Twin–twin transfusion syndrome affects 15% of monochorionic multiple pregnancies and results in amniotic fluid and growth discordance with signs of uteroplacental insufficiency and hypovolemia in the donor, and volume overload and cardiac dysfunction in the recipient. The high perinatal mortality, attributable to intrauterine death, preterm delivery, and neonatal complications, has recently fallen from 80%, with conservative management, to 40%, with modern treatment modalities.1 Various poor prognostic factors have been described in both donor (eg, anuria, absent end-diastolic flow in the umbilical artery) and recipient fetus (hydrops, abnormal venous Doppler).2,3 Based on 50 cases, Quintero et al4 suggested an intuitive staging system incorporating most factors found independently predictive of outcome. Staging is increasingly used to guide management, with simpler, relatively safe, but less effective procedures like amnioreduction and septostomy preferred for cases with good prognosis, and technically challenging procedures like cord occlusion and endoscopic laser ablation of anastomoses, which are seemingly more effective but with a higher intrinsic risk of fetal loss, employed in cases with poor prognosis.5
Although worst stage correlates with outcome, stage at presentation has not been shown to predict perinatal survival.5,6 There remains a need to predict survival based on features apparent at presentation or, at least, before treatment. Ultrasound detection of an artery–artery anastomosis7,8 is the only factor predictive of perinatal survival3 not incorporated into Quintero’s staging system.
Ex vivo, in vivo, and computer modeling studies have shown that bidirectional artery–artery anastomoses protect against transfusional imbalance mediated by unidirectional arteriovenous anastomoses7–13 by allowing compensatory countertransfusion. We and others have shown that artery–artery anastomoses are less frequent in twin–twin transfusion syndrome than in non–twin–twin transfusion syndrome monochorionic placentas.9,10,13 The most common anastomotic pattern is absent artery–artery anastomoses with one or more arteriovenous anastomoses, with twin–twin transfusion syndrome occurring in 75% or more of the cases with this configuration.13 Approximately 85% of artery–artery anastomoses can be visualized antenatally using power or color Doppler, and consistent with ex vivo studies, artery–artery anastomoses are identified less frequently in pregnancies with, versus those without, twin–twin transfusion syndrome.8 Nevertheless, artery–artery anastomoses are still found in 25–30% of pregnancies with twin–twin transfusion syndrome.13,14 The risk of developing twin–twin transfusion syndrome in monochorionic pregnancies is lower if an artery–artery anastomosis is detected antenatally than if one is not.8 Furthermore, in those who do develop twin–twin transfusion syndrome, detection of an artery–artery anastomosis by ultrasonography confers a survival advantage.3 Using computer models, even small-diameter artery–artery anastomoses appear to compensate for a wide range of arteriovenous anastomotic flow and thus protect, or at least mitigate against, unbalanced transfusion resulting in twin–twin transfusion syndrome.11,12 In support of this, we recently reported a case in which thrombosis of an antenatally detected artery–artery anastomosis unmasked the hemodynamic imbalance imposed by arteriovenous anastomoses, resulting in acute-onset, twin–twin transfusion syndrome.15
An artery–artery anastomosis, if present in twin–twin transfusion syndrome, is usually detected before invasive treatment is instituted. In addition, we have observed that progression is infrequent in twin–twin transfusion syndrome in the presence of an artery–artery anastomosis detected on Doppler. Thus, unlike other prognostic variables, the presence or absence of an artery–artery anastomosis should predict outcome at presentation and thus facilitate treatment selection. We estimated whether Doppler detection of an artery–artery anastomosis predicts perinatal survival in twin–twin transfusion syndrome independently of disease stage.
MATERIALS AND METHODS
We evaluated 105 consecutive cases of twin–twin transfusion syndrome diagnosed by standard criteria (discordant amniotic fluid volume with deepest vertical pool less than 2 cm and more than 8 cm in sacs of donors and recipients, respectively, in monochorionic twin pregnancy) from July 1995 and delivered by March 2003. Ten patients were excluded for intrauterine death of one or both fetuses at presentation (n = 2), termination of both twins before 24 weeks (n = 3), dichorionic triplets (n = 3), and laser treatment performed before referral (n = 2). The final analysis comprised 95 monochorionic diamniotic pregnancies.
