The purpose of this review is to analyze the current modalities available for the assessment of fetal cardiac function. The fetal heart differs from the ex-utero heart in both structure and function. The fetal heart represents two circulations, which effectively run in parallel, with two ‘shunts’ connecting them, i.e. the ductus arteriosus and foramen ovale. In ex-utero life, the two circulations are referred to as pulmonary and systemic; however, in the fetus this distinction is somewhat euphemistic, and it may be more accurate to discuss right and left circulations, a concept highlighted by Kiserud et al.1. Although studies disagree on the exact figure, there is a broad consensus that the fetal heart (in both the human and lamb model) exhibits right-sided dominance, with the majority (52–65%)2–5 of cardiac output (CO) going through the right ventricle (RV). Of the right ventricular output the majority (756–90% )4–6 is shunted through the ductus arteriosus to the systemic circulation. Thus, it is reasonable to say that in the fetus the RV is a systemic ventricle.
Extensive investigations in developmental anatomy show that the myocardial architecture is organized as myocardial muscle fibers running in geodesic curves around toroid (doughnut-shaped) nested layers. In addition, fibers penetrate from the epicardial to endocardial layers at oblique angles to the surface geodesics. The right and left ventricles differ in the arrangement of their respective nested tori; while the left ventricle (LV) follows a more regular arrangement, the RV torus is stretched and bent to bring the pulmonary orifice forward and left of the aortic orifice. The interface of the right and left tori creates the muscular septum. The entire heart can be envisioned as ‘a nested set of warped pretzels'7 and the seminal paper by Jouk et al.7 provides a detailed description.
Cardiac form develops to serve function8. The heart begins as a primitive tube, and the first contractions are seen at approximately 22 days in the human9. Next, the tube transitions to a looped heart. Soon afterward morphological differentiation of myocardium begins and chambers are formed. Finally, these stages are completed with the process of septation8. The geometry of the cardiac ventricles develops through the course of gestation; the myocardium undergoes a process of progressive compaction as coronary circulation develops and tissue can no longer be supplied by diffusion alone. As this ventricular architecture develops, so does the electrical excitation sequence. Activation follows blood flow, proceeding toward the outflow portion of the ventricle. As the trabeculae develop, they are activated first. Activation spreads from this interior tissue outward. In the mature trabeculated heart, activation proceeds apex-to-base10.
The right and left sides of the heart are in no way mirror images of each other. The atria are most accurately differentiated by the extensive distribution of the pectinate muscles in the right atrium, as opposed to their relative absence in the left atrium11; however, the constraints of ultrasound resolution mean that the atria are usually classified as right or left morphology, based on the pulmonary and systemic venous connections. The RV is trabeculated; in particular the ‘moderator band’, a prominently thickened trabeculation, is a sonographic landmark that identifies the RV. The RV is also shaped more like a banana, with the pulmonary and tricuspid valves at each end. The LV is shaped more conically, like a ‘ballerina's foot’, and is lined with much finer trabeculations, making it appear smooth-walled on ultrasonography12. The ventricles are also differentiated by their respective atrioventricular valves: the RV has the three-leaflet tricuspid valve, as opposed to the two-leaflet mitral valve of the LV, although the three leaflets may not be clearly distinguishable during early fetal echocardiography. Also, the septal leaflet of the tricuspid valve joins the interventricular septum more apically than does the mitral valve11, 13.
Despite these anatomical differences, Johnson et al.14 examined intracavitary pressures in 33 second- and third-trimester fetuses undergoing clinically indicated invasive obstetric procedures. The investigators found that the resulting waveforms were similar to those obtained in postnatal life. Both the LV and RV waveforms showed a small atrial component and rapid increase during systole, followed by a rapid decrease in diastole. Systolic and diastolic pressures increased linearly during gestation. There was no significant difference between left and right intraventricular pressures. The mean systolic and diastolic pressures measured approximately 20 mmHg and 5 mmHg at 20 weeks' gestation, respectively. Mean atrial pressure was approximately 3.4 mmHg in the left atrium and 3.6 mmHg in the right. No significant change was noted in intra-atrial pressure over the course of gestation. The authors concluded that changes in ventricular pressure mirror those seen in studies performed in fetal lambs and premature newborns of similar post-conceptional age at measurement. Such changes may correspond to those observed as resulting from the maturation of myocardial contractility in animal models14.
The orientation of the heart in the fetus differs from that in ex-utero life. The apex of the heart is displaced cranially by the relatively large liver, at least through the second trimester. This means that the long axis of the LV is more horizontal in the fetus than in the neonate15. It has also been shown that the geometry of the heart and great vessels changes during gestation. The angles between the ductal arch and fetal thoracic aorta, the ductal arch and aortic arch, and the left outflow tract and main pulmonary artery all change throughout gestation and this may have ramifications for the geometrical assumptions used in volume calculations16.
