Fetal Atrioventricular Flow-Velocity Waveforms and Their Relation to Arterial and Venous Flow-Velocity Waveforms at 8 to 20 Weeks of Gestation
Background Doppler ultrasonography was used to determine the nature and gestational age-related changes of human fetal AV flow-velocity waveforms and to establish their relationship with arterial impedance indexes and venous flow velocities in normal human fetuses between 8 and 20 weeks of gestation.
Methods and Results Flow-velocity waveform recordings were attempted in 318 singleton pregnancies. After the exclusion criteria were applied, data on 214 women were available for further analysis. Differentiation between E wave and A wave became possible at 9 weeks, whereas distinction between transmitral and transtricuspid valve flow velocities was first achieved at 10 to 11 weeks. A statistically significant nonlinear gestational age-dependent increase was established for all AV waveform parameters, which became linear when related to logarithmically estimated fetal crown-to-rump length. Transtricuspid valve flow velocities were significantly higher than transmitral valve flow velocities. Transmitral valve time-averaged flow velocities were positively correlated with peak diastolic velocities and time-velocity integral of late-diastolic reverse flow in the inferior vena cava. No correlation existed between AV time-averaged velocities and arterial impedance indexes.
Conclusions Monophasic AV flow-velocity waveforms can be recorded as early as 8 weeks of gestation and become biphasic as early as 8 weeks. They demonstrate a linear increase relative to logarithmically estimated fetal crown-to-rump length, suggesting that fetal growth-related increase in volume flow plays a role in this velocity rise. Transtricuspid valve A-wave and E-wave velocities suggest right ventricular predominance as early as the late first trimester of pregnancy. AV flow velocities are not related to arterial downstream impedance.
Cardiac development is characterized by morphogenesis, growth, and changing hemodynamics. In the human fetus, normal or abnormal cardiac anatomy can now be established with reasonable confidence by diagnostic ultrasound by 18 to 20 weeks of gestation.1 2 As a result of further experience in transvaginal ultrasound, some reports have appeared on the sonographic detection of congenital heart disease as early as 12 to 14 weeks of gestation.3 Since form and function are interrelated, knowledge of cardiac hemodynamics will be helpful in establishing the prognosis in a particular congenital cardiac anomaly.
Doppler echocardiography does not allow data collection on volume flow, but cardiac velocity recordings would be helpful in providing some insight into the intricate relationship between cardiac and extracardiac hemodynamics. Although several reports have appeared on fetal cardiac hemodynamics late in pregnancy,4 5 6 little information is available on early gestation. Preliminary information shows that cardiac flow velocities can be obtained as early as the first trimester of pregnancy.7 8 Moreover, a profound reduction in arterial downstream impedance has been observed at the fetoplacental level at 12 to 14 weeks of gestation.9 At the same time, changes occur in venous flow-velocity waveforms, notably in the ductus venosus and IVC.10 We propose that these extracardiac arterial and venous flow-velocity waveform changes are reflected in cardiac hemodynamics.
The present study focuses on cardiac diastolic filling characteristics and their relationship with venous inflow velocities and arterial downstream impedance.
The objective of the present study was threefold: (1) to establish the nature of normal human AV flow-velocity waveforms at 8 to 20 weeks of gestation, (2) to determine the gestational age- and fetal CRL-related changes in trans-MV and trans-TV flow velocities, and (3) to relate trans-MV and trans-TV flow velocities to arterial downstream impedance and flow velocities at the venous inflow level during this early period of normal fetal development.
Between August 1, 1992, and April 1, 1994, a total of 318 women with a normal singleton pregnancy between 8 and 20 weeks of gestation (median, 13 weeks) consented to participate in the study. The study protocol was approved by the Hospital Ethics Committee. Maternal age ranged between 16 and 40 years (median, 32 years). Pregnancy duration was estimated from the last menstrual period and confirmed by ultrasound measurement of the fetal CRL11 (8 to 12 weeks) or biparietal diameter12 (12 to 20 weeks). Each woman was included in the study once.
Only pregnancies that progressed uneventfully resulting in the term delivery of a normal infant with a birth weight between the 10th and 90th percentiles corrected for maternal parity and fetal sex13 were included in the data analysis.
Another 10 uncomplicated pregnancies were investigated both transvaginally and transabdominally at 12 and 13 weeks of gestation to compare fetal AV flow velocities obtained through both recording techniques.
