The reference ranges for fetal ductus venosus flow velocities and calculated waveform indices and their predictive values for right heart diseases

Study population

A total of 25,607 fetal data from the Maternal-Fetal Medicine Center in Fetal Heart Disease of Beijing Anzhen Hospital, spanning from November 2019 to April 2024, were retrospectively analyzed. The study process is illustrated in Figure 1. The gestational age was calculated from the last menstrual period and corroborated with the biparietal diameter and femur length.

Figure 1:

Study flow.

Inclusion criteria for low-risk fetuses: (1) maternal age≥18 years, (2) singleton pregnancy, (3) no maternal high-risk factors, (4) second or third trimester of pregnancy, (5) no fetal intracardiac or extracardiac abnormalities. The exclusion criteria included: (1) poor echocardiography or DV Doppler spectrum image quality, (2) fetal chromosome abnormalities, (3) fetal intrauterine growth restriction.

Inclusion criteria for fetuses with right heart diseases: (1) maternal age≥18 years, (2) second or third trimester of pregnancy, (3) diseases causing fetal right heart preload or afterload increase, or cardiomyopathy involving the right heart. The exclusion criteria included: (1) poor echocardiography or DV Doppler spectrum image quality, (2) fetal extracardiac abnormalities.

The study protocols followed the principles of the Declaration of Helsinki and were approved by the Ethics Committee, and the requirement for written informed consents from the prospective parents were waived by the Ethics Committee.

Acquisition of Doppler spectrum and measurement parameters for DV

The ultrasound diagnostic instruments used were the Voluson E8 (GE Healthcare, Milwaukee, Wisconsin, USA), Voluson E10 (GE Healthcare, Zipf, Austria), and HERA XM10 (SAMSUNG MEDISON, Gang-won-do, Korea), all equipped with 2–9 MHz frequency probes. The acquisition of fetal echocardiography adhered to the guidelines of the American Society of Echocardiography, and the International Society of Ultrasound in Obstetrics and Gynecology [6], [7], [8], [9], [10].

After adjusting the Doppler velocity scale, gain, and wall filter in the absence of fetal breathing and limb movements. Doppler measurements were performed in the sagittal plane through the upper fetal abdomen or the oblique transverse section of the fetal abdomen. The pulsed Doppler sampling volume was then adjusted according to the vessel diameter. When the sampling frame direction and the DV flow direction were less than 30°, the DV Doppler waveforms (over three cardiac cycles) were recorded: peak forward velocity (S) during ventricular contraction period, peak forward velocity (D) during early diastole period, and minimum forward velocity (A) during late diastolic atrial contraction period (Figure 2) [11]. Additionally, the time-averaged maximum velocity (Tamx) was calculated, and the following calculated waveform indices were measured: preload index (PLI), calculated as (S−A)/S; venous peak velocity index (PVI), calculated as (S−A)/D; venous pulsatility index (PI), calculated as (S−A)/Tamx [12]; S/D ratio and S/A ratio.

Figure 2:

Color Doppler flow imaging and spectral Doppler waveform of the ductus venosus. S and D represent peak ventricular systolic and diastolic forward velocities, respectively, and A represents minimum atrial systolic velocities.

All images, measurements, and clinical data were archived in the database.

Statistical analysis

All data were analyzed using R software (version 4.3.2) and SPSS software (version 26.0), with p<0.05 indicating statistical significance. Categorical variables were expressed as the numbers (percentage). Continuous variables were tested for normality by the Kolmogorov-Smirnov test. Continuous numerical variables were expressed as mean±standard deviation (SD) or median (interquartile range, IQR) regression model. The correlation between the DV Doppler parameter and gestational age was calculated by Spearman correlation analysis. Regression analysis was performed on DV hemodynamic parameters to determine the best-fitting equation. If the DV Doppler parameter was significantly correlated with gestational age, a univariate linear or univariate multiple linear mean regression model was established; otherwise, a quantile regression model was used to obtain the median, 5th percentile, and 95th percentile values.

