Evaluation of fetal foramen ovale blood fow by pulsed Doppler ultrasonography combined with spatiotemporal image correlation

Introduction

Shunting through the foramen ovale (FO), which constitutes the majority of left ventricular cardiac output(LVCO), is critical for the delivery of enriched oxygenatedblood to the coronary circuit, cerebral circuit, and upperbody of the fetus [1]. Premature closure or restriction ofthe FO leads to reduced or restricted right-to-left atrialshunting. Tere is substantial evidence that primary closure of the FO results in hypoplasia of left heart structures [2,3]. Studies of animal [4,5] and human fetuses[6–9] have also demonstrated that adaptive changes inFO blood fow volume occur during hypoxia or hypovolemia, suggesting that quantifying FO blood fow andits redistribution plays an important role in assessingfetal adaptation to nutrient and oxygen defciency. Tecombination of high-resolution imaging and Dopplerultrasonography has provided a method for calculating the volume of blood fow in the heart and great vesselsas the product of the Doppler fow velocity–time integral (VTI) and the cross-sectional area of the fow stream[10]. Because the FO has an irregular shape and a multiphasic blood velocity waveform during the cardiac cycle[11], FO blood fow volume was estimated previously bysubtracting pulmonary blood fow (QP) from LVCO [1,12–14]. However, indirect assessment of FO blood fowvolume resulted in varied data due to diferent measurement methods for QP used by diferent authors. Inaddition, errors arising from inaccuracies in vessel diameter measurement and Doppler recording are inevitable,especially in small-diameter vessels such as the ductusarteriosus or the left and right pulmonary arteries.

Four-dimensional ultrasound (4DUS) involves the capture of multiple adjacent two-dimensional planes to recreate a four-dimensional volume. Spatiotemporal imagecorrelation (STIC) software captures a fetal heart volume with a single automated sweep of the transducer in alimited time. Spatial and temporal information are combined to display dynamic images that can be extractedfrom volume datasets. STIC allows heart anatomy visualization through a cine loop sequence in multiplanar andrendering modes. Furthermore, fetal heart rate (FHR)can be detected and synchronized two-dimensionalimages. STIC can produce more standardized imagingof the fetal heart and reduce operator dependency compared with conventional two-dimensional ultrasound(2DUS) [15]. Visualization of specifc heart structures bysurface rendering may provide an additional method forassessing cardiac morphology and function and has beenused to calculate the area of many fetal cardiac structures, and these measurements present good intra- andinter-observer reproducibility [16–19].

Few published data exist on normal FO blood fow volume in relation to fetal gestational age. Tus, the presentstudy was designed to calculate FO blood fow volumedirectly as the product of foramen ovale area (FO-A) andforamen ovale fow velocity-time integral (FO-VTI). TeFO-A was measured by using 4DUS with STIC in rendering mode, and the FO-VTI was obtained by pulsed Doppler. Te objective of this study was to defne the normalreference range of FO blood fow volume for each gestational age.

Materials and Methods

Study Population.

Healthy women with a singleton fetus at 20–40 weeks ofgestation were recruited from the Second Xiangya Hospital, Central South University, and the Tird PeopleHospital of Yongzhou, Hunan, between April 2019 andAugust 2020 for a cross-sectional study. Te inclusioncriteria of the study population were as follows: the fetus had no cardiac or extracardiac malformation; each fetuswas appropriate for its gestational age in size (between 10and 90th percentile growth curves for local standards);the amniotic fuid volume was normal; and the motherwas in good healthand free of diabetes,

hypertension,proteinuria, smoking, and drug use. Te Research ReviewCommittee of our institution approved the research protocol, and each participant gave written consent to jointhe study.

Measurements.

Te participants were examined using a Voluson E8device (General Electric, Healthcare, Zipf, Austria)equipped with a 2.0 or 5.0 MHz RAB 4-8L convex probe.All measurements were performed in the state of fetalquietness without fetal respiratory movement. All therecordings were performed by an examiner (W.J.T) withexpertise in the 4DUS for obstetrics who received 1 yearof specifc training in the evaluating and measuringFO-A.

