Articles in PresS. Am J Physiol Heart Circ Physiol (January 6, 2006). doi: /ajpheart University, Gent, Belgium

Transcription

1 Articles in PresS. Am J Physiol Heart Circ Physiol (January 6, 2006). doi: /ajpheart AORTIC REFLECTION COEFFICIENTS AND THEIR ASSOCIATION WITH GLOBAL INDICES OF WAVE REFLECTION IN HEALTHY CONTROLS AND PATIENTS WITH MARFAN DISEASE P. Segers 1, J. De Backer 2,3, D. Devos 4, S.I. Rabben 5, T.C. Gillebert 3, L.M. Van Bortel 6, J. De Sutter 3, A. De Paepe 2, P.R. Verdonck 1 1 Cardiovascular Mechanics and Biofluid Dynamics, Institute Biomedical Technology, Ghent University, Gent, Belgium 2 Department of Medical Genetics, Ghent University Hospital, Gent, Belgium 3 Department of Cardiovascular Medicine, Ghent University Hospital, Gent, Belgium 4 Department of Medical Imaging, Ghent University Hospital, Gent, Belgium 5 Institute for Surgical Research, Rikshospitalet University Hospital, Oslo, Norway 6 Heymans Institute of Pharmacology, Ghent University Hospital, Gent, Belgium Patrick Segers and Julie De Backer equally contributed to the manuscript. Keywords: wave reflection, aorta, Marfan, MRI, augmentation index Running title: Local aortic and global wave reflection in control and Marfan disease Correspondence to: Patrick Segers, Cardiovascular Mechanics and Biofluid Dynamics, Hydraulics Laboratory, Institute Biomedical Technology, Ghent University, Sint-Pietersnieuwstraat 41, 9000 Gent, Belgium. Tel:+32/9/ , Fax: +32/9/ ; patrick.segers@ugent.be Abstract word count: 249 Total word count (excl. abstract): 6116 Copyright 2006 by the American Physiological Society.

2 Abstract Early return of reflected pressure waves increases the load on central arteries. In patients with Marfan syndrome (MFS), this may increase the risk of aortic rupture. To assess whether wave reflection is elevated in MFS, we measured central pressure and flow waveforms in 26 patients (age 13-54) and in 26 age and sex matched controls (CTRL) using ultrasound and magnetic resonance imaging. Aortic systolic and diastolic cross-sectional area was measured at 4 levels: ascending (AA) and descending (DA) aorta, diaphragm (DIA) and low abdominal aorta (AB). From these, local characteristic impedance (Z 0-xx ) was calculated as well as local reflection coefficients ( xx-yy ). Calculated global wave reflection indices were the augmentation index (AIx) and the ratio of backward-to-forward pressure wave (P b /P f ). The aorta was wider in MFS at AA (P<0.01) and DA (P<0.01). Aortic pulse wave velocity (PWV) was 42 cm/s higher in patients (P<0.05). Z 0-xx was not different between groups, except at DA where it was lower in MFS. In CTRL, AA-DA was 0.31±0.08, DA-DIA was 0.00±0.11 while DIA-AB was 0.31±0.16. Mean values of xx-yy were not different between patients and controls. In CTRL, ageing diminished AA-DA, while DIA-AB increased with age. Clear age-related patterns were absent in MFS. AIx or P b /P f were not higher in MFS than in CTRL. Nevertheless, there were indications for enhanced wave reflection in young MFS patients. Our data demonstrated that the major determinants of AIx were PWV and the effective length of the arterial system and, to a lesser degree, heart rate and P b /P f.

3 Introduction The propagation and reflection of pressure and flow waves along the arterial tree has been the subject of early fundamental biofluid mechanical research, but it was only in the early 1980 ies that the patho-physiological effect of pressure wave reflection was most clearly demonstrated by Murgo et al. (13). Early return of reflected waves boosts systolic pressure and presents an extra load for the heart and the central vessels (15, 22, 24). In the past few years, the study of arterial wave reflection has also reached the medical/clinical community, mainly due to the effort of O Rourke and colleagues who developed the augmentation index (9). This is an easy-to-use index which can be derived from central pressure (or diameter) waveforms and which formally quantifies the wave contour classification scheme of Murgo et al. (13, 16). The augmentation index (AIx) is, however, a composite index and therefore not always straightforward to interpret. It is dependent not only on the magnitude of wave reflection (the reflection coefficient), but also on the time delay between the forward and reflected wave. As such, the index is also determined by body stature, the stiffness of the aorta (aortic pulse wave velocity) and even heart rate. Wave reflection, however, is still not fully understood especially with respect to the origin of the reflected waves. In this in vivo study, we assessed both local and global reflection (coefficients) in normal subjects as well as in patients with Marfan disease, a genetically determined connective tissue disorder primarily affecting the (proximal) aorta. In addition to the effects of age, which spanned 4 decades in both groups, several aspects of the disease potentially alter the contributions of arterial wave reflection: (i) elevated aortic pulse wave velocity (PWV) due to global aortic stiffening has been reported in patients with Marfan disease (4, 5, 8), which would favour the early return of pressure waves from the periphery; (ii) the disease primarily affects the proximal part of the aorta (6) and may change the gradual proximal-to-distal evolution of the mechanical properties of the aorta and give rise to reflections originating from an impedance