The fetal heart also reacts to damage differently from the adult heart, responding with myocyte proliferation in addition to hypertrophy17–19. Perhaps, for these reasons, deterioration in clinical measure of cardiac function may often be the first sign of fetal pathology20. Therefore, the development of sensitive methods of quantifying fetal cardiac function is of extreme importance, with a particular emphasis on the RV, as this is effectively the systemic ventricle of the fetus. With the incidence of congenital heart disease now seemingly higher than was previously thought, in particular in the neonatal intensive care unit population21, the importance of tracking function along a time axis becomes critical for the timing of fetal interventions and other management decisions.
Cardiac Function: Basic Principles
Although a detailed survey of all components of normal cardiac pump function is beyond the scope of this review, it may be timely to delineate briefly the main features. The two pumps, left and right, are each made up of an atrium which receives venous blood and a ventricle which ejects blood into an arterial system. Once the pressure in the ventricle has fallen beneath the pressure in the atrium, the atrioventricular valve opens and blood enters the ventricle, at first passively and later actively, because of atrial depolarization and contraction (the so-called ‘atrial kick’). The terms ‘active’ and ‘passive’ of course refer to the macroscopic appearance, although both processes are active at the molecular level.
The atrial contribution to CO becomes more significant as heart rate increases, owing to the shortening of the passive ventricular filling time that occurs as heart rate rises. The atrial kick is one mechanism to ensure efficient ventricular filling across a spectrum of heart rates. Similarly, the sympathetic nervous system, while responsible for the positive chronotropic effect, also causes a decrease in the action potential duration, as well as an increase in the rate of cardiac relaxation, thereby reducing the loss of passive ventricular filling time at higher heart rates. As the ventricle depolarizes and contracts, pressure rises steeply within the ventricle, causing the atrioventricular (AV) valve to close. Then follows the period of isovolumetric contraction within the ventricle, leading to an increase in pressure, until the pressure in the ventricle exceeds that in the aorta or pulmonary artery, causing the semilunar valve to open and blood to be ejected forcefully into the arterial circulation. Eventually the ventricular pressure recedes, as the force of contraction decreases. Once ventricular pressure falls beneath arterial pressure, the semilunar valve closes. The time interval between the semilunar valve closing and the AV valve opening is known as the isovolumetric relaxation phase. The atrium fills continuously throughout ventricular systole, causing a gradual increase in atrial pressure until this exceeds ventricular pressure, at which point the AV valve opens and the cycle begins again. Although there is some debate in the literature, for the purposes of this article, systole comprises isovolumetric contraction and ventricular ejection, while diastole comprises isovolumetric relaxation and ventricular filling22.
Determinants of stroke volume: preload, afterload and contractility
Stroke volume, the amount of blood ejected by the heart in a single beat, is principally determined by three factors: preload, afterload and contractility. The pressure within the ventricle at the end of diastole is referred to as the ventricular preload, as this is a major determinant of the ventricular volume and therefore of cardiac muscle fiber length. Starling's law of the heart states that, in the non-failing heart, the increased length of the muscle fibers results in increased energy of contraction. In other words, increased end-diastolic volume causes increased stroke volume. Afterload refers to the pressure against which the cardiac muscle fibers are shortening, and in fact is the limiting factor that determines the extent to which they are able to shorten. Thus, increased afterload results in reduced cardiac muscle shortening and, therefore, reduced stroke volume. Systemic blood pressure is usually taken as a surrogate marker of afterload. Finally, the contractility (ability to shorten) of the cardiac muscle itself is controlled by the sympathetic nervous system. Noradrenaline release causes increased contractility of the fibers, for any given preload, causing increased stroke volume.
Over the years many parameters have been proposed in an attempt to quantitatively evaluate cardiac function. Most were first developed for adult heart evaluation and were adapted to the fetus. Some are based on Doppler flow mapping, others on heart biometry or on timing of cardiac cycle events, or a combination of these three. They include stroke volume (velocity time integral × valve area), CO (stroke volume × heart rate) and ejection fraction (EF) (stroke volume ÷ end-diastolic volume). Others are the shortening fraction ((end-diastolic ventricular diameter − end-systolic ventricular diameter) ÷ end-diastolic ventricular diameter); myocardial ejection force ((1.055 × valve area × velocity time integral of acceleration) × peak systolic velocity ÷ acceleration time); and myocardial performance index (MPI) ((isovolumetric contraction time + isovolumetric relaxation time) ÷ ejection time). These formulae are summarized in Table 1. As in evaluation of pediatric and adult heart function, any cardiac biometry, and functional measures based on heart or vessel dimensions, will necessarily correlate with body size.
