JOURNAL OF MAGNETIC RESONANCE IMAGING 21:53–58 (2005)
Original Research
Feasibility of Aortic Pulse Pressure and Pressure Wave Velocity MRI Measurement in Young Adults Eric Laffon, PhD, MD,1,2* Roger Marthan, PhD, MD,2 Michel Montaudon, MD,3 Vale´rie Latrabe, MD,3 Franc¸ois Laurent, PhD, MD,2,3 and Dominique Ducassou, MD1 sound (US) or MRI methods, typically by dividing the length of a particular aortic segment by the related “foot-to-foot” transit time (1,6). Local aortic PWV can also be obtained from the ratio of the systolic-diastolic difference in the aortic cross-sectional area (CSA) to an estimate of the aortic pulse pressure (APP). However, brachial pulse pressure is usually taken as a surrogate for APP, although it is not considered an accurate approximation (7–9). Finally, APP in the descending aorta has been measured in animals with MRI, but it required a previous estimation of the local PWV (10). Therefore, a noninvasive and direct method of estimating APP appears to be of interest, both in the clinical management of cardiovascular disease (3,4,9) and for assessing the central aortic PWV. The aim of this study was thus to examine the feasibility of such a direct assessment of APP in young adults using a novel, noninvasive MRI method. The APP value was derived from pairs of magnitude and phase images provided by velocity-encoded MR imaging in the ascending aorta with a 30-msec temporal resolution. Images were acquired between the onset of the blood flow in the aorta and the time of maximum blood flow. Furthermore, two different estimates of ascending aortic PWV were obtained. The first can be considered an improvement on the recently published method of Vullie´moz et al (5), whereas the second makes direct use of the APP estimate. Each estimate was compared to the other and to data from the literature. Finally, limitations, as well as possible improvements in APP and PWV assessment, are discussed.
Purpose: To investigate the feasibility of assessing, noninvasively, aortic pulse pressure (APP) and pulse wave velocity (PWV) in the ascending aorta of young adults by means of velocity-encoded magnetic resonance (MR) imaging. Materials and Methods: In a series of 11 healthy volunteers, velocity-encoded MR imaging provided pairs of magnitude and phase-contrast images. Blood flow velocity and aortic cross-sectional area (CSA) were determined with a 30-msec temporal resolution. A model analysis revealed that variation in aortic CSA and in maximal blood flow velocity throughout systole could be used to estimate APP and, hence, to derive PWV by means of two different methods. Results: Mean ⫾ SD values of the APP for the series were 54.2 ⫾ 16.4 mmHg (range 32.2– 84.1 mmHg). The ascending aortic PWV mean ⫾ SD values were 5.03 ⫾ 1.10 m/second and 5.37 ⫾ 1.23 m/second according to the two methods, and both estimates were not significantly different (95% confidence level). Conclusion: These results are in agreement with previously published data, suggesting that APP and PWV can be determined, noninvasively, in young adults using MRI. Key Words: velocity-encoded MR imaging; aortic pulse pressure; aortic pulse wave velocity; local aortic compliance; local aortic distensibility J. Magn. Reson. Imaging 2005;21:53–58. © 2004 Wiley-Liss, Inc.
AORTIC PULSE WAVE VELOCITY (PWV), a classic index of aortic stiffness, is considered to be one of the strongest predictors of cardiovascular mortality (1– 4). Recently, Vullie´moz et al (5) proposed a direct, noninvasive magnetic resonance imaging (MRI) method to evaluate the ascending aortic PWV. Alternatively, direct estimation of aortic PWV can be obtained using ultra-
THEORY In the ascending aorta of young normotensive adults, it has been shown that the pressure wave is unidirectional and reflectionless between the onset of the blood flow and the time of maximum blood flow (11,12). Under these mandatory conditions, one can consider the movement of an infinitesimally thin blood slice element in a vessel experiencing a pressure wave (Fig. 1a), assuming that gravity acts perpendicular to the vessel’s axis, and so plays no role. It can then be shown that, when the mass conservation and momentum conservation principles are applied to this slice, the following relationship can be established between the pressure
1 Service de Me´decine Nucle´aire, Hoˆpital du Haut-Le´veˆque, Pessac, France. 2 Laboratoire de Physiologie Cellulaire Respiratoire, Bordeaux, France. 3 Service de Radiologie, Hoˆpital du Haut-Le´veˆque, Pessac, France. *Address reprint requests to: E.L., Service de Me´decine Nucle´aire, Hoˆpital du Haut-Le´veˆque, 33604 Pessac, France. E-mail address:
[email protected] Received March 22, 2004; Accepted September 28, 2004. DOI 10.1002/jmri.20227 Published online in Wiley InterScience (www.interscience.wiley.com).
