Using speckle‐tracking echocardiography to assess fetal myocardial deformation: are we there yet?

Fetal echocardiography is now a fundamental part of antenatal care. In many countries, including the UK, ultrasound views of the fetal heart are part of the universal fetal anomaly screening program, with a more detailed fetal echocardiogram performed should any abnormalities be suspected. When initially introduced, fetal echocardiography focused on identifying congenital heart defects (CHDs). In recent years, however, attention has shifted towards the use of ultrasound in the assessment of fetal myocardial function1. This is clearly of value in the assessment of primary myocardial disease in the fetus, but in addition, given the central role of the heart in fetal adaptive response to several pathological processes, it may be of value in the management and prognosis of other non-cardiac disease states, such as twin-to-twin transfusion syndrome (TTTS)2. Subjective assessment of myocardial contractility has been the most commonly used method of assessing cardiac function in fetal life, but there has been steady development of methods to quantify this accurately. The aim of these methods is to reliably identify early changes, before subjective assessment alone can determine deterioration of function or the development of hydrops, and to monitor reliably changes in cardiac function over time. Many different echocardiographic modalities can be used to assess fetal cardiac function, including M-mode, pulsed-Doppler and tissue-Doppler techniques. Recently, speckle tracking (two-dimensional (2D) strain) has emerged as a further modality for assessing fetal cardiac function. All of these techniques are widely used in the assessment of cardiac function in children and adults, but the utility and limitation of their application to the fetal heart are less well understood. In this Opinion, we present techniques used to assess fetal cardiac function, with an emphasis on myocardial deformation measured by speckle-tracking echocardiography (STE). Functional assessment of the fetal myocardium is difficult, because the fetal heart is a small structure, within a small, highly mobile fetus. Fetal position is unpredictable, which can make image acquisition difficult, and this can be compounded by maternal body habitus. A further complication is that the fetus is both growing and maturing, meaning that normal reference ranges of myocardial deformation may need to be adjusted for gestational age or size. Given the extremely poor outcome of fetuses with evident heart failure (defined by the development of fetal hydrops), to be of any clinical use, fetal heart assessment must be sensitive enough to identify myocardial dysfunction in the preclinical state, before the development of such overt signs2-4. Further challenges in the assessment of fetal cardiac function arise from differences in the way the fetal myocardium and postnatal myocardium respond to hemodynamic changes. Fetal cardiomyocytes complete terminal differentiation only close to term. This means that the fetal heart can respond to changing pressure and volume conditions by proliferation, as well as hypertrophy. These responses are less well understood than in postnatal life, adding to the complexity of assessing myocardial function before birth. Several techniques for the measurement of myocardial function have been adapted from adult and pediatric cardiology for use in the fetus (Table 1). These include M-mode echocardiography and Doppler-based techniques (measuring either blood flow or tissue movement). M-mode echocardiography measures one-dimensional myocardial movement over time. It was the first ultrasound method used to assess the adult heart, and was introduced to fetal cardiology in the early 1980s5. By measuring the difference in end-systolic and end-diastolic ventricular diameter, the fractional shortening can be calculated6. In the longitudinal plane, the excursion of the mitral and tricuspid valve annuli can be measured, allowing an assessment of left-ventricular (LV) and right-ventricular (RV) longitudinal function7. M-mode echocardiography is simple and is readily available in most settings. However, there are several limitations to this technique. It is highly angle-dependent and hence reliant on suitable fetal lie4, 6. Calculation of ejection fraction, as opposed to fractional shortening, makes geometric assumptions about the heart that may not be applicable in smaller hearts, especially in CHD4. In fetal life, a fall in ejection fraction may appear only as a late manifestation of cardiac failure, reducing its clinical utility2, 6. Doppler techniques can be used to assess fetal cardiac function by measuring blood-flow velocity. When measured in combination with the semilunar valve annulus areas, outflow tract velocities can be used to estimate stroke volume and hence cardiac output8. Diastolic function can be assessed by examining atrioventricular valve inflow patterns4, 8, but this is made more complex by the potential for either atrium to ‘decompress’ through the foramen ovale. In addition to blood-flow velocity, Doppler echocardiography can be used to measure time intervals, especially the isovolumetric contraction and relaxation times2, 8. These intervals can be considered separately, or combined into formulas such as the myocardial performance index (Tei index), to give a global assessment of heart function8. The