Table 1. Formulae for fetal cardiac functional evaluation
Aortic or pulmonary (see text for details). ET, ejection time; ICT, isovolumetric contraction time; IRT, isovolumetric relaxation time; VD, ventricular diameter.
(1.055 × valve area × velocity time integral of acceleration) × peak systolic velocity ÷ acceleration time
Myocardial performance index (MPI)
(ICT + IRT) ÷ ET
Stroke volume (SV) is a calculation of blood flow out of the heart at systole. However, measurement of the ventricular volume is cumbersome. In the left heart, therefore, SV is based on measurement of diameter of the aorta at the valve annulus to determine valve area, multiplied by flow across the annulus, represented by the time velocity integral. Right SV is calculated from the diameter of the pulmonary artery. Any inaccuracy in diameter measurement will introduce considerable error into the calculation, since this number is squared to obtain the valve area. Thus, SV is necessarily an indirect measure of the blood volume exiting the ventricle. CO is SV multiplied by fetal heart rate. Mielke et al.3 examined SV and CO in 222 fetuses from 13 weeks' gestation to term. They showed that SV increases exponentially as gestation progresses. Median biventricular CO ranged from 40 mL/min at 15 weeks up to 1470 mL/min at 40 weeks; the median CO per fetal weight was 425 mL × min−1 × kg−1 and the median right/left CO ratio was approximately 1.4, remaining stable throughout gestation and underscoring right heart dominance in the fetus3. Three-/four-dimensional ultrasound (3DUS/4DUS) techniques to evaluate fetal heart function often are used to determine fetal ventricular volume, to enable direct quantification of fetal SV and CO and this is discussed below.
M-mode echocardiography is the study of two-dimensional motion of all structures along an ultrasound beam over time. It was first described for assessment of cardiac function in 1971 in adults23, and normal values in the fetus were published in 198224. It allows for calculation of the shortening fraction, the change in ventricular diameter between end diastole and end systole as a ratio of the end-diastolic diameter, which is a long-standing surrogate for function25. Disadvantages include the difficulty in obtaining the correct view in a fetus, a line perpendicular to the interventricular septum at the level of the AV valve leaflets26, depending on fetal lie (Figure 1). However, it is used as a component of other techniques, most notably annular displacement (see below).
Annular excursion/displacement uses M-mode echocardiography to measure the maximal excursion of the junction between the tricuspid annulus and the RV free wall, from end diastole to end systole. It is a measure of RV function that has been shown in adults to have prognostic significance independent of LV function27. In the fetus, M-mode evaluation of annular displacement of both the mitral and tricuspid valves is a feasible technique28, and amplitude has been shown to increase with gestational age29. Annular displacement techniques essentially use M-mode to measure long-axis function as opposed to short-axis function, which is more commonly associated with M-mode measurements29. This technique is most suited to RV examination because of the longitudinal orientation of the deep RV muscle fibers, as opposed to the mainly circumferential arrangement of LV muscle fibers29, 30. This technique has the advantage of utilizing M-mode technology that is readily available with modern ultrasound machines. However, it has not yet been evaluated fully or compared with other methods of assessing function in the fetus.
Early/atrial (E/A) ratio (atrioventricular flow)
The E/A ratio refers to the ratio of the two peaks in flow velocity observed over the atrioventricular valves during diastole. The E-wave is the early, passive diastolic filling, which is dependent on ventricular wall relaxation. The A-wave is the active diastolic filling known as the ‘atrial kick’31. It is measured using pulsed-wave (PW) Doppler echocardiography, with the cursor set on or just below the AV valve (usually the mitral) in a four-chamber view. Ex utero, under normal conditions, the E-wave is greater than the A-wave. In the healthy fetus, the A-wave is usually greater, although as gestation progresses, the E/A ratio increases, approaching the ex utero values (Figure 2). It is generally agreed that increase in the E/A ratio is due to increasing E-wave velocity, while the A-wave remains fairly constant throughout gestation, although there is some dispute as to whether the velocity increases linearly throughout gestation32, 33 or only in the last trimester34. The increase in E-wave is thought to result from improved ventricular relaxation35. Since ventricular relaxation allows for coronary blood flow36, it would be expected that this too is increased throughout gestation; however, experimental evidence so far has shown that coronary blood flow remains constant throughout gestation37.