© 2004 Wiley-Liss, Inc.
53
54
Laffon et al.
Figure 1. a: An infinitesimally thin, flat slice of a perfect liquid in a vessel whose CSA is S. b: Poiseuille’s flow profile of a real (viscous) liquid (dotted curve). The most rapid liquid slice elements are axial and flat.
Equation [4] indicates that the expression of ⌬P involves a term .(⌬U)2, which is related to the contractile performance of the ventricle, and another term, S/⌬S, which is related to the physical properties of the vessel wall. However, it should be noted that Eq. [1], and consequently Eq. [4], are valid only when: 1) the blood is assumed to be a perfect fluid (i.e., frictionless) and hence, the blood flow velocity profile is flat (Fig. 1a); and 2) ⌬P/P is small, which allows linearity assumptions (13). Both conditions are met during early systole, as assumed by Vullie´moz et al. (5). The method we propose considers that Eq. [1] and Eq. [4] still hold between the onset of the blood flow in the aorta and the time of maximum blood flow for the flat part of a flow profile, which involves the blood slices that move the most rapidly. Figure 1b shows this flat part in the theoretical case of a Poiseuille parabolic flow profile. Furthermore, it is considered that ⌬/, the relative increase in CSA of the most rapid liquid slice elements (Fig. 1b), is equal to ⌬S/S, the relative increase in the whole vessel CSA, i.e., all parts of the slice sketched in Fig. 1b undergo the same relative increase in CSA, whatever their velocity. Finally, the method we propose also considers that ⌬P/P is low between two measurement steps separated by 30 msec. Therefore, measurement of variation in aortic CSA and the highest blood flow velocity value (Umax) within the aortic lumen, between two successive pairs of MR magnitude and phase-contrast images, enables an estimate of the increase in pressure between these pairs. Equation [4] becomes: ⌬P ⫽
variation related to the pressure wave (⌬P) and the induced variation of blood flow rate (⌬Q). Hence, the relation to the velocity of the blood slice element (⌬U) follows (13): ⌬P ⫽
䡠 PWV 䡠 ⌬Q ⫽ 䡠 PWV 䡠 ⌬U S
(1)
where S is the vessel cross-sectional area (CSA), PWV is the pressure wave velocity value, is the blood density, and ⌬Q ⫽ S ⌬U. Furthermore, PWV is also given by (13,14):
冉 冊
S PWV ⫽ 䡠C
1/2
(2)
where C is the local vessel compliance, defined as: C⫽
⌬S ⌬P
(3)
and where ⌬S is the increase in vessel CSA and ⌬P the pressure increase. Eliminating PWV from Eq. [1] and substituting Eq. [3] for C yields the following expression for ⌬P: 䡠 共⌬U兲 2 䡠 S ⌬P ⫽ ⌬S
(4)
䡠 共⌬U max 兲 2 䡠 䡠 共⌬U max 兲 2 䡠 S ⫽ ⌬ ⌬S
(5)
From several successive calculations of ⌬P using 30msec steps between the onset of the blood flow in the aorta and the time of maximum blood flow, the sum of the whole pressure increases allows us to estimate APP. It should be noted that the use in Eq. [5] of Smax and Smin (maximal and minimal aortic CSA, respectively), and the Umax value at the systolic peak “⌬UmaxSys” (i.e., when the step-by-step approach is not used), leads to an erroneous value for APP (APP⬘), since ⌬P/P is too great (Eq. [4]): APP⬘ ⫽
䡠 共⌬U maxSys 兲 2 䡠 S min S max ⫺ S min
(6)
Moreover, two methods are available to derive a mean value of the ascending aortic PWV. The first is very similar to that of Vullie´moz, except that Umax replaces the mean flow across the aortic CSA. Combining Eq.[1] and Eq.[4] (corrected for ⌬Umax) yields the following expression for PWV: PWV ⫽
⌬U max 䡠 S ⌬S
(7)
Therefore, the mean value of the ascending aortic PWV can be obtained using a linear fit to the graph of ⌬Umax vs. ⌬S/S.
