Background
Echocardiography is the most frequently used modality for imaging and serial follow-up of the failing heart. As we have shown, standard measurements of ventricular thickness and function that are obtained using two-dimensional and M-mode echocardiography remain the mainstay of echocardiographic imaging. Although investigators have had the opportunity to critically study these modalities over the past several decades, the critical evaluation of the prognostic value and reproducibility of these modalities and measurements in children appears to have started only recently.Reference Parthiban and Shirali 1 – Reference Margossian, Chen and Sleeper 5
Over the past decade, advances in the speed of computerised processing and in the miniaturisation of electronics have led to an increasing momentum among the industry and investigators in search for more reliable, sensitive, and automated measurements of regional and global myocardial mechanics.Reference Pellikka, Douglas and Miller 6 These techniques have been utilised in the clinical context to varying degrees. Given the relatively recent development of these techniques, the inconsistency of adoption of each technique among centres, and the challenges that are unique to quantitation in CHD and paediatric echocardiography, it is unsurprising that none has been evaluated, to date, as critically as established echocardiographic techniques have been.
In this article, we shall discuss three-dimensional echocardiography and the imaging of myocardial deformation (strain). We describe the rationale and methodology, review available guidelines for practice, and discuss the advantages and limitations of each of these modalities. When available, we have also summarised the scientific evidence for the clinical application of these techniques to detect heart failure in children.
Considerations that are specific to paediatric and CHDs
Most of these techniques and modalities have been developed and studied in adults. Therefore, when they are used in children, consideration must be given to the varying range of body sizes, heart rates, and maturational changes in myocardial mechanics. The development of transducers and algorithms that are applicable to children typically lags behind similar developments in adults: a factor that compounds the challenges of new technology. Somatic growth is associated with changes in heart rates, loading conditions, and the size and thickness of the heart, all of which challenge the ability to interpret findings. For new measurements, there is paucity of meaningful normative data in children. The wide range of pathology that underlies the failing heart in children, such as the failing univentricular heart or the systemic right ventricle, further complicates the extrapolation of these techniques in children.
Three-dimensional echocardiography
At present, three-dimensional echocardiography enables the acquisition of the entirety of a cardiac chamber or valve in a single data set.Reference Nanda, Kisslo and Lang 7 Software that enables quantitation based on such data sets provides the potential for measuring the size and function of chambers and valves in a manner that would be free of geometric assumptions, independent of the size, shape, or morphology of a structure.Reference Corsi, Lang and Veronesi 8 Starting about 15 years ago, advances in piezoelectric technology have led to the development of single crystals of piezoelectric material that are cut into thousands of elements that are arrayed in a matrix.Reference Gururaja, Panda, Chen and Beck 9 Simultaneously, advances in the miniaturisation of transducers and faster processing have made three-dimensional echocardiography a reality with clinical applications.Reference Sugeng, Weinert and Thiele 10 The development of transducers that are capable of imaging at higher frequency and with smaller footprints has extended this application of this modality to children.Reference Acar, Abadir and Paranon 11
Furthermore, three-dimensional echocardiograms can be acquired using one of two main methods. The first of these is live imaging, in which a narrow pyramidal echocardiographic image is displayed in real time. At present, applications of this mode centre on interventional cardiology or localised abnormalities of cardiac structures or valves. The second method uses multiple narrow pyramidal images that are obtained from multiple cardiac cycles and are stitched together to develop a wide data set. This allows for acquisition of larger structures such as ventricles and enables the quantitation of volumetrics and function. Three-dimensional echocardiography has two main roles in the assessment of the dysfunctional ventricle – the measurement of ventricular volumes and ejection fraction and providing insights into the mechanism, and potentially the quantitation, of valvar regurgitation.