In adult life, reduction in the E/A ratio is a sign of diastolic dysfunction, associated with poor prognosis in patients with congestive heart failure38. In the fetus, a reduction in both mitral and tricuspid E/A ratio has been reported in recipient twins in twin–to–twin transfusion syndrome (TTTS), along with other markers of diastolic dysfunction39. However, other studies have shown an increase in E/A ratio in situations of cardiac compromise, including intrauterine growth restriction (IUGR) and hydrops due to congenital cystic adenomatoid malformation40, 41.
It has been shown that the E/A ratio shows poor correlation with isovolumetric relaxation time42, 43 as well as venous flow patterns44. However, the lack of an objective reference standard for measuring fetal cardiac function makes it difficult to compare one test with another, as each has its own methodological limitations. Qualitatively, monophasic AV flow patterns, with complete absence of the normal biphasic E/A morphology, indicates severe CO pathologies, such as aortic stenosis45 and TTTS46, and is a poor prognostic indicator in cases of IUGR47. This pattern should not be mistaken for fusion of E- and A-waves that occurs physiologically at high heart rates48. The E/A ratio has the advantage of allowing measurement of both sides of the heart independently, although it is a marker of diastolic function (Figure 2).
Myocardial performance index
The MPI or Tei index is the sum of the isovolumetric contraction and relaxation times, divided by the ejection time. It was first reported as a measure of global cardiac function in 199549. The index comprises both systolic and diastolic components, and can be used to analyze each ventricle independently. It has the advantage of not requiring a detailed anatomical survey in order to analyze function50. It is obtained by echocardiographic evaluation of the flow patterns through the AV valves and outflow tracts. The ejection time is measured as the duration of flow through the outflow tract, e.g. aortic valve. The isovolumetric contraction time is the interval between cessation of AV valve flow and the onset of outflow tract flow. The isovolumetric relaxation time is the interval between cessation of outflow tract flow and the onset of AV valve flow. The flow patterns are usually obtained with PW Doppler, but can also be obtained using M-mode and tissue Doppler imaging (TDI). As it utilizes only time intervals, it is independent of heart rate and ventricular structure 20, 51. Its use in fetal echocardiography was first reported in 1999, when Tsutsumi et al. showed that the Tei index can be used in fetuses, and that there is a decrease in the Tei index of both ventricles during gestation, with a transient increase immediately after birth52. Subsequently, it was suggested that the use of ‘clicks’ could help standardize the boundaries of the isovolumetric waveform durations. These clicks represent the Doppler echoes from the closure of the mitral and aortic valves, and provide a convenient objective standard for defining the boundaries of valvular flow (Figure 3). Thus, Hernandez-Andrade et al. demonstrated reduced inter- and intrauser variability with the incorporation of clicks into the calculation of the Tei index53. Using this ‘modified MPI’, they then demonstrated, contrary to Tsutsumi et al., that there is a slight increase overall in the LV Tei index from gestational week 19 onwards, with isovolumetric relaxation time increasing, ET decreasing and isovolumetric contraction time remaining constant54. These results are also contrary to those of van Splunder and Wladimiroff in 1996, who essentially measured the same variables, although not referring to the Tei index, and found that left ventricular ET and isovolumetric relaxation time both decrease with gestational age55, with no significant change in isovolumetric contraction time, admittedly with a much smaller number of subjects than were included in the study by Hernandez-Andrade et al. (52 vs. 557, respectively). Van Mieghem et al. showed the Tei index to correlate well with the EF in the fetus, with the advantage of less inter- and intraexaminer variability, thus validating the use of the Tei index in fetal echocardiography. Interestingly, they found no significant correlation between the E/A ratio and the Tei index, as well as no change in the Tei index with gestational age56.
The Tei index as measured in the fetus has advantages over its application in the adult heart. Friedman et al. showed that in the fetus one can measure the mitral and aortic valve flows simultaneously50, thereby removing the inaccuracy involved in measuring the time intervals across different heart beats. However, the right-sided valves, due to their different anatomical configuration, cannot be captured simultaneously. Perhaps for this reason the right side is less frequently included in MPI research. However, as has been mentioned, in adult cardiology, where the MPI was first utilized, neither side can be assessed simultaneously using flow Doppler techniques51. One technique to minimize the inaccuracy induced by measuring time intervals across different heart beats is taking the average of the MPI as obtained across several heart beats52. The application of tissue Doppler techniques to the MPI has been shown to enable simultaneous appraisal of inflow and outflow, both in the fetus57 and adult58. However, there is not yet a consensus that TDI and pulsed-wave Doppler evaluation of MPI give the same results59.
