Rate of blood pressure changes assessed by 24 h ambulatory blood pressure monitoring: another meaningful index of blood pressure variability?

Data collected over the past 60 years, thanks to the introduction of ambulatory blood pressure monitoring (ABPM) techniques, have clearly shown that blood pressure (BP) is a highly variable parameter [1]. Such a solid awareness, however, has not found a suitable translation into clinical practice yet, and the evaluation of BP-related cardiovascular risk is still largely confined, in the best case, to the assessment of average BP values from a series of measurements (obtained by ambulatory or home BPM), while most recommendations related to hypertension management are still based on isolated clinic measurements [2,3]. This despite the availability of growing evidence that BP variability is an independent predictor of organ damage and cardiovascular events [4–9]. A few reasons may explain the difficulties in introducing BP variability as a routine clinical tool. One, the complex nature of the mechanisms underlying an increase in BP excursions makes the clinical interpretation and practical use of estimates of BP variability difficult. Indeed, an increased amplitude and frequency of BP fluctuations may reflect alterations of different factors involved in cardiovascular control such as impairment of autonomic or hormonal regulation, renal dysfunction, and increased arterial stiffness [10]. Two, the lack of solid evidence supporting the efficacy of any specific treatment in reducing BP variability is associated with the even more important lack of evidence, except for a few animal studies [11], that its reduction may be actually beneficial in reducing cardiovascular risk. Three, the fact that simple conventional estimates of BP variability based on the calculation of 24 h, daytime or night-time standard deviation (SD) of either systolic or diastolic average BP, even if independently related to outcome, appear to have a rather limited predictive power compared to other more traditional risk factors [4–7,12]. Four, the awareness that the SD of BP average values, computed over a given time window, is a very rough indicator of BP dynamics, particularly if obtained with a discontinuous technique such as ABPM. This in spite of the previous demonstration that when the interval between measurements does not exceed 15 min, SD values computed from discontinuous BP monitoring may reflect with reasonable accuracy the ‘true’ SD obtained by analysis of beat-by-beat BP recordings [13]. The latter two problems have indeed stimulated the search of more refined BP variability estimates, with the aim of deriving parameters possibly characterized by a superior predictive power. One such approach is based on the observation that although an increased overall 24 h BP variability may be associated with adverse outcome, the magnitude of 24 h SD of BP is significantly affected by the contribution of day-night BP excursions in individuals with larger diurnal BP changes (i.e. in those defined as ‘dippers’, commonly believed to have a better prognosis) [14]. A few techniques have been proposed in order to mathematically remove any potential interference of the magnitude of the day-night BP difference, regarded as a ‘beneficial’ component of 24 h BP variability, with the assessment of overall 24 h BP variations, an interference that may prevent calculation of 24 h BP SD from reliably reflecting short-term BP fluctuations occurring during the day and the night. Aim of these methods, while excluding the contribution of day-night BP changes to overall 24 h BP variability, is to maintain at the same time the maximum amount of information on short-term BP changes occurring over the entire 24-h period, which are considered to represent the potentially ‘damaging’ component among the BP fluctuation occurring over a 24 h time. These methods, which allow the quantification of short-term BP changes during both the day and night, have been developed to overcome the limitations of previous approaches which, in order to disregard circadian BP changes, limited their focus only on the estimate of daytime [5] or of night-time BP SD [6]. One of these newer techniques is based on the simple calculation of the average of day and night BP SDs, weighted for the duration of the corresponding subperiod. A more sophisticated technique, yielding, however, similar results, is based on removal of the two major cyclic components of 24 h BP variability, identified through spectral analysis and respectively related to the day-night cycle and to the afternoon siesta, and thus focusing on the so-called ‘residual’ faster BP fluctuations. The prognostic relevance of the ‘residual’ 24 h BP variability estimates, deprived of the contribution of night-time dipping, obtained with these techniques was confirmed in outcome studies [8,9,14]. Other studies have attempted to further refine the analysis of short-term BP variability within a 24 h recording period, by focusing on the size of changes occurring over very short time intervals (i.e. between consecutive readings), rather than on measures of global variability. Such attempts are justified by our previous finding of a different speed of BP changes between normotensive and hypertensive individuals, with an increased steepness of SBP ramps, identified by means of intra-arterial beat-by-beat 24 h monitoring, in hypertensive as compared with individuals with normal BP levels [15]. They are also justified by the fact that although consistent shifts in BP levels over several hours may increase 24 h BP SD (similarly to what done by day-night BP changes), they may not reflect ‘true’ short-term BP variability. One such method, named ‘average real variability’ (ARV) assessment, is based on the estimate of the average of absolute changes in BP between consecutive readings and has been shown to be prognostically superior to 24 h or awake SD [9,16]. Along the same line, another index of BP variability, referred to as ‘time rate of BP variation’, was proposed by Zakopoulos et al.[17] and was defined as the first derivative of the BP values against time. This index is calculated as the mean of the absolute ratios of the differences between successive BP measures and the time (in minutes) between them. This parameter is aimed at quantifying how fast and in which direction SBP values change and, thereby, is claimed to offer an insight into how steep these changes are. Previous work of the same group showed that patients with increased left ventricular mass and impaired renal function presented higher rate of SBP variation values [18,19]. The paper by Manios et al.[20], published in the current issue of the Journal of Hypertension, adds further information on the clinical relevance of this method, by reporting the results of a study on the relationship between time rate of BP variation and the extent of coronary atherosclerosis. The main findings of this study, performed in normotensive individuals undergoing coronary angiography, are that the time rate of BP variation is higher in patients with coronary atherosclerotic lesions compared with controls with normal angiography result and that it correlates independently with the severity of these lesions assessed by Gensini score, which integrates the degree and importance of stenoses in coronary arteries and appears to have independent prognostic significance [21]. The ‘time rate of BP variation’ seems at the first glance an interesting index to assess short-term BP variability. In fact, conceptually, it might be thought to reflect the steepness of rapid changes in BP. However, one has to remember that, given the discontinuous nature of ABPM used for the determination of this index, no information can be obtained on the BP changes occurring between subsequent measurements (spaced at 15-min intervals in the study by Manios et al.[20]). As important BP oscillations are known to occur between minutes or even between seconds, the attempt to use discontinuous ABPM to assess rapid BP changes is unavoidably limited by the low frequency at which automated measures are scheduled, which represents a typical example of undersampling of a highly variable biological signal. In other words, the BP difference between two consecutive measurements taken 15 min apart seems to be unable, in the vast majority of cases, to reflect a consistent, ramp-like, BP trend (in which measurement of slope may be of interest), but is more likely an expression of the casual sampling of more or less random BP oscillations occurring over the time interval between scheduled measurements. Therefore, the concept of ‘time rate’ might be misleading, as the parameter so obtained reflects the overall instability of BP rather than the rate or speed of short-term BP changes. In fact, based on how this index is computed, one may expect that it may heavily depend on the time interval between measurements, although an accurate estimate of ‘true’ steepness of BP changes should reflect beat-by-beat changes and, thus, would require a sampling interval no longer than the beat-to-beat interval. These considerations, however, need to be verified and supported by experimental data. Given that the intervals between measurements that were considered to estimate the index used by Manios et al. are fixed [20] and do not have by themselves any physiological meaning, the ‘time rate’ approach seems to differ little from ARV indices (see above) [9,16] in terms of the information it provides while, compared with the former, has the disadvantage of being potentially influenced by the length of sampling interval. This disadvantage can be less relevant in the context of good quality ABPM recordings, with measurements obtained at standardized regular intervals, but it may become more important when one needs to compare recordings obtained at different frequency of measurements or in the presence of numerous artefacts (which artificially lengthen the average interval between valid measurements and, thus, may lead to an underestimation of rate of BP variation in some individuals). Time rate of BP variation obtained with a standardized ABPM methodology may, thus, be considered as an additional interesting estimate of overall BP variability, likely to be independent of day–night BP fluctuations, but because of the abovementioned pitfalls, it should only be interpreted as such. Keeping this in mind, the study by Manios et al.[20] should be regarded as providing a further indication that BP variability is associated with adverse changes in cardiovascular system, in this case with the extent of coronary atherosclerosis. Although the hypothesis that BP variability favors the development of lesions in coronary arteries is appealing, the cross-sectional nature of the study makes any direct inference on the cause–effect relationships between time rate of BP variation and coronary atherosclerosis impossible, as duly acknowledged by the authors in their paper [20]. In fact, alternative explanations may be proposed. One of them is the possibility that, rather than representing a consequence, coronary stenoses contribute to an increased BP variability. Although such hypothesis appears rather unlikely, one might argue that myocardial ischemia may impair the cardiac response to reflex stimuli aimed at stabilizing BP levels (e.g. autonomic nervous inputs) leading to an increased BP variability. More likely, coronary atherosclerosis and BP variability might both reflect other underlying pathophysiological factors, not fully accounted for by the multivariate analysis of this study. For instance, in case of occurrence of factors like diabetes or smoking, whose frequency differed significantly between patients with coronary disease and controls, and which are known to be associated with increased BP variability [22,23], a conventional statistical approach may not be sufficient to completely rule out their confounding effect (undoubtedly present, as shown in multivariate analysis). Other independent factors, which might contribute both to coronary atherosclerosis and to increased BP variability, include arterial stiffness and impaired autonomic cardiovascular regulation. In conclusion, notwithstanding the above methodological problems, the study by Manios et al.[20] should be regarded as a useful additional step toward a better understanding of the characteristics of BP variability and of its clinical correlates. Further research in this field is necessary, however, to provide a satisfactory answer to a number of still pending issues. These include first of all the definition of what the optimal (from both a methodological and a prognostic point of view) index quantifying BP variability might be (Table 1). The time rate of BP variation during a 24 h ABPM may certainly provide interesting information, but it does not fulfil all the requirements that should characterize such an ‘ideal’ index for the reasons highlighted above. Second, we still do not know what the ‘normal’ level of BP variability might be. In other words, we yet need to define the degree of BP variability above which the risk of cardiovascular events significantly increases. Another largely unresolved issue is the identification of the type of therapeutic intervention that might better reduce BP variability. Given the well known association between mean BP levels and the degree of BP variability, any time we reduce mean 24 h BP by treatment we also observe a reduction in 24 h SD. However, when we normalize BP variability by the average 24 h BP value, by computing the BP coefficient of variation (24 h SD×100/24 h mean BP), no reduction in normalized BP variability is usually observed in most studies [24–27]. Further insight in this direction might be provided by a more frequent use of the smoothness index, when assessing 24 h BP response to antihypertensive treatment, given the ability of this index to quantify not only the amount of BP reduction obtained by a given treatment over 24 h, but also the more or less satisfactory homogeneity of its hourly distribution throughout the recording period [28]. Finally, the definitive confirmation of the prognostic value of BP variability would need to be provided by outcome studies focusing on whether a therapeutic intervention reducing BP variability also carries additional prognostic benefit, by concomitantly reducing also the rate of cardiovascular events.Table 1: Comparison of the features of different techniques for blood pressure variability assessment with different blood pressure measuring methods