Relevance of measuring diastolic time intervals in the ductus venosus during the early stages of twin–twin transfusion syndrome




To determine if the discrete myocardial diastolic dysfunction documented previously in the recipient twin during the early stages of twin–twin transfusion syndrome (TTTS) has any repercussion on flow velocities through the ductus venosus (DV) and to investigate if this could allow early differentiation between TTTS and selective intrauterine growth restriction (IUGR).


Two groups of monochorionic twin pregnancies with growth discordance between twins were reviewed retrospectively. Group I was composed of fetuses in Stages I and II of TTTS; laser or amnioreduction was not performed in any instance. Group II twin pairs each included one fetus with IUGR due to placental circulatory insufficiency. Intertwin differences (smaller minus larger fetus) were analyzed for myocardial performance index of the right ventricle (MPI-RV) and for time variables in the DV.


There were 38 pairs of monochorionic twins (24 TTTS and 14 IUGR) in this study. In the TTTS group, the donors had a significantly lower MPI-RV (0.419 ± 0.18 vs. 0.596 ± 0.17, F(1, 19df) = 24.017, P < 0.001), a significantly longer total ventricular filling time (150.9 ± 25.6 ms vs. 124.0 ± 22.6 ms; F(1, 21df) = 19.631, P < 0.001) and a significantly longer early filling time (118.9 ± 22.9 ms vs. 92.6 ± 18.9 ms, F(1, 21df) = 28.419, P < 0.001) than had the recipient. None of these three differences was present in the IUGR group. Probability studies revealed that cut-off values of 12.75 for intertwin differences in total filling time and 8.5 for intertwin differences in early filling time had sensitivities of 71% and 92%, respectively. The false-positive rates were 23% and 15%, respectively, for the early diagnosis of TTTS.


In monochorionic twin pregnancies, shortening of the ventricular filling time in the recipient twin indicates diastolic myocardial dysfunction occurring early in the pathophysiology of TTTS. This early interwin difference in myocardial function is not found in pregnancies with IUGR in one twin due to placental circulatory insufficiency, allowing early differentiation between TTTS and selective IUGR. Copyright © 2007 ISUOG. Published by John Wiley & Sons, Ltd.


Twin–twin transfusion syndrome (TTTS) occurs in 10–20% of monochorionic diamniotic twin pregnancies1. If left unattended, this severe complication can lead to major cardiocirculatory or neurological morbidity or mortality in either twin, while early intervention can prevent or at least minimize these adverse outcomes2–4. In most cases, the differential diagnosis between TTTS and twin growth discordance due to placental circulatory insufficiency (intrauterine growth restriction (IUGR)) in one twin (selective IUGR) is easily established by close follow-up. Unfortunately, early in the process, differentiation between the two can be difficult, because both conditions can present with similar clinical signs, such as discordance in fetal weight and amniotic fluid volume. We have shown previously that the recipient twin with TTTS presents very early signs of subclinical diastolic myocardial dysfunction, characterized by isolated prolongation of the isovolumetric relaxation time5. This diastolic dysfunction is responsible for an abnormal myocardial performance index (MPI), a marker of global cardiac function6. In contrast, IUGR fetuses do not show any change in MPI at this early stage. The MPI is not, however, easy to quantify, because it requires Doppler recordings of both ventricular inflow and outflow velocities for the measurement of diastolic and ejection time intervals. Doppler flow velocities in the ductus venosus (DV), however, are simpler to record and are familiar to most sonographers involved with fetal circulatory monitoring. Since flow dynamics in the DV are influenced essentially by ventricular diastolic function, this study was undertaken with the objective of exploring whether the discreet subclinical diastolic myocardial dysfunction found in the early stages of TTTS can have repercussions on time parameters within the DV waveforms that might be useful in the evaluation and/or differential diagnosis of TTTS. We hypothesized that the prolongation of the isovolumetric relaxation period observed in the recipient twin of a TTTS pregnancy5 should reduce the ventricular filling time during the relaxation phase preceding the atrial contraction and that this would not occur in IUGR cases.


