Cervical length as a prognostic factor for preterm delivery in twin-to-twin transfusion syndrome treated by fetoscopic laser coagulation of chorionic plate anastomoses
To evaluate the risk of spontaneous preterm delivery in relation to cervical length in severe cases of twin-to-twin transfusion syndrome (TTTS) diagnosed before 26 weeks of gestation and treated by laser coagulation of the intertwin placental anastomoses.
This was an observational study of cases of severe TTTS diagnosed before 26 weeks of gestation and treated by fetoscopic laser coagulation between January 2002 and December 2003 in three centers. Laser was the first-line treatment in all cases. Elective preterm deliveries before 34 weeks' for maternal or fetal indications, including intrauterine death of both fetuses, were excluded. Transvaginal sonographic measurement of cervical length prior to laser treatment and other risk factors of preterm delivery were evaluated by univariate and logistic regression analysis.
Among the 137 women included, the mean cervical length was 32 and 38 mm in women delivering before and at or after 34 weeks, respectively (P < 0.001). For a cervical length of < 30 mm, the risk of delivery before 34 weeks was 74%. Severity of the disease, deepest vertical pool of amniotic fluid and amniotic fluid volume drained after the procedure were not associated with preterm delivery. Logistic regression analysis identified three independent factors to predict preterm delivery: short cervical length (increased risk), parity (increased risk) and intrauterine death of one twin (decreased risk).
In TTTS treated by laser, without indications for elective preterm birth, cervical length before treatment is significantly associated with gestational age at delivery, and this is independent of parity, intrauterine death of one fetus and other risk factors. Copyright © 2004 ISUOG. Published by John Wiley & Sons, Ltd.
Monochorionic twin pregnancies complicated by severe twin-to-twin transfusion syndrome (TTTS) before 26 weeks of gestation are associated with a high risk of fetal loss, perinatal death and subsequent handicap in survivors1–6. This is at least partly related to the high risk of preterm delivery and of intrauterine fetal demise of one or both twins. Preterm delivery is likely to be related to increased pressure in the amniotic cavity due to the development of severe polyhydramnios7, 8. Survival is improved by fetoscopic laser coagulation of the chorionic plate anastomotic vessels, but the risk of preterm delivery remains high9, 10. In our recent randomized trial comparing laser therapy with amnioreduction, 17%, 29% and 54% of women delivered before 24, 28 and 34 weeks' respectively10.
Maternal factors contributing to TTTS have not been well studied. However, it is known that uterine factors play a critical role in the duration of pregnancy, especially in cases with severe polyhydramnios, and that cervical length is a predictor of gestational age at delivery in uncomplicated twin pregnancies11. Cervical length measurment was therefore recently introduced in the pre-operative assessment of TTTS in the practice of our three centers.
The aim of our study was to evaluate the risk of spontaneous preterm delivery in relation to cervical length in severe cases of TTTS diagnosed before 26 weeks of gestation and treated by laser coagulation of the intertwin placental anastomoses.
Included in the study between January 2002 and December 2003 were consecutive women with monochorionic twin pregnancies complicated by TTTS, presenting at less than 26 weeks of gestation in three referral centers, who were managed by coagulation of intertwin vascular anastomoses on the chorionic plate and in whom cervical length was measured transvaginally before the procedure. Laser was the first line treatment in all cases. The only exclusion criterion was elective preterm delivery before 34 weeks' gestation for fetal or maternal indications, including death of both fetuses.
Data were recorded prospectively in a dedicated database maintained in each center, and included maternal characteristics such as a previous history of late miscarriage or preterm delivery before 36 weeks, and parity. Cervical length was measured transvaginally within 24 h prior to the procedure, as described previously12.
Fetal condition at diagnosis was assessed according to the Quintero staging system13. Briefly, in stage I the bladder of the donor twin is still visible and Doppler studies are normal. In stage II, the bladder of the donor twin is not visible during the entire examination period, but Doppler waveforms are not critically abnormal. In stage III, Doppler studies are critically abnormal in either twin, with absent or reversed end-diastolic flow in the umbilical artery, reverse flow in the ductus venosus or pulsatile umbilical venous flow. In stage IV, ascites, pericardial or pleural effusion, scalp edema or overt hydrops is present. Stage V is defined as the demise of one or both twins. The deepest vertical pool of amniotic fluid in the recipient sac was also recorded. Polyuric polyhydramnios in the recipient twin was diagnosed when the deepest vertical pool was at least 8.0 cm at or before 20 weeks' gestation or 10.0 cm after 20 weeks with a distended fetal bladder; oliguric oligohydramnios was diagnosed in the donor when the deepest vertical pool was at most 2.0 cm.