Ultrasound Doppler studies were performed with a Hitachi EUB 450 (Hitachi Medical Corp). We used a combined transvaginal real-time and pulsed Doppler system (carrier frequencies, 3.5 and 6.5 MHz, respectively) at 8 to 13 weeks of gestation or a combined transabdominal real-time and Doppler system (carrier frequencies, 3.5 and 3.0 MHz, respectively) at 14 to 20 weeks of gestation. The system operates at power outputs of <100 mW/cm2 spatial-peak temporal average in both imaging and Doppler modes according to the manufacturer's specifications. Doppler recordings were performed by one examiner (P.v.S.).
Flow-velocity waveforms at the fetal AV level were obtained from the cardiac “four-chamber” view. To obtain this view, first, a transverse cross section of the fetal chest at the level of the pulsating heart was obtained. An imaginary line was subsequently drawn from the fetal spine to the anterior chest wall, dividing the fetal chest into two equal parts. At an angle of 45° to the left or right of this imaginary line (depending on fetal position), the “four-chamber” view is presented, allowing identification of the atrial and ventricular parts of the heart and the AV valves, with distinction between the left and right sides of the heart and separate identification of the MV and TV, depending on gestational age (Fig 1⇓). Assuming that AV blood flow is parallel to this 45° axis, we positioned the Doppler interrogation beam as much as possible along this axis, and the Doppler sample volume was placed immediately distal to the MV and TV or immediately distal to the AV valve if distinction between MV and TV was not possible. Only waveforms with angles of insonation <30° and consisting of clear E and A waves were accepted. Previous experience7 has demonstrated that acceptable waveforms can be obtained in the majority of transvaginal examinations before 14 weeks of gestation. Flow-velocity waveforms from the umbilical vein and artery were obtained from a straight section of the free-floating loop of the umbilical cord. Flow-velocity waveforms from the ductus venosus, IVC, and descending aorta were recorded as previously described.14 15 16 The angle of insonation was always <30°, and for the descending aorta, <45°. Sample volume length for all flow-velocity waveform recordings ranged between 0.1 and 0.2 cm; the high-pass filter was set at 100 Hz. Although the range of motion of the transvaginal probe is limited, flow-velocity waveform recordings could be obtained in most cases because of recurrent changes in fetal position during the examination.
All Doppler studies were performed with the women in the semirecumbent position and during fetal apnea. The total examination time was limited to 15 minutes in each instance. All flow-velocity waveforms were recorded on hard copies. Waveform analysis was performed by one examiner (P.v.S.) using a microcomputer (Olivetti M24; Olivetti BV) linked to a graphical tablet.
Analysis of the AV waveforms consisted of calculation of (1) time-averaged velocity (cm/s), (2) peak velocity (cm/s) of the E wave (passive atrial filling) and A wave (atrial contraction), and (3) E/A ratio. Time-averaged velocity (cm/s) was determined in the umbilical vein. Waveform analysis in the ductus venosus and IVC consisted of calculation of (1) peak velocity (cm/s) during both systole and diastole and (2) time-averaged velocity (cm/s). In the IVC, the time-velocity integral of late-diastolic reverse flow, which is expressed as a percentage of total forward flow during systole and early diastole, was calculated. The PI was calculated for the umbilical artery and fetal descending aorta, which is expressed as the difference between the maximal and minimal flow velocity divided by the time-averaged velocity. Fetal heart rate (bpm) was determined from the time interval (m/s) between peak systolic velocities of two successive E-wave or A-wave flow velocities. Three consecutive flow-velocity waveforms with the highest velocity and similar appearance were used to calculate the different parameters across the AV valves and in each vessel. Earlier, in our center,17 acceptable intraobserver reproducibility was established for fetal flow-velocity waveforms during the late first and early second trimester of pregnancy. Intraobserver coefficient of variation was <6% for all AV and IVC flow-velocity parameters except for time-velocity integral of IVC late-diastolic reverse flow (8%). Intraobserver coefficient of variation was <5% for the umbilical artery and descending aorta PI and <4% for umbilical venous and ductus venosus velocity parameters. Sonographic measurement of fetal head and upper-abdominal circumference allows a crude approximation of fetal weight in late pregnancy; however, it is unreliable in early gestation. Instead, estimates of fetal CRL may serve as a measure of fetal growth in early pregnancy. CRL measurements can be reliably obtained by ultrasound up to 13 weeks of gestation,11 whereas after that, CRL data based on aborted specimens are available.18
The transvaginal technique of flow-velocity waveform recording was validated by comparison of transvaginally and transabdominally collected waveforms at the AV level in 10 normal singleton pregnancies at 12 to 13 weeks of gestation. This period of gestational age was selected because at that time, a transition from transvaginal to transabdominal scanning takes place. In half the women, the transvaginal scan preceded the transabdominal scan; in the other half, the transabdominal scan preceded the transvaginal scan. The total examination period was limited to 15 minutes.
