Roderick Van Grootel

Discussion and future perspectives

The aims of this thesis are twofold: One, to establish reference values for a variety of echocardiographic markers and determine influences of patient characteristics. And two, to identify prognostic markers which should consequently be utilized for risk-prediction, all to further reduce morbidity and mortality in patients with congenital heart defects.

Part I
Part I of this thesis assessed the first aim. The results are summarized in table 1 which presents an overview of the mean values that were found for each different echocardiographic measurement and per diagnosis group. This has resulted in interesting findings in the so-called “Navigator cohort” of healthy volunteers, which are underlined here: With regard to conventional echocardiographic assessment; as a rule of thumb echocardiographic linear and volumetric measurements in a Dutch population do not necessarily coincide with limits of normal as provided in European guidelines (1, 6). However, echocardiographic functional measurements such as left ventricular ejection fraction do neatly coincide with these guidelines. The latter can be explained by the fact these are often relative measures, where actual size is nullified due to dividing to measures with one another.

The Navigator cohort was also instrumental for the understanding and interpreting of strain analysis indices; what can be interpreted as normal, which patient characteristics influence strain analysis? This information is especially valuable when comparing them with patient data, as we have done in this thesis. For both systolic strain and diastolic strain rate, age, sex and blood pressure are significantly correlated. As a result of the Navigator data, we had good comparison data and we could conclude that both systolic and diastolic function are reduced in adult congenital heart disease patients. Congenital heart disease can roughly be divided in left- and right-sided pathology.

We found that in left-sided congenital heart diseases, left ventricular function is affected in different degrees; left ventricular function was reduced in patients with congenital aortic stenosis and repaired coarctation aortae, and we found slightly lower values for the first group: Mean LV GLS was -15.3% ± 3.2 vs -17.6% ± 2.9. A direct comparison is not viable between the groups given the different pathology, however age was roughly similar (table 1). Diastolic function measured using early diastolic strain rate was also more significantly reduced in congenital aortic stenosis patients: 0.66s-1 ± 0.18 vs 0.75s-1 ± 0.17. The fact that both systolic and diastolic function are worse, is less surprising; systolic and diastolic function are highly correlated with one another (31, 32).

Table 1 - Presenting an overview of the mean values found in this thesis:
Healthy controls: Age 44.6 ± 13.8, Vendor Qlab, Feasible 96%, 97%, 87%, 80%, 78%, 80%. Values: GLS -21.0 ± 2.0, GLSre 1.21 ± 0.20, GCS -25.4 ± 5.0, LA GLS 39.6 ± 6.3, GLSre -2.76 ± 0.63, GLSra -2.57 ± 0.62.
Systemic right ventricle: Tomtec, Feasible 90%, 90%, 90%, 80%, 75%. Values: RV GLS -13.8 ± 3.4, Free wall strain -15.5 ± 3.9, Septal strain -12.4 ± 3.3, GRS -13.6 ± 3.1, GCS -18.9 ± 13.5.
Tetralogy of Fallot: Age 33.2 [25.5 - 42.0], TomTec, Feasible 100%, 100%, 91%, 85%, 91%. Values: LV GLS -17.6 ± 3.6, GLSre 0.77 ± 0.22, RV GLS -16.5 ± 5.1, GCS -18.4 ± 4.1, GRS -15.4 ± 4.1.
Aortic stenosis: Age 33.4 [25.5 - 42.8], TomTec, Feasible 97%. Values: LV GLS -15.3 ± 3.2, LV GLSre 0.66 ± 0.18.
Aortic coarctation: Age 34 ± 13, TomTec, Feasible 95%. Values: LV GLS -17.6 ± 2.9, LV GLSre 0.75 ± 0.17.
(Measurements include Age, Vendor, GLS, GLSre, GCS, GRS, GLSra for LV, RV, and LA).