Gestational age was based on last menstrual period, if certain, or otherwise on first trimester ultrasonography. Monochorionicity was diagnosed by standard criteria.8 Ultrasonic surveillance was performed at least once every 2 weeks with an Acuson Sequoia or XP10 (Acuson, Mountain View, CA) for fetal growth, weight discordance, liquor volume, and phenotypic features in both donor (smaller size, reduced liquor, small bladder size) and recipient fetuses (increased size, hydramnios, chronically full bladder, and cardiomegaly). Doppler waveforms were obtained in each twin from the umbilical artery for the presence or absence of end-diastolic flow and from the umbilical vein and/or ductus venosus. Pulsatility in the umbilical vein synchronous with the cardiac cycle or absent/reversed end-diastolic flow in the ductus venosus during atrial contraction was deemed abnormal. Stage was determined at each visit (stage I, amniotic fluid discordance only; stage II, donor anuria; stage III, absence of end-diastolic flow in umbilical arterial or abnormal venous waveforms; stage IV, hydrops; and stage V, fetal death), prospectively from mid-1999 and, retrospectively, before this from data entered prospectively in an electronic database.4
An artery–artery anastomosis was sought by power or color Doppler ultrasonography at each visit until detected or until definitive treatment (cord occlusion or laser) was instituted. As previously described, the systematic search for artery–artery anastomoses was usually limited to 10 minutes per scan, and insonation was not pursued once an artery–artery anastomosis was detected.7,8 Artery–artery anastomoses were identified by their characteristic “speckled” appearance on color Doppler and their bidirectional periodic interference pattern waveform on pulsed wave Doppler (Figure 1). 16 Initially, artery–artery anastomosis detection was part of a research study approved by the institutional ethics committee to which patients gave informed consent, but later became part of routine clinical evaluation. Retrospective analysis was done as part of clinical audit. Per the Hammersmith, Queen Charlotte’s and Chelsea Research Ethics Committee, institutional review board approval for this analysis was not required.
The principles of management have been described previously.8 Briefly, in our institution, serial amnioreduction and/or septostomy (the latter as part of a multicenter randomized controlled trial) were the only treatments performed before mid-1999, and such treatment was indicated when the amniotic fluid index was 40 cm or greater. Afterward, treatment options were based on disease severity. Bipolar cord occlusion was offered as a preemptive procedure to save the co-twin when one fetus appeared preterminal,17 whereas, from 2002, highly selective laser ablation of placental vascular anastomoses was offered for stage III/IV disease. Delivery was considered at gestations of 28 weeks or more and was indicated in continuing pregnancies with 2 live fetuses at 32–34 weeks of gestation.
Placental dye injection studies were performed as described,13 and anastomoses were confirmed by chorionic plate inspection. Twenty-two placentas were excluded: 14 had intrauterine deaths (spontaneous or iatrogenic) remote from delivery, which precluded injection studies, and 8 subjects delivered abroad, declined consent, or had laser ablation performed. Seventy-three satisfactory placental injection studies were analyzed for sensitivity and specificity of antenatal artery–artery anastomoses detection. Neonates were followed for 28 days, and outcome data was confirmed from an institutional neonatal database or by correspondence with referring hospitals if delivered elsewhere.