The Akaike information criterion (AIC) was used to compare the relative goodness of fit of different models, and the model was selected according to the AIC value and the simplicity of the model. The Durbin-Watson test method was used to detect whether the residual term in the linear mean regression model was autocorrelated, the Q–Q Plot was used to assess the residual normality of the linear regression model, and the residuals vs. fitted values Plot was used to detect the residual variance homogeneity of the linear regression model. To construct the 5th and 95th percentile regression lines with the same slope as the mean regression line, the following calculation method was adopted: mean regression line plus/minus 1.645 times the SD of the residuals. Z score=(observation value−mean value [derived from the prediction model])/SD. The Z score of DV Doppler parameters of 90 low-risk fetuses (randomly selected) and 70 fetuses with right heart diseases were calculated according to Z score calculation. Receiver operating characteristic (ROC) curves were used to determine the best cut-off values for predicting right heart diseases. The intraclass correlation coefficient and Bland-Altman analyses were used to estimate intra- and inter-observer variation in DV Doppler parameters among 20 randomly selected subjects.

Clinical characteristics

After applying the inclusion and exclusion criteria, 8,953 low-risk fetuses and 70 fetuses with right heart diseases were finally included. The clinical characteristics of low-risk fetuses and fetuses with right heart disease are summarized in Table 1. The distribution of fetal right heart diseases was as follows:

1. Diseases with increased right heart preload (25 cases):

   1. Ebstein’s anomaly with varying degrees of tricuspid regurgitation (13 cases)

   2. Massive tricuspid regurgitation without Ebstein’s anomaly (12 cases)

2. Diseases with increased right heart afterload (27 cases):

   1. Pulmonary stenosis with reverse perfusion of the ductus arteriosus (nine cases)

   2. Pulmonary atresia with intact ventricular septum (18 cases)

3. Cardiomyopathy involving the right heart (18 cases):

   1. Excessive myocardial trabeculation (11 cases)

   2. Endocardial fibroelastosis (seven cases)

Normal reference ranges and Z score calculation formulas for DV hemodynamic parameters

The S, D, A, Tamx, PLI, PVI, PI, and S/A were all significantly correlated with gestational age (all p<0.001), whereas S/D was not significantly correlated with gestational age (p=0.051) (Supplemental Table S1). Univariate linear regression models were established for S, D, A, and Tamx; similarly, univariate quadratic linear regression models were also established for PLI, PVI, PI, and S/A (Table 2). For S/D, a quantile regression model was established, yielding the 5th percentile value of 1.00, the 50th percentile value of 1.10, and the 95th percentile value of 1.32.

Reference ranges for the 5th percentile, mean, and 95th percentile values of the S, D, A, Tamx, PLI, PVI, PI, and S/A for DV at 16–39 weeks of gestation were established, as shown in Figure 3, Figure 4, Table 3, and Table 4. As gestational age increased, the mean values of the S, D, A, and Tamx increased from 41.9 cm/s to 58.5 cm/s, from 38.2 cm/s to 51.5 cm/s, from 19.8 cm/s to 38.5 cm/s, and from 35.2 cm/s to 51.0 cm/s, respectively. Conversely, as gestational age increased, the mean values of the PLI, PVI, PI, and S/A decreased from 0.57 to 0.39, from 0.65 to 0.475, from 0.69 to 0.46, and from 2.37 to 1.77, respectively.

Figure 3:

Scatter plots and reference range interval plots of the S, D, A, and Tamx. The solid line represents the mean predicted value and the dashed lines represents the 5th and 95th percentile predicted values. A, atrial contraction wave velocity; D, ventricular diastole wave velocity; GA, gestational age; S, ventricular systole wave velocity; Tamx, time-averaged maximum velocity.

Figure 4:

Scatter plots and reference range interval plots of the PLI, PVI, PI, and S/A. The solid line represents the mean predicted value and the dashed lines represents the 5th and 95th predicted percentile values. GA, gestational age; PI, pulsatility index; PLI, preload index; PVI, peak velocity index; S/A, ventricular systole/atrial contraction wave velocity ratio.

Based on the best linear fitting regression equations and residual SD between DV blood flow velocity and calculated waveform indices against the gestational age, the Z score calculation formulas were established for 8 DV Doppler parameters (S, D, A, Tamx, PLI, PVI, PI, and S/A) (Table 5). Taking the Z score (S) as an example, Z score (S)=[observation value−(0.7196 × gestational age+30.3822)]/13.9695.

Diagnostic efficacies of Z score evaluation system in right heart diseases

Comparisons of the Z scores of DV hemodynamic parameters are shown in Table 6 between the fetuses in the control group and those in the overall group with right heart diseases, the group with increased right heart preload, the group with increased right heart afterload, and the group with cardiomyopathy involving the right heart. The Z score (A) of the overall group with right heart diseases was significantly reduced (p=0.001), and the Z scores of the PLI, PVI, PI, and S/A were all significantly increased (all p<0.001).