Less self-evident is that left ventricular function is reduced in patients with tetralogy of Fallot, given the fact that the congenital anomalies are located in the right-side of the heart. Strain analysis revealed that both systolic and diastolic function are reduced when compared to healthy controls. Previous studies have suggested several mechanisms for this finding, all of which relate to the right ventricle negatively impacting the left ventricle (33-36). For instance apical rotation is often reduced in tetralogy of Fallot patients, which means a reduced twisting and consequently untwisting motion. This results in reduced longitudinal shortening and lengthening of the left ventricle. Reasons for the impaired apical rotation are a dilated right ventricle and/or reduced right ventricular function. What is most striking about this, is that conventional echocardiographic markers fail to detect this deterioration (for instance left ventricular ejection fraction). Ultimately the information gained in Part I, made it possible to explore whether myocardial deformation contains prognostic information in adults with congenital heart disease, as was done in Part II of this thesis.

Challenges that are encountered when implementing new markers
New quantification tools can be useful for clinical practice, especially when these outperform current diagnostic tools. For interpretation purposes, adequate understanding of normal ranges is needed, as well as factors that can influence them. The first chapter assessed conventional echocardiographic markers in healthy controls, demonstrating the strong influences of demographic features. Correct reference values are essential in the search for novel diagnostic techniques. There is ample evidence already present and distilled into the ESC guidelines, however some interesting results were found. The main conclusion is that dimensional and volumetric measurements should be interpreted with care, whereas functional assessment coincides very well with guideline reference ranges.

In the following chapters strain analysis was used to establish reference ranges in order to be able to compare patient data in Part II. These papers are of critical importance if solid recommendations are to be endorsed by the European and American guidelines, which is as of yet not possible. Solid reference ranges have not yet been formulated, which is in part due to the difficulty that goes with getting these data published, which is interesting considering their clinical importance. Another explanation comes from heterogeneity between vendors, influence from age and, to a lesser extent, load dependency. The joint standardization initiative of both ASE/EACVI and the industry was started to tackle the first of these (12, 37). Since their formation significant steps have been made which has resulted in the current situation where the differences between vendors are still statistically significant, though one can argue whether these are also clinically significant. An increasing number of studies are being published concluding that global longitudinal strain is equal or even better than ejection fraction in regard to variability both in test-retest situations as well as between vendors (38, 39). Formulating a solid lower limit normal with regard to left ventricular global longitudinal strain is essential for adopting this technique, arguably the only obstacle against implementation into clinical practice is a systematic review and meta-analysis regarding the different strain measurements. That is why the data presented in Chapter VII is so important. In this meta-analysis we found that a global longitudinal strain value of less than -16% is indicative for myocardial dysfunction.

Diastolic strain rate analysis
For left ventricular diastolic function assessed with speckle-tracking software, few studies have been published (18, 19, 40-42). It is crucial that studies on normal values regarding speckle-tracking derived diastolic function are performed and published. In this thesis, we found that left ventricular diastolic function was reduced in most patients. Especially, those with congenital aortic stenosis and coarctation of the aorta. Both diagnoses result in a pressure loaded ventricle, which would explain these findings (31). It is reasonable that valvular function affects myocardial function and relaxation more severely, but it is impossible to know which variable attributes to reduced diastolic function and more importantly, to what extent.

Both previous diagnoses consist of left-sided pathology. However tetralogy of Fallot only affects the right ventricle. Though significantly lower values for left ventricular global longitudinal strain were found, (with a significant part was within normal ranges), left ventricular global diastolic strain rate was significantly reduced and below normal. Studies have identified the presence of an interaction between the left and right ventricle and vice versa (33-36). Just as the twisting movement of the left ventricle enhances systolic function, this untwisting enhances diastolic function (43, 44). We know from previous research that the right ventricle can impair this twisting motion (33, 45, 46). This could in part explain why diastolic function is reduced. As such, rotational movement contains valuable information but is not measured in daily practice. This may be in part because of the lower feasibility of especially the apical short axis view, however based on our results, we do recommend to try and obtain this data in clinical practice.

Part II
In Part II of this thesis the second aim was addressed. In the first part we concluded that overall, systolic and diastolic strain measurements were significantly reduced in adult patients with congenital heart disease. Table 2 presents an overview of the most important findings of Part II. In essence, we found that left ventricular systolic and diastolic strain measurements were associated with the occurrence of cardiovascular events and we can conclude that these variables are useful for risk-stratification, with the emphasis on “left ventricular”. Which is what would be expected in patients with left-sided pathology. However, also in patients with tetralogy of Fallot, where right ventricular dysfunction is often encountered, left ventricular apical rotation was associated with cardiovascular events. Right-sided problems are often well tolerated in Fallot patients, but as soon as problems of the left ventricle occur, the patient’s prognosis is at stake (47-49).

