Brief review on how to measure arterial stiffness in humans

Abstract

Despite the wide recognition of larger artery stiffness (LAS) as a highly clinically-relevant and independent prognostic biomarker, it has yet be incorporated into routine clinical practice and to take a more prominent position in clinical guidelines. An important reason may be the plethora of methods and devices claiming to measure “arterial stiffness” in humans. This brief review provides a concise overview of methods in use, indicating strengths and weaknesses. We classified and graded methods, highly weighing their scrutiny and purity in quantifying arterial stiffness, rather than focusing on their ease of application or the level at which methods have demonstrated their prognostic and diagnostic potential.

1. Arterial stiffness

The arterial wall is a highly organized structure composed of matrix proteins (collagen fibers oriented in different directions and elastic lamellae), smooth muscle cells (SMCs) and other matrix components such as glycosaminoglycans, along with endothelial cells in the intimal layer (which will not be further discussed)¹. The composition of the arterial wall changes from the central aorta towards the periphery; centrally, the media of large elastic arteries has an ultrastructure of concentric elastic lamellae (up to 60-80 in the human ascending aorta), interdigitated by connective tissue layers that contain SMCs. This microstructure gradually disappears, with the number of elastic lamellae decreasing and an increased SMCs contents in medium-sized, and particularly, smaller vessels (such as arterioles) ². Collagen is continuously degraded and deposited in a process of mechano-biological homeostasis, which allows blood vessels to grow and remodel in response to changes in the mechanical/biological environment³. The major constituents of the elastic fibers are elastin (core component of elastic fibers), fibrillin microfibrils (scaffold for elastic fiber formation and visible at the periphery of the elastin core), and associated proteins essential for the cross-linking and arrangement of the micro-constituents^{4, 5}. Elastin, however, is believed to be only deposited in the media of the arterial wall during fetal growth and infancy, and is not renewed thereafter.⁵ Consequently, elastin within the arterial wall becomes progressively stretched with somatic growth and is exposed to a permanent tensional load in the circumferential and longitudinal direction (well known to vascular surgeons, who see the arterial wall retracting after surgically incising them). The distensibility of a young, healthy artery at physiological pressures arises from the intrinsic distensibility of the vessel’s elastin and from the wavy, undulating nature of the collagen fiber families that helically spiral along the artery. Collagen fibers are progressively recruited with increasing pressure/stretch levels, explaining the pronounced nonlinearity in the mechanical response of an artery when subjected to stretching/pressurization over a large range, with progressive functional stiffening of the artery at higher pressures (Figure 1).

Figure 1.

Top panel displays the typical non-linear relation between luminal pressure and area in large elastic arteries, explained by the progressive shift in load bearing from elastin to collagen fibers. Data translates into distensibility and PWV with the use of the Bramwell-Hill relation. Distensibility decreases with increasing pressure with concomitant increase in PWV. The grey shaded area indicates the physiological pressure range.

Elastin has been reported to have a half-life of 40-50 years⁶, inherently predisposing large, elastic arteries to stiffening in late life, with adverse associated consequences on human health. These arteries progressively lose their low-stretch bearing component, shifting load to stiffer matrix components (collagen). At the same time, arteries also lose some of their longitudinal elasticity and geometry, leading to elongation and increased tortuosity. Together with an increase in stiffness and elongation, vessels tend to increase in cross-sectional size with age⁷. Other factors that may contribute to an overall increase in large artery stiffening (LAS) include an increase in smooth muscle cell stiffness, replacement fibrosis, cross-linking of elastin and collagen, and calcification of the media, which interact with each other and with overarching pathways (such as inflammation and endothelial dysfunction).