All available Doppler ultrasound examinations of monochorionic diamniotic twin pregnancies recorded in our unit from September 1998 to April 2005 were reviewed systematically. This report concerns only fetuses in whom intracardiac and DV Doppler recordings were available for the first ultrasound examination, before any therapeutic procedures had taken place. The patients were divided into two groups, depending on whether the final diagnosis was TTTS (Group I) or selective IUGR (Group II), based on the evolution of the pregnancy and perinatal data. Prenatally, monochorionicity was identified by the presence of twins of the same sex with a single placenta and sonographic evidence of a thin (< 2 mm) amniotic membrane. This was confirmed after delivery by examination of the placenta. The diagnosis of TTTS was based on discordance in fetal growth and the development of an oligopolyhydramnios sequence (deepest pool > 8 cm for the larger amniotic sac and < 1 cm for the smaller sac). Severity of TTTS at the time of the first echocardiogram was assessed according to the criteria proposed by Quintero et al.7: Stage I: bladder of the donor twin still visible; Stage II: bladder of the donor no longer visible but critically abnormal Doppler studies not yet present; Stage III: presence of critically abnormal Doppler studies; Stage IV: hydrops. Only fetuses at Stages I or II were ultimately included in this study. IUGR due to placental insufficiency was considered when the dimensions of the smaller fetus were below the 10th percentile8, with the pulsatility index of the umbilical artery elevated above the 95th percentile for gestational age9; both twins had normal amniotic fluid at this stage of their development. Fetal weight in relation to gestational age was ascertained by ultrasound measurement of head and abdominal circumferences and femur length10.

Ultrasound studies

Cardiocirculatory investigations were performed with 128 XP/10 c or Sequoia sonographic equipment (Acuson, Mountain View, CA, USA) with either a 5- or a 6-C2-MHz transducer. The following parameters were recorded in all fetuses: the pulsatility index of the umbilical artery at the abdominal end of the cord9; pulsed Doppler waveforms above the pulmonary valve and at the tip of the tricuspid valve; and the MPI of the right ventricle (MPI-RV), defined as the isovolumetric period (sum of isovolumetric contraction and relaxation times) divided by the ventricular ejection time6. On the DV Doppler waveforms, the following variables were measured during the absence of fetal breathing movements (Figure 1): the pulsatility index for veins (PIV); the ratio between the peak velocity during ventricular systole (S) and the nadir of the deceleration wave during atrial systole (A); the total diastolic filling time (FT) from the onset of the early filling wave (e-wave) to the nadir of the a-wave. The two components of FT were also measured: the early diastolic filling time (eT) from the onset of the e-wave to the beginning of fast deceleration caused by atrial contraction; and the late diastolic filling time (aT) obtained by subtracting eT from FT. The duration of the preceding cardiac cycle was measured systematically for fetal heart rate (FHR) calculations.

Figure 1.

Doppler tracings of flow velocities through the ductus venosus (DV) of the recipient (left) and donor (right) fetus in a monochorionic twin pregnancy. The technique used for time-interval measurements is illustrated schematically above each Doppler waveform. Note the shorter total diastolic filling time (FT) and early phase of diastole (eT) in the recipient twin. Heart rates were 137 and 134 bpm for the recipient and donor twins, respectively. a, deceleration wave caused by atrial contraction; e, forward wave during the early phase of diastolic ventricular filling; S, forward wave during ventricular systole.

Statistical analysis

All values used for analysis correspond to an average of three consecutive measurements. The data were analyzed in two stages. First, differences between the two conditions (TTTS and IUGR) were evaluated by analysis of covariance (ANCOVA), with one between-subjects factor (condition), one within-subject factor (large/small twin) and the intertwin difference in cardiac rate as a covariable. In order to identify significant interactions between factors, the full factorial model was used. In the case of significant interactions between factors, appropriate sub-analyses were planned. P-values < 0.05 were considered statistically significant, except for the Levene test of equality of variances, which was considered significant if P < 0.01. Second, the diagnostic characteristics were computed for relevant variables. Receiver–operating characteristics (ROC) curves were produced to identify the difference value (Δ) for each variable at which the diagnosis of TTTS could be established with confidence. All analyses were performed using SPSS 14.0.1 (SPSS Inc., Chicago, IL, USA).


During the study period, 48 monochorionic diamniotic twin pregnancies with fetal growth discrepancy were seen in our unit and ultimately diagnosed as having TTTS (Group I, n = 34) or selective IUGR (Group II, n = 14). Five of the 34 cases of Group 1 had to be excluded because of poor quality of the DV tracings. Five other cases were already Stage III of Quintero's classification at their first visit and were also excluded. Most TTTS pregnancies (21/24) were Stage I; three were Stage II. The first echocardiographic study was recorded at a mean ± SD of 24.4 ± 3.9 weeks of gestation for the TTTS group, and at 26.9 ± 4.4 weeks for the IUGR group (t36df = 1.787, P = 0.082). No twin showed phenotypic evidence of chromosomal abnormalities. In Group I, the mean ± SD FHR was 141.0 ± 8.1 bpm for the recipient and 145.7 ± 7.9 for the donor twin, and in Group II it was 144.2 ± 9.8 bpm for the bigger twin and 143.9 ± 6.1 bpm for the smaller twin. The larger fetuses had a significantly lower FHR in the TTTS group (Wilcoxon's rank Z = − 1.978, P = 0.048) but not in the IUGR group (Wilcoxon's rank Z = − 0.105, P = 0.916).