Left ventricular size and function
The ability to perform three-dimensional echocardiographic measurement of the left ventricular volumetrics requires the software to detect the chamber that is of interest, delineate its boundaries, and then track its borders throughout the cardiac cycle. Owing to the complexity of performing this on a rapidly moving structure, the algorithms that are in widespread clinical use are based on computerised modelling of cardiac structures and motion. They require that the user designate specific locations for the hinges of the mitral valve and the apex of the left ventricle, and then use a form of automated detection of endocardial borders followed by counting of voxels to determine the volume of the ventricle; three-dimensional measurements of left ventricular volumes and ejection fraction that are based on these algorithms have been validated using cardiac MRI as a reference standard. A recent meta-analysis of 95 such studies involving over 3000 adults showed that left ventricular ejection fraction obtained using three-dimensional echocardiography exhibited excellent correlation with MRI, whereas left ventricular volumes were systematically underestimated.Reference Shimada and Shiota 12 Similar studies have been performed in children, both with structurally normal hearts and with CHD, with similar results.Reference Friedberg, Su and Tworetzky 13 – Reference Riehle, Mahle and Parks 15
If three-dimensional echocardiography was, in fact, more accurate than two-dimensional techniques, then its ability to detect smaller changes in volume or ejection fraction of the ventricle in a manner that is more reproducible when compared with two-dimensional techniques would add incremental value; two recent studies have evaluated these features. Dorosz et alReference Dorosz, Lezotte and Weitzenkamp 16 published a meta-analysis of studies that compared left ventricular volumes and ejection fraction that were performed using two- and three-dimensional echocardiography with measurements obtained using cardiac MRI. They collated 23 studies that included over 1600 echocardiograms. They found that under controlled settings and in patients with adequate image quality, three-dimensional echocardiography provides better accuracy and precision in measuring the volume and ejection fraction of the left ventricle when compared with two-dimensional echocardiography; however, when compared with MRI, it underestimates true left ventricular volumes. Thavendiranathan et alReference Thavendiranathan, Grant and Negishi 17 compared the temporal variability of left ventricular ejection fraction that was measured using two-dimensional echocardiography with that obtained using three-dimensional echocardiography in the setting of normal global longitudinal strain. They interpreted the temporal variation in left ventricular ejection fraction as being attributable to physiological differences and intrinsic variability of the measurement, rather than to dysfunction. They showed that the temporal variability of two-dimensional echocardiographic measurement of left ventricular ejection fraction was ~10%, whereas the temporal variability of three-dimensional echocardiographic measurement of the left ventricular ejection fraction was ~6%. Baker et al,Reference Baker, Flack and Hlavacek 18 among others, have shown that left ventricular volumetrics in children can be performed quickly and that it is a reproducible technique.
The American Society of Echocardiography and the European Association of Cardiovascular Imaging published guidelines that delineate the role of three-dimensional echocardiography in adults in 2012;Reference Lang, Badano and Tsang 19 , Reference Lang, Badano and Mor-Avi 20 these guidelines were updated in 2015.Reference Lang, Badano and Mor-Avi 20 The routine use of three-dimensional echocardiography for measuring left ventricular volumetrics and ejection fraction is recommended in adults because it is more accurate and reproducible than two-dimensional techniques. This stance has been re-iterated in recent guidelines for serial monitoring of patients who have received cardiotoxic chemotherapy.Reference Plana, Galderisi and Barac 21 Guidelines do not exist, at present, for the use of three-dimensional echocardiography in children.
The right ventricle and the univentricular heart
Dilation and dysfunction of the right ventricle are frequently encountered in paediatric cardiology. The retrosternal position of the right ventricle, its complex pyramidal shape, and the prominence of trabeculations that obscure the border between the endocardium and the cavity are all factors that greatly complicate the echocardiography of the right ventricle. Although these factors pose a challenge to two-dimensional echocardiography, they should not, in theory, affect three-dimensional echocardiography, because with three-dimensional echocardiography there ought to be fewer assumptions about the shape of the ventricle. There are two echocardiographic approaches to volumetrics for the right ventricle: summation of discs and semi-automated detection of borders.
The method of summation of discs is most directly comparable with the technique that is used for MRI. It correlates well with volumes obtained by MRI, but underestimates volumes consistently. This method does not rely on left ventricular landmarks for the purposes of developing a model of the right ventricle. It can be applied to abnormal left ventricles, the systemic right ventricle, and the univentricular heart. Unfortunately, this method has been removed as an option from commercial platforms for three-dimensional quantification.