Tei index is computed as the sum of the isovolumetric contraction and relaxation times divided by the ejection time. It is one of the most widely used measures of fetal cardiac function, and is considered to be a measure of global myocardial performance because it includes both systolic and diastolic time intervals. It has been studied extensively in both healthy fetuses and in those with several disease states9. As well as blood flow, Doppler techniques can be used to measure the velocity of myocardial tissue4, 10. Measuring the longitudinal velocity of the mitral or tricuspid annulus gives information about both systolic and diastolic global function10. However, these techniques are limited by uncertainty regarding gestation-related normal values11. In addition, the measurement of accurate Doppler velocities is also highly angle-dependent, and so it may also be limited by fetal position. Deformation imaging aims to measure directly the lengthening and shortening of the myocardium throughout the cardiac cycle12. The amount of change in length of a particular segment of cardiac muscle between its end-diastolic and end-systolic length is known as strain12-14, which is a dimensionless quantity expressed as a percentage. Langrangian strain is defined as the instantaneous difference between the length of an object at a given moment and its original length12, 15. An alternative method of calculation, known as natural strain, expresses the difference in length at a given instant relative to the length at a previous timepoint, rather than the original length15. Langrangian strain and natural strain are similar at low strains, but become significantly different at larger values, and have a fixed but non-linear relationship15. Recent recommendations from the joint European and North American taskforce have stated that Langrangian strain is more suitable than natural strain in the context of STE16. Strain rate (SR) is the temporal derivative of strain, defined as the change in strain per unit of time13, 17. By convention, negative strain values describe myocardial fiber shortening, and positive values describe lengthening18. The orientation of myocardial fibers in the ventricles is complex, and strain occurs in three dimensions13, 19. To attempt a full assessment of this, strain can be measured as longitudinal (motion from base to apex), radial (motion perpendicular to the epicardium) and circumferential (change in short-axis radius)13. In addition to these three strain directions, there are also six shear strains that act in a parallel direction to one border of the myocardial region of interest, causing a skewing of the segment volume13. These shear strains are not currently measured in clinical practice. Color-encoded tissue Doppler can be used to derive strain and SR. By measuring the instantaneous tissue velocities at different points along a cardiac segment, a velocity gradient can be calculated18. From this, the SR can be estimated and then strain can be calculated as the time integral of SR18. This technique is severely limited by poor repeatability and angle-dependence, a particular restriction in fetal cardiology. It is also influenced by fetal and maternal movement, and cardiac translational movement4. Because of these limitations, color-encoded tissue Doppler has been largely superseded by alternative techniques, outlined below. STE is a relatively new echocardiographic technique, first described in adult subjects in 200420. It is a semi-automated process, performed offline on previously acquired 2D B-mode echocardiographic images, using proprietary software. When ultrasound images are acquired, irregular surfaces, such as the myocardium, give rise to a characteristic speckled pattern21-24. Although this pattern has a random appearance, it is characteristic of each particular area23. This pattern of speckles within a defined region, known as a kernel, can be conceptualized as an ‘acoustic fingerprint’. Through statistical correlation, the movement of the kernel can be tracked from one frame to another13, 21. The spatial movement of multiple kernels relative to each other can be used to measure strain, and then SR, using proprietary mathematical algorithms (Figure 1). By combining apical and short-axis echocardiographic views, longitudinal, radial and circumferential strain can be measured by STE, although longitudinal measurements (often the first dimension of systolic function to deteriorate) are the most widely used in clinical practice25. The postprocessing software can automatically divide the myocardium into equal segments, giving a quantification of regional strain (although this may be challenging in fetal practice)25. In postnatal practice, by combining the two-, three- and four-chamber apical views, the entire left ventricle can be assessed and visualized in the form of a 17-segment bullseye plot. These regional strain values can then be combined to give a value for global longitudinal strain25. STE can also be used to track rotational components of ventricular systole and diastole26. In the normal adult heart, when viewed from the apex, the base of the heart rotates clockwise and the apical portion anticlockwise. This causes a ‘twisting’ of the left ventricle, as the base and apex have different rotations. If LV twist is expressed as a function of the length of the ventricle, the LV torsion is derived. Estimation of these variables may be achieved by short-axis