Table 2 - Presenting an overview of the key results for each strain measurement per diagnosis group:
Systemic right ventricle: sHR 0.94 (0.81 - 1.09), 0.90 (0.79 - 1.01), 0.89 (0.55 - 1.45), 0.85 (0.58 - 1.25), 0.73 (0.47 - 1.14), 0.58 (0.39 - 0.86), 0.77 (0.46 - 1.30), 0.69 (0.46 - 1.05). Endpoints: Death or HF, Death or arrhythmia.
Tetralogy of Fallot: sHR 0.43 (0.30 - 0.62), 0.63 (0.49 - 0.82), 0.31 (0.18 - 0.53), 0.62 (0.46 - 0.83), 0.69 (0.53 - 0.90), 0.86 (0.71 - 1.04), 0.34 (0.20 - 0.58), 0.64 (0.49 - 0.84). Endpoints: Death or HF, Any event.
Aortic stenosis: sHR 0.62 (0.47 - 0.81), 0.62 (0.47 - 0.83). Endpoint: Any event.
Aortic coarctation: sHR 1.43 (0.81 - 2.54), 0.75 (0.45 - 1.25). Endpoint: Any event.
Strain measurements included GLS, GLSre, GCS for LV and RV (Free wall and Septal strain).

The main exception in this thesis were patients with a systemic right ventricle. Strain analysis was not associated with cardiovascular events in our study in this cohort. Though there are studies that have shown that systemic right ventricular strain does hold prognostic information also in these patients (50). Several causes in combination may have resulted in our negative findings. In these patients the anatomy is quite complex which makes visualizing the entire ventricle difficult. On top of that the ventriculo-arterial mismatch has induced changes such as hypertrophy and a functional shift away from longitudinal shortening, favoring a more enhanced radial and circumferential function (51). As such we found a severely reduced longitudinal strain, which is in line with other studies (50-53), confirming our findings. However the severely reduced values also mean a reduced spread, which could be a part of the reason why we found no prognostic value. Circumferential and radial strain were also assessed. However none of these were associated with cardiovascular events. Possible explanations are difficulty acquiring adequate views, in case of radial strain the reliance on two strain measurements, and lastly probably also the limited number of patients and events. Other, for instance clinical or blood biomarkers, have better prognostic value in this specific patient population.

Overall we can conclude that measurements derived from speckle-tracking echocardiography contain valuable information. Of course it is not “the one” measurement to replace all currently used markers, but it should be incorporated in future models to assess risk, evaluate treatment effects and support clinical decision making. Currently strain analysis is undervalued and underestimated, resulting in a limited implementation in clinical practice, which is a shame considering it being readily available, being relatively cheap (as opposed to CT or CMR), and as such being ideal for serial measurements within an individual. As a result, also in our experience, the prognostic value is limited. It is our opinion that a better integration of strain analysis in the guidelines would greatly improve this.

The future of myocardial deformation imaging
Myocardial function is a complicated process where systolic and diastolic forces actively and passively play a role. It is influenced by pre- and afterload, age, valvular function, electro-mechanical (dys)synchrony and previous interventions. Diastolic function is currently assessed by several different variables, which are combined to approximate left ventricular filling pressures. This is done by measuring mitral valve velocities, tissue Doppler derived e’ and atrial volume. The E-wave is load dependent, the A-wave unmeasurable for instance in patients with atrial fibrillation and atrial dilatation can occur in the absence of diastolic dysfunction. Not to mention that the absolute cut-off value used for atrial volume regardless of BSA-indexation, cannot be extrapolated to the Dutch population, as is shown in chapter II. Strain analysis can overcome most of these limitations, making it a promising alternative. We showed in chapter III and IV that strain analysis can assess diastolic parameters, without angle-dependency or geometric assumptions. The e’ would be replaced by global longitudinal early diastolic strain rate, the E- and A-wave by atrial early and late longitudinal strain rate, and atrial volume by atrial peak systolic longitudinal strain.