The presence of an antenatally detected artery–artery anastomosis was determined at presentation, at first invasive treatment, and before delivery. At similar time points, disease stages at presentation, at first treatment, and at worst stage were also determined. Disease progression was defined as an increase in stage from that at presentation. In conservatively managed cases, delivery was considered the first invasive treatment. Principal outcome measures were perinatal survival rate (number of fetuses surviving 28 postnatal days/total number of fetuses), double survival rate (number of pregnancies with both fetuses surviving 28 postnatal days/total number of pregnancies), and any survival rate (number pregnancies with 1 or more fetuses surviving 28 postnatal days/total number of pregnancies). Statistical analysis was performed with SPSS 9.01 (SPSS Inc, Chicago, IL) and Stata 7.0 (Stata Corporation, College Station, TX). Pearson χ2 tests were used to test associations between nominal variables, and χ2 tests for linear-by-linear associations, with stage as an ordinal variable. Fisher exact test was used for low expected cell frequencies. Multiple logistic regression analysis was performed on a main effects model, consisting of 4 factors (artery–artery anastomosis detected before first treatment, Quintero stage at treatment, progression of disease, and first-treatment modality), to determine whether stage and artery–artery anastomosis independently predicted survival. Generalized estimating equations, which adjust for correlation among data points,18 were used to analyze the differences in perinatal survival per fetus in these twin pregnancies to avoid violation of the assumption of independence. Difference in log likelihood ratio χ2 tests by comparison of a reduced model (reducing the final model by 1 factor) and the final model was used in turn to determine the strength of association of each individual factor with survival. Stage V, as an adverse outcome variable itself, was excluded from analysis. One-tailed tests were used for known independent predictors of survival. Values of P < .05 are considered significant. Outcomes from some of these patients have been reported elsewhere.3,6–8,17
The maternal demographic distribution of gestational ages and stages at presentation, first invasive treatment, and worst stage are shown in Table 1. About half were stage I/II at presentation and at treatment. Disease progressed during pregnancy in 32 (34%). Seventy women (74%) had at least one invasive treatment procedure; 25 (26%) underwent cord occlusion (n = 22) or laser ablation (n = 3) with or without other prior treatments, whereas 45 women (47%) underwent only amnioreduction and/or septostomy. Twenty-five women (26%) were managed expectantly and/or delivered when indicated. The overall perinatal survival rate was 61% (115 of 190 fetuses). Twenty-two (23%), 31 (33%), and 42 (44%) pregnancies resulted in 0, 1, and 2 neonatal survivors respectively.
An artery–artery anastomosis was visualized in 24 pregnancies (25%), of which 18 (75%) were detected before or at diagnosis of twin–twin transfusion syndrome, and almost all (23 or 96%) before treatment. The sensitivity of Doppler for artery–artery anastomosis detection was 67% (22 of 33, 95% CI 48%, 81%) with 100% specificity (40 of 40, 95% CI 89%, 100%). No placenta had more than one artery–artery anastomosis.
Perinatal and double survival rates were better when an artery–artery anastomosis was detected at presentation (78% and 72%, respectively) than when one was not detected (56% and 38%, P = .046 and .008, respectively, Table 2). These differences increased as the rate of artery–artery anastomosis detection increased, both at first invasive treatment (P = .003 and < .001 for perinatal and double survival rates, respectively) and before delivery (P = .002 and < .001 for perinatal and double survival rates, respectively). In contrast, rates of any survival were not influenced by the presence of an artery–artery anastomosis. Disease progressed in 21% (5 of 24) with and 38% (27 of 71) without an artery–artery anastomosis (P = .1).
As shown in Table 3, perinatal and double survival were also influenced by stage, with rates decreasing as stage increased at all 3 time points. As with artery–artery anastomosis, stage did not influence the rate of any survival. Accordingly multiple logistic regression analysis was used to derive well-fitting models (P ≤ .001, using difference in log likelihood ratio χ2 between models), which demonstrated that both artery–artery anastomosis detection and stage at treatment were independent predictors of both perinatal and double survival rates. Disease progression was also an independent predictor of perinatal and double survival rates, whereas first-treatment modality was not. In the multiple logistic regression models, Doppler detection of artery–artery anastomosis at treatment yielded odds ratios of 5.1 (95% CI 1.6, 15.9) and 19.3 (95% CI 2.7, 138), respectively, for perinatal and double survival, independently of stage.