Correlation coefficients (rs) were calculated with the Spearman rank correlation test to establish the relationship between gestational age and (1) descending aorta and umbilical artery PI and (2) venous flow-velocity parameters. Linear regression analysis was used to assess (1) fetal heart rate relative to gestational age and (2) AV flow-velocity waveform parameters relative to gestational age as well as estimated fetal CRL. In the latter instance, logarithmic transformation of the fetal CRL data was carried out because fetal CRL shows a 1.7-fold increase at 10 to 12 weeks and a 1.1-fold increase at 18 to 20 weeks of gestation. The difference between the slopes of two regression lines was tested by a simple z test based on the difference of the two estimated slopes and their corresponding standard errors. The paired t test was applied to establish the difference in time-averaged velocity, peak E-wave and peak A-wave velocity, and E/A ratio between trans-MV and trans-TV flow-velocity waveforms. Multiple regression analysis was used to establish the relationship between AV flow-velocity waveform parameters and fetal heart rate, adjusted for gestational age. Partial correlation coefficients19 were calculated to assess the relation between AV flow-velocity waveform parameters and (1) descending aorta and umbilical artery PI and (2) venous flow-velocity waveform parameters. The paired t test was used to establish the difference in arterial, cardiac, and venous flow-velocity waveforms between the transvaginal and transabdominal approaches at 12 to 13 weeks of gestation. Limits of agreement between the transvaginal and transabdominal approaches were calculated according to Bland and Altman.20 Limits of agreement were defined as the range in which ≈95% of the differences between the transvaginal and transabdominal approach are situated. Data are reported as mean±SD. Values of P<.01 were considered statistically significant.
Of the 318 women who consented to participate in the study, 18 were excluded from analysis because no Doppler signals could be obtained as a result of maternal obesity, fetal position, or fetal body movements. Of these 18 women, 12 were investigated transvaginally and 6 transabdominally. An additional 48 women were excluded independently of the study protocol because of a fetal birth weight below the 10th percentile or above the 90th percentile for gestational age, and 38 women subsequently dropped out because of pregnancy abnormalities. These exclusions were made independently of the knowledge obtained from the study protocol. Flow-velocity waveform recordings from 214 women were available for further analysis.
Validation of Transvaginal Flow-Velocity Waveform Recordings
Comparison of transvaginal and transabdominal flow-velocity waveform recordings at 12 to 13 weeks revealed no statistically significant difference for the AV flow-velocity parameters (Table⇓).
AV Flow-Velocity Waveforms
At 8 weeks of gestation, only monophasic AV flow-velocity waveforms could be obtained. Differentiation between early-diastolic filling velocities (E wave) and late-diastolic velocities (A wave) became feasible in 6 of 17 cases (35.3%) at 9 weeks, in 17 of 20 cases (85%) at 10 weeks, in 13 of 15 cases (86.7%) at 11 weeks, and in all cases after 11 weeks of gestation. Differentiation between trans-MV and trans-TV velocities was not possible at 8 to 9 weeks but was achieved in 1 of 21 cases (4.8%) at 10 weeks, in 14 of 29 cases (48.3%) at 11 weeks, and in all cases after 11 weeks of gestation.
