Aortic wall thickness

MRI allows to visualize the aortic wall in any plane orientation (Fig. 1). Most studies are based on black-blood spin echo T1-weighted acquisitions which generate a sharp contrast between the signal from the wall and the blood. An age-related increase in average wall thickness of the descending aorta was reported in 1053 general population individuals aged 45–85 years (average 2.35 ± 0.5 mm) but was not found for the ascending aorta (average 2.8 mm).^3,4 In this study SBP and hypertension were also independent correlates of increased wall thickness. Good reproducibility for this measurement has been reported.^5,6 However, the spatial resolution of most acquisitions (0.7–1 mm) being about half of the expected thickness for the aortic wall (1.5–2.5 mm) we believe that partial volume remains a major drawback that will require higher spatial resolution and refined automated quantification methods to be accurate. No large MRI studies are available on the wall thickness of young and healthy individuals.

Static cross sectional aortic dimensions

Numerous MRI methods allow to measure aortic size. We will not review them here extensively. One of the strengths of MRI is to provide a 2D or 3D approach allowing to accurately measure aortic area or volume instead of relying on a 1D measure of diameter prone to higher measurement bias. Cross-sectional area of the aorta can also be used to calculate a mean diameter value. Recently, the distribution of the mean diameter of the ascending aorta was reported according to age and gender in 3573 general population individuals aged 45–85 years according to age and gender.³ In this study the median value of AA luminal diameter increased by 1.1 mm per each decade for both men and women. Overall, age was the strongest associate of increased ascending aortic diameter along with gender and body surface area. The subgroup study of the 1612 « healthy » individuals without risk factors showed a mean and 95th percentile (upper limit) of AA diameter of 31 mm (38 mm) vs. 32 mm (39 mm) in the global cohort. The inter-reader intraclass correlation for the measurement of mean ascending aortic area on modulus images of MRI phase contrast acquisitions on 100 individuals was 0.98 (95%CI: 0.98–0.99) with a mean difference of 0.14 mm (limits of agreement −1.1–1.4 mm). Interestingly, in the Dallas Heart study, increased abdominal aortic wall thickness was associated with a high lifetime predicted risk of cardiovascular disease in individuals from 30 to 50 years with a low short-term risk.⁷

The constitutional decrease in aortic diameter along the aorta or aortic tapering has been well documented by MRI.^8,9 However, the age-related widening of the aortic lumen and rate of diameter increase for each respective location is not homogeneous along the aorta. Indeed, the progression of aortic diameters markedly predominates in the ascending compared to descending aorta. Secondly, the thoracic aorta elongates with increasing age.^9,10 We have recently shown that the age-related elongation of the aorta is also much more marked in the ascending aorta and associated with widening of the arch and unfolding of the aorta⁹ as illustrated on Fig. 2. Reasons for this inhomogeneity in regional structural alterations likely include the relative content in elastin fibers, higher in the ascending aorta and decreasing along the aorta leading to the more muscular and resistive peripheral arteries. This constitutional difference between the ascending aorta and more peripheral locations is well known but not often functionally explored. We have shown Aortic arch unfolding due to widening and elongation with increased tortuosity to be independently related to increased peripheral and central blood pressure components as well as to cardiac hypertrophic remodeling i.e. increased left ventricular mass and mass to volume ratio in healthy individuals.

Figure 2

Age-related unfolding of the aortic arch. Comparison of aortic arch morphologies between 16, 40 and 80 year old asymptomatic individuals demonstrating the transversal and longitudinal enlargement of the aortic lumen and lengthening of the arch with progressive unfolding and increased tortuosity.

Which MRI method should be used to measure aortic strain?

Accurate measurement of aortic strain requires high spatial, temporal and contrast resolution. These features are achieved by current SSFP (steady state free precession) cine acquisitions which can achieve 0.7 mm of spatial resolution, less than 10 ms of temporal resolution and high contrast between wall and blood with constant signal. Such images allow to perform robust automated segmentation of the aortic wall.^11,12 Herment et al. validated an automated (2D + t) snake method to measure circumferential area to time curves in the aorta from either SSFP or phase contrast images with comparable results and good accuracy when compared to a compliant phantom.¹³ However, interobserver agreement was better with SSFP compared to phase-contrast images. The dependency of the segmentation methods on signal variations at the wall/blood interface favors the use of flow compensated SSFP images rather than gradient echo or phase-contrast modulus images where near wall signal is greatly influenced by local flow with lower near wall velocities generating lower signal. This phenomenon may be aggravated at higher magnet field strengths. Spatial resolution in phase contrast is also typically lower than in SSFP imaging. Consequently, we recommend using high spatial and temporal resolution SSFP cine images to measure aortic strain.