Because both factors independently predicted perinatal and double survival, we subcategorized stage by the presence (denoted “a”) or absence (denoted “b”) of an antenatally detected artery–artery anastomosis. Both perinatal and double survival rates were better with than without an antenatally detected artery–artery anastomosis for stages I–III at all 3 time points, with the clearest difference at time of first treatment (Figure 2). Stage IIIa at all time points was associated with survival at least comparable with that of stage Ib, and, indeed, at the time of first treatment, with better perinatal (83% versus 63%) and double survival (78% versus 52%, Figure 2). Empirically, rates of any survival were also higher for stages Ia, IIa, and IIIa, than for Ib, IIb, and IIIb, respectively (Figure 2). However, this did not apply in stage IV, in which there were no double survivors. Indeed, the presence of an artery–artery anastomosis on antenatal Doppler in pregnancies complicated by spontaneous intrauterine death was associated with worse perinatal survival (0 of 2 [0%] versus 6 of 10 [60%]), although this trend did not reach significance (P = .2).
This study demonstrates that Doppler detection of a functional artery–artery anastomosis confers a survival advantage in twin–twin transfusion syndrome and that this is independent of stage at presentation, at treatment, and at worst stage. This is consistent with existing lines of evidence suggesting that artery–artery anastomoses have a protective role in twin–twin transfusion syndrome,7,12,13,15 and supports a clinical role for Doppler detection of artery–artery anastomoses in monochorionic pregnancies.8
A staging system should discriminate at presentation between those with good and bad prognoses. The Doppler presence of artery–artery anastomoses and stage can be assessed at presentation and before treatment, allowing both to be incorporated into a modified staging system. Our suggested subcategorization of Quintero stage, in which the presence or absence of an antenatally detected artery–artery anastomosis is denoted “a” and “b,” respectively, not only allows distinction of good from bad prognosis cases within each stage, but also an artery–artery anastomoses Doppler positive subgroup within stage III (IIIa), which is associated with better perinatal and double survival than artery–artery, anastomosis-Doppler negative stages I and II (Ib and IIb) cases.
This large series of twin–twin transfusion syndrome classified by both stage and artery–artery anastomosis detection was managed at a single tertiary referral center. Although staging was first described in 1999,4 our unit had documented antenatal prognostic factors before 1999, including artery–artery anastomoses and those for staging, which allowed accurate retrospective assignment of stage for cases assessed. Because this series covers an 8-year period, a potential criticism is that changes in treatment paradigm over the course of the study, in particular the offer of selective feticide from mid-1999 for stage IV, or stage III with preterminal Doppler studies, might have biased the results. However, there was no substantive change in treatment paradigm for milder disease (stage I to nonprogressive stage III). Further, as selective feticide was only offered for preterminal fetuses not likely to have survived the neonatal period anyway, we consider cord occlusion unlikely to have contributed greatly to increased mortality in advanced disease. Laser ablation of placental vascular anastomoses was also used in a few cases. As discussed elsewhere,6 it is not possible to exclude entirely any effect of treatment paradox, unless all cases are managed conservatively, which would be unethical. However, statistical analysis with multiple logistic regression models failed to show that first-treatment modality predicted survival after correcting for stage, artery–artery anastomosis, and progression. Nevertheless, the majority was treated conservatively or by amnioreduction and/or septostomy or by delivery.
The 70% sensitivity of artery–artery anastomosis detection in the subgroup with placental injection studies in this series is less than the 85% reported in an earlier series, which included both twin–twin transfusion syndrome and non–twin–twin transfusion syndrome monochorionic pregnancies.8 Artery–artery anastomoses may be more difficult to detect in twin–twin transfusion syndrome, because of difficulty insonating posterior placentas in the presence of hydramnios and/or artery–artery anastomoses with smaller diameters. However, antenatal detection of artery–artery anastomoses still predicted survival independent of stage at each time point. Furthermore, we have previously shown that artery–artery anastomoses not detected ultrasonically are of narrower caliber than those detected,8 suggesting that missed artery–artery anastomoses are of less hemodynamic, and thus clinical, significance.
Quintero stage predicted perinatal and double survival at all 3 time points in this series, in contrast to an earlier series from the same center.6 Methodological differences in the current study that may explain this include 1) the larger number of cases, 2) extension of follow-up to 28, rather than 7, postnatal days, 3) exclusion of stage V at presentation, and 4) statistical analysis with stage as ordinal, rather than nominal, data.