The MPI has been widely studied in a number of fetal pathologies. In TTTS the MPI has been shown to be pathological in the recipient twin, due to a prolongation of the isovolumetric relaxation time, implying diastolic dysfunction60. This is in contrast to results in fetuses suspected of suffering from fetal inflammatory response syndrome due to infection secondary to premature rupture of membranes. In the latter case the MPI is also increased, implying reduced function, but the increase is due to a shortening of the ejection time, the denominator of the Tei index formula61. In fetuses with homozygous α-thalassemia (Hb Bart's fetal edema), a cause of fetal demise with progressive cardiac dysfunction, the MPI is elevated as early as week 20, long before ventricular and atrial enlargement occur57. The Tei index has also been shown to be elevated in hydrops fetalis, as well as in large-for-gestational age fetuses of diabetic mothers62, although the clinical significance of the latter finding is not clear. Crispi et al. demonstrated an increase in MPI with increasing severity of IUGR, with both diastolic and systolic components affected, as well as an association between increased MPI and fetal death41.
Thus, it has been shown that a raised MPI value is a sensitive, albeit non-specific, marker of fetal cardiac dysfunction. However, the Tei index has been shown to have some limitations. Firstly, competence in the technique itself seems to be somewhat challenging to acquire, as demonstrated by Cruz-Martinez et al.63. It takes on average 65 exams before the inexperienced practitioner can achieve competence in this skill. This limits the applicability of the Tei index to high-volume referral centers, in which practitioners will be able to develop and maintain competence.
There have also been certain clinical contexts in which the Tei index has been shown to be unreliable. In a study of adults with aortic stenosis and reduced LV function, Sud and Massel showed that the Tei index remains unchanged in severe aortic stenosis despite worsening EF, and that the index paradoxically decreases as the aortic stenosis gets more severe in patients with reduced EF64. Using ROC curves, the index was shown to be unable to identify accurately patients with reduced LV function in the presence of severe aortic stenosis, as well as patients with severe aortic stenosis in the presence of reduced LV function. Another concern has been raised regarding the use of the Tei index in cases of pulmonary hypertension. The index has been reported to correlate with pulmonary hypertension in both adults65 and children66. Translating these concepts to fetal medicine, the paradigm of increased fetal RV afterload is ductal constriction. Mori et al. found the Tei index to be increased in cases of ductal constriction, and suggest that it can be used as a sensitive marker of RV dysfunction in this condition48. However, concerns have been raised as to the legitimacy of using time intervals to assess function in pulmonary hypertension, since time intervals are themselves related to RV afterload. Thus, given an abnormal Tei index in cases of elevated pulmonary artery pressure or ductal constriction, it may be that the index is caused by the elevated afterload, and not directly by RV myocardial dysfunction67. Similarly, Cheung et al. found that the Tei index was not sensitive to dobutamine infusion in anesthetized pigs, but was sensitive to changes in preload and afterload68. This is in contrast to the results published by Eidem et al. which showed no effect of loading conditions on the Tei index in adults and children with congenital heart disease before and after surgery69. As such, it has been suggested that the Tei index be viewed as a ‘sedimentation rate of the heart’, an indication of pathology, with perhaps limited ability to reflect causality69. However, the reproducibility56 of the MPI, as well as its sensitivity, make it, in the opinion of the authors, an important tool in fetal echocardiography, including the timing of interventions.
Three- and four-dimensional ultrasound
3DUS/4DUS technologies have been used over the last decade to evaluate fetal cardiac function, chiefly with the aid of spatiotemporal image correlation (STIC)70. STIC is based on a sweep of the fetal heart comprising the five transverse planes approach71, and delivers a volume dataset containing a complete reconstructed cardiac cycle, made up of approximately 1500 images. The operator can navigate both spatially and temporally within the saved dataset. Thus, for example, the four-chamber view at end diastole and end systole can be identified by valve movement, and from these starting points either manual or semi-automated volume measurement of the cardiac ventricles is possible. The studies of 3DUS/4DUS applied to fetal heart function evaluation are based primarily on cardiac ventricular volumetry and cardiac valve clicks. The goal of all these methods has been extrapolation of fetal stroke volume, EF and CO from the resulting ventricular volumes72–76 (Figures 4 and 5).
Manual volumetry based on STIC volumes obtained by segmentation was studied by Uittenbogaard and colleagues75. The ventricles are manually traced in multiple serial slices 1 mm apart; these were obtained by scrolling through the saved volume in multiplanar reconstruction, and Simpson's rule was applied75. Alternatively, semiautomated segmentation involves specialized algorithms applied to the STIC volume72–74, 76–78. Virtual Organ Computer-aided AnaLysis (VOCAL), which measures the volume of a defined area by reconstructing planes around a fixed central axis, is initiated and the volume of the organ of interest is determined (Figure 4). VOCAL may be combined with inversion mode (IM)73, which isolates fluid-filled areas (black) from tissue (gray) and inverts their representation (Figure 5).
