Why the study?
Do arterial transfer functions improve the estimation of aortic blood pressure characteristics compared to peripheral waveforms?
Do arterial transfer functions improve the estimation of aortic blood pressure characteristics compared to peripheral waveforms?
Answers are grounded in this paper and cite their sources.
Arterial transfer functions fail to significantly reduce random error in estimating central blood pressure from peripheral waveforms, limiting their clinical utility.
Introduction It is widely recognized that deviating aortic blood pressure parameters (e.g. diastolic, pulse and systolic blood pressure) are important risk factors for cardiovascular disease [1]. However, there are no non-invasive methods to supply these aortic blood pressure characteristics directly. Therefore, worldwide research is aiming to achieve an indirect assessment of the aortic pressure waveform, which subsequently can be evaluated to supply relevant characteristics. A possible option might be the use of arterial transfer functions, which transforms peripheral blood pressure waveforms back to their assumed original aortic shape by correcting for the linear filter properties of the intermediate arterial segment [2,3]. Ideally, the transfer function should be obtained from a large test population to obtain an estimate for the averaged ratio of the spectral characteristics (magnitude and phase) of the aortic and peripheral blood pressure waveforms. The aortic pressure waveform is obtained by a microtip pressure catheter while the peripheral waveform is assessed non-invasively with a tonometer (e.g. at the radial artery). The merit of the concept of arterial transfer functions should follow from an improved estimate of aortic blood pressure waveform characteristics with a small bias and narrow margins of agreement (mean ± 2 SD, i.e. 95% confidence interval) compared to characteristics derived from the peripheral waveform. The study by Hope et al. [4] performed on 78 subjects, clearly demonstrates that, for some of the characteristics, the bias between peripheral and central waveform has a systematic nature. The use of arterial transfer function reduces this bias but hardly reduces the limits of agreement [4]. The authors conclude that the transfer function might be useful for some characteristics, but that it will not provide additional clinical benefit over the evaluation of the peripheral waveform. These conclusions corroborate the earlier results of Segers et al. [5] concerning the benefit of the radial–carotid transfer function for estimating the carotid augmentation index as a measure of the relative magnitude and timing of peripheral pressure wave reflections. Sources of error in transfer function analysis To appreciate the results, the various factors affecting the shape of the assessed peripheral blood pressure waveform (e.g. at the radial artery) need to be considered. As the pressure waveform travels from the heart to the periphery, its pulsatile nature is cushioned by the site-dependent compliance of arteries, predominantly that of the ascending and descending aorta [6]. The compliance of central arterial segments varies considerably over the population and is modulated by systemic pressure and age. On the other hand, the compliance of the brachial artery is independent of age and exhibits an intersubject variability of the order of 10% [7]. In the lower frequency range, the transfer function mainly depends on the combined effect of aorta and brachial compliance whereas, for higher frequencies, the inertia gains influence. Moreover, at the site of pressure assessment, the waveform will be modified by both the pressure dependent filter characteristics of the passed arterial segments and by reflections from the distal arterial bed. The latter depend not only on the relative changes in haemodynamic impedance of arterial segments (e.g. at bifurcations), but also on the relative magnitude of the peripheral resistance, which may vary with recording and psychological conditions. Peripheral reflections introduce another source for intrasubject and intersubject variability. The reflected pressure waves travel in the opposite direction and will arrive in the aorta in late systole, modifying the local pressure waveform. However, at the radial artery, a possible reflection cannot be discerned from the incident pressure wave traveling from the aorta to the periphery and, accordingly, will be treated the same way as in the forward wave. Consequently, the application of the transfer function to the radial artery pressure waveform to estimate the aortic pressure waveform causes the reflection to appear earlier in the cardiac cycle than is to be expected from the round trip transit time. Comparison of the reconstructed and the invasively assessed pressure waveform will elicit the resulting discrepancy. Usually, the peripheral pressure waveform is assessed with a tonometer, partly compressing the artery of interest against a solid background structure [8]. For a truthful registration, the tonometer should be maintained in a stable position, independent of the arterial pressure. However, it will be difficult to maintain the probe position steady throughout the cardiac cycle. Especially in late systole, in reaction to the peak systole, the probe position might be forced outwards, causing a temporary and variable underestimation of the local pressure. The pressure bounce explains the trailing of the diameter with respect to pressure in late systole and early diastole. The modification of the assessed pressure