LAS results in multiple adverse consequences for the organism, as central arteries lose their ability to cyclically distend and store elastic energy in systole, that can be released to promote blood flow in diastole (the “windkessel” function)⁸. Since less blood can be locally stored in the less distensible aorta from the ejecting left ventricle, more blood has to be transported over longer distances in systole, requiring higher driving pressures and increased energy demands for the heart, while leading to higher diastolic-systolic pressure differences (i.e., widening of pulse pressure). Increased arterial pressures and pulsatility impose higher mechanical stress on the vessels and organs, leading to strong associations between LAS and organ damage in the heart, kidney or brain⁹. With peripheral arteries being less susceptible to stiffening (possibly because of their much lower elastin content), the “stiffness gradient” between the central and peripheral arteries decreases, leading to alterations in arterial hemodynamics and the transmission and reflection of pressure and flow waves in the circulation and affecting the physiological amplification of the pressure pulse from the heart to the periphery^{9, 10}. All in all, LAS is considered as a key biomarker of vascular health, believed to integrate the accumulated effect of aging and exposure of the arterial system to insults and mechano-chemical stressors, and has repeatedly been demonstrated to predict cardiovascular mortality and morbidity above and beyond conventional cardiovascular risk factors^11-13. Given the strong effect of aging on the arterial wall, which can be accelerated or modified by other factors or insults, the concept of “early vascular aging” (EVA) has been introduced, indicating subjects exhibiting a more rapid decline in arterial distensibility than can be expected on the basis of their chronologic age¹⁴.

Although LAS is a highly clinically-relevant and powerful prognostic biomarker, it is remarkable that it has yet be incorporated into routine clinical practice and to take a more prominent position in clinical guidelines published by various professional cardiovascular societies. An important reason, besides the current non-existence of proven therapies to “rejuvenate” arteries, may be the plethora of methods and devices claiming to measure “arterial stiffness” in humans. It is the aim of this brief review to provide a concise overview and classification of methods in use, addressing their applicability, strengths, limitations and potential pitfalls. Rather than focusing on devices, we will mainly address principles and techniques that form the basis of various devices. Note also that this review is limited to methods providing measures of “functional stiffness”, incorporating both intrinsic material parameters (e.g. Young elasticity modulus) and geometry (vessel diameter and wall thickness). As such, when referring to stiffness, we refer to functional stiffness and not to intrinsic soft tissue material parameters as one would derive from multi-axial tensile testing on tissue specimens. We will also, deliberately, not address methods that aim to calculate volume or total arterial compliance (via windkessel model-based methods), because although valuable to understand (patho)physiology in individuals, their values are highly dependent on body size and mass and therefore less practical in clinical or translational research.

2. Arterial stiffness and Pulse Wave Velocity (PWV): closely related but not synonyms

For many biomedical professionals, arterial stiffness has virtually become synonymous to PWV, the speed at which the arterial pulse propagates along the arterial wall. It can be shown, theoretically, that in a uniform elastic tube with a lumen cross-sectional area A and filled with a liquid with density ρ, a perturbation to the system propagates as a wave along this tube with a speed (PWV) given by the Bramwell-Hill equation¹⁵:

$$PWV=\sqrt{\frac{A}{\rho }\frac{\partial P}{\partial A}}$$

with ∂A the change in lumen area in response to a change in pressure ∂P. With $\mathrm{D}=\frac{\partial A}{A\partial P}$ representing the distensibility (D) of the tube (defined as the relative change in cross sectional area in response to pressure). Therefore, $PWV=\sqrt{\frac{1}{\rho D}}$. High vessel stiffness (lower D) thus translates into a higher PWV. Assume an artery that increases its luminal diameter by 10% (and its cross-sectional area by 21%) under an incremental pressure load of 40 mmHg (5.3 kPa). With blood density in the order of 1050 kg/m³, the equations above yield a theoretical PWV of 4.91 m/s. At this PWV, the transit time over an arterial segment of 0.5 m is 102 ms. An alternative scenario in which the diameter distension is reduced by half (i.e., 5% distension under the same loading conditions) increases PWV to 7.04 m/s (+43%) and reduces the transit time to 71 ms. It is therefore an important observation that PWV scales to 1/$\sqrt{D}$ such that a decline in distensibility translates in a blunted (square root) increase in PWV. It is also worth mentioning that when the relation between pressure and area is non-linear (as is the case in arteries), PWV depends on operating pressure. As such, when using the foot of the wave as the fiducial point to assess transit time, one obtains PWV at diastolic pressure. Last, the above theoretical relation holds for a uniform elastic tube, free of reflections (presumed infinitely long). The inter-relations between the pressure-area diagram, distensibility and PWV are illustrated in Figure 1.