4 mismatch along the aorta; (iii) Marfan subjects are in general taller than the normal population (one of the visual landmarks of the disease), hence affecting the distance to reflection sites; (iv) it has been suggested that the global wave reflection coefficient may be elevated in patients with Marfan disease (25). Local reflection coefficients along the aorta will be assessed through calculation of changes in characteristic impedance along the vessel, with characteristic impedance estimated from Magnetic Resonance Imaging (MRI) recordings of the systolic and diastolic cross-sectional area of the aorta at different levels along the aorta. Global reflection will be estimated both via the augmentation index and linear wave separation analysis. The ratio of the amplitude of the backward (P b ) and forward (P f ) wave, P b /P f, will be used as an estimate of the global reflection coefficient (24). The in vivo data should thus provide local aortic reflection coefficients and have the potential to reveal a possible relation between local reflection along the aorta and the global wave reflection indices such as AIx and P b /P f.

5 Materials and Methods The population studied consisted of twenty-six patients with confirmed Marfan syndrome (MFS; age range yrs) and twenty-six age and sex matched control subjects (CTRL) (see also Table 1). All subjects underwent a 1-day measurement protocol including magnetic resonance imaging (MRI) and echocardiography for the assessment of systolic and diastolic dimensions of the aorta at different levels as well as central pressure and flow waveforms. MRI: PWV, aortic dimensions and local reflection coefficients All subjects were scanned on a 1.5T MR system (Magnetom Symphony, Siemens, Erlangen, Germany) with ECG gating. Aortic systolic (A xxs ) and end-diastolic (A xxd ) cross-sectional area was measured using truefisp images (Fast Imaging with Steady-state Precession; temporal resolution of 25 milliseconds; spatial resolution of 1.33 mm/pixel in the x and y direction) obtained at 4 levels (indicated by xx ) along the aorta: the ascending (AA) and descending thoracic aorta (DA), thoracicabdominal aorta near the diaphragm (DIA) and low abdominal aorta (AB) (see Figure 1). The distance between the different aortic levels was assessed. Through-plane phase-contrast images were taken at these same levels to assess the flow curves, which were calculated with the Siemens Mean Curve software. The curves were then interpolated to obtain a temporal resolution of 1 ms, and the time-to-half-peak (i.e., the time delay between the R- top of the ECG and the moment when flow reaches half of its peak value) were calculated. With time and distance travelled by the propagating flow front known at 4 locations, PWV (cm/sec) was calculated as the slope of the regression line through these data points (Matlab, The Mathworks, Natick, Massachusetts) (Figure 1)

6 Local reflection, arising from impedance mismatch between levels xx and yy, was quantified using local wave reflection coefficients ( xx-yy ) calculated as xx yy = Z Z 0 yy 0 yy Z + Z 0xx 0xx (1) where Z 0-xx is the characteristic impedance at level xx, approximated as SBP DBP Z0 xx = (2) Axx Axxs Axxd with the density of blood (assumed 1030 kg/m 3 ) and A xx the average value of A xxs and A xxd. SBP and DBP are central systolic and diastolic blood pressure, respectively (see further). Assessing central blood pressure waveforms (P ao ) Central blood pressure waveforms were obtained via calibration of the common carotid artery diameter distension waveforms (see also Figure 2) (17). With the subject in supine position, a sequence of common carotid artery diameter distension tracings typically containing 3 to 5 complexes was measured with a commercially available ultrasonographic system (Vivid 7, GE Vingmed Ultrasound, Horten, Norway) and a 12 MHz vascular probe (12L). The tracing was averaged to obtain one representative waveform which was subsequently transformed into a carotid artery pressure waveform (20) that was further used as a surrogate of the central pressure waveform (P ao ). To do so, it was assumed that the relation between pressure and diameter was linear and that diastolic (DBP) and mean arterial pressure (MAP) was similar at the brachial and carotid artery. Central systolic blood pressure (SBP) was taken as the maximum value of P ao. MAP was assessed following a procedure as recently described by Verbeke et al. that includes three steps: (1) applanation tonometry at the brachial artery to obtain the brachial artery waveform; (2) calibration of this waveform using diastolic and systolic blood pressure as measured with a brachial cuff sphygmomanometer; (3) the average of the scaled brachial artery pressure wave yields MAP (21).