Aortic Pulse Pressure With MRI
55
Figure 2. Velocity-encoded MRI applied perpendicular to the axis of the ascending aorta, at the level of the pulmonary artery bifurcation, in one volunteer. The magnitude (upper) and the correlated phasecontrast images (lower) allowed measurement of variations in aortic CSA and blood flow velocity, respectively, with a 30-msec temporal resolution.
The second PWV estimation uses the APP estimate, assuming (Eq. [3]): C⫽
⌬S S max ⫺ S min ⫽ ⌬P APP
(8)
PWV is then alternatively given by Eq. [2]: PWV ⫽
冉
S min 䡠 APP 䡠 共S max ⫺ S min 兲
冊
1/2
(9)
It should be noted that Eq. [9] still assumes that the blood should be considered a perfect fluid. MATERIALS AND METHODS Subjects The feasibility of the proposed method was assessed in a series of 11 healthy young adults (nine men and two women, age range 21–30 years, mean 24.5 years). An Ethics Committee of this university approved the study and informed consent was obtained from all the subjects. None had any history of respiratory or cardiac disease, as assessed by a questionnaire and a medical examination by a physician that included a measurement of the arterial pressure using a sphygmomanometer. This brachial pulse pressure measurement was not obtained in the course of the MR acquisition. MRI and Data Processing Experiments were performed with a 1-T Magnetom Expert Imager (Siemens, Erlangen, Germany). A T1weighted turbo fast low-angle shot (FLASH) sequence, with a dark-blood preparatory scheme, was used in a
sagittal oblique orientation involving the thoracic aorta. Then, a slice was selected perpendicular to the ascending aorta, at the level of the pulmonary artery bifurcation, in order to perform the phase mapping sequence. The shortest temporal resolution provided by the manufacturer was 30 msec, which was previously validated (14). The sequence was electrocardiogram (ECG)-triggered, with repetition time 30 msec, echo time 6 msec (5,10), and flip angle 30 degrees. The field of view size was 338 ⫻ 450 mm, the matrix size 192 ⫻ 256 pixels (pixel size 1.76 ⫻ 1.76 mm2), the slice thickness 10 mm (6), and the encoding velocity 1.50 m/second. The flow quantification software displayed successive pairs of magnitude and phase-contrast images, separated by 30 msec, on the screen (Fig. 2). The time duration from the onset of the blood flow in the aorta to the time of maximum blood flow was 120 – 150 msec, and this allowed collection of four or five such image pairs. The aortic CSA was manually outlined four times in each magnitude image of each timeframe, and therefore a CSA mean value (⫾ SD) was obtained for each time-frame. From each magnitude/ phase-contrast image pair, the corresponding mean value of blood flow velocity was also derived. However, in the present work, we used the maximal blood flow velocity, which was measured in the phase-contrast image of each time-frame. Tuning the windowing of the phase-contrast image allowed identification of the highest velocity area within the vessel lumen. A circular region of interest (ROI) of area 0.06 – 0.24 cm2 was then used four times in this highest velocity area (of irregular shape), and the Umax mean value (⫾ SD) was obtained. Four or five Umax values between the onset of the blood flow in the aorta and the time of maximum blood flow were acquired.
56
Laffon et al.
Table 1 Data for Each Volunteer Volunteer
PWV (Eq. [7]) (m/second)
1 2 3 4 5 6 7 8 9 10 11 Mean SD
4.15 5.00 6.32 4.34 5.11 4.47 7.58 4.48 4.50 3.83 5.53 5.03 1.10
r 0.98 0.99 0.97 0.99 0.95 0.96 0.99 0.99 0.99 0.97 0.99
PWV (Eq. [9]) (m/second)
APP (mmHg)
APP⬘ (mmHg)
4.29 5.65 7.12 4.64 6.42 5.03 7.82 4.88 4.55 4.26 4.46 5.37 1.23
49.2 59.6 84.1 50.1 75.4 51.9 70.4 40.9 32.2 41.4 40.9 54.2 16.4
30.7 54.9 51.6 43.0 35.4 27.8 54.1 32.5 30.1 30.9 24.8 37.8 11.1
PWV ⫽ pressure wave velocity, r ⫽ correlation coefficient in the graph Umax vs. ⌬S/S, APP and APP⬘ ⫽ aortic pulse pressure from Eq. [5] and Eq. [6], respectively.
A first mandatory condition to implement the method was that the phase delay between the time of the maximum blood flow and that of the maximum CSA (hence pressure) did not exceed 30 msec (i.e., the temporal resolution of the experiments), because the method required that the pressure wave could be considered unidirectional and reflectionless between the onset of the blood flow and the time of maximum blood flow as discussed above (11,12). Indeed, the less elastic the vessel wall, the larger this time delay, so the earlier the reflected pressure wave, the less valid the method. Then, the value of APP was the sum of the pressure increases obtained from Eq. [5] as follows. ⌬P1 was calculated from 0 to 30 msec, ⌬P2 calculated from 30 to 60 msec, ⌬P3 calculated from 60 to 90 msec, etc. Due to the temporal resolution of the present experiments, three or four steps were necessary (from four or five time-frames, respectively). A first estimation of the PWV in the ascending aorta was obtained from the graph showing Umax vs. ⌬S/S (Eq. [7]). A significant correlation coefficient for the results of this fit (95% confidence) was considered a second mandatory condition, in order to validate the estimate of APP. The minimal and maximal aortic CSA values, and the APP estimate, were used in the calculation of the second ascending aorta PWV estimate (Eq. [9]). The method of Bland and Altman was used to assess the agreement between these two estimates of ascending aorta PWV (15). RESULTS Table 1 shows all measurements in all individuals. Figure 3a shows typical plots of maximal and mean velocity of the blood flow (over the aortic CSA) plotted vs. time (volunteer 1). Figure 3b shows maximal blood flow velocity and aortic CSA plotted vs. time in the same volunteer. As shown in Fig. 3b, in 7 of the 11 volunteers the time delay between the velocity peak and the CSA peak was 30 msec (and 0 msec in the remaining four). The average SD of aortic CSA measurements was 0.14 cm2 (maximal SD ⫽ 0.35 cm2). A minimal average aortic CSA, equal to 5.24 cm2, leads to a relative uncertainty of 3% (and a maximal uncertainty of 7%). The mean SD
Figure 3. a: Maximal (circles) and mean (triangles) velocities of blood flow plotted vs. time for a volunteer, from the onset of the blood flow wave to the early descending part of the systolic peak. b: Maximal blood flow velocity (circles) and related aortic CSA (rhombi) plotted vs. time for the same volunteer. (The relative scaling of the two data sets is arbitrary.)
Aortic Pulse Pressure With MRI
Figure 4. Umax vs. ⌬S/S for the same volunteer as in Fig. 3; the slope of the linear fit gives an estimate of the ascending aortic PWV: Umax ⫽ 4.15 ⌬S/S – 0.03 with r ⫽ 0.98 (r0 ⫽ 0.88 for N ⫽ 5 and 95% confidence).
of the maximal blood flow velocity measurement was 1.61 cm/second. Figure 4 shows the significant linear fit to the graph of Umax vs. ⌬S/S for volunteer 1: Umax ⫽ 4.15⌬S/S – 0.03 (r ⫽ 0.98), where the slope of 4.15 m/second is an estimate of the ascending aorta PWV. The mean APP value (⫾ SD) over the series was 54.2 ⫾ 16.4 mmHg (range 32.2– 84.1 mmHg). For comparison (Table 1), an APP estimate was made from Eq. [6] (APP⬘), and, as expected, it was found to be significantly lower than the previous value, with a mean difference between the two estimates (and in the 95% confidence interval) of 16.4 ⫾ 8.0 mmHg (P ⬍ 0.01; graph not shown). The mean PWV values ⫾ SD for the two estimates of the ascending aorta PWV over the series were 5.03 ⫾ 1.10 m/second and 5.37 ⫾ 1.23 m/second, from Eq. [7] and Eq. [9], respectively. The two estimates for the series are compared in Fig. 5 according to the method of Bland and Altman (15). The plot indicates that these estimates were not significantly different (95% confidence). No dependence on the PWV value was observed.