Which pressures should be used to calculate distensibility?

Most aortic distensibility studies rely on brachial cuff pressure measured during the MRI exam to normalize aortic strain. However, because of the amplification phenomenon in the periphery this may lead to overestimate central PP and therefore underestimate distensibility, especially in younger individuals. Accordingly, we proposed to use central PP instead of brachial PP to calculate distensibility. Central pressures can be estimated either using carotid tonometry or radial tonometry and a transfer function. Ideally pressure and strain should be measured simultaneously. However, tonometry devices are not MRI compatible and measurements have to be performed immediately after the MRI exam and can be normalized by the brachial diastolic pressure measured during the strain acquisitions to approximate similar pressure conditions. As an alternative, we have shown that using central pressures estimated by the vicorder™ device during the MRI exam to calculate distensibility was feasible and resulted in higher correlations with age, cfPWV and augmentation index than using carotid or brachial pressures.¹⁴

Aortic wall dynamics and age

Aortic distensibility decreases non-linearly with advancing age. Marked age-related decrease in aortic distensibility has been shown to occur in asymptomatic general population individuals from 20 to 50 years of age with a continued decrease thereafter albeit at a slower pace.¹⁵ Ascending aortic distensibility measured using central pulse pressure is on average 74 ± 23 kPa⁻¹ 10⁻³ between 20 and 30 years, 18 ± 7 kPa⁻¹ 10⁻³ between 50 and 60 and 10 ± 6 kPa⁻¹ 10⁻³ after 70 years with an average decrease of 14 ± 1 kPa⁻¹ 10⁻³ per decade. Importantly, this change is due to a decline in aortic strain as pulse pressure is not significantly increased in this age group. In multivariate analysis including cf-PWV, carotid distensibility and Aix, ascending aortic distensibility was the strongest correlate of age making it a powerful independent and early infraclinical marker of functional aortic dysfunction. Regional heterogeneity is also found for functional indices such as distensibility as only the youngest individuals before the age of 30 have ascending aorta distensibility exceeding descending aorta distensibility.^15,16 This early reversal in proximal to distal distensibility in healthy individuals could be modified by early onset cardiovascular disease. Through aortic–ventricular coupling, age-related changes in aortic distensibility is positively related to left ventricular elastance and total arterial elastance and to infraclinical changes in systolic cardiac strain and diastolic function.

Regional aortic pulse wave velocity (Fig. 2)

Phase contrast MRI allows to acquire spatially registered velocities simultaneously with morphological data throughout the cardiac cycle thus allowing to obtain local flow values in any aortic location. Velocity mapping is usually performed in one through-plane direction (perpendicular to the vessel of interest) but can also be performed in-plane¹⁶ or in 3D at the cost of lower spatial and temporal resolutions and higher scanning and post-processing times.^17,18 A single acquisition plane, typically placed perpendicular to the ascending aorta at the level of the mid right pulmonary artery simultaneously provides flow in the ascending and descending thoracic aorta during the same heartbeats. Pulse wave velocity is measured as the transit vascular length divided by the blood flow transit time. Flow curves should be generated using a robust automated vascular segmentation method reducing manual bias. Transit time measurement also requires caution as the temporal resolutions used in most 2D-MRI phase contrast methods (<10–20 ms) are not as resolved as in tonometry (2 ms) making the foot-to-foot measurements hazardous. Several solutions have been proposed including using a least squares minimization of the full upslope of normalized mean velocity curves corresponding to the two aortic locations of interest. The advantage of using the full upslope is to rely on several instead of one data points in the initial portion of the flow curve when flow is unidirectional and reflectionless.