Initially, a 2DUS examination was performed to evaluate fetal biometry, including the biparietal diameter, headcircumference, abdominal circumference, and femorallength, and to calculate the estimated fetal weight (EFW,in kilograms). Gestational age was determined by thedate of the last menstrual period and confrmed by sonographic biometry in early pregnancy; Image-directedpulsed and colour Doppler equipment was used to obtainthe blood velocity waveforms at the level of the FO, andthe sample volume was placed on the left atrial side ofthe four-chamber view (Fig. 1). A 120 Hz high-pass flterwas used, and the spatial-peak temporal average poweroutput for colour and pulsed Doppler was kept at<100mW/cm2. Te sample volume was set at 2-3 mm. If therewas an angle of<15° between the direction of FO bloodfow and the Doppler beam, the data were included inthe analysis. From Doppler traces, the FHR (in beats perminute) was measured, and the FO-VTI (in millimeters)was determined by measuring the area underneath theDoppler spectrum with a planimeter. Only the rightto-left fow measurements were reported. At least threeconsecutive cardiac cycles were analysed, and their meanvalue was used for further analysis. Next, the right andleft atrial transverse diameter at end of systole (in millimeters) and the length of atrial septum (in millimeters)were measured in the same four-chamber view.

Subsequently, the fetal FO-A (in millimeterssquared) was calculated using the cardio-STIC application. Te four-chamber view was used as a reference. Te optimal position of the fetal spine forimaging was approximately 3 or 9 o’clock. Te sampling frame was adjusted to encompass the entire fetalheart with all its vascular connections at the smallest

possible dimension. An aperture angle of 20–40°wasused according to gestational age, and the acquisitiontime was 10–12.5 s. Each mother was asked to holdher breath during acquisition time until a single fourdimensional volume was collected. Later, ofine evaluation of the STIC volumes was performed by one of theauthors (W.J.T). Te four-chamber view was selected asa reference plane, and it was rotated around the Z-axisuntil the atrial septum and ventricular septum wereboth in a horizontal position. Te rendering buttonwas pressed, and the device was set to surface mode.Te reference dot was positioned at the level of the FOand the region of interest (ROI) was adjusted to encompass the entire atrial septum and ventricular septumwith the green line of ROI positioned inside the rightatrioventricular cavity. Te size of the ROI was modifed to include half of the left and right atriums. Eachsection was shown layer by layer from the right atrialside until the FO was clearly visible (Fig. 2a). FO-A wasmeasured at its maximum opening. Next, the renderedimage was enlarged and shown on the full screen, themeasure button was activated, and the area was manually painted along the margin of FO (Fig. 2b).

Blood fow calculations.

Te absolute blood fow of foramen ovale (QFO, in milliliters per minute) was calculated with the following formula: QFO=FHR×FO-A×FO-VTI.

Next, QFO was normalized to EFW. Weight-indexedforamen ovale blood volume (iQFO, in milliliters perminute per kilogram) was calculated by using the following formula: iQFO=QFO/EFW.

For the assessment of inter-observer reliability, a second examiner (J.W.Y) with the same level of experiencein 4DUS in obstetrics performed a second measurementon QFO of 80 randomly chosen fetuses. Te examinerswere not allowed to share results.

Statistical analysis

For descriptive analysis, the means, medians, standarddeviations, and maximum and minimum values of theFO-A and FO blood volume for each gestational age weredetermined. A curvilinear regression model was ftted tothe data versus gestation age, and to interpolate FO-Aversus EFW. Bland–Altman plots were used to assessinter-observer variability. Statistical analysis was performed using SPSS 25.0. For all the analyses, statisticalsignifcance was defned as P<0.05.