Despite its limitations, measuring PWV rather than D offers several advantages. When measured over a segment (for example, the aorta), it provides an average value of its stiffness (which may be preferred over a single local value) and avoids the need for simultaneous measurement of pressure and area (or diameter) which is problematic from a methodologic and technical standpoint, particularly for deeper arteries such as the aorta.

3. Direct transit time methods to measure aortic PWV (or its best possible proxy)

In clinical practice, PWV is most commonly calculated as PWV = ΔL/ΔT, with ΔL the distance between two measuring sites, and ΔT the time it takes for the arterial pulse to travel from the proximal to the distal measuring site¹⁶. With the aorta being the major elastic vessel in the body, aortic PWV (or even PWV in segments of the aorta) likely represents the most informative measurement.

Aside from invasive pressure catheter recordings that are only occasionally used in technical validation studies because of their complexity, cost and ethical constraints, Magnetic Resonance Imaging (MRI) is the only imaging modality to provide, within one examination, the 3D aorta and path length (anatomical imaging) and transit times (phase contrast sequences), both needed to assess PWV¹⁷. Caution remains warranted, however, regarding the temporal resolution that is required to resolve the relatively short transit times between imaging planes when measuring through-plane signals. This comment pertains particularly to measurements of PWV over the aortic arch. With a path length of about 15 cm, transit times are in the order of 15-30 ms (for PWV of 6-12 m/s) or less, requiring a temporal resolution approaching 5-10 ms, which has not been accomplished in most published studies¹⁸. In addition, the cost and logistic complexity does not allow for MRI based PWV measurements in the general population for the time being.

Other than MRI, various transit-time based methods are available that are based on detectable signals that arise from the propagating pulse (pressure, flow or distension) and that can be sampled at a sufficiently high temporal resolution and with discernable fiducial points (systolic upstroke, dicrotic notch) at their respective measuring sites. The most widespread used proxy for aortic PWV is carotid-femoral PWV, with transit times assessed from signals measured at the carotid and femoral arteries¹⁶. These sites provide superficial arteries that are still relatively close to the aorta. Several devices aim for measurements at these locations, including commercially available systems such as Complior¹⁹, Sphygmocor²⁰, Pulsepen²¹ and others, as well as custom-built data acquisition systems (as used in the Framingham¹³ and Asklepios²² population studies). Ideally, measurements are performed simultaneously; ECG-gated sequential measurements are the second-best alternative. In principle, any clinical ultrasound machine with a vascular probe can also be used assuming methodological scrutiny and the development of appropriate software to identify fiducial points and compute time delays with sufficient precision. Less trivial is the assessment of the carotid-femoral path length: (i) intra-arterial distances need to be estimated from body surface measurements, and (ii) the pulse does not directly travel along a single path from the carotid to the femoral measuring site, creating ambiguity. Many methods have been proposed to calculate distance, complicating standardization of the measurement, but the current consensus is to simply calculate PWV using 0.8 times the distance between the carotid and femoral measuring site²³. Carotid-femoral PWV is, no doubt, considered the reference for clinical studies in Europe and the US, due to the availability of a large reference value database acquired from data across Europe²⁴ and studies demonstrating its prognostic power^{12, 25}. An important drawback of carotid-femoral PWV as a phenotype of LAS is, however, that it does not account for the wave travel in the ascending aorta and part of the aortic arch, the section of the aorta that contributes most to the buffering function of the aorta. We repeat that MRI may be most suitable to study aortic arch properties, although also 2D ultrasound holds promise as a widely available technique to assess thoracic aortic PWV (from the aortic root to the diaphragmatic hiatus)⁹.