Semi-automated detection of borders is a validated methodReference Jenkins, Chan and Bricknell 22 – Reference Tamborini, Marsan and Gripari 24 that is in widest usage for right ventricular volumetrics. It requires that the user first define both right and left ventricular anatomic landmarks, and then manually draw the contours of the ventricle in three orthogonal views in both end diastole and end systole. This method has been validated in adults, but the need for the definition of left ventricular landmarks – in locations that are anatomically correct – poses a major challenge to the application of this technique in many situations that are encountered commonly in paediatric practice. Nevertheless, the feasibility, reproducibility, and accuracy of three-dimensional echocardiographic quantitation of the size and function of the right ventricle has been evaluated in a variety of CHDs including tetralogy of Fallot, the univentricular heart, and the systemic right ventricle.Reference Khoo, Young and Occleshaw 25 – Reference Soriano, Hoch and Ithuralde 31 Overall, three-dimensional echocardiography is feasible in this population with acceptable inter- and intra-observer variability. Correlation with MRI is good when the right ventricle is normal in size and function, and is moderate when the right ventricle is enlarged or dysfunctional. Moreover, three-dimensional echocardiography yields right ventricular volumes that are typically underestimated compared with cardiac MRI; the two modalities are not interchangeable.
Valvar structure and function
Routine three-dimensional imaging of the mitral valve is recommended in adults because it provides the best physiological and morphological assessment of the valve;Reference Lang, Badano and Tsang 19 three-dimensional colour Doppler imaging of the mitral valve allows for quantitation of the number of jets and measurement of the regurgitant orifice.Reference Thavendiranathan, Phelan and Collier 32
Imaging of myocardial deformation (strain)
Strain is a dimensionless index that reflects the regional deformation of ventricular myocardium during a cardiac cycle as a function of its initial length or thickness. This form of quantitation uses deformation as a direct measure of ventricular contraction and relaxation, in contrast to indices that measure the overall change in dimensions of the cavity of the ventricle, such as shortening fraction or ejection fraction. The rationale behind the use of deformation imaging is that the direct measurement of myocardial mechanics may detect early changes in the health of the myocardium, before overall changes in function become evident using shortening fraction or ejection fraction. The measurement of deformation of the myocardium may be useful in recognising pre-clinical disease, in surveillance for disease, and in stratifying patients for their risk of developing disease. As strain does not rely on assumptions regarding the geometry of cardiac chambers, this method could be useful in quantifying myocardial function in the context of CHD, particularly in quantifying the function of the morphological right ventricle or of the univentricular heart. Strain imaging can also identify patterns of dyssynchronous contraction of the ventricles.
Strain is expressed as a percentage; the value may be positive or negative depending on whether there is shortening or lengthening of the myocardial segment. Strain is measured in each of the three planes – longitudinal, circumferential, and radial – in which a chamber deforms during the cardiac cycle. Segmental strain refers to deformation within a segment of the ventricle, whereas global strain refers to the average of all strains for all segments of the ventricle; in practice, global longitudinal strain is the type of global strain measurements that is used. With any measurement of strain, a corresponding strain rate can be measured. Strain rate is the rate of change in strain over time; it is expressed as per second. Both strain and strain rate can be measured in systole or diastole. Strain is imaged using speckle tracking echocardiography, a technique that tracks “speckles” that are ultrasonic patterns of reflection within the myocardium, throughout the cardiac cycle to determine deformation.
There are some caveats to the imaging of strain. It requires high temporal resolution: frame rates between 40 and 80 frames/second are recommended in adults, and even higher frame rates are required in patients who have high heart rates. In addition, the quality of the image has to be optimal, such that it includes the apex of the ventricle without foreshortening and ensures that views of the short axis of the left ventricle are not off-axis. Although measurements of strain that are based on two-dimensional speckle tracking have been believed to be independent of the angle of insonation, a recent study in childrenReference Forsha, Risum and Rajagopal 33 demonstrated that left ventricular peak longitudinal systolic strain is modestly dependent on the angle of insonation as well as on the depth of the target. Although strain is, conceptually, independent of the angle of insonation or the depth of the target, echocardiography itself is not – for example, if a structure is imaged in the axial plane when imaged from one window, and in the lateral plane in another window, then it is unrealistic to expect images that have identical resolution from both windows. The same logic holds for the depth of the target. As a corollary to this, the ability to track speckles may well depend on the well-known differences in the resolution of an echocardiographic image in different planes and at differing depths. Two-dimensional analysis of strain is fundamentally challenged by the loss of speckles of ultrasound due to motion outside the plane of imaging,Reference Geyer, Caracciolo and Abe 34 which is inherent to the complex, three-dimensional nature of myocardial contraction. Three-dimensional speckle tracking is in its infancy; it faces many challenges,Reference Gayat, Ahmad and Weinert 35 and is therefore not discussed further.