7 Assessing central flow waveforms (Q ao ) Blood flow velocities were acquired in the left ventricular outflow tract (LVOT) using pulsed wave Doppler (3.5 MHz probe) in the apical 5-chamber view. Images were stored as DICOM files for off-line analysis, where contours were semi-automatically traced with a dedicated software interface written in Matlab (The Mathworks, Natick, Massachusetts). An ensemble average was constructed of minimally three cycles, and the average curve was scaled so that the area under the curve matched stroke volume (SV) as determined from MRI (left ventricular volumes were acquired with an ECG triggered truefisp sequence and SV was calculated as the difference between end-diastolic and end-systolic volume). We judged this approach to be the most accurate because LVOT diameters were difficult to assess accurately in Marfan patients with dilated aortic roots. The aortic flow waveform is further indicated as Q ao. Cardiac output (CO) was obtained as the product of SV and heart rate (HR). It was verified that heart rate was similar (+/-5 beats/min) during MRI and ultrasound measurements, which was the case in all subjects. The augmentation index (AIx) and distance to apparent reflection site The augmentation index (AIx) is calculated as P DBP AIx = (3) P DBP 1 where P 1 and P 2 are either SBP, either the pressure associated with an inflection point visually identified on P ao (see also Figure 2, right panel). The pressure occurring first is labelled as P 1. AIx<100% indicates arrival of the pressure wave in late systole; AIx > 100% is indicative for arrival in early systole. We also measured the time delay between the foot of P ao, and the moment of occurrence of the inflection point, T f-b. This time interval is associated with the time needed for a

8 wave to travel forth and back from the ascending aorta to its apparent reflection site (the effective length of the arterial system, x), calculated as PWV. T x = f -b (4) 2 with PWV the aortic pulse wave velocity, assessed with magnetic resonance imaging (see further). We refer to Figure 2 for illustration of the inflection point and timing intervals. Linear wave separation analysis: P b /P f As demonstrated by Westerhof et al.(23), the pressure wave is composed of a forward (P f ) and a reflected or backward (P b ) travelling component, which can be separated from each other provided that P ao and Q ao are known, as well as characteristic impedance (Z 0 ): P f ( P + Z Q ) ( P Z Q ) ao 0 ao ao 0 ao = ;Pb = (5) 2 2 One can then define the global wave reflection coefficient as the ratio of the amplitudes of P b and P f (P b /P f ). Z 0 was estimated as the average value of the modulus of the high frequency components of input impedance(12, 14). Statistical analysis Data are reported as mean ± standard deviations. Population means were compared using Student s t-test. To study the evolution of parameters with age, data were organized in tertiles of age (27 years, >27 and 40 years, > 40 years). Relation between parameters was studied using Pearson correlation and linear regression analysis. When appropriate, the differences between Marfan patients and control subjects were studied using analysis of variance with age tertile and disease as fixed factors. All analyses were performed in SPSS (Version 11.5, SPSS Inc., Chicago, Illinois).

9 Results The patients with Marfan disease were taller, had a higher weight and body surface area (BSA) than controls (Table 1). There was no difference between patients and controls for age, body mass index, brachial and central blood pressure, heart rate (HR), stroke volume and cardiac output. General hemodynamic data and patient characteristics are summarized in Table 1. Mean age was 22.7± 4.6, 34.6 ± 4.7 and 49.5 ± 4.8 years in tertile 1 to 3. Aortic dimensions, characteristic impedance and local reflection coefficients Population-averaged aortic cross sectional area measured at four levels along the aorta are given in Table 2. On average, the aorta was significantly wider in Marfan patients than in controls at the 2 most proximal measuring locations (AA; P<0.01 and DA; P<0.01). To better appreciate the evolution of aortic size with age, data are also plotted as a function of age in Figure 3. Aortic cross sectional area progressively increased with age both in Marfan patients and controls (P<0.01) at all levels. For the ascending aorta, the progression of dilatation with age was higher in Marfan patients than in controls (P<0.05), leading to a significantly higher aortic cross sectional area in the third tertile (P<0.05). The characteristic impedance was similar at the ascending aorta and the 2 most distal locations, and significantly lower in Marfan patients for the descending aorta (P<0.05) (Table 2). For the lower abdominal aorta, the evolution of Z 0-AB with age is significantly different between the two groups (P<0.01), increasing with age in controls and decreasing in Marfan patients (Figure 3). The difference was statistically significant in the 3 rd tertile (P<0.05). Local reflection coefficients are displayed as a function of age in Figure 3. In the control population, a positive reflection coefficient was found, AA-DA being on average 0.31 ± In the