Over the past decades, several other devices have been developed to record signals at more peripheral locations using cuffs at the brachial, ankle and thigh or photo-plethysmograph devices on the finger and toe²⁶, combining measurements on an upper and lower body location and deriving time delays in arrival on these two sites to obtain “transit times”²⁷. Combined with some distance measurement, these can provide a PWV (e.g. “brachial-ankle PWV²⁸ or finger-toe PWV²⁶). Brachial-ankle PWV is frequently used in Asia, and large studies have demonstrated the prognostic power of this metric to assess risk of cardiovascular disease²⁹. Nonetheless, the very large distance of the measuring sites to the aorta leads to an uninterpretable ambiguity in the path travelled by the wave along elastic and muscular arteries, making it virtually impossible to reconcile how derived values effectively relate to aortic pulse wave velocity. A recent invasive validation study of several devices confirms that, moving away from carotid-to-femoral PWV reduces the correlation and agreement with invasively measured aortic PWV³⁰. The better approach may be the use of methods aiming for heart-femoral PWV, detecting the time delay between closure of the aortic valve (detectable via a microphone) and arrival of the pressure wave picked up by a femoral cuff. The cardio-ankle vascular index (CAVI) can be added to this list but it “extends” on PWV and uses the non-linear relation between diameter and pressure (assumed exponential) for its calculation ³¹. CAVI has the theoretical advantage of providing a pressure-independent stiffness metric (after correcting the original formulation)³², but the arterial path used for measurement of PWV will largely impact its physiological meaning and interpretation. When measured heart-to-ankle, it stretches beyond the aorta and includes elastic and medium-sized muscular arteries, inevitably leading to complex phenotyping, showing only modest correlations with more anatomically relevant measures of aortic stiffness³⁰ .

4. Indirect estimation of aortic PWV from waveform analysis (upper arm cuff-based devices)

Some devices claim to assess pulse wave velocity from a single brachial cuff pressure recording. It is important to understand that it is physically impossible to measure PWV from only a pressure tracing at one single measuring site, and that devices can only provide some estimate that may or may not be based on legitimate underlying models, and involve algorithms trained on datasets to ensure that plausible PWV values are provided by the device, which in itself is not sufficient for validation. One such device is the Arteriograph³³, intrinsically based on the flawed assumption that the arterial tree can be simplified to a uniform tube with a single discrete reflection site at some distance³⁴. Upon inflation of the cuff to supra-systolic blood pressure, the device enforces an absolute reflection site at the upper arm, leading to a specific double peaked pattern; the time delay between some points on the two peaks are then ascribed to the travel time of a wave travelling down the aorta and reflecting at the distal end of the tube. Using a validated one-dimensional arterial network model, Trachet et al. explained the measured double-peaked pressure pattern on the basis of wave propagation and re-reflection in the upper arm arterial segment itself³⁵. Another device on the market is the Mobil-O-Graph. This is a 24hr ambulatory (central) blood pressure monitor, but the device also claims to provide an estimate of aortic PWV³⁶. Recent data demonstrate that the estimate provided by the device is largely driven by the subjects’ age (which is entered into the device) and measured systolic blood pressure, accounting for over 99% of the variance in the estimated PWV (the same variables accounting for only 40% in variance of measured carotid-femoral PWV)³⁷. Furthermore, the device performed poorly in a study comparing device estimates with invasively measured aortic PWV³⁰.

With the booming of artificial intelligence and machine learning techniques, it is to be expected that other “black box” devices and methods will emerge that will, without any doubt, demonstrate an excellent capability to predict trends in PWV in populations, given that devices will have been trained on large datasets that contain these trends. It is, however, important to keep in mind that such devices may no longer be based on measurements, and do no more than translate an integrative measure of any available clinical data into a PWV-value that the algorithm expects on the basis of the data it has been trained on. From a scientific-technical point of view, such devices should intrinsically be less useful for clinical practice and research purposes than actual measurements, as it is doubtful that they would detect those individuals that deviate from the general trend, among whom measurements of PWV may be particularly useful, providing information that may guide therapy above and beyond conventional clinical variables. These considerations also have implications for current guidelines on the validation of methods to assess PWV³⁸, which should go beyond demonstrating that devices can replicate age and blood pressure driven trends in PWV in a representative population sample.