Table 1 gives descriptive cardiocirculatory data from the first Doppler assessment in both groups as well as for intertwin differences within each group. The first repeated measures ANCOVA was performed to assess whether the changes in MPI-RV were similar in this sample to those previously reported5. Since the results indicated a significant interaction between condition (TTTS/IUGR) and twins, subanalyses for each condition taken individually were then performed. The two repeated measures ANCOVAs indicated that, in the TTTS group, the donors had a significantly lower MPI-RV than had the recipients (F(1, 19df) = 24.017, P < 0.001). This difference was not present in the IUGR group (F(1, 12df) = 0.006, P = 0.938). The PI of the umbilical artery (F(1, 32df) = 0.208, P = 0.651), the S/A (F(1, 31df) = 0.243, P = 0.625) and the PIV (F(1, 32df) = 0.208, P = 0.651) did not show any significant association with the condition group. The analysis of FT and eT, as for MPI-RV, indicated significant two-way interactions between twins and condition. As shown in Table 1, the donor in TTTS had a significantly longer FT than had the recipient ( F(1, 21df) = 19.631, P < 0.001). This divergence was not present in the IUGR fetuses (F(1, 12df) = 1.919, P = 0.138). Sub-analysis also indicated that the donor in TTTS had a significantly longer eT than had the recipient (F(1, 21df) = 28.419, P < 0.001). This divergence was not present in the IUGR fetuses (F(1, 12df) = 0.040, P = 0.845).

Table 1. Descriptive cardiocirculatory data from the first Doppler assessment of twin pregnancies affected by twin–twin transfusion syndrome (TTTS; n = 24) and intrauterine growth restriction (IUGR; n = 14)
ParameterTTTS groupIUGR group
  1. Values are given as mean (SD). Mean intertwin difference was calculated as the donor or small fetus value minus the recipient or large fetus value. A, deceleration wave during atrial contraction; aT, duration of ventricular filling during atrial contraction; eT, duration of the early phase (relaxation) of ventricular filling; FT, total diastolic filling time; MPI-RV, myocardial performance index of the right ventricle; PIV, pulsatility index for veins; S, forward wave in Doppler recording of flow velocities through the ductus venosus during ventricular systole.

PIV0.728 (0.274)0.948 (0.799)− 0.194 (0.790)0.1490.808 (0.229)0.685 (0.148)0.123 (0.247)0.073
S/A2.733 (1.340)3.166 (2.800)− 0.528 (2.988)0.7802.982 (1.153)2.320 (0.527)0.666 (1.128)0.037
FT (ms)150.958 (25.596)124.042 (22.594)26.917 (25.114)< 0.001137.464 (22.259)140.750 (21.511)− 3.143 (35.828)0.138
eT (ms)118.958 (22.909)92.625 (18.897)26.375 (21.250)< 0.001117.821 (22.379)118.786 (31.077)− 0.964 (27.311)0.845
aT (ms)32.000 (10.350)31.417 (9.995)0.583 (10.834)0.62619.643 (25.151)21.964 (23.123)− 2.321 (17.318)0.844
MPI-RV0.419 (0.181)0.596 (0.169)− 0.222 (0.137)< 0.0010.479 (0.093)0.481 (0.085)− 0.067 (0.069)0.938

In order to verify the hypothesis that myocardial impairment was at the core of the problem, we compared the filling times between the two groups. Controlling for FHR, we found that the bigger fetuses in the IUGR group had significantly longer filling times than had the recipient TTTS twins (Table 1; F(1, 34df) = 8.661, P = 0.006 for eT and F(1, 34df) = 4.301, P = 0.046 for FT). On the other hand, there were no significant differences between the smaller fetuses of the two groups (F(1, 34df) = 0.401, P = 0.531 for eT and F(1, 34df) = 3.360, P = 0.076 for FT).

The two differences (ΔFT and ΔeT) had a Pearson's correlation coefficient of 0.915 (P < 0.001) that did not vary greatly between the two conditions (r = 0.904 for TTTS and 0.882 for IUGR, both P < 0.001). The correlation between ΔeT and ΔaT was not statistically significant (r = 0.215, P = 0.196), indicating that as eT increased (or decreased), the late filling phase was not affected systematically. The differences between fetuses for MPI-RV and FT (ΔMPI-RV and ΔFT) had a Pearson's correlation coefficient of − 0.473(P = 0.020).