10 mid aorta, DA-DIA was close to zero while the most distal reflection coefficient, DIA-AB, was again positive (0.31 ± 0.16). When analyzed as a function of age, AA-DA decreased with age (P<0.05), while the value of DIA-AB increased with age (P<0.05). For the Marfan patients, a similar global pattern was observed, with a positive proximal and distal reflection coefficient, and absence of reflection in the mid aorta. In contrast with the control population, no correlation with age was found in any segment. In simple t-test analysis, there were no differences in xx-yy between controls and patients with Marfan disease. Using ANOVA, a marginal difference between control subjects and Marfan patients was found for AA-DA (P = 0.049), with the reflection coefficient being lower in the patient group. Global reflection: AIx and distance to reflection site Augmentation index, AIx, was on average virtually identical in controls and patients with Marfan disease (102.1 ± 15.0% versus ± 10.6 %; P=0.75). Displaying the data as a function of age (Figure 4, panel A), AIx tended to be higher in MFS than in CTRL in tertile 1, and lower in MFS than in CTRL in tertile 3, but the differences were not statistically significant. PWV was 486 ± 110 cm/s in control, versus 519 ± 100 cm/s in Marfan patients, the difference being non-significant in t- test analysis (P=0.27). Analysis of variance, however, indicated a significant offset between both groups (P=0.03) estimated to be 42 cm/s (Figure 4, panel C). As for the timing of forward and backward waves, T f-b was not different between controls and Marfan patients (0.165 ± 0.033s vs ± 0.039s, P=0.70). The effective length of the arterial system, x, on the other hand was shorter in the control group than in Marfan patients (38.7 ± 6.0 cm vs 43.4 ± 9.3 cm; P<0.05). Data split per tertile of age are shown in Figure 4 (panel D).

11 Global reflection: linear wave separation: P b /P f Input impedance derived from P ao and Q ao is plotted in Figure 5. There were no differences between both groups for none of the harmonics. Characteristic impedance estimated from central pressure and flow was ± mmhg.ml -1.s in the control group, and ± mmhg.ml -1.s in the patients, the difference being non-significant (P=0.39). There was no difference in P b /P f between patients with Marfan syndrome and controls (0.45 ± 0.08 versus 0.47 ± 0.09; P=0.44). Data are displayed in Figure 4 (panel B). Determinants of augmentation index To assess the relative importance of the major determinants of AIx, multiple linear regression analysis (stepwise forward model) was performed with age, gender, central blood pressure (systolic, diastolic and mean), stroke volume, heart rate, length, patient/control, PWV, amplitude of P f, amplitude of P b, P b /P f, x, and the three local reflection coefficients as possible determining independent factors. The model obtained was AIx = PWV x HR P b /P f, r 2 = 0.80 with PWV in cm/s, x in cm and HR in beats/minute. The parameters are displayed in the equation in the order of their relative importance, PWV being the most important contributor. Table 3 displays the standardized coefficients and statistical significance of the different contributors, and the increase in predictive value of the model (r 2 ) upon additional inclusion of the parameter into the model. Local reflection coefficients did not contribute to the model.

12 Discussion The major findings of this study can be summarized as follows: (i) The morphological and functional changes in the (proximal part of the) aorta in patients with Marfan disease do not lead to a change in local aortic characteristic impedance of the aorta; (ii) In normal subjects, ageing appears to diminish the local reflection coefficient in the proximal aorta, while it increases with age in the distal part. Clear age-related patterns are absent in the Marfan patient group; (iii) On average, global wave reflection as quantified by the augmentation index or the ratio of backward-to-forward wave is not higher in patients with Marfan disease than in a control population. Nevertheless, there are indications for elevated wave reflection in young patients with Marfan disease; (iv) The major determinants of the augmentation index are pulse wave velocity and the effective length of the arterial system and, to a lesser degree, heart rate and the ratio of the backward-to-forward wave. Conform to common knowledge (4, 6, 8), we found that the aorta was widened in Marfan patients at the ascending and descending level only. In both controls and Marfan patients, aortic enlargement with age was observed, and this at all levels (Figure 4). Enlargement appeared to progress at the same rate, except for the ascending aorta, where aortic dilatation takes place at a higher pace in Marfan patients. In this study, focusing on wave reflection, it is characteristic impedance and, more importantly, the changes in characteristic impedance that deserve attention as these may locally provoke wave reflection. Both vessel calibre and stiffness affect characteristic impedance (see also formula (2)) and our data suggest that both effects counterbalance each other, with no net effect on characteristic impedance for the most proximal part of the aorta. This observation also supports the findings of Yin et al., who derived aortic characteristic impedance from central pressure and flow and who found characteristic impedance within the normal range in Marfan patients. In this study, characteristic impedance was assessed both from central pressure and flow as well as from the changes in aortic cross-sectional area along the aorta measured with MRI, which allowed us to

