What are the real determinants of the ambulatory arterial stiffness index?

Over the last few years, increasing attention has been devoted to the pathophysiological role of large arteries in functionally coupling the left ventricle with peripheral vessels [1]. Particularly, an increasing number of articles have focused on the investigation of the functional properties of the large arteries, through the availability of devices able to allow the noninvasive analysis of arterial pulse waveforms in normal conditions, as well as the assessment of their modifications in a number of diseases [2,3]. Making use of this approach, evidence has been provided that arterial stiffness, assessed through the quantification of aortic pulse–wave velocity (PWV), may be a predictor of cardiovascular organ damage as well as of cardiovascular events [4,5]. On the basis of such evidence, aortic PWV has been included among those factors influencing prognosis in hypertensive patients, the evaluation of which may allow a more precise assessment of cardiovascular risk [6]. In spite of its universally acknowledged pathophysiologic and clinical value, however, aortic PWV has not become yet a routine procedure to assess the level of cardiovascular risk of hypertensive patients in daily clinical practice. This is mainly due to the need of dedicated and often expensive equipment for its assessment, as well as to the requirement of careful training of the personnel in charge of its measurement. These difficulties have stimulated research and development of new technologies to estimate arterial stiffness in humans in an easier manner, based on the obvious consideration that a technique able to assess arterial wall properties, without the need of using specially designed and costly equipment and without the mandatory requirement of carefully trained and highly qualified observers, would allow a much larger diffusion of this assessment in daily practice. The attention of several investigators has, thus, focused on the possibility to use ambulatory blood pressure monitoring (ABPM) as a tool to explore arterial distensibility, although somehow indirectly. On the one side, a number of specially modified devices for ABPM have been designed, which allow information on arterial stiffness and central BP to be obtained through use of special sensors included in the arm cuff of the ABPM device, i.e. devices such as Arteriograph24 (TensioMed Ltd, Budapest, Hungary) and Mobil-O-Graph (I.E.M. GmbH, Stolberg, Germany). On the other side, it has been proposed that information on arterial stiffness could be more simply obtained from the analysis of simultaneously occurring variations in SBP and DBP. More specifically, it has been proposed to consider repeated ABP measurements obtained over 24 h in a given individual, possibly under different circumstances, and to plot SBP against DBP values. The slope of the relationship between SBP and DBP has been proposed as a simple and theoretically intriguing means of investigating large-artery functional properties [7,8]. In principle, a change in DBP in a normal individual is generally accompanied by a parallel change in SBP. However, for a given increase in DBP, SBP might be expected to increase more in a stiffer artery than in a more elastic one. Thus, the increase in SBP for a given increase in DBP (SBP-on-DBP slope) could be related to arterial stiffness. The opposite holds for the increase in DBP accompanying a given increase in SBP (DBP-on-SBP slope), which could be considered as a measure of arterial compliance. On the basis of these considerations, a parameter obtained from 24-h ABPM that is based on the slope of the relation between DBP and SBP changes over 24 h, namely the ambulatory arterial stiffness index (AASI) [7], has been recently proposed and has received great attention from the scientific community. This index is computed as one minus the slope of the linear regression line between DBP and SBP values, and previous articles have shown that AASI correlates well with PWV [7]. AASI has also been shown to have an important prognostic value, being able to predict cardiovascular mortality [9,10] and, particularly, stroke [11]. In spite of the interest raised by this approach, however, the results of a number of recent studies have emphasized that caution is needed in interpreting its physiologic and pathophysiologic background, questioning its actual ability to represent a reliable surrogate for more direct measures of arterial rigidity [12,13]. In fact, it is still controversial to which extent the AASI actually reflects arterial stiffness or whether it is rather majorly affected by other parameters (Table 1). Indeed, a number of previous studies have seriously questioned the ability of AASI to represent an accurate measure of arterial stiffness [12,14–17] and the possibility that this index might be used as an alternative for PWV [18]. These studies have in fact suggested that AASI is largely influenced by peripheral resistance, heart–vessel coupling [15,16] and by the degree of nocturnal BP fall [14]. It has also been suggested that the relation between AASI and arterial stiffness could be explained by the nonlinear behavior of the arterial wall [19,20].TABLE 1: Determinants of the ambulatory arterial stiffness indexIn the present issue of the Journal of Hypertension, Kips et al.