Central and 24-h blood pressure: dwarfs standing upon the shoulders of giants?

Abstract

Since Riva-Rocci and Korotkoff gave us the technique for measuring blood pressure over a century ago, we have landed men on the moon, encircled Mars, invented the automobile and aeroplane, and most importantly revolutionised the technology of science with the microchip. Why, we might ask, has medicine ignored scientific evidence for so long and perpetuated a grossly inaccurate measurement technique in both clinical practice and hypertension research? Eoin T. O'Brien, 2009 [1] Introduction Brachial blood pressure (BP) measurements obtained in the physician's office using the traditional mercury sphygmomanometer or, more recently, the more environment friendly oscillometric devices have long been the cornerstones of our current understanding of the relation between BP and cardiovascular risk. Long-term epidemiological studies have unequivocally shown that office-based brachial BP has a strong, independent, graded, log-linear relationship with the risk of a wide array of adverse cardiovascular events and with mortality [2]. Indeed, data from the WHO show that high BP, defined on the basis of office brachial BP measurement, represents the single most important risk factor for mortality on a worldwide basis, with an estimated 8 million deaths attributable to high BP out of 56 million total deaths worldwide in the year 2001 [3]. The fundamental technology of sphygmomanometric BP measurement has changed little since its development by Korotkoff and Riva-Rocci over one century ago. Despite its universally acknowledged clinical and prognostic value, office brachial BP measurement has two major limitations. First, a few BP readings obtained in the physician's office may not be representative of an individual's ‘true’ average pressure during daily life, mainly due to the marked spontaneous random variability of BP measurements and the so-called ‘white-coat effect’ [4]. Second, systolic BP at a given site of the arterial tree may not be representative of the BP values at other sites, and in particular brachial systolic BP may differ substantially from systolic BP measured at the level of the ascending aorta [5]. Although it had not been technically possible to overcome the above limitations for a long time, the last decades have seen an impressive escalation of technological and conceptual progress in the field of BP measurement. Ambulatory vs. office blood pressure By 1966, Sokolow et al.[6] had used a semi-automated device to compare casual office BP with readings taken during awake hours in 124 hypertensive patients and were able to show that the ambulatory BP was a better predictor of hypertensive complications than the casual office BP. The development of more sophisticated, fully automated recording devices made it later possible to perform 24-h ambulatory BP monitoring in routine clinical practice and to describe the complex behavior of BP during daily life, thus addressing random and circadian variability, which could not be properly addressed by office BP. Today, there is general consensus that 24-h ambulatory BP monitoring is a better method for diagnosing hypertension and predicting BP-related cardiovascular risk than conventional office brachial BP measurement. Twenty-four-hour ambulatory BP is more closely correlated than office BP with several markers of target organ damage [7,8] and, more importantly, is a significantly better predictor of cardiovascular and cerebrovascular events, with many studies providing strong evidence to support this belief [9–16]. The prognostic information offered by 24-h BP is independent of, and incremental to, that provided by office BP [10–12,15,17]. Moreover, ambulatory BP has the unique ability to provide information on nocturnal BP and on day–night BP changes, which have been shown to provide independent prognostic information over and above that provided by office-based and daytime BP, both in the general population [18] and in hypertensive patients [19]. Central vs. peripheral blood pressure The periodically oscillating pulse wave generated in the arterial tree by left ventricular contraction undergoes a progressive distortion with distance. Given the fact that large conduit arteries offer no significant resistance to blood flow, mean and diastolic pressures vary little from the aorta to the brachial artery; in contrast, systolic and pulse pressures tend to increase progressively from the aorta to the peripheral arteries. Two main reasons underlie this phenomenon, which is known as amplification. First, as peripheral arteries are usually stiffer than the aorta, the forward pressure wave created by ventricular contraction generates a higher pulse pressure in peripheral than in central arteries. Second, reflection sites are closer to peripheral arteries than to the aorta; thus, reflected waves generally merge with forward waves earlier in peripheral than in central arteries. The earlier arrival of the reflected wave superimposes to the forward wave in early systole rather than in late systole, and further augments the systolic pressure more in periphery than in the aorta. The net result of these two events is that the amplitude of the pressure wave is higher in peripheral than in central arteries. In the last two decades, several devices have been made available that aim at estimating noninvasively central BP from the combination of peripheral BP measurement and carotid or peripheral waveform [20]. The principles underlying the estimation of central from peripheral BP are based on either mathematical models of the arterial path in the upper limb [21–24], or noninvasive recordings of the carotid diameter waveforms [25] or carotid pressure pulse [26]. Due to anatomical closeness, the stress imposed on the heart, kidneys, and the brain is driven more directly by the central aortic pressure than by the peripheral pressure [27]. Central BP has a closer relation to left ventricular mass and concentric geometry [28], and to carotid intima–media thickness and glomerular filtration rate [29], than peripheral BP. Knowledge of central hemodynamics may also have important implications in the proper assessment of antihypertensive drugs. While treatment with β-blockers effectively lowers peripheral systolic BP, central systolic pressure is lowered to a much lesser extent due to a different spectral content of the arterial pressure waveform associated with the lower achieved heart rate [30]. The lower effect on central BP might contribute to explain why β-blockers are inferior to other antihypertensive drug classes in preventing stroke [31] and in determining regression of left ventricular hypertrophy [32–34]. Although there are strong reasons to believe that central BP might be a more robust and physiologically relevant parameter than brachial BP, its prognostic superiority over conventional BP is less well established. Most [29,30,35–39], but not all [40,41], event-based studies have shown that central BP, mainly pulse pressure, is a stronger predictor of cardiovascular outcomes than the corresponding brachial value. In a recent overview of the available longitudinal studies, Vlachopoulos et al.