Assessing aortic strain and stiffness: don't forget the physics and engineering

Bart Bijnens, ICREA Research Professor,

Other Contributors:

August 10, 2010

There is more and more evidence that the mechanical properties of the aorta, and arterio-ventricular coupling, are of great clinical importance in cardiovascular diseases. Therefore, we have read the manuscript of Vitarelli et. all with great interest [1]. Whereas, given the recent evolution of ultrasound equipment and processing tools, the idea to assess the aorta using velocity and deformation information is interesting, the way it is used by the authors is very confusing. They measure the velocity of one point on the aortic wall and suggest this is related to stiffness of the wall. However, the velocity describes the motion of the wall within the thorax rather than only the extension of the wall due to the internal pressure (the intrinsic deformation, determined by the pressure within and the stiffness of the vessel). As opposed to the carotid artery, that is not showing substantial overall motion, to measure aortic distension, the difference in motion of proximal and distal wall has to be assessed, as described in e.g. [2]. This can easily be seen on the m-mode (fig. 1A [1]), where the whole aorta is moving substantially during the cardiac cycle and the velocity thus mainly describes this motion. Additionally, from aortic pressure curves, there is no evidence of the biphasic diastolic decrease of the aortic diameter, with a large atrial phase, as shown in fig 1B. Whereas this motion might be an interesting parameter to assess in pathological cases in clinical practice, it does not directly describe the properties of the aorta. It is much more related to the motion of the aorta by the LV and therefore rather represents LV systolic and diastolic function. Distensibility was calculated as D=2(As-Ad)/[Ad(Ps-Pd)]. Most authors would not use the multiplication by 2. There also seems a problem with the units since they report a normal average D=79 Pa-1 while published values are in the order of 37x10-3 kPa-1 [3]. A more important problem is the assessment of radial strain (aortic wall thinning), which is calculated from the temporal integration of strain rate, and strain rate is assessed based on spatial differences in velocities. To calculate this, a region of interest and a calculation length has to be chosen. The authors used a 3mm strain length and a ROI of 2-4 mm. Looking at the implementation of the strain analysis in the system, the experience learns that this corresponds to a range of >5 mm used for the quantification [4]. Considering that the wall of the aorta is not 5 mm and a scan plane close to the valve is used, it is clear that (aliased) velocities of the blood in the RV outflow and aortic lumen are included. Therefore, the calculated strain is not the radial deformation (or thinning) of the aortic wall. This is clearly seen from the strain curve in fig 1C, where it is difficult to explain why the aortic wall would thin substantially during atrial contraction and not start thinning fast and gradually from aortic valve closure as would be expected when analysing pressure traces in the aorta (keeping in mind that aortic strain is directly related to pressure so it should have a profile similar to local aortic pressure). Additionally, the measured strain values are not realistic. To show this, we simulated the radial strain (wall thinning) in a tube resembling the aorta and assuming the wall is incompressible (conservation of volume). We used a wide range of realistic parameters: end-diastolic diameter ranging from 20-40 mm; wall thickness of 2-4 mm; distensibility between 4x10-3 and 9x10-3 mmHg-1 for normal individuals and between 1x10-3 and 3x10-3 mmHg-1 for hypertensives [3]; and a pressure gradient of 45 mmHg (normals) and 70 mmHg (hypertensives) (the extreme values from [1]). Additionally, a longitudinal strain of 1% was assumed to account for the lengthening of the aorta during systole. Fig. 1 shows the resulting ranges for the radial strain. In extreme conditions (certainly not the average patient), the strain can range from 6-15 % in normals and 3-9 % in hypertensives. This is clearly substantially less than the in [1] reported average values of 23.1 and 8.8%.

Fig 1: Left: the range of radial strain values as a function of end- diastolic (ED) radius and wall thickness for a large distensibility range in normals (blue) and hypertensives (red). Middle: radial strain as a function of ED radius for a wall thickness of 2.5 mm; Right: radial strain as a function of wall thickness for an ED radius of 12 mm

A thorough analysis of what is measured using these techniques would be appropriate before reporting clinical results. As reported by the authors, the measured values might correlate with changes induced by hypertension, but they surely do not specifically describe changes in the aortic wall. Dedicated analysis of aortic properties should investigate measurements that directly reflect aortic wall mechanics or blood hydrodynamics instead of statistical analysis of a set of observations.

1. Vitarelli A, Giordano M, GermanÃÆ’ƒÃâ€Â Ãƒ‚’ÃÆ’‚² G, Pergolini M, Cicconetti P, Tomei F, Sancini A, Battaglia D, Dettori O, Capotosto L, De Cicco V, De Maio M, Vitarelli M, Bruno P. Assessment of ascending aorta wall stiffness in hypertensive patients by tissue Doppler imaging and strain Doppler echocardiography. Heart, doi:10.1136/hrt.2010.198358, 2010. 2. Long A, Rouet L, Bissery A, Rossignol P, Mouradian D, Sapoval M. Compliance of abdominal aortic aneurysms evaluated by tissue Doppler imaging: Correlation with aneurysm size. Journal of Vascular Surgery, 42(1): 18-26, 2005. 3. Resnick LM, Militianu D, Cunnings AJ, Pipe JG, Evelhoch JL and Soulen RL. Direct Magnetic Resonance Determination of Aortic Distensibility in Essential Hypertension: Relation to Age, Abdominal Visceral Fat, and In Situ Intracellular Free Magnesium. Hypertension. 1997;30:654-659 4. Bjastad T, Aase SA and Torp H. Velocity Sensitivity Mapping in Tissue Doppler Images. Proceedings of the IEEE Ultrasonics Symposium, 4:1968- 1971, 2005.

Conflict of Interest:

None declared

Conflict of Interest

None declared