AV Diastolic Velocities Relative to Fetal Heart Rate, Gestational Age, and CRL
A statistically significant negative regression coefficient (r=−.81, P<.001) was established for fetal heart rate relative to gestational age, with a mean value of 175.0±6.1 bpm at 8 weeks, 158.0±6.4 bpm at 15 weeks, and 152.0±5.1 bpm at 20 weeks of gestation. A statistically significant negative regression coefficient (P<.001) was found between fetal heart rate and the following trans-MV and trans-TV flow parameters: (1) E-wave velocity (MV, r=−.39; TV, r=−.47), (2) time-averaged velocity (MV, r=−.42; TV, r=−.48) and (3) early/late-diastolic (E/A) ratio (MV, r=−.47; TV, r=−.42). However, when adjusted for gestational age, these regression coefficients were no longer significant. A statistically significant linear increase (P<.01) relative to gestational age was established for early-diastolic (E wave; Fig 2⇓), late-diastolic (A wave; Fig 3⇓), and time-averaged velocities (Fig 4⇓) as well as E/A ratio (Fig 5⇓) in the absence of MV and TV differentiation (AV, 9 to 12 weeks), at the MV level (10 to 20 weeks), and at the TV level (10 to 20 weeks). Reference ranges depicted in Figs 2 through 5⇓⇓⇓⇓ are based on the assumption that trans-MV and trans-TV regression lines run parallel.
Also, the slope for late-diastolic (A-wave) and time-averaged velocities in the absence of MV and TV differentiation (9 to 12 weeks) was statistically significantly different from the slope at the MV level (10 to 20 weeks; A-wave velocity, P=.0003; time-averaged velocity, P=.003) and at the TV level (10 to 20 weeks; A-wave velocity, P=.001; time-averaged velocity, P=.0005).
A statistically significant positive linear regression coefficient (P<.001) was, however, established for time-averaged velocities (AV, y=−8.24+9.28 log10 x; MV, y=−3.59+6.48 log10 x; TV, y=−5.30+8.07 log10 x; cm/s) relative to logarithmically transformed fetal CRL (mm) (Fig 6⇓). The slope for AV time-averaged velocities at 9 to 12 weeks (AV) and at 12 to 20 weeks of gestation (MV, TV) was not statistically significantly different when related to logarithmically transformed fetal CRL.
Trans-MV Relative to Trans-TV Diastolic Velocities
Throughout the study period, trans-TV flow velocities were significantly higher than MV velocities for E-wave velocities (mean difference, −4.67; SD, 5.37; P<.001), A-wave velocities (mean difference, −3.69; SD, 4.74; P<.001), E/A ratio (mean difference, −0.02; SD, 0.08; P<.01), and time-averaged velocities (mean difference, −1.82; SD, 1.79; P<.001). In all cases, the difference between trans-MV and trans-TV flow-velocity waveform parameters was not related to gestational age.
AV Flow Velocities Relative to Arterial and Venous Flow-Velocity Waveform Parameters
A statistically significant decrease with gestational age was established for fetal descending aorta PI (rs=−.79, P<.001) and umbilical artery PI (rs=−.83, P<.001). The decrease was most pronounced after 11 weeks of gestation, with mean values of 2.72±0.33 and 2.61±0.30 at 12 weeks and 1.58±0.17 and 1.22±0.10 at 20 weeks. No correlation existed between AV flow velocities and descending aorta and umbilical artery PI.
A statistically significant increase (P<.001) with gestational age was found for (1) the time-averaged velocity in the umbilical vein and (2) the peak systolic velocity, peak diastolic velocity, and time-averaged velocity in the ductus venosus and in the IVC. A statistically significant gestational age-related reduction was established for the time-velocity integral of late-diastolic reverse flow in the IVC (P<.001). The slope for the time-averaged velocity in the umbilical vein and ductus venosus at 8 to 12 weeks of gestation was significantly different from the slope at 13 to 20 weeks of gestation (umbilical vein, P<.001; ductus venosus, P<.01). After adjustment for gestational age, a statistically significant correlation was established between trans-MV time-averaged velocity and peak diastolic velocity (rs=+.48, P<.01) and time-velocity integral of late-diastolic reverse flow (rs=−.50, P<.01) in the IVC. No correlation existed between AV waveform velocities and ductus venosus flow velocities.
Embryonic and early fetal development is characterized by rapid growth and cardiac morphogenesis. The embryonic heart develops from a smooth-walled cardiac loop into a septated trabecular heart. Characterization of the functional aspects of the embryonic and early human fetal cardiovascular system is important in the eventual understanding of normal and abnormal cardiovascular development. The present article describes diastolic filling characteristics between 8 and 20 weeks of gestation and provides a perspective of the changes in these characteristics, including their relationship to venous, descending aortic, and umbilical artery flow-velocity waveforms. It should be emphasized that noninvasive Doppler studies of the human fetal circulation allow assessment only of flow velocities. The absence of intracardiac and extracardiac volume flow and pressure measurements puts a restriction on the interpretation of our data.