[42] have found that central systolic and pulse pressures are independent predictors of future cardiovascular events and all-cause mortality. However, the predictive ability of central pulse pressure was only marginally and not significantly superior to that of peripheral pulse pressure in a meta-analysis of the five studies that reported relationships of risk of cardiovascular events with both central and peripheral pulse pressure [42]. The present study In the current issue of the Journal, Huang et al. [43] examine the prognostic value of central office and peripheral 24-h BP in 1014 nondiabetic normotensive or untreated hypertensive Asian individuals without overt cardiovascular disease at baseline. Over an average follow-up period of 15 years, 201 all-cause deaths and 55 deaths from cardiovascular causes were reported. All the systolic BP measures that were taken into account – peripheral office, central office, and peripheral 24-h – were found to predict future cardiovascular mortality, both in unadjusted analyses and after considering the effect of some major cardiovascular risk factors (age, sex, smoking, fasting plasma glucose, and the ratio of total to high-density-lipoprotein cholesterol). More importantly, after adjustment for office peripheral systolic BP, both central office systolic BP (hazard ratio, 1.71 for each standard deviation increment; 95% confidence interval, 1.21–2.40) and peripheral 24-h systolic BP (hazard ratio, 2.01; 95% confidence interval, 1.42–2.85) remained significant predictors of cardiovascular mortality. The same trend was observed for pulse pressure, although the excess cardiovascular risk for central office and peripheral 24-h pulse pressure only bordered statistical significance after adjustment for office peripheral pulse pressure. As expected, the various BP measures also predicted all-cause mortality in univariate analyses, although the respective odds ratios were lower than those for cardiovascular mortality. Central office pulse pressure was the only BP measure that predicted all-cause mortality after adjustment for other cardiovascular risk factors in a multivariate model. The authors also compared the prognostic value of central and 24-h BP. After adjustment for central office systolic BP, peripheral 24-h systolic BP still remained predictive of future cardiovascular mortality (hazard ratio, 1.71; 95% confidence interval, 1.16–2.52). No further prognostic value was observed for all-cause mortality. The impact of central and 24-h BP on prognosis had never been compared directly previously. In this regard, the study by Huang et al.[43] is original and conveys new, clinically relevant information, which suggests that both 24-h and central BP might provide comparable additional prognostic value. Nonetheless, the study exhibits several limits and leaves some questions open. First, central office and, respectively, peripheral 24-h systolic BP were significant predictors of cardiovascular mortality in Cox multivariate models in which office peripheral systolic BP was simultaneously included but failed to enter the final models. This is not equivalent to demonstrating that a predictive model which includes central office or peripheral 24-h systolic BP is necessarily better than one including office peripheral pressure. Formal testing of this hypothesis would need comparing the goodness of fit of different models with appropriate statistics. This limitation also applies to the comparison of central and 24-h BP. Despite the fact that peripheral 24-h systolic BP predicted cardiovascular mortality and central office pulse pressure predicted all-cause mortality in multivariate models, the present study cannot establish superiority of one over another BP measure. On the other hand, the limited sample size and the relatively low number of events (only 55 cardiovascular deaths) make it difficult to draw definite conclusions on the incremental contribution of highly correlated variables. Second, in multivariate models, the coefficient estimates may change erratically in response to small changes in the model or the data due to multicollinearity, which can make it difficult to come up with reliable estimates of the individual regression coefficient of the parameter defined as ‘independent’ in the model. This can be a serious problem in studies, such as the one by Huang et al.[43], which involve multiple highly correlated predictor variables. The authors assume an acceptable degree of multicollinearity given that all the examined variance inflation factors, indicating potential multicollinearity problems, range between 1.26 and 3.01. However, although cut-off values of 4–10 have been proposed, variance inflation factor values in the range between 2 and 4 may also be found in the presence of serious multicollinearity [44,45], and in general any ‘rule of thumb’ cut-off value should be put into the context of the effects of other factors that affect the variance of the regression coefficient of the independent variable under scrutiny [46]. This limits the accuracy of studies like the present one in estimating the contribution of individual predictors to cardiovascular risk. In summary, this is the first published study that compares directly the prognostic impact of the three different BP measurement techniques – peripheral office, central office, and peripheral 24-h. The results are in agreement with the previous body of literature, which convincingly demonstrates the superiority of 24-h over office BP [9–17] and suggests the potential superiority of central over brachial BP measurement [42]. The study has not the power to establish which of the two measures, peripheral 24-h BP or central office BP, is prognostically superior. Interestingly, the advantages of the two techniques can be theoretically combined by using devices that now offer the possibility to estimate central BP over the 24 h (Mobil-O-Graph 24 h PWA Monitor; I.E.M. GmbH, Stolberg, Germany). It is time for studies aimed at providing conclusive documentation that strategies based upon lowering of central office and peripheral 24-h BP lead to better care of patients and reduce cardiovascular morbidity and mortality and target-organ damage more effectively than traditional strategies based upon peripheral office BP lowering. Resistance to abandoning the mercury sphygmomanometer as the gold standard for diagnosing hypertension has been quite remarkable. Nevertheless, the invaluable amount of knowledge that has been built upon conventional cuff sphygmomanometry over the last century and the affection for one of the most ubiquitous symbols of the medical profession should not detract us from pursuing biomedical innovation in this field, with the aim of developing better approaches to BP measurement.