Since 2011, the American Society of Echocardiography and the European Association of Echocardiography have published three statements that reflect a consensus on the methodology and clinical applications of myocardial deformation imaging.Reference Lang, Badano and Mor-Avi 20 , Reference Plana, Galderisi and Barac 21 , Reference Mor-Avi, Lang and Badano 36 Together, these statements point to the potential value of deformation imaging in the recognition of sub-clinical diseases such as cardiotoxicity from chemotherapeutic agents, serial monitoring of cardiomyopathy, and pulmonary artery hypertension as well as in the assessment of myocardial ischaemia and viability. They conclude that global longitudinal strain measured by two-dimensional speckle tracking echocardiography is reproducible and provides incremental prognostic data over left ventricular ejection fraction, although the variability of measurements among vendors and among different versions of software makes it difficult to recommend normal values or lower limits of normal range. In a recent meta-analysis of over 1500 adults who had received cardiotoxic chemotherapy, Thavendiranathan et alReference Thavendiranathan, Poulin and Lim 37 found that a decline in left ventricular global longitudinal strain precedes a decrease in left ventricular ejection fraction. Similar findings have been reported in children with Duchenne’s muscular dystrophy,Reference Mertens, Ganame and Claus 38 Friedrich’s ataxia,Reference Dedobbeleer, Rai and Donal 39 and after cardiotoxic chemotherapy for childhood cancer.Reference Poterucha, Kutty and Lindquist 40 , Reference Cheung, Hong and Chan 41 It is unknown, however, whether using strain as a primary marker of cardiotoxicity to initiate cardioprotective therapy is superior to an approach that is based on ejection fraction, or whether a sub-clinical decrease in ejection fraction predicts subsequent heart failure.
Measurements of strain have been found to be predictive of outcomes in adults with hypertrophic cardiomyopathy: left ventricular global longitudinal strain below 15% has been shown to independently predict adverse cardiac events, adding incremental value to the stratification of risk that is possible with the use of traditional variables.Reference Reant, Reynaud and Pillois 42 , Reference Hartlage, Kim and Strickland 43 Studies on adults with heart failure suggest that left ventricular global longitudinal strain and circumferential strain add to the ability of a model that uses ejection fraction and tissue Doppler measurements to predict re-hospitalisation or death.Reference Cho, Marwick and Kim 44 In adults with pulmonary hypertension, right ventricular longitudinal strain has been shown to add incremental value to a model that predicts mortality.Reference Fine, Chen and Bastiansen 45 Such data are lacking in children. Guidelines do not exist at present for the use of the imaging of myocardial deformation (strain) in children.
Other measurements
The list of additional echocardiographic measurements that can be performed on children with ventricular dysfunction is long and can be overwhelming for the sonographer and echocardiographer, to say nothing of providing a potentially vast amount of numeric information for the clinician. Some such measurements, such as the ratio of the peak transmitral inflow Doppler velocity to the peak mitral annular velocity in early diastole – E:E' ratio – and the index of myocardial performance – Tei index – are in wide clinical usage. Others, such as the slope of propagation of transmitral inflow as measured by colour M-mode, and the slope of isovolumic acceleration of the myocardium at the level of the annulus of the mitral or tricuspid valve, or at the septum, are not. To date, there are neither practice guidelines nor high levels of evidence that would make a strong case for performance of these measurements in either children or adults. On the contrary, all the measurements that are listed in this paragraph have, in fact, been evaluated criticallyReference Tamborini, Marsan and Gripari 2 , Reference Tamborini, Marsan and Gripari 4 and have been found to have poor reproducibility and/or a lack of ability to predict outcomes. Similarly, the echocardiographic assessment of mechanical ventricular dyssynchrony is a controversial topic that lacks guidelines or high levels of evidence that might help build a case for its routine use.
The future
New echocardiographic modalities and technologies have great potential for clinical applications in children. Maturity of hardware and software platforms, coupled with operability across vendors, would be an important foundation to establish the reproducibility and normal paediatric values of the various new parameters. It is likely that the automation of measurements would help improve their reproducibility. Finally, large prospective studies with follow-up over the long term – with analysis of echocardiograms in core facilities – would be needed to establish the clinical significance, prognostic, and incremental value of these modalities in the evaluation and management of heart failure in children.
Acknowledgement
None.
Financial Support
This research received no specific grant from any funding agency, commercial, or not-for-profit sectors.
Conflicts of Interest
There are no relevant conflicts of interest.
Ethical Standards
The authors assert that all procedures contributing to this study comply with the ethical standards of the relevant national guidelines on human experimentation and with the Helsinki Declaration of 1975, as revised in 2008.