Transvaginal pulsed Doppler ultrasound allows flow-velocity waveform analysis as early as 8 weeks of gestation. At that time, only monophasic velocities were obtained, as was recently reported by Leiva et al.8 This could be the result of increased heart rate with reduced time for early-diastolic filling or reduced ventricular compliance relative to older fetuses, with very low, perhaps immeasurable or absent, early-diastolic filling. Another explanation of the failure to identify separate early-diastolic filling (E-wave) and late-diastolic atrial contraction (A-wave) velocities at 8 weeks may be the result of limitations in image resolution, despite the high-quality ultrasound equipment used in this study. After 8 weeks, AV velocity waveforms increasingly resemble those observed in late pregnancy, with a well-defined early-diastolic E-wave and late-diastolic A-wave component. The absence of retrograde flow at the AV level confirms that AV cushions function as valves during the cardiac cycle this early in pregnancy.
Differentiation between trans-MV and trans-TV flow velocities was first achieved at 10 to 11 weeks. After 11 weeks of gestation, E-wave and A-wave velocities can be recorded at both MV and trans-TV levels in every instance. Trans-MV and trans-TV E-wave and A-wave velocities display a marked rise with advancing gestational age, reflecting an increase in early-diastolic filling and atrial contraction and as a result an ≈1.6-fold rise in time-averaged velocities at the AV level at 8 to 20 weeks of gestation, reflecting the larger volume of blood entering the ventricles. The gestational age-dependent rise in E/A ratio suggests a shift of blood flow from late diastole toward early diastole, which may be due to increased ventricular compliance and/or raised ventricular relaxation rate. In the chick embryo, average diastolic ventricular wall stiffness decreases geometrically with development.21 Our findings of a gestational age-dependent rise in E/A ratio are in agreement with Tulzer et al.22 They demonstrated, however, that despite a changing relation between early and late inflow velocities, the proportion of ventricular filling contributed by atrial contraction remains constant, indicating unchanged ventricular compliance.
Of interest is the significantly steeper slope of increment for time-averaged velocities from data before sonographic differentiation between trans-MV and trans-TV velocity waveforms (9 to 12 weeks) compared with data collected after differentiation became possible (13 to 20 weeks). A similar change in slope of increment was established at around 12 weeks for the time-averaged velocity in the umbilical vein and ductus venosus but not in the IVC. These changes are unrelated to the technique of measurement. No difference between transvaginal and transabdominal flow-velocity waveform recordings could be established at 12 to 13 weeks of gestation. Comparison with fetal CRL estimates revealed that AV, umbilical venous, and ductus venosus time-averaged velocities increase parallel to an increase in fetal CRL. This indirectly suggests that fetal growth-determined increase in volume flow plays a part in these gestational age-related flow-velocity changes. This is further supported by a study in the chick embryo, in which the increase in cardiac output as represented by dorsal aortic volume flow is comparable to the increase in body weight beyond Hamburger and Hamilton23 stage 12 (50 hours of 21-day incubation).24 For direct information on AV volume flow, data on both AV velocities and valve area are needed. However, the latter cannot be reliably measured at this early stage of fetal cardiac development.
Fetal heart rate showed a significant reduction, which has been explained by parasympathetic development.25 With an ≈1.2-fold drop, the change in fetal heart rate was most pronounced before 15 weeks of gestation. When it was adjusted for gestational age, no correlation between fetal heart rate and AV diastolic velocities could be established, suggesting independence of these velocities from heart rate at this stage of parasympathetic nerve development. A nearly twofold reduction in descending aorta and umbilical artery PI was observed after 12 weeks of gestation, reflecting a marked drop in fetoplacental vascular resistance, which may be determined by the process of angiogenesis taking place in the developing placenta.26 No relation could be established between AV flow velocities and PI in the descending aorta and umbilical artery, indicating trans-MV and trans-TV flow velocities to be independent of arterial downstream impedance at the fetal trunk and placental levels.
Animal experimental27 and human fetal studies using color-coded Doppler ultrasound28 indicate that blood flow from the IVC is directed primarily through the TV to the right ventricle and blood flow from the ductus venosus through the MV to the left ventricle, although some mixture may exist. Diastolic components of venous flow velocities are subject to intrinsic cardiac properties such as the degree of atrial filling and atrial contraction force. However, we found no relationship between ductus venosus and AV flow velocities, whereas early-diastolic (forward) and late-diastolic (reverse) velocities in the IVC were related only to trans-MV time-averaged velocities. These data suggest that variables other than volume flow, such as the pressure gradient across the AV valves, may also be responsible for the observed AV flow-velocity changes with advancing gestational age.