By ROC analysis, the Z scores of each parameter were determined to be able to predict different types of right heart diseases in the fetus (Figure 5). ROC analysis of statistically significant parameters:

1. The best cut-off values for the Z scores of the A, PLI, PVI, PI, and S/A in the overall group of right heart diseases were −0.191 (area under receiver operating characteristic curve [AUC]=0.655, sensitivity=77.1 %, and specificity=50.0 %), 0.415 (AUC=0.720, sensitivity=67.1 %, and specificity=67.8 %), 1.189 (AUC=0.697, sensitivity=44.3 % and specificity=90.0 %), 0.546 (AUC=0.740, sensitivity=64.3 % and specificity=81.1 %), and 0.226 (AUC=0.689, sensitivity=64.3 %, and specificity=67.9 %), respectively (Supplemental Table S2).

2. The best cut-off values for the Z scores of the PLI and PI for right preload diseases were −0.365 (AUC=0.638, sensitivity=96.0 %, and specificity=33.3 %) and 0.716 (AUC=0.660, sensitivity=52.0 %, and specificity=81.1 %), respectively (Supplemental Table S3).

3. The best cut-off values for the Z scores of the A, PLI, PVI, PI, and S/A for right heart afterload-increasing diseases were −0.998 (AUC=0.734, sensitivity=63.0 %, and specificity=78.9 %), 1.381 (AUC=0.803, sensitivity=59.3 %, and specificity=93.3 %), 1.240 (AUC=0.791, sensitivity=63.0 %, and specificity=90.0 %), 0.840 (AUC=0.830, sensitivity=81.5 %, and specificity=84.4 %), and 0.987 (AUC=0.806, sensitivity=63.0 %, and specificity=91.1 %), respectively (Supplemental Table S4).

4. The best cut-off values for the Z scores of the PLI, PVI, PI, and S/A for cardiomyopathy involving the right heart were 1.201 (AUC=0.708, sensitivity=50.0 %, and specificity=92.2 %), 1.189 (AUC=0.712, sensitivity=55.6 % and, specificity=92.2 %), 0.525 (AUC=0.717, sensitivity=61.1 %, and specificity=75.6 %), and 1.043 (AUC=0.654, sensitivity=44.4 %, and specificity=91.1 %), respectively (Supplemental Table S5).

Figure 5:

ROC analysis. (A) Receiver operating characteristic analysis of Z scores of ductus venous hemodynamic parameters for predicting right heart disease. (B) Receiver operating characteristic analysis of Z scores of ductus venous hemodynamic parameters for predicting increased right heart preload. (C) Receiver operating characteristic analysis of Z scores of ductus venous hemodynamic parameters for predicting increased right heart afterload. (D) Receiver operating characteristic analysis of Z scores of ductus venous hemodynamic parameters for predicting cardiomyopathy involving the right heart.

Intra-observer and inter-observer variabilities

Intra- and inter-observer variabilities of DV hemodynamic parameters are summarized in Table 7, Supplemental Figures S1 and S2, showing good reproducibility.

Influencing factors of fetal right heart function and hemodynamics

The primary influencing factors of right heart function and hemodynamics in the fetal period are increased right heart preload, increased right heart afterload, and cardiomyopathy involving the right heart. When the right heart preload increases, the right heart expands to compensate, leading to an increase in myocardial tone. If the preload remains elevated for an extended period, the afterload also increases. When the right heart afterload increases, the initial response of the right ventricle is adaptive remodeling, with the myocardium thickening to enhance contractility. However, if the pressure is too high or persists for too long, the right heart decompensates, and the right heart function significantly decreases, potentially resulting in right heart failure [24]. Cardiomyopathy affecting the right heart directly impairs the normal systolic and diastolic function of cardiomyocytes, reducing right heart function. This, in turn, hinders the flow of peripheral blood back into the atrium and its efficient pumping out of the ventricle. The fetal right heart myocardium is immature, which inherently elevates the risk of right heart failure and adverse pregnancy outcomes in the presence of right heart diseases [1]. During the fetal period, due to the presence of the foramen ovale and the ductus arteriosus, previous partial function evaluation parameters cannot comprehensively evaluate the overall right heart function and hemodynamics. The DV is located at the opening of the inferior vena cava entering the right atrium. The DV integrates the overall pressure and flow shunt outcome of the right atrium, right ventricle, foramen ovale, and ductus arteriosus, which can theoretically accurately and comprehensively evaluate right heart function and hemodynamics. Moreover, the S, D, and A in the DV Doppler spectrum reflect the function and hemodynamics of ventricular systole, ventricular diastole, and atrial systole during the cardiac cycle. Furthermore, multiple calculated waveform indices of the blood flow Doppler spectrum can also indirectly reflect differences in the right heart volume load, pressure load, and primary myocardial abnormalities.

This study suggests that the S, D, and Tamx may not be sensitive parameters for evaluating right heart disease changes. When the right heart afterload increases, ventricular end-diastolic pressure also rises, leading to an increase in forward blood flow impedance. Under this condition, the A wave exhibits a significant reduction, absence, or reversal [25]. Regarding the DV calculation waveform indices, the Z scores of the PLI and PI were significantly higher than those of the control group when the right heart preload increased; and the Z scores of the PLI, PVI, PI, and S/A were significantly higher in the increased right heart afterload group and the group with cardiomyopathy involving the right heart. This indicates that the angle-independent DV calculation waveform indices are more sensitive to the hemodynamic changes in the DV caused by different right heart diseases than the flow velocity parameters. Therefore, these indices should be used clinically as valuable evaluation indices for assessing the hemodynamic states of the fetal right heart.

The Z scores of the PLI and PI have moderate diagnostic values for the overall right heart diseases (PLI: AUC=0.720, PI: AUC=0.740, respectively). However, for screening overall right heart diseases, the efficiency of DV Doppler parameters is not high. This may be because when screening for right heart diseases as a whole, which encompasses various causes of right heart function and/or hemodynamic changes, the changes in DV hemodynamic parameters are inconsistent. The Z scores of the PLI and PI have low diagnostic values for diseases with increased right preload (PLI: AUC=0.638, PI: AUC=0.660, respectively). For the specific diagnosis of diseases with increased right preload, the diagnostic efficacies of DV hemodynamic parameters are low. This may be attributed to the compensatory enhancement of right myocardial contractility, increased blood output, and no significant change in right heart blood return when the right heart volume load increases, resulting in insignificant changes in DV hemodynamic parameters of the right heart [24], 26]. The Z scores of the PLI, PI, and S/A have high diagnostic values for diseases with increased right heart afterload (PLI: AUC=0.803, PI: AUC=0.830, S/A: AUC=0.806, respectively). Compared with increased right heart volume load, right heart pressure load is insufficient to compensate for right heart function, leading to significant blockage of right heart forward blood flow and notable hemodynamic changes in each phase, which results in significant changes in DV hemodynamic parameters [27]. The Z scores of the PLI, PVI, and PI have moderate diagnostic values for diseases involving right heart cardiomyopathy (PLI: AUC=0.708, PVI: AUC=0.712, PI: AUC=0.717, respectively). For the specific diagnosis of diseases involving right heart cardiomyopathy, the diagnostic efficacy of DV hemodynamic parameters is not high. This may be due to the low degree of fetal right heart function and hemodynamics change caused by right heart cardiomyopathy, thus resulting in no high diagnostic efficacy for diseases involving right heart cardiomyopathy.

Strengths and limitations

This study is the largest sample size to date in investigating the hemodynamic parameters of low-risk fetal DVs. The DV measurement parameters were relatively comprehensive, and the established parameter Z score calculation formulas provide convenient references for effectively assessing whether the DV Doppler parameters of different gestational ages are normal in clinical practice. Previous studies have only observed the A reduction and direction reversal of fetal DVs with partial right heart afterload increase, and only used PVI as the parameter to predict increased right heart afterload [28]. We are the first to validate the predictive values of the Z score calculation formulas for multiple hemodynamic parameters of the DV in fetal right heart diseases.

This study is a retrospective, cross-sectional analysis (only select one recording for the included fetuses). The sample size of each gestational age is not uniformly distributed (predominantly concentrated around 25 weeks), which may influence the accuracy of the regression equation construction. All fetuses included in the study were Chinese; thus, the established normal reference ranges may not be applicable to populations in other countries. The number of cases included is small and the diversity of case types is relatively small. Further expansion of the sample size and inclusion of a wider range of case types is necessary. Lastly, although our center is the national referral center for fetal heart disease, prospective multi-center validation is the next step in our research plan.