5. Measuring local stiffness or PWV

5.1. Measuring distensibility from pressure and diameter

The above-cited Bramwell-Hill equation indicates that arterial stiffness is fully defined by the relation between intra-arterial pressure and lumen cross-sectional area (or diameter). When pressure and area/diameter are available as waveforms, the pressure-area or pressure-diameter relation can be constructed over the available pressure range, which is most often limited to the physiological pressure range in clinical practice⁹. It is common practice to only consider the diastolic-to systolic changes in pressure and diameter/area (denoted as ΔP and ΔDia or ΔA, respectively), discarding any non-linearity. One can then calculate the distensibility coefficient (DC) and use DC in the Bramwell-Hill equation to calculate one single value for PWV (PWV_DC) rather than a pressure-dependent value:

$$DC=\frac{\Delta A∕A}{\Delta P}=\frac{(2Dia\Delta Dia+\Delta Di{a}^2)∕D^2}{\Delta P}\approx \frac{2\Delta Dia∕Dia}{\Delta P};PW{V}_{DC}=\sqrt{\frac{1}{\rho DC}}$$

DC is best calculated using local blood pressure, which may differ substantially from brachial cuff values due to pulse amplification as it travels from the heart to the periphery. Arterial area and diameter can be obtained using ultrasound for superficial vessels (typically the carotid and femoral arteries); MRI can in principle provide such data all along the systemic and pulmonary vasculature, but the spatial resolution sets limits in practice to the larger vessels where changes in area can still be measured with reasonable accuracy. Obviously, pulmonary artery pressure cannot be obtained in a non-invasive way. Of note is that in ultrasound, some algorithms (especially wall tracking algorithms) may measure the diameter using the media-to-adventitia transitions rather than lumen diameter. This method yields smaller ΔDia and larger Dia, such that DC will be somewhat underestimated.

5.2. Loop methods for measurement of local PWV (without pressure measurement)

In a uniform and infinitely long elastic tube, i.e. in absence of any wave reflection, there is a constant relationship between a given change in pressure (dP) and the associated change in mean flow velocity (dU) which is given by the water hammer equations, generally expressed as³⁹ dP_± = ±ρPWVdU_±

The +/− sign indicates the direction of propagation of the wave, ρ is the density of the liquid inside the tube (about 1050 kg/m³ for blood) and PWV is the local pulse wave velocity. For a tube with a PWV of 5 m/s, a step increase in flow velocity of 1 m/s thus leads to a step increase in pressure of 5250Pa (36.4 mmHg).

With knowledge of simultaneous pressure (pressure catheter or applanation tonometry) and flow velocity (ultrasound, MRI) at a given location, this equation can be used to estimate the “local” PWV⁴⁰ – provided absence of wave reflection. It is a frequently made assumption that early systole is a period where wave reflections are absent (presuming that it takes some time for waves to travel back and forth to any reflection site), and that the water hammer equation can be safely applied. Most often, when plotting the relation between pressure and flow velocity (forming a loop), one can indeed identify a linear section in the loop in early systole, seemingly supporting the reflection-free assumption.

The original formulation using P and U (called the PU-loop method) has been reformulated whereby P and/or U have been replaced by variables that are easier to obtain in a non-invasive way, giving rise to non-invasively applicable methods such as the flow-area method (QA-method⁴¹) and the ln(Dia)U method, Dia being diameter⁴². Especially the QA-loop method is alluring as both variables are fairly easily measured with MRI and can also be applied to the pulmonary artery⁴³.

Unfortunately, these most appealing methods have proven to be susceptible to error induced by wave reflections⁴⁴. Reflections, having an opposite effect on pressure and flow (or velocity), interfere with methods that combine differentially affected signals. Experiments on hydraulic bench set-ups demonstrated that, in the vicinity of reflections of the closed end type, the PU-loop method tends to overestimate PWV (with errors proportional to the strength of the reflection), while the QA and ln(Dia)U loop methods tend to underestimate it⁴⁵. Trends reverse when the nature of the reflection reverses (i.e., reflections of the open type). It is therefore important to keep in mind that while absence of reflection implies a linear relationship between the considered variables, the inverse is not true: a linear relationship between e.g. P and U does not imply absence of reflections. One will indeed find that reported PWV-values estimated using the QA and ln(Dia)U loop method will generally be lower than what one would expect, while the opposite may be true for the PU-loop method.