Results

A total of 550 participants underwent fetal echocardiography, of whom 110 were removed from the data set dueto the poor technical quality of their fetal cardiac images,which was caused by maternal conditions including obesity and abdominal scars. Te concentration of losers inthe third trimester led to a 69% (253/363) of the viewing rate for FO-A in late pregnancy. Adequate datasets

for subsequent analysis were ultimately obtained in 440fetuses (80% rate of acceptance). Each fetus was deliveredat term and had an uncomplicated neonatal course. Temean gestational age (with SD) was 29.6±5.7 weeks, themean maternal age was 27.9±6.0 years (range 17–42),the mean maternal weight was 55.49±4.18 kg, and themean maternal height was 159.97±4.37 cm.

Te mean FO-A was 20.53±2.79mm2(range15.25–25.47) in the 20th week of pregnancy and57.85±7.81mm2(range 49.12- 76.14) in the 40th week ofpregnancy. In STIC rendering mode, the FO presented around, oval, or irregular shape. Te growth of the FO-Aincreased signifcantly before 30 weeks, followed byslower growth (Fig. 3a). Figure 3b showed increasing areaof the FO versus EFW. Te power exponent of the FO-Acurve indicated that the increase in FO-A was approximately the square root (0.49) of the fetal body weight.

The median FHR derived from Doppler recording inthe FO was 141 bpm (mean 142 bpm; SD 9.11 bpm).Figure 4demonstrated the trend in FO VTI withadvancing gestational age.

General echocardiographic findings, 5th, 50th, and95th percentiles for the QFO and iQFO at each gestational age evaluated were shown in Table 1.

Fetal QFO increased more than threefold from 20to 40 weeks of gestation. Figure 5a showed the QFOincreased significantly from 20 to 30 weeks of gestation (the mean growth rate was 67%) and then showeda slow upward trend between 30 and 40 weeks of gestation (the mean growth rate was 17%). Figure 5bpresented the trend of iQFO with gestational weeks.The negative slope of iQFO was significantly from 20to 40 weeks of gestation, mean iQFO decreased by59% (from 450.30 ml/min/kg at 20 weeks of gestationdropped to 184.91 ml/min/kg at term). Median iQFOwas 320.82 ml/min/kg (mean 319.1 ml/min/kg; SD106.33 ml/min/kg).

Through the Bland–Altman plots, good agreementin the measurement for the QFO was observed, witha mean difference of 2.84 (SD±24.92 ml/min and95% CI±48.84 ml/min) for inter-observer reliability(Fig. 6).

Discussion

Our study identifed gestational age-specifc referenceranges for FO blood volume in normal fetuses from 20to 40 weeks of gestation using STIC in rendering modecombined with pulsed Doppler. Te fndings of this

study suggest that QFO increases with advancing gestational age, showing an exponential increase from 20 to30 weeks of gestation and a fat growth trend during thelast trimester, while iQFO decreases with advanced gestational weeks.

Te present study showed that QFO increased morethan threefold from 20 to 40 weeks of gestation. QFOshowed an exponential increase between 20 and 30 weeksof gestation, the mean growth rate was 67%, whereas themean growth rate was only 17% from 30 to 40 weeks ofgestation. In fetal circulation, enriched oxygenated bloodoriginating from the placenta enters the fetus via theumbilical vein, and part of this constitutes the majority of LVCO through the FO. Terefore, the growth ofQFO was associated with an increase in LVCO withadvancing gestational age. Te emergence of a period ofslow growth may be related to the increase in QP in thethird trimester. In a study investigating the relationshipbetween QFO and QP in the proportion of LVCO composition in human fetuses during pregnancy, Rasanenet al. [14]found that QP constituted 27% of the LVCO atthe beginning of the second half of pregnancy and 50% ofthe LVCO during the third trimester. Te flling of the leftatrium is enhanced by an increase in pulmonary venousreturn, such that right-to-left shunting through the FOmay be limited due to the elevation of left atrial pressure.Animal studies have also demonstrated that QFO cannotfully compensate for impaired pulmonary venous return[20,21], and FO has limited ability to increase its volumeblood fow at near-term gestation [22].