[21] provide a further contribution to this ongoing controversy, by exploring the relative importance of possible different determinants of AASI through use of a previously validated computer model of the arterial circulation [22]. Aim of the study by Kips et al. was to explore the determinants of AASI and to assess their relative importance in a strictly controlled setting, with the final goal to investigate to which extent AASI can be used as a measure of arterial stiffness. The authors convincingly show that an increase in heart rate, arterial distensibility or peripheral resistance from 80 to 120% of their ‘initial’ value caused AASI to decrease by 37, 21 or 9%, respectively. Whereas there was no overlap in the ‘true’ arterial distensibility ranges for the three theoretical participants considered by the model, the amount of overlap between the AASI distributions was substantial. On the basis of these data, obtained through a computer model of the left heart and the systemic circulation, Kips et al. conclude that arterial stiffness, vascular resistance and heart rate are the main determinants of the AASI, which seriously limits the use of AASI as a reliable and specific marker of arterial stiffness only, due to the confounding effects of vascular resistance and heart rate on its values. Indeed, also based on previous papers, the determinants of AASI appear to be multifold. Further information in this regard has been recently provided by Gavish et al.[23] through the analysis of three small but specific experimental studies, showing that the SBP-on-DBP slope can be affected also by seasonal changes, salt intake and paced breathing. Paced slow breathing and the summer season were both associated with a reduction in the 24-h SBP-on-DBP regression slope. Further variability in the results was also depending on how data were analyzed, namely on whether the SBP-on-DBP slope-related parameters were calculated using symmetric or asymmetric regression methods. The authors found that indexes based on symmetric regression (leading to the quantification of the so-called ‘symmetric AASI’, i.e. symmetric SBP-on-DBP slope), or on the assessment of an equivalent parameter, the SBP-DBP variability ratio [24], showed a larger and more significant response to interventions or changes in environmental conditions than those based on asymmetric SBP-DBP regression (‘traditional’ AASI) did. Further insights into the actual meaning of an index such as AASI might come from the evaluation of the effects of drug treatment on SBP-on-DBP slope, an issue, however, which has received little attention so far. In 188 hypertensive individuals, treatment with an angiotensin-receptor blocker was associated with a reduction in AASI, whereas patients treated with calcium channel blockers showed no significant changes, despite a similar reduction in BP [25]. In a randomized study, the impact of 1-year treatment with perindopril/indapamide combination on AASI and aortic PWV was not significantly different from that of atenolol [26]. Finally, in a small study carried out in treated hypertensive patients, both asymmetric and symmetric AASI were lower in the individuals with high adherence to treatment as compared with noncompliant participants [27]. In spite of the interest of these observations, however, it is evident that so few and small studies on the impact of treatment on AASI values are unable to cast any further meaningful light on the clinical significance of this index. In spite of all the articles so far available on AASI, a number of important issues yet remain unexplored, and the ongoing debate in this field could be summarized by two important research questions, which are related to, first, how we can best quantify the SBP-on-DBP regression slope and, second, what the actual relationship of AASI with arterial stiffness is. Several different indexes have been proposed to assess the slope of the relationship between SBP and DBP (Table 2). AASI is classically defined as one minus the DBP-on-SBP regression slope, the latter obtained using a standard asymmetric regression of DBP-on-SBP readings during a 24-h ABPM session [7]. Regardless of the actual relationship between this index and arterial stiffness (see below), a number of studies have supported the clinical relevance of AASI, a relevance that is suggested by the cross-sectional correlation of AASI with preclinical target-organ damage [28–30] and by its predictive value for future cardiovascular morbidity and mortality [9–11,31]. However, the superiority of AASI in this regard over more widely accepted surrogate measures of arterial stiffness such as 24-h pulse pressure [32–34] has not been established, yet.TABLE 2: Clinical parameters based on the slope of the relation between SBP and DBPSome important methodological issues concerning AASI need to be taken into account, however [13]. First, AASI is strongly dependent on the degree of nocturnal BP reduction [14,35,36]. By definition, individuals with a low nocturnal BP fall (nondippers) have a narrower range of 24-h DBP values (a range that is the dependent variable in the regression equation that generates AASI). As a consequence, the coefficient of regression B of DBP on SBP decreases, and its complement (AASI, or one minus B) increases [14]. Notably, the relationship of AASI with left ventricular hypertrophy [14] and microalbuminuria [37] becomes nonsignificant after adjustment for nocturnal BP reduction. Second, due to the above relationship between AASI and day–night BP reduction, AASI also depends on the ratio of nighttime over daytime BP readings number, which is determined by the intervals between consecutive measurements during the day and the night [38,39]. Moreover, AASI is intrinsically dependent on the correlation coefficient between SBP and DBP [24]. SBP-on-DBP slope and DBP-on-SBP slope calculated using standard asymmetric regression are reciprocal to each other only when the correlation coefficient between SBP and DBP is one. In all the other cases, use of an asymmetric regression leads to artifactual dependence of AASI on the SBP–DBP correlation coefficient [24]. However, the artifactual dependence of AASI on the correlation between SBP and DBP may be eliminated if AASI is calculated with a symmetric regression model. The resulting ‘symmetric’ AASI has been associated with all-cause mortality better than conventional AASI [40,41]. The method, however, has not been compared with direct measures of arterial stiffness, and its clinical and prognostic relevance awaits further confirmation. Notably, symmetric AASI is mathematically connected to the ratio between the SD of 24-h SBP and the SD of 24-h DBP (defined as BP variability ratio [24]) (Table 2). Home BP monitoring may also provide a reasonably large number of BP readings on which to assess the slope of SBP-on-DBP changes and is cheaper and more universally available than 24 h ABPM. The resulting home ASI has been associated with age and pulse pressure, although such associations were invariably weaker than those observed with AASI [42]. Possible reasons for the inferiority of home ASI as compared with AASI are a lower number of BP readings on which it was usually computed, and the fact that home ASI is based on BP measurements taken at rest and in a standardized position, whereas AASI captures BP changes during a wider spectrum of activities over the 24 h. AASI, the most widely investigated measure of DBP-on-SBP slope, has been related, in the study in which it was proposed, to carotid-to-femoral PWV, a direct measure of aortic stiffness [7]. However, most subsequent studies did not confirm this finding in larger cohorts of individuals [14,18,43,44]. In these studies, the relation between AASI and aortic PWV was weak and, importantly, was affected by or entirely disappeared after adjustment for age and other factors [14,18,43,44]. The physiological significance of AASI is, thus, currently debated. Westerhof et al.[15] suggested that AASI expresses ventriculo-arterial coupling rather than arterial stiffness. Craiem et al.[19] showed that AASI can be accurately predicted by a model, which takes into account the nonlinear behavior of the arterial wall. Gavish [20] proposed that the SBP-on-DBP slope could be related more to the increase in arterial stiffness from diastolic to systolic values (‘systolic stiffening’) than to diastolic stiffness, the latter being typically measured by PWV. Interestingly, it has been known for many decades that advanced age and prevalent atherosclerotic disease are accompanied by an increase in the slope of the change in pulse pressure over the change in diastolic pressure after inhalation of amyl nitrite [45,46]. The resulting parameter, called ‘arterial rigidity index’ [46], was taken as a measure of the change in the slope of the pressure–volume curve (dP/dV) with pressure. Thus, the relationship between SBP and DBP might describe a pathophysiological process that is different from and only partially related to diastolic arterial stiffness as measured by PWV. The need of caution in considering AASI as a reliable index of arterial stiffness is further emphasized by the results of the study by Kips et al.[21], based on their model of arterial circulation, which clearly shows a dependence of AASI not only on arterial stiffness but also on factors such as vascular resistance and heart rate. The results of this study also lend further support to the hypothesis that AASI depends in part on a functional increase in arterial stiffness ongoing from diastolic to systolic BP. However, a few limitations affecting the study by Kips et al. need to be underlined, too. This study is entirely based on a theoretical computer model, which, as any theoretical model, has its advantages and disadvantages. As acknowledged by the authors, the model is only an approximation of real physiological or pathological conditions, and although it was validated [22] by comparison to noninvasively assessed pressure and flow waveforms of young, healthy volunteers, it cannot be completely excluded that some of the observed effects in this study are exaggerated or masked when compared with reality. Furthermore, the model does not explicitly include the microcirculation, with vascular resistance modeled as a Windkessel model approximation at the distal end of each arterial branch. Finally, it should be considered that, although the study by Kips et al. clearly demonstrates that AASI is not an unequivocal marker of arterial stiffness, AASI might nevertheless still be a useful marker of cardiovascular risk, as it was shown in previous articles to predict outcome over and beyond classical risk factors. Additional studies are, thus, needed to better clarify the physiological and pathophysiological correlates of AASI, besides arterial stiffness, as well as the factors responsible for its prognostic value in predicting outcome. ACKNOWLEDGEMENTS Conflicts of interest The authors declare no conflict of interest.