During the entire study period, both E-wave and A-wave velocities are responsible for the higher time-averaged velocities at the TV level compared with the MV level. Since volume flow is equal to mean velocity multiplied by vessel area, the higher trans-TV time-averaged velocities may reflect increased right ventricular stroke volume and output. This would be in agreement with observations of right ventricular predominance in normal late pregnancies.4 Further support for this was provided by the observation22 that atrial contribution to ventricular filling was higher at the TV than at the MV. It was suggested that the right ventricle might be less compliant than the left ventricle because of a larger right ventricular muscle mass.
It can be concluded that flow-velocity waveforms at the AV level can be recorded as early as 8 weeks of gestation. These waveforms are mostly monophasic before 9 weeks of gestation and biphasic thereafter. Normal late first and early second trimester pregnancies are characterized by marked changes in trans-MV and trans-TV flow velocities. These changes may be determined primarily by increased volume flow in the developing fetus, which would explain the absent relation between AV flow velocities and arterial downstream impedance at the fetal placental level. Trans-TV A-wave and E-wave velocities suggest right ventricular predominance as early as the late first and early second trimester of pregnancy.
Selected Abbreviations and Acronyms
|IVC||=||inferior vena cava|
This study was supported by a grant from the Netherlands Organization for Scientific Research, NWO (grant 900-516-139).
- Received November 1, 1995.
- Revision received March 21, 1996.
- Accepted March 26, 1996.
- Copyright © 1996 by American Heart Association
Allan LD, Crawford DC, Chita SK, Tynan MJ. Prenatal screening for congenital heart disease. Br Med J. 1986;292:1717-1719.
Reed KL, Meijboom EJ, Sahn DJ, Seagnelli S, Valdes-Cruz LM, Shenker L. Cardiac Doppler flow velocities in human fetuses. Circulation. 1986;73:41-46.
Kenny JF, Plappert T, Doubilet P, Salzman DH, Cartier M, Zollars L, Leathermann GF, St John Sutton MG. Changes in intra-cardiac blood flow velocities and right and left ventricular stroke volume with gestational age in the normal human fetus: a prospective Doppler echocardiographic study. Circulation. 1986;74:1208-1216.
Allan LD, Chita SK, Al-Ghazali W, Crawford DC, Tynan M. Doppler echocardiographic evaluation of the normal fetal heart. Br Heart J. 1987;57:528-533.
Leiva M, Tolosa J, Binotto C, Weil S, Huhta J. First trimester normal human cardiac Doppler. J Am Coll Cardiol. 1994;843:53. Abstract.
Kloosterman G. On intrauterine growth. Int J Gynecol Obstet. 1970;8:895-912.
Reed KL, Appleton CP, Anderson CF, Shenker L, Sahn DJ. Doppler studies of vena cava flows in human fetuses: insight into normal and abnormal cardiac physiology. Circulation. 1990;81:498-505.
Eik-Nes SH, Brubakk AO, Ulstein MK. Measurement of human fetal blood flow. Br Med J. 1980;280:283-284.
Moore KL. The eighth week to birth. In: J. Dusseau, ed. Before We Are Born: Basic Embryology and Birth Defects. Philadelphia, Pa: WB Saunders Co; 1974.
Altman DG. Relation between two continuous variables. In: Practical Statistics for Medical Research. New York, NY: Chapman & Hall; 1993.
Hu N, Keller BB, Taber LA, Clark EB. Diastolic wall modulus and ventricular stiffness in the stage 16 to 27 chick embryo. Circulation. 1990;82(suppl III):III-605. Abstract.
Hu N, Clark EB. Hemodynamics of the stage 12 to stage 29 chick embryo. Circ Res. 1989;65:1665-1670.
Wladimiroff JW, Seelen J. Doppler tachometry in early pregnancy: development of fetal vagal function. Eur J Obstet Gynecol Reprod Biol. 1972;2:55-63.
Edelstone DI, Rudolph AM. Preferential streaming of ductus venosus blood to the brain and heart in fetal lambs. Am J Physiol. 1979;237:H724-H729.