Aortic PWV and distensibility, non-linearly related through the Bramwell Hill equations,^19,20 are not interchangeable measures of stiffness. Indeed, when distensibility varies most in young to middle aged individuals, aortic arch PWV rises with advancing age although more rapidly after 50 years.¹⁵ Arch PWV is on average 3.5 ± 0.5 m/s between 20 and 30 years, 7.2 ± 2.3 m/s between 50 and 60 and 11.1 ± 4.6 m/s after 70 years with an average decrease of 1.6 ± 0.1 m/s per decade. Arch PWV is strongly and independently related to age as its strongest determinant. In particular it is not related to BMI and body morphology such as cf-PWV as it does not require surface approximation of aortic length.¹⁰ Arch PWV values are lower than gold standard cf-PWV values for several reasons including potential over estimation of vessel length with surface measurements and exclusion of the ascending aorta with low PWV and inclusion of peripheral arteries with higher PWV values with cf-PWV. Further potential benefits of arch PWV lie in not requiring central pressure measurement to be assessed and in being a valuable marker of proximal aortic stiffness in studies including older individuals, when distensibility may be less relevant.

Regional differences have also been described for aortic PWV, in particular by Hickson et al. showing a lower rate of age-related increase in PWV in the ascending aorta as compared to more distal locations on the aorta.⁸ Taken together, the aforementioned changes in aortic geometry and function probably explain most of this difference. The relatively contained age-related increase in ascending aorta PWV is probably offset in part by an increased capacitance through elongation and dilation of the proximal aorta. Subsequently, cardiovascular disease and specifically atheroma with an increased distal prevalence across the arterial tree also play an important role in explaining such regional heterogeneity. Through aortic–ventricular coupling, increased arch PWV is related to impaired diastolic function and increased left ventricular filling pressures.

Backward and forward flow components (Fig. 4)

MRI mapping of flow in 4D in the thoracic aorta can illustrate the combined effects of altered local aortic geometry, compliance and pressure on age-related flow pattern changes. In young healthy individuals, backward flow has a late onset, appearing o nly in diastole when pressure reversal due to the end of ejection and closure of the aortic valve leads to physiological local flow reversal (Fig. 4) which predominates in the inner portion of the aortic arch curvature²¹ (Bensalah et al. in press). Hence, even in the absence of aortic disease, the curved shape of the aorta determines in great part the flow patterns along the proximal aorta. Age-related alterations of geometry such as dilation, elongation leading to unfolding of the aorta or disease such as aneurysms, stenosis or severe atheroma will further alter blood flow patterns in the aorta potentially leading to aggravated disease in a vicious circle. In addition to these alterations of the conduit function of the aorta, functional impairment in particular the loss of the cushioning function of pulsatile flow (distension and recoil) evidenced by the loss of aortic distensibility may participate in the creation of major flow dispersion and disorganization in the proximal aorta. These processes lead to increased left ventricular afterload and result in hypertrophy to normalize wall stress and loss of cardiovascular efficiency as cardiac work will increase to maintain a constant output and vascular–ventricular coupling.

Figure 4

4D-MRI mapping of backward and forward flow components. Streamlines of flow in multiple locations across the thoracic aorta (ascending aorta: white arrow, descending aorta: black arrow) in a middle aged healthy individual (Gyrotools™ software). Panel A: during the early systolic phase (flow curve upslope) flow is well organized with a predominant forward component and characterized by high velocities except at the very edges where velocities are lower. Panel B: during late systole (after peak flow and pressure) we can observe flow reversal corresponding to local pressure reversal and lower velocities in the inner portion of the arch curvature preceding a similar process in the descending aorta towards the posterior wall. Panel C: during early diastole we see a marked and organized backward flow component near the inner portion of the arch curvature with relatively low velocities and disorganized flow in the descending aorta with almost nulled net instantaneous flow.

Combining pressure and flow curves

R>elating temporally normalized flow curves from MRI and pressure curves from tonometry allows to non-invasively measure other indices of aortic biomechanics. Aortic reflection indices such as the time difference between flow and pressure peaks increasing with age as been recently described. Furthermore, reflection magnitude as the ratio between backward and forward flow components can be calculated as well as the characteristic impedance, a measure of pulsatility²² (Bollache et al. Artery 2013). Further studies will determine the added value of such indices in comparison to established markers of aortic biomechanics.