The problem is that the impact of reflections on the accuracy of the method is variable and differs from one location to the other^{46, 47}. Attempts to correct for the error have been undertaken, but without real applicable success to date, and it is recommended to calculate local PWV using the Bramwell-Hill equation or equivalent expressions combining pressure and diameter, such as the ln(Dia)P method that is not susceptible to reflections with $PWV=\sqrt{\frac{1}{2\rho }\frac{dP}{dln(Dia)}}$ calculated from the slope of the linear segment of the loop constructed from plotting pressure as a function of the natural logarithm of diameter⁴⁸. Interestingly, it has been shown that the method works well with a surrogate pressure waveform, whereby a pressure waveform is obtained from recalibrating and scaling the diameter waveform to brachial diastolic and mean systolic pressure⁴⁸. Clearly, this approach will depend on the accuracy of the non-invasively measured pressures. All of the above cited loop methods also require adequately time-aligned signals.

5.3. Pulse wave imaging and elastography using ultrasound

In recent years, most interesting evolutions in ultrasound imaging have emerged that may open up a whole new toolset for direct measurement of arterial stiffness and local PWV⁴⁹. While conventional ultrasound imaging sequences were too slow to detect the propagation of an arterial pulse along an artery, the use of unfocused plane waves enables measurements with a temporal resolution of several kHz that are thus fast enough to monitor the propagation of a pulse within the artery over the width of the imaging window (taking 8 ms for a wave travelling 4 cm at a speed of 5 m/s). In so-called pulse wave imaging, spatiotemporal plots of wall displacement, velocity or acceleration indeed visualize the propagating wave front, with PWV given by the slope of the propagating wave front⁵⁰. It is important that these techniques capture PWV at a discrete instant in time, and hence at a given pressure level (diastolic pressure when tracking the early systolic wave front; the dicrotic notch pressure when tracking the wavefront generated upon closure of the aortic valve). This may explain why the correlation with other measures such as PWV calculated from distensibility may be low or even non-existent, as the latter accounts to some extent for the degree of “stiffening” as the artery pressurizes and gets stretched, while this is unaccounted for in pulse wave imaging⁵¹.

A final word goes to elastography techniques that are, in fact, aimed to characterize intrinsic material properties rather than functional stiffness. Within the ultrasound community, shear wave elastography for soft tissue characterization is receiving a lot of attention⁵². Although there are some variants, methods rely on the induction of (a) small amplitude mechanical perturbation(s) within the tissue (typically generated using focused energy from the ultrasound probe itself), and tracking the propagation of the generated shear waves, the shear wave speed being proportional to the stiffness of the tissue. With primary efforts targeted towards staging of liver fibrosis or breast tumor characterization, shear wave elastography is being explored for the quantification of stiffness of cardiovascular tissue. With one single measurement taking a few tens of milliseconds, multiple measurements can be done over one heart cycle⁵³. For the heart, this implies that the technique would allow direct measurement of cardiac muscle stiffening (contractility) and de-stiffening (diastolic stiffness), while direct quantification of stiffening of the arteries would become feasible⁵⁴. Methods are still under investigation, and the complex anisotropic and layered nature of the thin walled tissues pose extra challenges, but initial results are highly promising⁵¹. Application is mainly directed towards the carotid artery that would be used as a proxy for LAS. However, the material properties of the carotid wall may not change in tandem with the aortic wall, and the clinical and prognostic value of this approach remains to be established.

6. Concluding remarks

This mini-review is, no doubt, incomplete and focused on those methods that the authors deemed most relevant for the journal readership. We limited ourselves to methods that are either established and widely used, emerging or in research with the potential to emerge. The methods referred to are graphically summarized in Figure 2. In the text, we have tried to indicate strengths and weaknesses of methods and to classify and grade them, highly weighting their scrutiny and purity in quantifying arterial stiffness, rather than focusing on their ease of application or the level at which methods have demonstrated their prognostic and diagnostic potential.

Figure 2:

Visual overview of (the concept of) methods to assess larger artery stiffness discussed in this mini-review. The top row addresses methods aiming to measure PWV, classified as direct and indirect (from a single cuff recoding). The background and label color scale ranges from green (clear phenotype) to red (unclear phenotype) to indicate the author’s qualification of methods, the arrow pointing in the direction of level of recommendation. The bottom row includes methods to assess local stiffness or local pulse wave velocity from a single site recording, using the same background color scale and arrow. The figure displays figure panels adapted from ⁵⁵, ⁵⁰ and ⁵². See Table 1 for accompanying text.

Table 1.

Advantages and disadvantages of methods aiming to measure aortic PWV (top half) and local arterial stiffness or PWV (bottom half), with our appreciation of the method on a five-star scaling scale. Main criteria in the assessment are the method’s accuracy in phenotyping aortic PWV or local stiffness and the physical principles underlying the method. The table directly supports the color scale used in Figure 2.

METHODS AIMING TO MEASURE AORTIC PULSE WAVE VELOCITY
	Aorta PWV		Carotid-femoral PWV	Peripheral cuffs/sensors		Single cuff
	Pressure catheters	Non-invasive imaging (MRI, US)		Brachial-ankle PWV	Finger-toe PWV
Pro	well-specified phenotype when both catheters are positioned within the aorta; gold standard method for PWV; high temporal resolution	3D anatomy and accurate path length when using MRI; non-invasive; well-specified phenotype	non-invasive; flexibility in signal used (pressure, diameter, flow velocity); best possible proxy for aortic PWV; applicable to large populations; relatively inexpensive	non-invasive and automated; fast; relatively inexpensive	non-invasive and automated; fast; relatively inexpensive	non-invasive, automated, easy and fast, inexpensive
Con	invasive; expensive; confined to cath-lab; only in patients scheduled for catheterization; no direct view on aortic anatomy and position of catheters	expensive; confined to radiology department when using MRI; temporal resolution is limited for MRI	requires trained staff; measurement excludes the ascending aorta and aortic arch; ambiguity in travel path of the wave; path length/distance measurement is not precise	ambiguous phenotype; limited correspondence with aortic PWV	very ambiguous phenotype; very limited correspondence with aortic PWV	no measurement of PWV; devices based on trained algorithms and/or unfounded model assumptions
Our grade

METHODS AIMING TO MEASURE LOCAL STIFFNESS: PULSE WAVE VELOCITY
	Pressure- diameter/area	New ultrasound technologies		Loop methods
		Pulse wave imaging	Shear wave elastography
Pro	non-invasive; robust and direct measurement of local distensibility; directly transferrable into local PWV using Bramwell-Hill equation; possibility to assess pressure-dependency of stiffness	non-invasive; direct assessment of the locally propagating pulse wave; possibility to measure PWV at diastolic (foot) and end-systolic pressure (dicrotic notch) when applied to the carotid artery	non-invasive; direct assessment of material stiffness (shear modulus); measurements can be performed at different pressure levels (stiffening of the artery)	direct assessment of PWV at diastolic pressure; some methods are straightforward to implement using non-invasive techniques (MRI, ultrasound);
Con	combination of two distinct signals and measurements; may be time consuming; ideally simultaneously measured; straightforward on peripheral arteries but more complicated for deeper arteries; requires local pressure measurement	more expensive ultrasound equipment; requires ultrafast imaging; not yet on (all) commercial systems and still in research phase; validation pending; applicable to superficial arteries only	more expensive ultrasound equipment; requires ultrafast imaging; not yet on (all) commercial systems and still in research phase; validation pending; applicable to superficial arteries only; yields material stiffness and may be difficult to link to functional indices as PWV.	methods are susceptible to wave reflections; some methods systematically overestimate, while others underestimate; theoretical conditions underpinning the method are likely never satisfied in vivo; presence of a linear segment in the loop does not exclude the presence of reflections
Our grade

Highlights.

• Mechanistic origin of pressure-dependency of arterial stiffness and relation between stiffness and pulse wave velocity

• Pathophysiological consequences of arterial stiffening

• Overview of methods to assess regional and local arterial stiffness

• (Graphical) appreciation of scrutiny of methods to quantify arterial stiffness

Abbreviations

A

cross-sectional area

CAVI

cardio-ankle vascular index

D

distensibility (D)

DC

distensibility coefficient

Dia

diameter

EVA

early vascular aging

LAS

large artery stiffening

P

pressure

PWV

pulse wave velocity

Q

volume flow

U

flow velocity