We found that ventricular volumetry successfully differentiated between normal and anomalous hearts. For example, Figure 6 shows a case of pulmonary stenosis at 32 weeks' gestation, with RV volume significantly decreased from the mean (RV end-diastolic volume measured 0.7 (mean, 2.71) cm3 and RV end-systolic volume measured 0.37 (mean, 1.34) cm3, both below the 5% CI for the mean for gestational age)73. We extended the combination of VOCAL and IM used for ventricular volumetry73 to calculate ventricular mass. Applying the algorithm we were able to deduct the intraventricular volume from the total ventricular volume automatically, the remainder being the volume of the myocardium. This was multiplied by estimated fetal cardiac density (1.050 g/cm3)79 to obtain the mass80. While these methodologies show promise for clinical application, they are still in their infancy, requiring a long learning curve and considerable operator expertise. If a viable automated program for ventricular volumetry and its related measures are introduced, making the techniques less operator-dependent, they may provide additional alternatives for fetal cardiac functional evaluation.
Tissue Doppler imaging, strain and strain rate
TDI refers to the application of Doppler principles to the measurement of the velocity of the myocardium rather than that of intracardiac blood flow (Figures 7 and 8). Since the cardiac apex remains relatively stationary throughout the cardiac cycle, analysis of the motion of the mitral valve annulus relative to the apex gives a good approximation of the longitudinal contractility of the ventricle81. Pulsed-wave tissue Doppler examination of the mitral annulus longitudinal motion gives three waveforms: S′, the velocity of the systolic downwards motion of the annulus towards the apex—a positive deflection waveform; E′, the velocity of the early diastolic movement away from the apex—a negative deflection waveform; A′, the velocity of the movement of the annulus associated with atrial contraction—a negative deflection waveform. The prime (′) notation is used to differentiate from the E and A waveforms of mitral Doppler inflow velocities; however, some researchers prefer the nomenclature Sa, Ea and Aa, and others use Sm, Em and Am. TDI measures the peak velocity of the myocardial segment being interrogated, unlike color TDI (see below) which measures the mean velocity82.
Broadly speaking, S′ corresponds with LV systolic function, and has been shown to correlate with EF as measured by 3DUS83. Changes in the S′ waveform have been demonstrated as soon as 15 seconds after the onset of ischemia in experimental animal models84. E′ corresponds with diastolic function, and has been shown to be less preload-dependent than the E/A profile85. It can be combined with the mitral inflow, as the E/E′ ratio, which is an even more sensitive measure of diastolic dysfunction86. The A′ waveform has been shown to be more sensitive than the AV valve inflow profile in detecting atrial mechanical dysfunction87.
TDI of fetal myocardium was first reported as a feasible technique in 199988. Since then there have been conflicting reports as to its usefulness in the assessment of fetal heart function. RV TDI alone could not differentiate between fetuses with and without heart failure, although incorporating TDI into other techniques – the E/E′ ratio, and use of TDI to measure the Tei index, did differentiate between the groups89. A more recent study showed a significantly reduced LV-S′, as well as an increased E/E′ ratio and RV-S′/LV-S′ in a group of fetuses with hydrops fetalis, as compared to normal controls90. Similarly, it has been reported recently that TDI is more sensitive than ‘conventional’ AV flow and MPI measurements in detecting systolic and diastolic dysfunction in IUGR fetuses91. It is unclear if there is any clinical significance in the increased sensitivity of TDI in IUGR, e.g. if it marks out a subgroup of fetuses that have a poorer prognosis. Alternatively, TDI may be just a more sensitive technique, picking out the lower end of ‘normal’, with no functional or clinical significance.
The main disadvantages of PW-TDI are that it can provide information about only one area of the myocardium at any one time48 as well as being very angle-dependent, i.e. only those areas of the myocardium that are parallel to the angle of insonation can be analyzed92. The application of color Doppler to TDI enables the assessment of strain rate (change in length per unit time), and, by mathematical derivation, myocardial strain (change in length) itself93. These modalities have the advantage of directly measuring myocardial segments, as opposed to chamber-dimension changes, and thus should reflect myocardial contractility more accurately20, 29, 40. However, they suffer from the same drawbacks as PW-TDI, namely angle dependency and assessment of individual segments rather than global function.
