Method of calibration of measurement of central aortic pressure and prediction of all-cause mortality in chronic kidney disease

The classic experiment of Stephen Hales in 1733, in which he obtained a reading of blood pressure by measuring the height of the blood column in a tube inserted in the crural artery of a horse, is commonly used as a starting point of descriptions and discussion related to physiological measurement of blood pressure. In addition to being the first known attempt of a quantitative measure of blood pressure, an important attribute of this experiment is that the quantity is a true measurement of arterial pressure in relation to the reference level of atmospheric pressure. The measurement does not depend on any secondary transduction process of a surrogate parameter and the pressure value is directly proportional to the height of the fluid column in the tube with uniform bore. There is no simpler or more direct measure of arterial pressure. All other methods, both direct and indirect, depend on processes of signal transduction, associated or surrogate parameters and methods of calibration, all with inherent sources of error. The wide-ranging attempts at measuring pressure noninvasively in humans converged to the development of the cuff sphygmomanometer in the late nineteenth century by Riva-Rocci, a device that, together with the stethoscope, has not undergone fundamental changes since its inception, and is among the most ubiquitous devices in clinical medicine. The original auscultatory technique, utilizing the surrogate quantities of the Korotkoff sounds to detect systolic and diastolic pressure, has persisted to become the standard methodology against which other techniques are validated and so has become a de facto standard for regulatory agencies and professional societies [1,2]. The approval of new devices is done by demonstrating ‘substantial equivalence’ with existing regulated devices based on the ausculatory (Korotkoff sounds) technique such that the values are within an accepted range when obtained from a specific number of individuals [1]. The pioneering work of Les Geddes in medical instrumentation during the 1970s resulted in the consolidation of the earlier technique (described by Marey in 1876) of detecting oscillations in the cuff to obtain indirect measurement of mean arterial pressure [3–5]. Due to the unloading of the artery wall by changing transmural pressure by cuff deflation from a supra-systolic pressure, the mean pressure is taken as the pressure in the cuff corresponding to the maximum amplitude of the cuff oscillations [6]. Systolic and diastolic values are then determined by algorithmic procedures applied to the cuff oscillogram [7]. ‘Substantial equivalence’ is then determined in relation to manual ausculatory devices giving systolic and diastolic pressure values, which at some earlier time were in turn compared with direct measurements. These approval procedures enabled oscillometric devices to be used as automated devices, thus facilitating the measurement of blood pressure. However, although the actual pressure measurement in oscillometric devices is mean pressure, the values that have become relevant for assessment of blood pressure as a cardiovascular risk are systolic and diastolic pressure, as those are the values that are validated against the manual (auscultatory) brachial blood pressure measurement. The adoption of device-based mean arterial pressure measurement in clinical medicine would require some system of validation of the measure. Such a validation is made more difficult, given that there is no standard, accurate and noninvasive measure of mean arterial pressure. Because of physical parameters related to cuff size in relation to limb size and shape, tissue compliance, difference in subcutaneous fat distribution, variable and nonuniform tissue compression and arterial stiffness, the shape of the oscillogram during cuff deflation is not consistent, thus producing errors in estimation of mean pressure and consequently systolic and diastolic pressure [7–10]. In addition, many conventional oscillometric devices do not give an explicit measure of mean pressure and, if required, it is computed using approximating formulas from the derived systolic and diastolic pressure values, that is, a procedure exhibiting error accumulation. The study by Wassertheurer and Baumann [11] in this issue of the Journal of Hypertension describes the effects of calibration methodology applied to the volume pulse wave detected in the brachial cuff on the estimated central aortic pressure using a mathematical transfer function. The study assesses noninvasive measurement of central aortic pressure in a cohort of 159 patients with chronic kidney disease (CKD). At the end of a follow-up period of 30–50 months (mean 42 months), there were 13 deaths, nine of which were related to cardiovascular events. The study aimed to assess whether mortality could be predicted by metrics related to brachial or central aortic pressure. The central aortic pressure estimated from the brachial cuff volume wave was calibrated using either brachial systolic and diastolic pressure or mean and diastolic pressure, all measurements obtained by the oscillometric technique. The finding of the study was that all-cause (although mainly cardiovascular) mortality was predicted only by the value of central aortic systolic pressure using the brachial mean and diastolic