All procedures were performed percutaneously under local or regional anesthesia9. Prophylactic tocolytics (indomethacin 100 mg intrarectally) and antibiotics (cefazoline 1 g IV) were administered perioperatively; no laparotomy was performed. A 3.3-mm cannula with trocar under continuous ultrasound guidance was used and a 2-mm fetoscope (Karl Storz 11 630, 50 000 pixels, Storz Gmbh, Tuttlingen, Germany) was introduced. Primarily, coagulation selectively targeted both superficial and deep anastomoses whenever identified on the chorionic plate in the recipient sac. Vessels whose end could not be identified were also coagulated. At the end of the procedure, excessive amniotic fluid was drained through the cannula to obtain a maximum deepest pool of 5–6 cm on ultrasound, and its volume was recorded. Postoperative occurrence of a single fetal death following the procedure was also recorded. In cases with a cervical length of 20 mm or less, it was left to the obstetrician in charge either to perform a cerclage or to follow these cases expectantly.
The primary outcome of the study was spontaneous preterm delivery at less than 34 weeks of gestation. We also evaluated risk factors for spontaneous delivery before 28 and 32 weeks. Spontaneous delivery was defined as delivery following spontaneous labor or preterm rupture of the membranes.
Statistical analysis was aimed at identifying independent risk factors for preterm delivery. Univariate analysis was performed, with results reported as mean and SD, or as median and interquartile range (IQR) (25th–75th percentile) when the distribution was not normal. Categorical outcomes were reported as numbers and percentages. We used a logistic regression model to identify the independent predictors for preterm delivery. A P-value of less than 0.05 was considered statistically significant.
Three centers experienced in fetoscopic treatment of TTTS participated in the study. One hundred and thirty-seven women were included, of whom 33 had been included in a randomized clinical trial of amniocentesis versus laser therapy10. Table 1 summarizes the characteristics of the population. Gestational age at the time of treatment ranged from 16 to 26 weeks. One woman had a previous history of late miscarriage following cerclage and seven (5.1%) women had a history of preterm delivery before 36 weeks. The majority of pregnancies were classified as Quintero stage II or III. After treatment, 63 (46.0%) women had intrauterine death (IUD) of one fetus. Only ten patients required either repeat laser or cord coagulation for recurrence of TTTS or a major anomaly in one fetus. Gestational age at delivery was not different from that in the entire group.
Table 1. Characteristics of included pregnancies
|Maternal age (years)||30.2 (4.9)|
|History of delivery before 36 weeks' gestation||7 (5.1%)|
|Quintero stage|| |
| Stage I||17 (12.4%)|
| Stage II||56 (40.9%)|
| Stage III||61 (44.5%)|
| Stage IV||3 (2.2%)|
|Cervical length before treatment (mm)||35.1 (10.3)|
|Gestational age at treatment (weeks, median [IQR])||20.7 [18.9–22.4]|
|Gestational age at delivery (weeks, median [IQR])||34.0 [29.1–35.3]|
|Cervical cerclage||9 (6.6%)|
Eleven (8.0%) and 20 (14.6%) women, respectively, delivered within 7 and 28 days after the procedure. Premature rupture of the membranes occurred in 10 (7.3%) and 16 (11.7%) women, respectively, within 7 days and 28 days after the procedure. Among those, 4/10 (40%) and 1/16, respectively, had a cervical length of less than 20 mm before the procedure. Premature rupture of the membranes before 34 weeks occurred in 44 (32.1%) women. Of these, 11 (25.0%) patients had a miscarriage (delivery before 24 weeks) and 40 (90.0%) delivered before 34 weeks; 16 (36.4%) had a cervical length of less than 30 mm.