The ROC curves for ΔFT and ΔeT are shown in Figure 2. The areas under the curves (AUC) were similar and statistically significant for both: the AUC (area ± SE) for ΔFT was 0.737 ± 0.084(P = 0.016) and for ΔeT it was 0.804 ± 0.077(P = 0.002). From the ROC analysis we found that the best cut-off values for differential diagnosis between TTTS and IUGR would be ≤− 12.75 ms for ΔFT and ≤− 8.5 ms for ΔeT. In this sample, for those pairs of twins with an absolute difference in FT of 12.75 or more, 17 of the 24 TTTS cases would be classified correctly, for a sensitivity of 71% and nine of the 14 cases of IUGR would be classified as such, for a specificity of 64%. With this criterion, the false positive rate for TTTS would be 23%. We also found that for an absolute difference in eT of 8.5 or larger between the twins, 22 of the 24 cases of TTTS would be classified as such, for a sensitivity of 92% and 10 of the 14 cases of IUGR would be classified as such, for a specificity of 71%. With this criterion, the false positive rate for TTTS would be 15%. The ROC curves for DV-PI and S/A ratio differences (ΔDV-PI and ΔS/A) are shown in Figure 3. Neither the AUC for ΔS/A (0.587 ± 0.105, P = 0.403) nor that for ΔDV-PI (0.569 ± 0.103, P = 0.506) was statistically significant.

Figure 2.

Receiver–operating characteristics curves for the prediction of twin–twin transfusion syndrome based on differences (small minus large twin) in total diastolic filling time (ΔFT; equation image) and the early phase of ventricular diastolic filling (ΔeT; equation image).

Figure 3.

Receiver–operating characteristics curves for the differences (small minus large twin) in ductus venosus pulsatility index (equation image) and the S/A ratio (—). The areas under both curves are not significant.


Blood flow velocities through the central venous system are modulated by dynamic events occurring within the heart11. During ventricular systole, atrial filling is responsible for an acceleration of venous flow (S-wave). With ventricular diastole, the two phases of ventricular filling have opposite effects on the venous flow pattern. During the early part of diastole, corresponding to the ventricular relaxation phase, a second acceleration of venous flow is observed. On the DV Doppler tracing, this forward wave is generally called the D-wave (‘diastolic wave’), despite the fact that the wave does not encompass the entire diastole. In our study, we elected to refer to this instead as ‘e’ (‘early’ part of diastole), in accordance with the widely accepted terminology for identification of the same two Doppler waveforms through the atrioventricular valves. During the second part of diastole, atrial contraction causes a rise in the intracardiac pressures responsible for a deceleration wave in the veins (A-wave). The better the ventricular compliance, the lower is the rise in pressure and the smaller the A-wave deceleration. In the present study, we speculated that measurements in the DV representing ventricular filling periods, instead of the usual velocity indices, would better reflect the reported changes in the isometric relaxation phase in the recipient twin of a TTTS pregnancy5; a shorter eT was indeed found in the recipient compared with the donor twin in the TTTS group, despite a slower FHR in the former12. This shorter eT can be explained by the prolonged isovolumetric relaxation period: the longer the duration of isovolumetric relaxation, the later the atrioventricular valves will open and the shorter will be the time left for the early diastolic filling period, which ends with atrial contraction (whose timing remains unaltered). None of the variables influenced by the A-wave showed specific changes that could be used to identify either TTTS or IUGR, suggesting that the second part of diastole was not significantly affected by either condition.

In TTTS, the hypertrophic cardiomyopathy observed in the recipient twin is not fully explained by a mere transfer of blood from the donor twin. Endothelin, a powerful vasoconstrictor, has been found to be twice as high in the blood of the recipient compared with the donor twin13. Furthermore, as an adaptive mechanism against hypovolemia, upregulation of the renin–angiotensin system has been well documented in the donor twin14. Consequently, in addition to the elevated level of endothelin, the recipient twin receives passively from his/her co-twin blood with a high angiotensin II concentration. The accumulation of these vasoactive agents could cause hypertrophic cardiomyopathy both by direct action on cardiomyocytes15 and by increased cardiac afterload created by peripheral vasoconstriction. In postnatal life, during the early stage of a hypertrophic cardiomyopathy, subclinical changes in myocardial diastolic function before impairment of systolic performance are well documented16. In the present study, the hypothesis that, from the very beginning of TTTS, a subclinical disturbance in diastolic myocardial function could alter the flow pattern through the DV, proved to be valid but only for the first part of diastole, the relaxation phase.

From this observation, it follows that the characteristics of Quintero's Stages I and II, classically based on the absence of Doppler abnormalities, should be updated. Specifically, abnormally elevated ΔFT or ΔeT are features of Stages I and II of TTTS. The appearance of alterations in the variables influenced by the A-wave remain specific to Stage III. This information should also help in establishing the prognosis and in choosing therapeutic approaches early in the process. However, the restrospective nature of this investigation on a limited number of subjects warrants a prospective study on a larger number of monochorionic pregnancies in order to establish with more confidence the sensitivity and specificity of the proposed indices.