 % lignocaine) was injected into the maternal abdomen down to the myometrium. Under continuous ultrasound visualisation, a rigid 2mm diameter 0° paediatric cystoscope (Olympus, Hamburg, Germany or Storz, Tuttlingen, Germany) housed in a 9.8 Charriere cannula was introduced transabdominally into the amniotic cavity of the recipient twin. A 400 pm diameter Nd:YAG laser fibre (Dornier, Munich, Germany) was then passed down the side-arm of the cannula to 1 cm beyond the tip of the fetoscope. A combination of ultrasonographic and direct vision was used to examine systematically the chorionic plate along the whole length of the inter-twin membrane to identify the crossing vessels, which were coagulated using an output of 30 to 50 watts for one to three seconds from a distance of about 1 cm16. Subsequently, amniotic fluid was drained through the endoscope cannula over a period of 10 to 15 minutes to obtain a subjective normalisation of the amniotic fluid volume on ultrasonographic examination. The women remained in hospital for 6 to 48 hours for observation.
Demographic characteristics, ultrasound findings and details on the endoscopic surgery or any further procedures were kept on a computerised database. Data on maternal and neonatal outcomes were obtained from the referring physicians and paediatricians involved in the follow up of the infants. Pregnancy outcome was examined in relation to the gestational age at presentation, previous amniodrainages, placental position, inter twin difference in size, presence or absence of fetal hydrops, umbilical artery Doppler study results, operative difficulties or complications and post-operative interventions. Analysis of variance, χ2 and Student's test were carried out to determine the significance of differences between groups.
Pre-operative characteristics are listed in Table 1. In 13 cases (10%) the recipient fetus was hydropic and in one of these there was intracerebral periventricular hyperechogenicity. In another seven cases, the donor fetus had cerebral ventriculomegaly.
Table 1. Characteristics describing the population and the procedure in relation to the number of survivors. Values are given as n(%), mean [SD] or median range. GA = gestational age; NS = no statistically significant difference (P > 0.05); AC ratio = difference in abdominal circumference between the two fetuses in relation to the abdominal circumference of the larger one; AEDF = absent end-diastolic flow.
No. of survivors
No. of cases
GA at presentation (weeks)
A:C ratio (%)
Deepest pool of amniotic fluid (cm)
Velamentous cord insertion
AEDF before the procedure (donor)
Amniotic fluid drained (L)
Operative time (min)
In 123 cases the endoscope was introduced into the amniotic cavity through a placenta-free area; in 57 cases (43%) the placenta was entirely anterior and a very lateral insertion was used. Even then, in nine cases the entry was transplacental. Good visualisation of the inter-twin membrane and of the crossing vessels was achieved in all cases at the first attempt. However, in five cases, a greenish-black amniotic fluid, presumably due to a previous intra-amniotic haemorrhage originally impaired visualisation, and the fluid was replaced with pre-warmed Hartmans solution before the communicating placental vessels could be visualised. In four of the 132 cases there was intraoperative bleeding from a chorionic plate vessel. In three cases this was controlled by laser coagulation of the vessel proximally and distally to the bleeding site and the procedure could be completed, but in another the procedure was abandoned and repeated one week later because of a recurrence of the polyhydramnios. Table 1 presents the intraoperative characteristics.
Following the initial endoscopic surgery, further interventions were carried out in 26 cases (20%)including repeated endoscopic laser coagulation, fetal blood sampling with or without intrauterine transfusion, amniodrainage, selective fetocide, and cervical cerclage.
In 12 cases there was recurrence of polyhydramnios 1 to 11 weeks (median 3 weeks) after the original procedure, and in one of these the recipient fetus became hydropic. The subsequent management included endoscopic laser in five cases, amniodrainage in six and delivery in one (Table 2). In six cases, 1 to 28 days following the endoscopic surgery, one of the fetuses was noted to have mild ascites and skin oedema, which was most prominent in the upper part of the body The donor was involved in four cases and the recipient in two. Cordocentesis was performed, which demonstrated severe anaemia in two, severe thrombocytopenia with mild anaemia and acidaemia in one, and normal indices in the other three cases. The first three cases were treated with exchange blood transfusions and the other three expectantly. Five of these 12 infants survived, including two of the three who received a blood transfusion.
Table 2. Management and pregnancy outcome at gestational age (GA) of patients with recurrence of polyhydramnios after the initial endoscopic surgery. GA = gestational age in weeks, LB = live birth, NND = neonatal death, IUD = intrauterine death.
Outcome (GA) Recipient/donor
*A cordocentesis demonstrated a haemoglobin concentration of 3.3 g/dL in the former recipient and an intrauterine blood transfusion was given.
LB (32)/IUD (30)
In six of the cases where the donor fetus had ventriculomegaly, selective fetocide was performed at the parents’ request after laser coagulation. In two cases the normal baby survived, but in four there was either intrauterine death of the normal twin or spontaneous abortion within one week.
In one case despite endoscopic surgery at 20 weeks, there was progressive deterioration in the condition of the hydropic recipient with signs of congestive heart failure (bilateral atrioventricular regurgitation and absent flow in the pulmonary trunk); the severely growth restricted donor fetus remained anuric over the following three weeks. In view of the likely poor prognosis for both twins, termination of pregnancy was carried out at the parents’ request.
In one case successful endoscopic surgery was carried out at 19 weeks of gestation. At 24 weeks during a routine antenatal visit, the mother was found to have a cervical dilatation of 4 cm with hourglass prolapse of the membranes. The membranes were reduced and a cervical suture inserted under general anaesthesia. At 30 weeks the suture was removed after spontaneous onset of labour and healthy twins were born.
There were no maternal deaths, amniotic fluid embolism or placental abruption related to the procedure. Two women had severe pre-eclampsia. In one woman, proteinuria and hypertension were noted from 2 1 weeks of gestation. Endoscopic surgery was carried out unevenfully at 22 weeks. Over the subsequent four weeks, the amniotic fluid level normalised and both fetuses grew normally and had normal biophysical profile. At 26 weeks, the patient developed disseminated intravascular coagulation and despite intensive resuscitation attempts, she died of multiple organ failure within six hours. The post-mortem examination demonstrated that there was no evidence of chorioamnionitis or amniotic fluid embolism. It was therefore concluded that maternal death was not related to the procedure. In the second woman, an uneventful endoscopic surgery was carried out at 21 weeks. At 27 weeks, she developed fulminating preeclampsia and an emergency caesarean section was performed. Eclamptic fits and adult respiratory distress syndrome necessitating intensive care therapy complicated the immediate postpartum period. She made a full recovery and was discharged three weeks later.
Immediately after the procedure in four women there was abdominal pain that lasted for up to five hours; this was associated with the development of ascites, presumably due to leakage of amniotic fluid into the abdominal cavity (Table 3). In three women there was spontaneous resolution of the ascites within 12 hours, but in the fourth ultrasound examination demonstrated an intraperitoneal blood clot measuring 6 cm in diameter and maternal haemoglobin concentration fell to 5g/dL. This woman was transfused with seven units of blood and developed pulmonary oedema, which responded to diuretic therapy. A pulmonary arterial sampling via a Swan Ganz catheter excluded amniotic fluid embolism and amniocentesis excluded chorioamnionitis. The woman was discharged from hospital after eight days.
Table 3. Maternal, pregnancy and neonatal complications related, or possibly related to, endoscopic laser surgery in 132 cases.
Maternal or pregnancy complications:
Intraperitoneal haemorrhage or loss of amniotic fluid
Miscarriage within 2 weeks of surgery
Preterm prelabour rupture of membranes
Recurrence of polyhydramnios
Delivery before 32 weeks (156 live births)
Neonatal death (156 live births)
Confirmed neurological deficit (144 survivors)
Possible neurological deficit (144 survivors)
In 97 of the 132 pregnancies (73%) there was at least one survivor. Overall, 144 out of 264 fetuses (55%)survived. In 47 pregnancies (36%) both babies survived, in 50 one baby (38%) survived, and in 35 (26%) there were no survivors (Table 4). Fig. 2 illustrates the interval between endoscopic laser treatment and delivery for the three outcome groups. Pregnancy complications included spontaneous abortion and recurrence of polyhydramnios (Table 3). In 13 women (10%) there was preterm prelabour rupture of membranes with evidence of oligohydramnios in the presenting twin 3 to 16 weeks after endoscopic surgery. These pregnancies were man- aged expectantly, and the women were delivered one to five weeks (median two weeks) after amniorrhexis; none of these women developed chorioamnionitis.
Table 4. Cross-tabulation expressing fetal outcome in the donor twin (Y-axis) and in the recipient twin (X-axis). SA = spontaneous abortion, IUD = intrauterine death, TOP = termination of pregnancy, SF = selective fetocide, NND = neonatal death.
Outcome of donors
Outcome of recipients
Follow up of survivors
There were 156 live births, but 12 babies died in the neonatal period from prematurity-related complications (n= 10) and congenital anomalies (one case with severe pulmonary stenosis and one case with renal agenesis) (Table 3). One hundred and forty-four survivors were followed for 12 to 51 months, with regular examinations; and 138 (95.8%) are developing normally. However, six are handicapped or will potentially become handicapped (Table 5).