The median gestational age at delivery was 34 weeks. This was similar in the three centers (33.3 weeks in Paris and Leuven, 34.6 weeks in Hamburg). Delivery occurred in 13.1% (n = 18), 22.6% (n = 31), 36.5% (n = 50) and 49.6% (n = 68) of the women before 24 weeks, 28 weeks, 32 weeks and 34 weeks, respectively. Sixty-six (48.2%) women delivered vaginally and 71 (51.8%) by Cesarean section.
A total of nine women (six nulliparous and three parous) had cervical cerclage after the procedure because of a short cervix (< 20 mm). The three parous women had a previous history of cervical cerclage, late miscarriage and preterm delivery. In this group, four women miscarried, three delivered before 28 weeks, one at 33 weeks and one at 34 weeks.
Univariate analysis showed that mean cervical length was shorter in women delivering before 34 weeks (32.0 mm), compared with women who delivered at or after 34 weeks (38.1 mm; P < 0.001). The mean cervical length measured before the procedure was 27.2 mm, 32.0 mm, 36 mm and 38.1 mm in women delivering before 24, at 24–27, at 28–33 and at or after 34 weeks, respectively (P < 0.001). The risk of miscarriage or spontaneous preterm delivery decreased with increasing cervical length (Figure 1); a cervical length of less than 20 mm carried a poor prognosis. Of the 10 women with a cervical length < 20 mm, six had a miscarriage and only two delivered at or after 33 weeks.
There was a trend towards a higher risk of preterm delivery before 34 weeks in parous women (Table 2). IUD of one twin decreased the risk of preterm delivery before 34 weeks', this effect was not statistically significant for preterm delivery before 28 weeks' and 32 weeks' (Table 2).
Table 2. Risk of preterm delivery (before 34, 32 and 28 weeks of gestation) according to pregnancy characteristics
|Parity|| || ||0.07|| ||0.48|| ||0.76|
| Nulliparous||63||26 (41.3)|| ||21 (33.3)|| ||15 (23.8)|| |
| Parous||74||42 (56.8)|| ||29 (39.2)|| ||16 (21.6)|| |
|Quintero stage|| || ||0.15|| ||0.56|| ||0.07|
| Stage I–II||73||32 (43.8)|| ||25 (34.2)|| ||12 (16.4)|| |
| Stage III–IV||64||36 (56.3)|| ||25 (39.1)|| ||19 (29.7)|| |
|Cervical length (mm)|| || ||0.02|| ||0.03|| ||< 0.001|
| < 20||10||9 (90.0)|| ||8 (80.0)|| ||8 (80.0)|| |
| 20–24|| 8||5 (62.5)|| ||3 (37.5)|| ||2 (25.0)|| |
| 25–29||16||11 (68.8)|| ||7 (43.8)|| ||4 (25.0)|| |
| 30–34||21||10 (47.6)|| ||8 (38.1)|| ||4 (19.0)|| |
| ≥ 35||82||33 (40.2)|| ||24 (29.3)|| ||13 (15.9)|| |
|Survival|| || ||0.03|| ||0.08|| ||0.18|
| Death of one fetus||63||25 (39.7)|| ||18 (28.6)|| ||11 (17.5)|| |
| Both fetuses surviving||74||43 (58.1)|| ||32 (43.2)|| ||20 (27.0)|| |
There was no significant difference in the severity of the disease as assessed by the Quintero staging system between women delivering before and after 28 weeks, 32 weeks and 34 weeks. The mean deepest vertical pool in the recipient's sac was 118 (SD, 25) mm and 117 (SD, 27) mm and the mean amniotic fluid volume drained after the procedure was 2114 (SD, 1069) mL and 2010 (SD, 1158) mL in women delivering before and after 34 weeks, respectively. The mean deepest vertical pool of amniotic fluid was larger (124 mm) in parous women than it was in nulliparous women (110 mm; P = 0.001). Gestational age at diagnosis was on average 8 (SD, 32.12) days later in parous women (148 days vs. 140 days, P < 0.001).The mean cervical length before the procedure was similar between nulliparous and parous women (34.8 mm and 35.3 mm, P = 0.4).
Logistic regression analysis (Table 3) identified three independent risk factors for preterm delivery before 34 weeks: cervical length < 30 mm and parity were associated with an increased risk. However, parity was not a risk factor for preterm delivery before 32 or 28 weeks. IUD of one twin reduced the risk of delivering before 34 weeks, and was not associated with a higher risk of delivering before 28 or 32 weeks.