By measuring atrial contraction in late diastole, left ventricular global longitudinal strain and early strain rate, each of the phasic functions of the left atrium can be measured. Another advantage of this approach is, that the systolic strain and diastolic strain rate measurement are acquired in a single measurement, meaning that you only have to perform two measurements to acquire the necessary data. Of course there are other measurements, which would have to be evaluated, such as left ventricular rotation and twist.

It is clear that systolic and diastolic function are closely related to each other, just as it is clear that left atrial function influences left ventricular function. It is clear that the right ventricle affects the left ventricle and vice versa. However in current practice and research, systolic and diastolic function are assessed separately and cardiac function is being divided per cardiac chamber. Strain analysis offers a unique opportunity to really start shifting focus from this compartmentalized approach to a more wholesome approach.

When measuring longitudinal strain, strain rate is automatically calculated, as it is the integral of strain. Previously only the systolic data was used, but the software traces the myocardium throughout the entire cardiac cycle, meaning it contains data on the diastolic phase as well. Therefore, both systolic and diastolic indices are given by one measurement. Furthermore, atrial contraction can be measured using ventricular strain rate, as it causes a lengthening of the ventricle. Taking it one step further, one could visualize a measurement that encompasses both the ventricular and atrial myocardium, making it possible to investigate atrio-ventricular properties. And previous research as well as our own data have shown that there is a ventricular interaction, which makes an integrated approach on a ventricular level a logical next step.

To be able to analyze cardiac function in the way described above several factors need to be addressed. Ideally such new markers would need to be validated, and determined which view would be used to measure both ventricle and atrium. The longitudinal axes of these two chambers are often misaligned. Software would need to be rewritten; preferably in such a way that post-processing of data with spreadsheets would no longer be necessary, as is currently the case for diastolic strain rate measures. And the most ideal modality to assess cardiac function in such an integrated approach would have to be three-dimensional echocardiography. However improvements would have to be made to the currently existing hardware to improve both temporal and spatial resolution.

Future research
Several markers have been measured in this thesis, in different patients groups, but there are still boxes left unchecked after the completion of this thesis. Left atrial strain analysis was not done except in the Navigator cohort of healthy volunteers, although it is a promising marker. Another such marker would be rotation at the basal and apical level of the left ventricle, which was assessed for the tetralogy of Fallot patients but not in other patient groups. Apical rotation would be of special interest in patients with sub endocardial ischemia, as the endocardial fibers usually inhibit apical rotation to a certain degree due to the orientation of myocardial fibers. In patients with left ventricular hypertrophy that would mean that apical rotation is enhanced, but this was not studied in this thesis.

Identifying relevant prognostic markers does not directly translates to clinical benefit which is ultimately the goal. Additional research should be performed, focusing on the discriminative abilities of these markers. When these variables would be included in a model, the model would have to be calibrated and validated. In order to assess clinical benefit, the model would have to be implemented and the effect of the implementation would have to be assessed. The assumption is that if a patient with a higher risk can be identified, a more stringent follow-up would be used while low-risk patients would be followed less intensively. All of which would ideally result in a better and longer quality of life. Ideally we would also have clear treatment options for patients at high-risk, such as medical therapies, surgical or percutaneous interventions or advanced pacing options such as biventricular pacing. Unfortunately in adults with congenital heart disease all these options are open for use but typically have no proven benefit. Finally, also transplantation or ventricular assist devices are options for very sick patients, but clearly more research is needed to identify the patients who have best benefit from these therapies and especially optimal timing is still difficult to determine. Quite possibly serial measurements can help answer both the question when to implement therapeutic options, and whether such therapeutic options have clinical benefit.

Conclusion
This thesis provides insight in myocardial deformation in adult patients with congenital heart disease, influences of baseline characteristics on myocardial deformation, and their predictive value of cardiovascular events. Cross-sectional analyses have shown that systolic and diastolic function are reduced in congenital heart patients, while the prospective studies reveal that strain analysis is associated with cardiovascular events. Though these results are promising, the role of strain analysis in clinical practice seems to be limited due to a lack of normal values. The new echocardiographic markers presented here contain the building blocks for future research needed to translate this data into treatment strategies, ultimately leading to better healthcare for this complex patient group.