57
method could not be directly determined. We therefore suggest that the validation of the method should be obtained from animals, as done by Urchuk et al (10) for evaluating pulse pressure wave forms in porcine thoracic aortas. Nevertheless, it should be pointed out that the SD value reflects measurement variability, as well as actual physiologic differences involving volunteers’ emotions (14), between the subjects in the series. Moreover, it should be noted that the present study, which is based upon theoretical considerations (Eq. [5] and Eq. [7]), emphasizes the measurement of highest velocity values that meet the conditions of Tasu’s methodology in measuring with MRI the highest acceleration and velocity values for estimating the left ventricular performance through temporal pressure variations (17). Also, the method allowed us to derive two different estimates of the ascending aortic PWV, which were not significantly different (15) (Table 1 and Fig. 5). The mean PWV values (⫾ SD) from the present study, of 5.03 ⫾ 1.10 m/second and 5.37 ⫾ 1.23 m/second, respectively, are in very good agreement with that of Vullie´moz et al (5) (4.9 ⫾ 1.1 m/second), even though healthy volunteers in that study were older than ours (mean ages 34.3 and 24.5 years, respectively). The noninvasive MRI method we used to estimate APP in young adults has a principal limitation. The ascending aorta is not a perfectly rectilinear tube, as was assumed in deriving Fig. 1b. In particular, this explains why we observed that the location of the slices containing the most rapid blood flow moved toward the internal part of the aortic arch during the ascending slope of the systolic peak (18,19). We suggest that this problem cannot be overcome, and therefore, with respect to this
DISCUSSION A novel, noninvasive MRI method for assessing APP in the ascending aorta of young adults has proven feasible. This central pulse pressure influences the cardiac afterload, and it has been clearly shown that the brachial pulse pressure was an inadequate surrogate (7–9). The mean APP value ⫾ SD for the series was 54.2 ⫾ 16.4 mmHg (Table 1). This experimental estimation of APP is therefore in close agreement with the value of 60 mmHg, a central pulse pressure value usually found in the literature (13,16). In the framework of the feasibility of a noninvasive study, comparison with a gold standard value obtained from invasive left-sided catheterization was precluded in young healthy adults, by ethical considerations, and therefore the accuracy of the
Figure 5. Comparison of the two estimates of the ascending aortic PWV (see text) by means of the method of Bland and Altman (15), indicating that they were not significantly different. Each point is calculated with data from one volunteer in the series. The mean value of the difference between the two estimates (bold horizontal line) and the 95% confidence interval (two dotted lines) was – 0.35 ⫾ 0.39 m/second.
58
point, the accuracy of the method (which nonetheless is in good agreement with published data) most likely cannot be improved. Another limitation is that, even in young adults, the aortic wall is likely not to be perfectly elastic, as implicitly assumed in Eq. [4]. This limitation has been taken into account in the present study, by implementing the method only when the time delay between the velocity peak and the CSA peak was less than or equal to 30 msec (for four and seven volunteers, respectively). However, we suggest that the method could be improved by increasing the temporal resolution. Indeed, the shorter the resolving time, the lower the increase in pressure between two successive measurement times and, consequently, the more valid is Eq. [5]. This suggestion is supported by the comparison between APP and APP⬘ related to Eq. [5] and Eq. [6], respectively (Table 1). Further developments to improve APP estimation could implement Vullie´moz’s method to reduce the temporal resolution of the acquisition (5). The present method for assessing ascending aortic APP has proven valid for young adults. Nevertheless, whatever the patient’s age, it improves upon the method of assessing central PWV proposed by Vullie´moz et al (5) (Eq. [7]), which used the mean blood flow velocity value over the aortic CSA, that can only be applied during early systole. Indeed, the present work shows that, when Umax is used in Eq. [7] instead of the value of the mean blood flow velocity over the aortic CSA, the time range for the linear fit in the graph showing Umax vs. ⌬S/S, is increased. Consequently, increasing the number of time-frames involved in that linear fit leads to an increase in its accuracy and hence in the accuracy of the PWV assessment. In conclusion, a novel noninvasive MRI method to estimate the central pulse pressure has proven feasible in young adults. Values of both aortic pulse pressure and two derived, local PWV estimations were in good agreement with published data. In older patients, it is suggested that the principle of the present method could improve a recently published method (5) to assess the local ascending aortic PWV, if the number of useful time frames is increased. ACKNOWLEDGMENT The authors gratefully acknowledge valuable suggestions from S. Wynchank.