views of the ventricle at the base or apex, or by the use of three-dimensional (3D) echocardiography to acquire the full volume of the ventricular myocardium. STE is now an accepted modality in pediatric and adult echocardiography, with several clinical applications. It has been shown that changes in myocardial strain and SR precede changes in ejection fraction in patients with hypertrophic cardiomyopathy, and that STE can also be useful in distinguishing hypertrophic cardiomyopathy from other forms of ventricular hypertrophy27, 28. The quantification of regional strain is useful in coronary artery disease, to identify ischemic and infarcted myocardial segments21. It is also useful when assessing cardiac dyssynchrony and the response to resynchronization therapy29. STE has several advantages over alternative methods of assessing fetal cardiac function. It is considered less angle-dependent than conventional methods (although this is still under debate, as outlined below), allowing more flexibility in image acquisition than Doppler-based techniques30, 31. It is also less affected by fetal and maternal movement. The accuracy of STE has been validated in postnatal life in animal models by comparing it with more invasive experimental methods such as sonomicrometry and magnetic resonance imaging tagging, but such validation experiments have not been performed in fetal life32. It has been shown that the calculation of strain and SR from routinely acquired B-mode fetal images is feasible, and several studies have confirmed good inter- and intraobserver reproducibility, partially as a consequence of the semiautomated nature of the technique1, 13, 33-36. Some investigators have reported a lower interobserver variability at 24 weeks' gestation than at 20 weeks, but overall, the reliability of fetal STE appears to be good, and equal to or better than that of Doppler-based methods1, 34. Evaluation of the rotational mechanics of the left ventricle, including twist and torsion, has also been attempted in fetal life and found to be feasible (although technically challenging)26. The rotational patterns of the heart are different during fetal life, and STE is beginning to provide novel insights into myocardial mechanics during early human development. In addition to myocardial deformation, STE can also be used as an alternative (and potentially more reproducible) method of measuring conventional myocardial function parameters such as fractional shortening37. Fetal STE has been used as a research tool in several clinical scenarios. However, as yet, there are few data demonstrating an increased clinical benefit of using STE over and above the standard measures of myocardial function already in widespread use. STE has been used in pediatric patients with CHD, and has been found to have some clinical utility38. Its application to the assessment of CHD in fetuses has also been described and it has been shown to be feasible31, 39-42. Although all the studies published thus far have been small in size, this raises the possibility that STE may be of use in the future as part of the assessment of the fetus with heart disease. Published studies vary in terms of case selection, making comparisons difficult, and the results are varied. One study including 12 fetuses with aortic coarctation (confirmed postnatally) found that global LV longitudinal strain and SR in those fetuses were significantly lower than in a control group40. Another study found no effect of coarctation on LV strain, although it included just two cases of fetal coarctation39. The same study also included fetuses with several other congenital heart lesions and found that fetuses with hypoplastic left heart syndrome had severely reduced LV strain, while those with Ebstein's malformation of the tricuspid valve had severely reduced RV strain39. A further report found that, when a heterogeneous group of fetuses with left-heart obstruction were analyzed as a group, there was no difference in LV strain compared with controls41. This contradicts the findings of another study examining the effect of fetal intervention in pregnancies with evolving hypoplastic left heart syndrome, in which a group of 57 fetuses with severe aortic stenosis were found to have significantly reduced longitudinal and circumferential strain. This remained low after fetal balloon valvuloplasty, but interestingly, there was a significant difference between the group that ultimately had a biventricular repair and the univentricular group42. These findings suggest that myocardial functional changes can be identified early and that STE may have a role in decision-making for fetuses with CHD in the future. The effect of non-cardiac disease on fetal cardiac function has been studied extensively, but the use of STE to assess this has had only limited attention thus far2. TTTS has been investigated in two studies utilizing STE43, 44. Previous methods have shown significant changes in cardiac function in TTTS fetal pairs, especially in the recipient twin, presumably as a result of pressure and volume overload2. This has been replicated in the STE studies. One study, focusing on recipient fetuses only, found significant decreases in RV strain compared with controls44. Another study of both donor and recipient fetuses found that, in the donor twin, LV strain was increased while RV strain was decreased, while in recipients, the strain of both ventricles was reduced43. As