Table 3. Multiple logistic regression analysis for prediction of delivery before 34, 32 and 28 weeks of gestation
|Cervical length < 30 mm||3.53 (1.55–8.03)||0.003||2.18 (1.01–4.70)||0.047||2.87 (1.24–6.66)||0.01|
|Parous||2.27 (1.09–4.74)||0.03||1.40 (0.68–2.90)||0.36||0.96 (0.42–2.21)||0.93|
|Death of one fetus||0.45 (0.22–0.93)||0.03||0.53 (0.25–1.09)||0.08||0.57 (0.24–1.34)||0.20|
No fetuses delivered before 24 weeks survived. Among the 19 fetuses (of 13 pregnancies) born alive between 24 and 27 weeks, eight (42.1%) were alive at follow-up (1 month to 4 years). Among the 60 fetuses of 37 pregnancies born alive between 28 and 34 weeks, 54 (90%) were alive at follow-up (1 month to 4 years).
Our study shows that a short cervical length before treatment is an independent risk factor for preterm delivery in TTTS treated by fetoscopic laser coagulation of placental anastomoses.
This is in agreement with the results of previous studies describing the association between short cervical length and preterm delivery in uncomplicated twin pregnancies11, 14–16. A cervical length of less than 20 mm carries a particularly poor prognosis since six of the 10 cases had a late miscarriage and only two delivered at or after 33 weeks. However, since the only alternative treatment available is amnioreduction, which has been shown to lead to preterm delivery more often than does laser therapy10, a short cervical length should not disqualify these cases from surgery.
Parity has been reported to decrease the risk of preterm delivery in selected cases of uncomplicated multiple pregnancies17. However, this study excluded parous women with a previous history of preterm delivery. We found that parity was a risk factor for preterm delivery before 34 weeks, but not before 32 weeks or 28 weeks in TTTS. This might, therefore, have been due to chance, as this was found in a subgroup analysis and not pre-specified. However, there was a significantly larger deepest pool of amniotic fluid in the recipient's sac in parous women and similar cervical length in parous and nulliparous women. A larger deepest pool of amniotic fluid with a similar cervical length could reflect a better tolerance to uterine distension. However, although the later diagnosis in this group could also be explained by a better tolerance to uterine distension, this may contribute to increasing the risk of preterm delivery due to prolonged overdistension. Similar cervical length in both groups may mask a lower competence of the cervix in parous women under these extreme circumstances.
This paper reports on a subgroup of pregnancies which were not electively delivered before 34 weeks, excluding cases with double IUD as well as those with single or double survival delivered for maternal or fetal indications. Therefore, it does not allow calculation of survival rates following laser therapy. IUD of one twin was found to decrease the risk of preterm delivery before 34 weeks. Furthermore, a single IUD did not increase the risk of very preterm delivery before 28 or 32 weeks, and is even likely to decrease it at these gestational ages. The absence of statistical significance is likely to be due to the small size of these subgroups. This could best be explained by the influence of a single IUD on uterine distension. One should not extrapolate from this result that selective fetocide in TTTS could be considered as a first-line approach for at least three reasons other than ethical ones: i) a gain in severe prematurity is uncertain, ii) in the majority of these cases with single fetal demise following laser treatment, it was impossible to predict which of the donor or the recipient was going to die, and iii) selective fetocide in monochorionic twins remains a challenging procedure with uncertain outcome for this indication. Therefore, survival is unlikely to approach even 50% of cases, whereas laser therapy in severe TTTS leads to the survival of at least one twin in 76% of cases and a favorable long-term outcome in 90% of the survivors10. Our population did not allow us to evaluate if the death of one twin could be beneficial in cases with a short cervix before the procedure, since among cases with a cervix less than 20 mm in length, there were only two with a single IUD, which delivered at 23 and 36 weeks.
These data provide useful counseling information. In addition to factors predicting fetal survival, such as fetal Doppler or Quintero staging, cervical length is a strong predictor of preterm delivery. The risk is particularly high in women presenting with a cervix shorter than 20 mm. However, interventions which would prevent severe prematurity in these cases need further evaluation.
This study was supported by the European Commission in its fifth framework programme (EuroTwin2Twin contract QLG1-CT-2002-01632). L. Lewi and A. Huber are recipients of a research fellowship of the EC.