A relatively recent approach to studying myocardial motion as a surrogate for cardiac function is the use of speckle-tracking techniques. These use 2D B-mode echocardiography, and are based on identifying ‘speckles’. Speckles are natural acoustic markers, spread randomly throughout the myocardium, which are generated by stable interference and backscatter of the ultrasound signal80, 94. These speckles are identified, and their positions are noted in subsequent frames in a cineloop. With the frame-rate a known quantity, the velocity vectors for each speckle can be calculated and, thereby, the strain and strain rate can be evaluated segmentally as well as for the whole chamber. Speckle tracking is usually coupled with an automated border recognition program, so that speckle tracking occurs within the context of the ventricle under investigation. This combination of software allows for estimation of the EF as well as direct measurement of strain and strain rate. Speckle tracking essentially measures myocardial deformation (change of shape) as opposed to the point changes in velocities measured by TDI95.
Speckle tracking is limited, however, to speckles which remain within the imaging plane throughout the cardiac cycle. Speckles which pass through the plane of insonation cannot be tracked, at least not by the majority of current systems96. However, a recent study by Matsui et al. has shown that speckle tracking, which requires offline processing with dedicated software, is no better than is M-mode for measuring annular displacement techniques, which is readily performed on any modern ultrasound machine97.
Perk et al. compared EF estimation by speckle tracking, by applying the standard ‘Simpson’s rule’ (measuring end-diastolic and end-systolic volumes manually) and by visual examination by an experienced echocardiographer. They found strong correlation among all three methods94. This too begs the question as to the usefulness of a technique which is equivalent to, but less accessible than, older technology. The authors do not provide data about the total time required to perform and process the speckle-tracking studies as opposed to the M-mode assessment of EF. Matsui et al.97 also noted that, conceptually, there is a problem in applying color Doppler techniques to the fetus, where the ratio of pixel to myocardial volume is much higher than in the adult. This is presumably true of speckle-tracking techniques as well. Also, the technique is dependent on frame-rate of the ultrasound machine, and it has been suggested that the frame-rates in most current machines are too low to allow accurate speckle tracking in the fetus. This is in part owing to the lack of ECG gating in fetal echocardiography. The authors demonstrated that interposing a metronome to artificially generate simulated ECG spikes, thereby enabling the images to be stored at a higher frame rate, caused an increase in successful speckle-tracking acquisition97. As has been mentioned, speckle tracking, as well as Doppler-derived measures of strain, have the advantage of providing both segmental and global functional information concerning both the left and right heart in systole. Speckle tracking has not yet been studied sufficiently for use in evaluation of diastolic function, and the use of TDI for diastolic evaluation is limited to certain subgroups in adults98.
Magnetic resonance imaging
Since the advent of magnetic resonance imaging (MRI) in the 1980 s as a research, and subsequently clinical, tool it rapidly became an important adjunct for assessment of cardiac structure and function (both systolic and diastolic) ex-utero99. Fetal circulatory physiology and many congenital lesions make accurate depiction of RV structure and function critically important, and MRI is currently considered the reference standard for ex-utero RV assessment100, 101. MRI, both in utero and ex utero, enables measurement and calculations of ventricular volumes and mass, as well as EF and CO/cardiac index. Unlike ultrasonographic techniques, MRI is not affected by maternal obesity or oligohydramnios102, and image quality is not dependent on gestational age103. Since it does not rely on assumptions, but rather on true real-time measurements, it is useful for the examination of abnormal hearts that do not conform to the geometric models used in ultrasound techniques104. Other advantages include better image quality and structural detail105. Technical disadvantages include the expense of the technique, the relatively long duration of the examination (although this is reported to be as short as 15 minutes in some studies101) and the lack of availability of both the technology and expertise to perform the examination. Some centers advocate using a sedative premedication to reduce fetal movements; however, as technology improves and study times shorten, this will no longer be required106.
Historically, fetal MRI techniques have been hampered by the lack of ECG-triggering, which is typically used in ex-utero cardiac MRI. However, experimental techniques have been developed in the chick embryo which could bypass this requirement107. Another concern is the problem of temporal resolution, due to the time required to acquire images, in the context of rapid fetal heart rate; however, there are feasibility studies showing that modern MRI sequences are able to acquire fetal cardiac MR94, 104.
Venous flow assessment
Analysis of the flow (by PW Doppler) within venous channels contiguous with the RA (ductus venosus, inferior vena cava, hepatic veins and pulmonary veins (DV, IVC, HV and PV, respectively), excluding the umbilical vein (UV) which is non-pulsatile from the end of the first trimester108), gives a good approximation of the pressure gradients within the atrium itself (Figure 9). The major veins all exhibit a pulsatile flow waveform, representing changes in pressure during the cardiac cycle, with forward venous flow facilitated by low atrial pressures. Thus, at those points within the cycle where atrial pressure is lowest, forward venous flow will be maximal, and where atrial pressure is highest, venous flow will be minimal or even reversed. The normal waveform is the S-wave (maximal forward flow corresponds to ventricular systole, with rapid descent of the closed AV valves causing a drop in atrial pressure), v-descent (ventricular relaxation with rising AV valves, causing a temporary increase in atrial pressure), D-wave (early ventricular diastole, with blood rushing forward into the ventricles, causing a drop in atrial pressure) and a-wave (atrial systole, or atrial kick with pressure in atrium rising steeply)109.