abnormalities in fetal cardiac function have been shown to be sensitive early predictors of outcome in non-cardiac fetal disease, such as fetal growth restriction, this is an area in which STE is likely to be of considerable interest in future research2. The impact of maternal disease on fetal cardiac function is another area in which the use of STE is at an experimental stage. Maternal diabetes is known to result in fetal cardiac hypertrophy, with evidence of diastolic dysfunction based on Doppler assessment45, 46. One study used STE to assess the effect of maternal diabetes on fetal cardiac function, and documented significant decreases in both global longitudinal and circumferential strain compared with controls47, 48. Decreases in longitudinal strain were also seen in the fetuses of obese, non-diabetic women47. Maternal cholestasis has also been shown to have an adverse effect on LV longitudinal strain and SR49. As well as systolic cardiac function, STE can also be used to assess diastolic function. This is of relevance in fetal cardiology because diastolic dysfunction precedes impairment in systolic 34. values of fetal diastolic SR throughout gestation have been STE has been used to diastolic function, for in addition to the systolic impairment seen in fetuses of 48. in the use of STE in fetal several that currently its As with any technique, it is to that and identify which further study before the technique can use. are regarding changes in fetal longitudinal strain and SR with Some examining fetuses from 20 weeks' gestation have found that strain from that have found a significant in strain as gestation (Figure when assessing SR, some studies found no change found a significant with In a research each disease to be studied a control group, as there is no as to These data regarding normal reference ranges and changes in strain and SR during gestation make it to This is one area in which the proprietary nature of the STE algorithms may be a tracking is performed using each of which different algorithms to track speckle patterns and determine myocardial from different may deformation values that are significantly different from each and the should be of this when a of the European of the American of and the is to and and this is an An of this taskforce is that global longitudinal strain should be calculated along the entire myocardial rather than be a of values that have been as opposed to tissue STE is an measure of deformation, as is in STE research This is because kernels are tracked in direction of the of this may be to one must the nature of ultrasound B-mode and are not and this may a of studies have investigated this in One study in children found that STE had a significant of when measuring longitudinal This is an for those use STE in the fetal cardiology as the of to use image is no studies have the effect of of on deformation in fetal life, an area that further based on rate is another that is challenging in fetal A relatively frame rate compared with standard B-mode imaging is to that the speckles can be tracked from one frame to the In adult STE, a frame rate of to per is As the fetal heart rate is far than that of an a frame rate may be However, the frame rate to provide both temporal and spatial has not been A recent study that a frame rate of or was used had no significant impact on fetal longitudinal strain however, have shown a significant in LV longitudinal strain when low were compared with of One of which of STE need to be is that the frame rate of B-mode used for offline STE is far lower than the original frame In the of an in fetal data using the and in occurs at a frame rate of may also be a for the use of STE in on the method the spatial is between and this may not be a in adult echocardiography, the extremely small of the fetal heart may into fetal STE should the software attempt to track kernels that are than this This may the use of STE to fetal radial and circumferential function, given that the ventricular of the fetus be less than When the combination of heart rate and small heart in the fetus, may be about the of If we image a left ventricle in a fetus with a gestational age of 20 weeks, then it is that this throughout the cardiac by only If we a fetal heart rate of per and a frame rate of then each cardiac to If systole is to of a cardiac then this means that the shortening over or per This value be lower we attempt to measure strain in cardiac segments, or strain, and at the present time it is not this is the of An in fetal cardiac is the use of of relatively low compared with those used in pediatric with a of These further lower the spatial and temporal the in image in STE based on 2D B-mode imaging also has the limitation that kernels may of the imaging plane, meaning that are no this be by the use of This has only been introduced in adult and pediatric practice, and there are only limited of its use in fetal Although this speckle tracking in three low frame rate is likely to be a the Although not a part of clinical in fetal STE has the potential to of fetal myocardial function. It has been shown to be feasible and in fetal echocardiography and has several advantages over other methods of functional STE can use in antenatal further is to normal values of strain and SR throughout identify and quantify differences in the of frame rate and angle-dependence, and further the of fetal and maternal cardiac and non-cardiac disease on myocardial