The optimal treatment strategy for twin–twin transfusion syndrome remains unclear, although there are now compelling arguments for a stage-based approach.5,6 Stages Ia, IIa, and IIIa form a group with good survival prognosis, which can be managed conservatively or with temporizing measures such as amnioreduction and/or septostomy and then delivered electively by 32–34 weeks of gestation. Stages Ib and IIb compose a group with intermediate survival prognosis. Although amnioreduction is associated with better perinatal survival than laser ablation for stages I–II (87% versus 71%, P = .01),5 some of this group nevertheless progress to more advanced stage disease, eventually requiring definitive treatment.6 Stages IIIb and IV form the group with the poorest survival, in which more invasive, albeit definitive, treatment such as selective laser or cord occlusion appears to be indicated. Selective laser ablation of anastomotic vessels has been shown to result in higher perinatal, double, and any survival rates than amniocentesis procedures for stage IV,5 whereas cord occlusion in stage III/IV disease is associated with empirically higher survival than amnioreduction.17 We acknowledge that our analysis is limited to survival. Imaging and neurodevelopmental series suggest a high incidence of neurological sequelae with all treatment modalities,19–21 and future studies will now need to stratify by modified stage.
Although this study supports an ameliorative role for artery–artery anastomosis detection in stages I–III twin–twin transfusion syndrome, survival was poorer in stage IV disease with an artery–artery anastomosis on Doppler. This supports the findings of an earlier study22 suggesting that, when intrauterine death occurs, artery–artery anastomoses may mediate agonal transfusion from the co-twin into the dying twin, predisposing to death and/or intracranial lesions with neurological sequelae.23,24 The question of whether artery–artery anastomoses should be ablated during fetoscopic laser remains controversial. On the one hand, there is a risk of death or injury to the co-twin if one fetus dies in utero in the presence of an unablated artery–artery anastomosis, but on the other hand, ablation of artery–artery anastomoses might worsen twin–twin transfusion syndrome by reducing the hemodynamic compensatory ability in the event of incomplete arteriovenous ablation and thus continued intertwin transfusion.25 Notwithstanding this, modified stage IVa is rare, accounting for only 2 of 96 patients in this series.
We conclude that incorporation of artery–artery anastomoses detection into a modified twin–twin transfusion syndrome staging system improves prediction of perinatal survival and may facilitate treatment selection to optimize outcome in twin–twin transfusion syndrome.
1. Wee LY, Fisk NM. The twin-twin transfusion syndrome. Semin Neonatol 2002;7:187–202.
2. Mari G, Roberts A, Detti L, Kovanci E, Stefos T, Bahado-Singh RO, 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.
3. Taylor MJ, Denbow ML, Duncan KR, Overton TG, Fisk NM. Antenatal factors at diagnosis that predict outcome in twin-twin transfusion syndrome. Am J Obstet Gynecol 2000;183:1023–8.
4. Quintero RA, Morales WJ, Allen MH, Bornick PW, Johnson PK, Kruger M. Staging of twin-twin transfusion syndrome. J Perinatol 1999;19:550–5.
5. Quintero RA, Dickinson JE, Morales WJ, Bornick PW, Bermudez C, Cincotta R, et al. Stage-based treatment of twin-twin transfusion syndrome. Am J Obstet Gynecol 2003;188:1333–40.
6. Taylor MJ, Govender L, Jolly M, Wee L, Fisk NM. Validation of the Quintero staging system for twin-twin transfusion syndrome. Obstet Gynecol 2002;100:1257–65.
7. Denbow ML, Cox P, Talbert D, Fisk NM. Colour Doppler energy insonation of placental vasculature in monochorionic twins: absent arterio-arterial anastomoses in association with twin-to-twin transfusion syndrome. Br J Obstet Gynaecol 1998;105:760–5.
8. Taylor MJ, Denbow ML, Tanawattanacharoen S, Gannon C, Cox PM, Fisk NM. Doppler detection of arterio-arterial anastomoses in monochorionic twins: feasibility and clinical application. Hum Reprod 2000;15:1632–6.