Table 5. Follow up of the survivors with confirmed or potential significant handicap. R = recipient; D = donor; IUGR = intrauterine growth restriction; IUD = intrauterine death; GA = gestational age; CT scan = computerised tomography scan; IVH = intraventricular haemorrhage.
GA at Fetoscopy (weeks)
GA at Delivery (weeks)
Perinatal findings and follow up
R: Hydrops, Periventricular hyperechogenicity
IUD at 27 weeks
CT scan in the neonatal period: periventricular leukomalacia. Spastic quadriplegia at 36 months of age.
Alive and well
Recurrence of polyhydramnios; Grade 3 IVH and severe anaemia. Mild spasticity with normal CT scan at 12 months of age.
IUD at 21 weeks
Progressive porencephaly Cerebral palsy at 22 months of age
Alive and well
Progressive microcephaly Cerebral palsy at 28 months of age
IUD at 25 weeks
Placenta previa. CT scan at 8 months of age: Partial atrophy left hemisphere Clinically normal at 18 months of age
IUD at 20 weeks
Obstructive hydrocephalus shunted early after birth. Clinically normal at 40 months of age
Comparison of groups
There was no statistically significant difference in the characteristics analysed in the three outcome groups (Table 1). The survival or intrauterine death of the donor or the recipient appeared to be influenced by three preoperative and one perioperative conditions. The A:C ratio was lower in the group in which the donor rather than the recipient survived (17.4% vs 23.5%; P < 0.01) signalling a lower inter-twin discrepancy in size, although no cutoff threshold was clearly identified. The survival rate in the donors was also higher when umbilical artery end-diastolic flow was present rather than absent (56% vs 25%, P < 0.03). Prior amniodrainage was carried out in 18% of cases where the recipient died in utero compared with 6% in the cases where the recipient survived (P= 0.06). In both donor and recipient survivor groups, the duration of the procedure was shorter when compared with the group with no survivors (38 vs 45 minutes; P < 0.05). There was no significant differences in the energy used, the number of shots or the number of vessels coagulated between the two groups.
In the cases where the polyhydramnios recurred, compared with those with no recurrence, the number of shots and the energy used were significantly lower (31 vs 69 shots, 1694 J vs 3355 J; P= 0.01 and P < 0.01, respectively) but the duration of the procedure was not significantly different (38 vs 40 minutes, P= 0.7). No factors were predictive of the development of hydrops in either the donor or the recipient following the procedure.
This multicentre study confirms the feasibility of an endoscopic technique for the coagulation of anastomotic placental vessels and interruption of the twin-to-twin transfusion syndrome. Irrespective of the vascular nature of the anastomoses or their depth within the placenta, their afferent and efferent branches are superficial and can be seen on the chorionic surface of the placenta7,17. Although the intertwin membrane does not necessarily overlie these common cotyledons, the systematic coagulation of all crossing vessels should inevitably include the branches of these anastomoses. A preliminary report I suggested that this treatment may be associated with a higher survival rate and a lower perinatal morbidity compared with treatment based on serial drainage of amniotic fluid4,6,8,13–15.
With endoscopic surgery, survival both with time and in different centres remains stable at 55% for fetuses and more than 70% for pregnancies with at least one survivor. In contrast, with the alternative method of treatment, serial amniodrainage, there are marked differences in reported results from different series. As reported by Saunders et al.4, in a series before 199 1 the survival was 30% to 40%, whereas in the last six years, using apparently the same technique the survival improved up to 83%6. Thus, three centres in five different papers6,12–15 have reported that in a total of 53 pregnancies treated with serial amniodrainage and in 14 managed with watchful waiting, the fetal survival rates were 76% and 36%, respectively, suggesting that the condition was milder than in previous studies. Also of importance, the largest series included six dichorionic pregnancies12,15. The most recent publication on serial amniodrainage reported the treatment of 26 pregnancies and the survival rate was 57%8.
An important issue in the management of severe twin-to-twin transfusion syndrome is the rate of severe handicap in survivors. With endoscopic laser the rate of neurological handicap is < 5% which is much lower than in pregnancies treated with serial amniodrainage. Thus, Pinette et al.6 reported cerebral palsy in 36% of their survivors, Bajoria et al.7 noted the presence of porencephalic cyst in 29% and cardiac dysfunction in another 29% of their 14 survivors and Trespidi et al.8 reported severe handicap in 15% of their survivors. The main underlying mechanism of perinatal handicap in twin-to-twin transfusion syndrome is thought to be an acute haemodynamic imbalance between the twins, following a hypotensive period or fetal demise, leading to an acute and significant transfer of blood from the normotensive to the hypotensive or dead fetus. This continues until the transfusing twin's mean arterial pressure equals the transfused twin's systemic filling pressure. In two series including a total of 32 monochorionic diamniotic pregnancies with death of one fetus, 17 (53%) of the surviving co-twin were handicaped.18,19.