Tere was a declining trend in iQFO with advancinggestational age in our study. Mean iQFO decreased by59%. Te decrease of iQFO with advanced gestationalweeks may be related to the decrease of ductus venosus shunt in addition to the increase of QP. Bellotti et al.[23] found the weight-indexed amniotic umbilical fowdid not change signifcantly during gestation, whereasweight-indexed ductus venosus fow decreased signifcantly, the percentage of umbilical blood fow shuntedthrough the ductus venosus decreased signifcantly(from 40 to 15%), consequently, the percentage of fowto the liver increased between 20 and 38 weeks gestation. Rudolph et al. [24] found that 55% of the umbilicalblood fow was shunted through the ductus venosus in33 exteriorized human fetuses at 10 to 20 weeks gestation by using isotope-labelled microspheres, and Kiserud et al. [25] found only 20% to 30% shunting throughthe ductus venosus for the direct supply of the FO during the second half of pregnancy by ultrasonic measurement; correspondingly, an average of 70% to 80% of theoxygenated umbilical venous blood perfused the hepatictissue. As fetal growth accelerates, an increased proportion of umbilical blood is directed to the liver, and theproportion that is shunted through the ductus venosus decreases, although the blood fow in the umbilical vein grows as pregnancy progresses. Te reductionof shunting through the ductus venosus to the inferiorvena cava limits, to some extent, the capacity of theFO to increase its volume of blood fow in the thirdtrimester. At the same time, the increasing diversion

from cardiac to pulmonary circulation enhances theoutput of left ventricle as gestation advances, althoughthe refux of deoxygenated blood from the lungs to theleft atrium through the pulmonary veins is marginal inearly pregnancy. Phase contrast magnetic resonanceimaging (PC-MRI) is a promising new technique for thestudy of the fetal circulation. It has been shown to befeasible in measurement of the vessel blood fow in thefetal lamb [26] and late-gestation human fetus [27–30],and has been used successfully to make initial observations of redistribution of the fetal circulation in humanfetuses with congenital heart disease [31,32]. Te meanvalues of iQFO in the third trimester calculated in thepresent study are slightly higher than previous PC-MRI

measurements in a small series of human fetuses in latepregnancy [27,28]. Tis may be because ultrasoundtechnique assumed a constant fow velocity across thevessel lumen, where in fact the fow was likely sloweraround the vessel periphery than the fow velocitysampled in the middle of the vessel. Te Doppler fowparameter used to quantify blood fow in our study isVTI which is the distance that the maximum fow velocity of the stream travels per unit time regardless of thedistribution gradient of blood fow velocity in the vessel, this parameter may lead to the overestimation ofiQFO to some extent. In addition, intrinsic inaccuraciesin ultrasound measurements of fow may also accountfor discrepancy, such as the measurement of the crosssectional area and ultrasonic beam angle correction [33].Te mean iQFO (319.1±106.33 ml/min/kg) in the present study is higher than fetal lamb study [26] in whichthe mean iQFO was 164±20 ml/min/kg. Te possiblereason is the diferent proportion of brain mass in animal and human fetuses in addition to the above factors.

Analysing the growth pattern of FO is helpful tounderstand the variation trend of QFO. In the presentstudy, the FO-A was based on direct measurements inSTIC rendering mode. Te data showed linear increaseswith advancing gestation, but slow growth was seenafter 30 weeks of gestation. Tis growth pattern isdetermined by changes in blood fow through the FOduring diferent gestations, the FO-A increases proportionally to the absolute blood fow to the FO. Te datafor the FO-A were slightly higher than those in a previously published report [34], in which the horizontalarea between the FO valve and atrial septum, as calculated by 2DUS, grew from 15 to 50 mm2between18 and 42 weeks of gestation. However, Patten et al.found that the area of the orifce in the atrial septumwas larger than the horizontal area between the FOvalve and atrial septum in fetuses and new-borns [35],which corresponds closely to our results. Te diferencein measurement methods may be a factor contributing to the discrepancy. Te sagittal view of the FO inthe atrial septum not routinely feasible with 2DUS canbe generated by STIC. It is possible to directly observeand measure the size of FO by using STIC. Yagel et al.[36] have also suggested that the rendering mode enables superior evaluation of FO in the interatrial septum. Furthermore, STIC ofers the possibility of ofineanalysis without the need for the pregnant woman tobe present [37,38]. In our study, a section including allfour heart chambers was easily obtained as the initialscanning surface. As long as the professional has beenappropriately trained to sample cardiac volumes, thesevolumes can be obtained without the need for a specialist, which does not become essential until later, toidentify possible abnormalities of heart morphology[38]. An interesting observation was made when wecompared FO-A with EFW, FO-A increases approximately as the square root (0.49) of the EFW. Our fnding clearly demonstrates that the increase in FO-A isproportional to body size.

Te Doppler methods had good accuracy in evaluatingblood fow when suitably designed equipment was usedin appropriate situations, with less than 6% systematic errors. However, measurement of cross-sectional areaand beam angle correction led to random errors [33].Fetal FO blood fow has been studied by Doppler ultrasound by some scholars [1,13,14], while these studieswere mainly to establish the concept of fetal combinedcardiac output, the research on FO blood fow wasfocused on the change trend of its proportion in combined cardiac output with gestational weeks, so thatthere was no specifc value of FO blood fow for clinical reference. Terefore, the results of our study cannotbe compared with these studies. Previously, FO bloodfow was estimated indirectly by subtracting QP fromLVCO. QP was expressed as the diference between theright cardiac output and the ductus arteriosus or theleft and right pulmonary arteries. In these methods,the inner diameter and blood fow spectrum of multiple vessels need to be measured, there is no doubt thaterrors were inevitable. It was reported that errors indiameter measurement constitute the main obstacle toreliable fow measurement in small vessels when ultrasonographic techniques are used [39]. Te inherent inaccuracy of these methods limits routine clinical practice.Our study shows that the direct measurement of bloodfow in the FO is feasible. In our study, the four-chambersection which is used as the measured section is easilyaccessible. All the paraments needed to calculate FOblood fow volume can be obtained in a single section.Te measurement errors resulting from multiple vesselsinner diameters and beam angle adjustment in diferentsections are likely to be greatly reduced. Furthermore,the error caused by measurement in diferent periods inwhich the physiological state of the fetus is diferent canbe reduced as much as possible. Te QFO measurementsobtained in this study showed good inter-observer reliability. Terefore, we believe that our current resultsprovide an accurate presentation of QFO from 20 to40 weeks of pregnancy and that the range provides a reasonable refection of the biological variation. However,there are several limitations to the present work. Temajor limitation of the current study is its possible selection bias. FO-VTI was measured specifcally in fetuseswith only right-to-left shunting, contributing to theoverestimation of QFO to some extent. Te novelty ofthis approach warrants the reporting of this initial sample, and further studies with larger numbers of subjectswill ensure the accuracy and reliability of the currentstudy. Our initial experience suggests that this techniqueis best suited to fetus in the middle of gestation becauseadequate image acquisition and display with STIC arelimited by early gestational age, fetal positioning, maternal obesity, and previous lower abdominal surgery, especially in the third trimester, when only 69% of data in thecurrent were accepted despite the overall qualifcation rate of 80%, however, these disadvantages are also inherent to conventional ultrasonography. Fetal movementand sudden changes in FHR during data collection areadditional factors afecting the technique, possibly causing serious mismatch of the information demanded foraccurate reconstruction of moving cardiac structures.

Conclusions

In summary, this study provides reference data on FOblood volume by means of pulsed Doppler combinedwith STIC. Te fndings of this study suggest that QFOincreases with advancing gestational age, while iQFOdecreases with advanced gestational weeks. Te FOmight have a limited capacity to increase its volume ofblood fow during the last trimester. We believe thatestablishing the normal reference range of FO bloodvolume can not only improve the understanding of fetalcirculatory physiology but also help to assess the risklevel of the fetus at an early stage and facilitate perinatalmanagement.