The most significant change in venous Doppler with cardiac dysfunction is reversal or absence of the a-wave, which portends serious consequences in cardiac pump function, with a subsequent daily risk of worsening fetal wellbeing and intrauterine death109 (Figure 10). Reversal of the a-wave in the DV in fetuses aged 11–14 weeks has been shown to be associated with a 25% chance of congenital heart defects110, and in high-risk fetuses aged 26–34 weeks, absent or reversed a-wave was associated with a 63% risk of fetal or neonatal death44. This finding is also indicative of Stage III in the Quintero staging of TTTS111. Another venous waveform with prognostic significance is pulsatile flow in the umbilical vein, which has been shown to correlate with the presence of myocardial dysfunction112. Although the UV is non-pulsatile under normal conditions, the presence of pulsatile flow in the UV beyond the first trimester is yet another sign of cardiac dysfunction. It is a marker of progressive placental dysfunction113, and has been shown to correlate with elevated troponin levels in the neonate114.
Various indices of venous flow profile have been devised. One of these, the pulsatility index for veins, is the peak systolic velocity minus the peak diastolic velocity, divided by the time-averaged maximum velocity. In fetuses with IUGR, a raised ductus venosus pulsatility index for veins (DV-PIV) is indicative of a ten-fold acceleration of deterioration113. In a study examining neonates with elevated levels of troponin or N-terminal pro-atrial natriuretic peptide (NT-pro-ANP), markers of myocardial damage and dysfunction, respectively, elevated PIV in the fetal DV, left hepatic vein and IVC was shown to correlate with elevated umbilical artery NT-pro-ANP in specimens drawn immediately after delivery. The PIV was also highest in the subgroup of neonates with elevated levels of both troponin and NT-pro-ANP115. A more recent study by the same group showed that N-terminal pro-brain natriuretic peptide (NT-pro-BNP), another marker of cardiac dysfunction, was also correlated with elevated PIV116.
Another way of examining cardiac function, as expressed in the venous system, is by analysis of the vessel pressure waveform. Mori et al. have shown that one can measure the changes in vessel diameter, providing a waveform that is equivalent to the central venous pressure waveform, with ‘A’ and ‘V’ peaks, and ‘X’ and ‘Y’ troughs117. Elements of the morphology of the waveform, in particular shortening of the A-X-V time and reduction in the X nadir, can be indicative of fetal cardiac dysfunction118.
The utility of fetal venous Doppler examination lies in its predictive powers. Pathological changes in the venous Doppler results in growth-restricted fetuses precede changes in the cardiotocogram and biophysical profile, in some cases by a period of weeks119, 120. Although there remains controversy among obstetricians regarding the benefits of early vs. delayed delivery in IUGR, Doppler studies (including arterial) are reported to be the most accurate non-invasive modality for assessing placental function, and therefore provide the information on the basis of which these decisions can be taken121.
The focus of ultrasound scanning is shifting from the purely descriptive towards a functional, quantitative modality122. This results from both technological advance in ultrasound machines as well as in the progress of dedicated examiners. The techniques described above are all designed to establish markers of fetal cardiac dysfunction, in the absence of an accepted reference standard. Essentially, they all suffer from the shortcoming that, while they may strongly imply the presence of cardiac dysfunction, they do not necessarily pinpoint the etiology or causal mechanism. These modalities are perhaps most appropriate when there is a known pathology that is being followed. Many of the modalities have been tested extensively on the LV, but less so on the right. Conversely, some, such as short-axis shortening fraction, are used for analysis of the RV, even though anatomically and geometrically they are less applicable to that ventricle.
The ideal test of fetal cardiac function should be applicable mainly to the RV which, as mentioned above, is the dominant ventricle in the fetal cardiovascular system; it should be accessible using standard ultrasound machines, without relying on offline post-processing; and it should be able to predict cardiac dysfunction before there are clinical signs of fetal distress. It would seem to the authors that simple modalities such as M-mode annular displacement, precordial venous Doppler flow assessment and MPI are the only ones to have truly crossed the translational divide between the experimental and clinical, and that can be recommended for clinical practice.