Attempts to model the inter-twin circulation in order to define high and low risk vascular patterns7,20 remain without clinical relevance because there is no clear relation between placental histological findings and morbidity in the survivors. Bejar et al.21 reported necrosis of the cerebral white matter in surviving twins from monochorionic diamniotic pregnancies. This occurred in six of seven cases with combined venovenous (VV), arterioarterial (AA) and arteriovenous (AV) anastomoses, in three of nine with combined AV and AA, in two of eight with AA, one of twelve with AV, and none with single venovenous anastomoses. In addition, no clear in vivo vascular mapping of the placenta has yet been demonstrated to be feasible. Our attempt to coagulate all vessels crossing the inter-twin membranes is therefore the only treatment that can potentially prevent the severe haemodynamic imbalance when both twins are alive or in cases of single fetal demise.
The survival rate in the three centres involved in this study is similar to both our first report and the results of De Lia et al.22 who use a more invasive technique for introduction of the endoscope (general anaesthesia, laparotomy and hysterotomy), but a selective coagulation of suspected anastomoses. In contrast to our technique of systematic coagulation of all vessels crossing the inter-twin membrane, De Lia23 used a larger endoscope to get a more panoramic view of the placental surface and confined coagulation only to the anastomotic vessels in the vascular equator of the placenta. There are three main causes of fetal death associated with endoscopic surgery. First, miscarriage with varying contributions from the polyhydramnios of the disease and the procedure itself. Secondly, intrauterine death, most commonly of the donor fetus. The severely growth restricted donor fetus, with features of primary placental insufficiency and hypoxia, is often subjected to the additional trauma of acute obliteration of part of its placenta as a result of coagulation of all vessels crossing the inter-twin membrane. The third cause of death, often presenting as intrauterine death of both twins, may reflect failure of endoscopic surgery to obliterate all anastomotic vessels. Since there are many anastomoses and VA anastomoses between the twins, incomplete coagulation of all vessels could hasten the demise of the fetuses by worsening the inter-twin shunting of blood.
In this study we have been able to demonstrate the development of hydrops and anaemia following endoscopic surgery and death of one fetus. Since in such cases the surviving fetus was treated by intrauterine blood transfusions, it is possible that in pregnancies complicated by death of one fetus soon after the procedure, the use of cordocentesis in the survivor with the option of blood transfusion could reduce the cases where both fetuses die.
In our series 9% had prior amniodrainages. Since this group did not have a significantly worse outcome, it could be argued that the primary method of treating severe twin-to-twin transfusion syndrome is amniodrainage and only when this fails should endoscopic surgery be considered. However, the condition of both twins is likely to worsen in the meantime. This might be reflected by the tendency towards a higher incidence of previous amniodrainage in the group of recipient twins dying in utero. This would support the hypothesis that an earlier surgical approach could have prevented the recipient's demise.
In this series there was one maternal death. Although this death was unrelated to the procedure, it highlights the fact that twin-to-twin transfusion presenting with acute polyhydramnios during the second trimester of pregnancy is a severe pregnancy complication with potential risks for the mother. The only procedure-related serious complication was intra-abdominal bleeding from the uterine puncture site that required blood transfusion. In this case amniotic membrane separation made it impossible to carry out amniodrainage as usual. Presumably, the presence of severe polyhydramnios and high intrauterine pressure prevented the closure of the puncture orifice which usually occurs by a localized contraction at the site of entry of the trocar. Another potential complication that was not observed in this series is pelvic haematoma from puncture of the lateral uterine wall vessels. Although in 43% of our cases an extensive anterior placenta necessitated a very lateral entry of the fetoscope, the uterine vessels were easily visualised by colour Doppler and were therefore avoided. Consequently, in such cases of anterior placentas we did not find it necessary, as De Lia et al.22 and Deprest et al.24 to perform laparotomy and insert the fetoscope through the fundus or the posterior wall of the uterus. The position of the placenta did not appear to be a major contributor to the success or failure of the procedure.
The results presented have been stable over the last five years and the fetal survival rate compares favourably with the results presented in the most optimistic series using serial amniodrainage. Importantly, the rate of handicap among survivors appears to be much lower with endoscopic surgery. Nevertheless, the only accepted way of comparing endoscopic treatment with serial amniodrainage is a randomised study. On the basis of our results and an average of those of the recent series on serial amniodrainage6–8,12–15, a randomized controlled trial to demonstrate significant difference in survival between the two methods (Power go%), it would require a minimum of 162 women. Such a number of participants would also allow to detect a difference of 25% in the rate of handicap among survivors.
We wish to thank Mr G. Barki and Storz Endoscope for their technical support in developing the instruments necessary to the procedure.