waveform will thus depend on the subject, due to local pulse pressure, and on the ability of the operator to keep the tonometer probe steady. A possible way to reduce, or even to eliminate, this recording artifact is to use a fixture that stabilizes the probe position with respect to the radial artery. This may also reduce failure to acquire a reliable pressure waveform, as reported by Chen et al. [2]. The compression of the artery by the probe will also cause wave reflections. This will produce an operator-dependent reduction in the pulse pressure as sensed by the tonometer. The calibration and scaling of the diastolic and pulse pressure, according to the brachial diastolic and mean pressure, correct for the underestimation. Conclusions When comparing the peripherally assessed pressure waveform with the directly measured or reconstructed aortic waveform, it should be realized that, in this study, the radial waveform was scaled using the diastolic and mean pressure derived from the original aortic waveform rather than those from non-invasively measured brachial pressure. Hope et al. [4] acknowledge that this will circumvent a possible source of error and will result in a lower bias and narrower margins of error. As expected, the systolic pressure in the radial artery was 12.6 mmHg (limits of agreement ±23 mmHg) higher than in the aorta. The application of the transfer function converted this to an underestimation of 3 mmHg (±14.6 mmHg) and 8 mmHg (±5 mmHg) for the systolic and diastolic pressure, respectively, while the pulse pressure was overestimated with 5 mmHg (±16.2 mmHg). From the results for the systolic pressure, it can be concluded that the transfer function indeed suppresses the scatter but, nevertheless, a large margin of error remains. The same conclusion can be made for the time to peak pressure (±73 versus ±49 ms), although there was no difference for the systolic and diastolic pressure integral. The results for central blood pressures, as reported above, are less favourable than those obtained in a similar study [9], considering that, in the latter study, peripheral blood pressures were used for calibration and that the transfer function was tested prospectively. In both studies, the random error in the estimate for systolic pressure (±20 mmHg 95% confidence interval) was comparable. Neither study provided direct evidence about the correlation of the errors in systolic and diastolic pressure. However, in the study by Hope et al. [4], the variance of the pulse pressure approximates the sum of the variances of systolic and diastolic pressure, indicating a poor cross-correlation. Whether it is allowed to remove the systematic error (bias in the estimate) by a simple subtraction remains unclear [4], because the bias may be group dependent. Consequently, the procedure to allocate a subject to a group will largely determine the outcome. Sophisticated signal processing to improve the estimate for the transfer function might further reduce the bias, but this will hardly affect the random error. The disappointing results could be explained by the frequency characteristics of the water-filled pressure catheter. However, if the relevant upper frequency components exceed the spectral noise level, the filter characteristics of the pressure catheter will be automatically corrected for by the transfer function. Indeed, the noise level may have been a problem, as suggested by the wide range of individual transfer function components for frequencies beyond 5 Hz, although the large group size will attenuate this considerably. Aortic pressure wave reconstruction was disastrous for the augmentation index, corroborating earlier results by Segers et al. [5] and supporting the conclusion that the transfer function does not offer additional benefits. For all of the above parameters, the random error was independent of the mean value. Alternatives The pessimistic conclusion that the transfer function hardly offers additional benefits, as expressed by Hope et al. [4], prompts the question of the possible alternatives. The tonometer pressure recording can be made at the common carotid artery (CCA), where it may be assumed that its pressure characteristics closely resemble that of the ascending aorta. The CCA has no solid background, introducing additional recording errors in the instantaneous amplitude. However, this does not affect timing or relative parameters such as the augmentation index. If the absolute values of the pressure are of interest, then evaluating the distension waveform of the CCA might be considered, as observed with the Wall Track System, after scaling it in accordance with the observed difference between brachial mean and diastolic pressure [10]. The scatter in the eventual results is strongly related to the error in brachial pressure measurements [9], which should be minimized by repeated measurements. The above scaling protocol is only valid if the diameter changes linearly with the pressure over the operating range. A more sophisticated calibration protocol utilizes the known relationship between pressure and arterial cross-section [11], modified by a subject-dependent parameter (e.g. the rigidity index or distensibility), which is estimated in a recursive way. The procedure is also suitable for the aortic arch [12]. The calibrated distension waveform then supplies all individual parameters of interest, without the need for correction for group averages.
No takes yet. Share an insight, caveat, or question.
Hoeks et al. (2003) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: