Misconceptions About Arterial Stiffness May Lead to Erroneous Conclusions

The present commentary discusses some frequently observed misconceptions about arterial stiffness. It will address the importance of choosing the right vascular territory, the importance of standardization of the methods, the use of the cutoff value of normality, the correction for blood pressure, and the interpretation of augmentation index. The arterial stiffness of different vascular territories can be measured noninvasively. In contrast to medium sized muscular arteries of the limbs, the stiffness of centrally located more elastic arteries like the aorta and carotid arteries has been shown to predict cardiovascular morbidity and mortality.1 Carotid–femoral pulse wave velocity (cfPWV) mainly measures the stiffness of the aorta and is therefore regularly used as a surrogate for aortic (also called central) stiffness. Many epidemiologic studies using cfPWV have demonstrated the predictive value of aortic stiffness for cardiovascular events.1,2 For this reason cfPWV is considered the “gold standard” method for arterial stiffness. cfPWV has shown to be a robust measure and to have an added value to classical Framingham and SCORE CV risk scores.3 Nowadays also other noninvasive methods have demonstrated predictive value for cardiovascular events. However, direct head-to-head comparisons of their predictive value with cfPWV are scarce or not existing. In addition, most of them measure stiffness of the aorta and medium sized muscular arteries altogether, leading to less specific phenotyping.4 This may be misleading in diseases like migraine where the alteration in arterial stiffness is very likely limited to the peripheral muscular arteries. This may explain why Ikeda et al.5 found an increased arterial stiffness in migraineurs using brachial-ankle pulse wave velocity (baPWV), while contradicting results are reported in this issue of the Journal. Here, Apelbaum et al. published the results of cfPWV in the largest population sample of migraine patients so far (n = 1.355) as part of the large ELSA-Brasil study, and did not find an altered cfPWV compared with controls.6 This is in line with another small study in migraine patients which also did not find an altered stiffness of the aorta but a smaller diameter of the muscular brachial, femoral, and temporal arteries, and not of the elastic carotid artery in migraine patients.7 These observations suggest that the increased baPWV in migraine patients may be only due to an increase in muscular artery stiffness and not of the elastic aorta. Although cfPWV is nowadays often used, regularly misconceptions about its measurement are encountered. Whereas the time delay between the measurement points is given by the algorithm embedded in the devices, the distance to calculate cfPWV has to be entered by the operator. Many different distances have been used to calculate cfPWV (Table 1), each method producing a different cfPWV value. In an attempt to assess the real traveled distance. Huybrechts et al. studied the traveled path length by magnetic resonance imaging (MRI) and found 80% of the full direct distance between the measurement points at the carotid and femoral arteries as the best assessment of the real traveled distance.8 This distance has been accepted in a consensus document as the standard distance to get the standard cfPWV for three reasons: (i) it has been found to be the best assessment of the real traveled distance, (ii) it limits the measurement errors since it may be less influenced by a big abdomen, and (iii) in contrast to subtracted distances—only one distance has to be measured.9 Authors advise to always show the standard cfPWV side-by-side with the observed cfPWV. This not only allows to judge normality in clinical practice (see cutoff value), but also allows comparison between studies.10 Measured distances can be converted to standard distances using Table 1 of the consensus document.9 Conversion factors are presented in Table 1 of this manuscript. In this issue of the Journal Apelbaum et al. applied a less frequently used distance method: from the suprasternal notch to the groin. This distance can decrease measurement errors by not using a subtracted distance, but since the distance tract goes over the abdomen it can be influenced by a big abdomen. For this distance method the use of a caliper has been proposed.2,9 The conversion factor for this distance method has been added to Table 1 (SSN-FA). Comparison of MRI measured path length (reference distance) with different tape measure distances (adapted from ref. 8, 9 and 10) Abbreviations: AA, ascending aorta; FA, femoral artery; CA, carotid artery; SN, sternal notch; SSN, suprasternal notch; UMB, umbilicus. aConversion factor: multiply observed distance by conversion factor to obtain standard distance. Comparison of MRI measured path length (reference distance) with