Laffon et al.
REFERENCES 1. Safar ME, Henry O, Meaume S. Aortic pulse wave velocity: an independent marker of cardiovascular risk. Am J Geriatr Cardiol 2002;11:295–298. 2. Ohmori K, Emura S, Takashima T. Risk factors of atherosclerosis and aortic pulse wave velocity. Angiology 2000;51:53– 60. 3. Wilkinson IB, Webb DJ, Cockcroft JR. Aortic pulse-wave velocity. Lancet 1999;354:1996 –1997. 4. Lehman ED. Aortic pulse-wave velocity versus pulse-wave analysis. Lancet 1999;355:412. 5. Vullie´moz S, Stergiopulos N, Meuli R. Estimation of local aortic elastic properties with MRI. Magn Reson Med 2002;47:649 – 654. 6. Mohiaddin RH, Firmin DN, Longmore DB. Age-related changes of human aortic flow wave velocity measured non invasively by magnetic resonance imaging. J Appl Physiol 1993;74:492– 497. 7. Karamanoglu M, O’Rourke MF, Aviolo AP, et al. An analysis of the relationship between central aortic and peripheral upper limb pressure waves in man. Eur Heart J 1993;14:160 –167. 8. Lehman ED. Non invasive measurements of aortic stiffness: methodological considerations. Path Biol 1999;47:716 –730. 9. O’Rourke MF. Isolated systolic hypertension, pulse pressure, and arterial stiffness as risk factors for cardiovascular disease. Curr Hypertens Rep 1999;1:204 –211. 10. Urchuk SN, Fremes SE, Plewes DB. In vivo validation of MR pulse pressure measurement in an aortic flow model: preliminary results. Magn Reson Med 1997;38:215–223. 11. Murgo JP, Westerhof N, Giolma JP, et al. Aortic input impedance in normal man: relationship to pressure wave forms. Circulation 1980;62:105–116. 12. Nichols WW, Edwards DG. Arterial elastance and wave reflection augmentation of systolic blood pressure: deleterious effects and implications for therapy. J Cardiovasc Pharmacol Ther 2001;6:5– 21. 13. Comolet R. Wave travelling without damping “Propagation d’ondes non amorties.” In: Comolet R, editor. Circulation biomechanics “Biome´canique circulatoire.” Paris: Masson; 1984. p 159 –174. 14. Laffon E, Valli N, Latrabe V, et al. A validation of a flow quantification by MR phase mapping software. Eur J Radiol 1998;27:166 – 172. 15. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986;1:307– 310. 16. Hurst JW, Logue RB, Rackley CE, et al. (translation by G. Drobinski et al.). Physiology of the normal cardiovascular system “physiologie du syste`me cardiovasculaire normal.” In: Masson, editor. The heart, arteries and veins “Le cœur, arte`res et veines.” Paris: Masson; 1984 (French translation). 17. Tasu JP, Mousseaux E, Colin P, et al. Estimation of left ventricular performance through temporal pressure variations measured by MR velocity and acceleration mappings. J Magn Reson Imaging 2002;16:246 –252. 18. Bogren HG, Mohiaddin RH, Klipstein RK, et al. The function of the aorta in ischemic heart disease: a magnetic resonance and angiographic study of aortic compliance and blood flow patterns. Am Heart J 1989;118:234 –247. 19. Bogren HG, Klipstein RH, Firmin DN, et al. Quantitation of anterograde and retrograde blood flow in the human aorta by magnetic resonance velocity mapping. Am Heart J 1989;117:1214 –1222.