pressure calibration. The study by Wassertheurer and Baumann [11] presents challenging findings as well as significant limitations. The effect is found in a small number of CKD patients (13 out of 159, 8.2%) with a hazard ratio of 1.027. The finding is also limited to a specific device (Mobil-O-Graph PWA monitor; IEM, Stolberg, Germany) and it is not known whether there are any specific biases in measurement or computation. Device-specific errors can be important in relation to the difference in random errors. Although some devices may show low mean differences between invasive and noninvasive measurements of systolic pressure, it has been shown that the random errors (quantified as standard deviation) can be quite large, of the order of 10 mmHg [12]. In terms of effect of errors in measured and computed quantities, this study was at variance with other studies assessing the effects of calibration. In a validation cohort of 100 patients using invasive measurements, Shih et al.[13] have shown that calibration of a volume plethysmography wave from a brachial cuff using systolic and diastolic values produced a much lower error than when using mean and diastolic pressure for calibration when transfer functions are used to estimate central aortic pressure. It is not possible to uncover the discrepancy, but it has been shown that the measurement errors inherent in brachial systolic and diastolic pressure can reliably predict the errors in central pressure when using a transfer function [14]. The effects of systematic errors have been shown to be highly significant in clinical measurement methodologies and this has highlighted the important difference between correlation and agreement of measured quantities, as formalized by the now classic work of Bland and Altman [15]. An important aspect that is illustrated by the study of Wassertheurer and Baumann [11] is the accumulation of potential ‘noise’ that can increase the effect of both systematic and random errors. If systolic and diastolic pressures are derived variables from oscillograms with variable shapes, the error will accumulate in the secondary derived variable such as central aortic pressure. The explanation as to why the mean pressure and diastolic calibration gives a significant predictive power for central systolic pressure is that even though there are random errors in the mean pressure estimation from the cuff oscillogram, these are smaller than those for systolic pressure [11]. However, with such a small effect and only limited to a specific device, these assertions require further confirmation. To date, mean arterial pressure does not feature in the pressure quantities that are associated with cardiovascular risk and it is not included in guidelines from professional societies for measurement of blood pressure [16]. One explanation for this might be that it is not readily feasible to obtain consistent measurements of mean pressure given the large variability of the oscillogram envelope [17] and so substantial processing requiring advanced algorithms of pattern recognition and neural networks need to be employed [18]. This would presumably affect the procedures required to assess ‘substantial equivalence’ among devices for acceptable noninvasive measurement of mean arterial pressure. An important result of the difference in calibration methodology is the difference in pulse amplitude between central and peripheral sites found in the study by Wassertheurer and Baumann [11]. The systolic and diastolic calibration method gives a higher brachial pulse pressure (average increase of 32%) than central aortic pulse pressure, whereas the mean and diastolic method gives a lower brachial pulse pressure (average decrease of 4%), resulting in a higher central aortic systolic pressure than brachial systolic pressure. This effect is exaggerated at high pressures wherein systolic pressure in the brachial artery is in excess of 10 mmHg lower than central aortic systolic pressure. This latter (potentially nonphysiological) result is presumably due to the effects of error accumulation and distribution, and the fact that mean pressure is only used for the calibration of the central aortic pressure wave and not for the brachial measurement [11]. In addition to illustrating the effects of calibration, the study also shows the potential prognostic superiority of central aortic systolic pressure in comparison to brachial systolic pressure. In the past two decades, there has been an increasing interest in assessing whether central aortic pressure can provide additional information with respect to assessment of cardiovascular risk and treatment and management of hypertension. Although definitive studies are yet to appear that show that treating to central aortic pressure as target values can improve overall outcome, there are significant indications that central aortic pressure is assuming substantial clinical importance [19]. It was shown that using central aortic pressure as a guide for treatment, target brachial values can be achieved with reduced antihypertensive medication [20], and that measurement of central aortic pressure can inform treatment to improve exercise performance in heart failure [21]. The study by Wassertheurer and Baumann [11] complements these recent studies and shows the potential predictive effects of central aortic systolic pressure in CKD. ACKNOWLEDGEMENTS Conflicts of interest There are no conflicts of interest.