different tape measure distances (adapted from ref. 8, 9 and 10) Abbreviations: AA, ascending aorta; FA, femoral artery; CA, carotid artery; SN, sternal notch; SSN, suprasternal notch; UMB, umbilicus. aConversion factor: multiply observed distance by conversion factor to obtain standard distance. In a scientific statement, the American Heart Association recommends besides the 80% of the full direct carotid–femoral distance also the use of the subtracted distance using the suprasternal notch (see Table 1, (SSN-FA) − (SSN-CA)).2 Authors are not in favor of this recommendation because: (i) the subtracted distance underestimates the real traveled distance by 10%,7 (ii) the cutoff value of 10 m/s is not applicable when using this distance, (iii) it can increase measurement error since two distance measures are needed instead of one, and (iv) recommendation of two different distance methods goes against standardization. A cutoff value of 10 m/s was proposed in the consensus document9 and accepted in the hypertension guidelines11 with values above 10 m/s considered as pathological. This cutoff value is based on the “normal and reference values” study.12 These normal and reference values are all expressed as standard cfPWV using the standard distance method: 80% of the full direct distance between the measurement points at the carotid and femoral arteries. All data in this study coming from different epidemiologic studies using different distance methods have been converted to the standard distance to obtain the standard cfPWV. Thus, applying the cutoff value of 10 m/s for normality of cfPWV is only valid on the standard cfPWV. cfPWV values obtained using other distance methods have to be converted first to the standard cfPWV before this cutoff value can be used. For example, applying the cutoff value to observed cfPWV instead of the standard cfPWV in the study by Apelbaum et al. will erroneously classify a substantial part of the population sample as having an abnormally high cfPWV. Blood pressure is a major determinant of arterial stiffness. Therefore, when comparing arterial stiffness of population samples with different levels of blood pressure, correction for blood pressure is mandatory. Regularly arterial stiffness is erroneously corrected for systolic blood pressure (SBP), probably because SBP, more than diastolic blood pressure (DBP) is determining cardiovascular risk. But by correcting for SBP, arterial stiffness is overcorrected since an increase in arterial stiffness will increase SBP and to a smaller extent decrease DBP. The best way to correct arterial stiffness for blood pressure, is to correct for mean arterial pressure (MAP). To understand this we should not look to blood pressure as a SBP and DBP, but approach it as being composed by a static component and a dynamic component oscillating around this static component. The static component is the MAP, determined by cardiac output (CO) and peripheral resistance (TPR) with MAP = CO × TPR. The peripheral vascular resistance is mainly determined by the small arteries and arterioles. The dynamic component is the pulse pressure (PP), largely determined by stroke volume, arterial volume, and stiffness of the large arteries. PP ≈ SV × stiffness/V. In contrast to PP, MAP is not directly determined by arterial stiffness. Devices are validated for SBP and DBP, but not for MAP. As a consequence, MAP has to be calculated from SBP and DBP. Bos et al. found that in adults MAP is best calculated by the formula: MAP=DBP+0.4×PP⁠.13 In the past augmentation index (AIx) has been advocated as a surrogate of arterial stiffness. Despite reports countering this idea,14,15 this erroneous claim can still be found in recent literature.16 It became clear that the increase in AIx with age is leveling off from the age of 50 years, while at that age the increase of cfPWV with age is accelerating.17 In addition, Lemogoum et al. showed that isoprenaline did not change cfPWV in healthy volunteers, while AIx significantly dropped.15 These studies clearly show that AIx is not a good surrogate of arterial stiffness. This misconception comes from the fact that a higher pulse wave velocity will indeed speed up the forward and backward wave with early wave arrival back to the heart. But AIx is not only depending on the pulse wave velocity but also on cardiac stroke properties and the distance of the reflection points from the heart. Reflections occur at places of changes in vascular impedance. Changes in vascular smooth muscle tone can change vascular impedance. As a result vasodilation will move the reflection points further from the heart and decrease AIx, while vasoconstriction does the opposite. Nowadays AIx usually is seen as a measure of wave reflection, being a rather poor one and not to be used as a proxy for wave reflection magnitude as derived from pressure-flow analyses.18–20 The authors declared no conflict of interest.