9. Bajoria R, Wigglesworth J, Fisk NM. Angioarchitecture of monochorionic placentas in relation to the twin-twin transfusion syndrome. Am J Obstet Gynecol 1995;172:856–63.
10. Machin G, Still K, Lalani T. Correlations of placental vascular anatomy and clinical outcomes in 69 monochorionic twin pregnancies. Am J Med Genet 1996;61:229–36.
11. Talbert D, Bajoria R, Sepulveda W, Bower S, Fisk N. Hydrostatic and osmotic pressure gradients produce manifestations of fetofetal transfusion syndrome in a computerised model of monochorionial twin pregnancy. Am J Obstet Gynecol 1996;174:598–608.
12. Umur A, van Gemert MJ, Ross MG. Does amniotic fluid volume affect fetofetal transfusion in monochorionic twin pregnancies? Modelling two possible mechanisms. Phys Med Biol 2002;47:2165–77.
13. Denbow ML, Cox P, Taylor M, Hammal DM, Fisk NM. Placental angioarchitecture in monochorionic twin pregnancies: relationship to fetal growth, fetofetal transfusion syndrome, and pregnancy outcome. Am J Obstet Gynecol 2000;182:417–26.
14. Diehl W, Hecher K, Zikulnig L, Vetter M, Hackeloer BJ. Placental vascular anastomoses visualized during fetoscopic laser surgery in severe mid-trimester twin-twin transfusion syndrome. Placenta 2001;22:876–81.
15. Tan T, Denbow ML, Cox PM, Talbert D, Fisk NM. Occlusion of arterio-arterial anastomosis manifesting as acute twin-twin transfusion syndrome. Placenta 2004;25:238–42.
16. Taylor MJ, Talbert DG, Fisk NM. Mapping the monochorionic equator: the new frontier. Ultrasound Obstet Gynecol 1999;14:372–4.
17. Taylor MJ, Shalev E, Tanawattanacharoen 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.
18. Cheung YB, Khoo KS, Karlberg J, Machin D. Association between psychological symptoms in adults and growth in early life: longitudinal follow-up study. BMJ 2002;325:749–52.
19. Denbow ML, Battin MR, Cowan F, Azzopardi D, Edwards AD, Fisk NM. Neonatal cranial ultrasonographic findings in preterm twins complicated by severe fetofetal transfusion syndrome. Am J Obstet Gynecol 1998;178:479–83.
20. Sutcliffe AG, Sebire NJ, Pigott AJ, Taylor B, Edwards PR, Nicolaides KH. Outcome for children born after in utero laser ablation therapy for severe twin-to-twin transfusion syndrome. BJOG 2001;108:1246–50.
21. Banek CS, Hecher K, Hackeloer BJ, Bartmann P. Long-term neurodevelopmental outcome after intrauterine laser treatment for severe twin-twin transfusion syndrome. Am J Obstet Gynecol 2003;188:876–80.
22. Bajoria R, Wee LY, Anwar S, Ward S. Outcome of twin pregnancies complicated by single intrauterine death in relation to vascular anatomy of the monochorionic placenta. Hum Reprod 1999;14:2124–30.
23. Fusi L, McParland P, Fisk N, Nicolini U, Wigglesworth J. Acute twin-twin transfusion: a possible mechanism for brain-damaged survivors after intrauterine death of a monochorionic twin. Obstet Gynecol 1991;78:517–20.
24. Tanawattanacharoen S, Taylor MJ, Letsky EA, Cox PM, Cowan FM, Fisk NM. Intrauterine rescue transfusion in monochorionic multiple pregnancies with recent single intrauterine death. Prenat Diagn 2001;21:274–8.
© 2004 by The American College of Obstetricians and Gynecologists. Published by Wolters Kluwer Health, Inc. All rights reserved.
25. Wee LY, Taylor MJ, Vanderheyden T, Wimalasundera R, Gardiner HM, Fisk NM. Reversal of twin-twin transfusion syndrome: frequency, vascular anatomy, associated anomalies and outcome. Prenat Diagn 2004;24:104–10.