13 study the aorta at different levels. It is generally accepted that in the normal population, there is a gradual increase in impedance along the aorta (due to geometric and elastic taper (12, 14, 18)) but that impedance mismatch is most important in the periphery, where small-sized arteries make the transition to arterioles and capillaries. For the descending aorta, the dilatation in the Marfan patients seems to overcompensate for an increase in stiffness, with lower characteristic impedance in the Marfan patients. Despite the absence of significant differences in mean values of many calculated parameters, there are trends in the data when they are studied as a function of age (Figure 3). In the ascendingdescending section, there is a more pronounced gradient in characteristic impedance (increasing distally) in young controls when compared to old controls and Marfan patients. The result is a positive reflection coefficient in the proximal aorta in young controls that decreases with age when the proximal aorta stiffens and the difference in characteristic impedance with adjacent sections becomes less. In the lower abdominal aorta, opposite changes were observed. In the control subjects, the age-related increase in Z 0-AB results in increasing local positive reflection coefficients in the distal aorta. Again, we did not observe any age related changes in the Marfan patients. We speculate that inter-patient differences in severity of the disease complicate the detection of eventual age-related patterns in this patient group. The literature on the subject of local reflection coefficients in the aorta in the general population - and in Marfan patients in particular - is scarce. Ting et al., analyzing apparent phase velocity, reported that local wave reflection can differ remarkably along different regions in the aorta, with pronounced reflections in the ascending aorta and from just proximal to the renal arteries to the aorto-iliac bifurcation, but not in the mid thoracic region (19). This is in accordance with data presented in our Figure 3, where DA-DIA is indeed much lower than in the other sections. Also,

14 more rapid mechanical aging near the aortic bifurcation has been reported by Gillesen et al (2). Greenwald et al. studied the effect of ageing on the local reflection coefficient of the aortic bifurcation (3). They concluded that the progressive increase of lower aorta characteristic impedance (as also found in our study) decreases the impedance mismatch with the iliac arteries, decreasing the reflection coefficient from the bifurcation. The question remains in how far local aortic properties and local reflection coefficients have an impact on the global picture of arterial wave reflection, as quantified with indices such as the augmentation index. Our data seem to suggest that, if any, the effect is marginal. The only (relatively weak) correlation that we found between local and global indices of wave reflection was between AIx and DIA-AB (r=0.29, P<0.05). However, none of the local reflection coefficients entered the model for AIx in multiple linear regression analysis. The impact of local reflection properties along the aorta thus seems to be negligible compared to the other co-existing sources of wave reflection. At present, there is growing clinical interest in arterial wave reflection and the augmentation index. Meijboom et al. found elevated AIx in patients with Marfan disease(10), but their patients (n=4) had had surgical repair of the aorta with a Dacron prosthesis, which may cause a drastic increase in PWV and compliance mismatch at the site of the anastomosis. Although it is recognized that the augmentation index is a composite measure (14), no study has truly focused on dissecting the index into its determining factors. We could demonstrate that the main determinants of AIx were PWV and the effective length of the arterial tree and, to a lesser extent, reflection coefficient and heart rate (Table 3). This analysis also reveals why we could not demonstrate a (anticipated) difference in AIx between control subjects and the Marfan patients. The elevated pulse wave velocity and the lower heart rate in the patient group are counterbalanced by the fact that the effective length was

15 larger in Marfan patients. Note that this is not necessarily a patho-physiological consequence of the disease, but probably rather a reflection of the difference in body length between both groups. The taller stature of the Marfan patients thus seems to have a protective effect in terms of wave reflection, delaying the return of the reflected wave. These mechanistic determinants of AIx also apply to other diseases affecting the functional properties of the aorta such as atherosclerosis, hypertension or diabetes (1) where we speculate that the hemodynamic burden caused by early wave reflection may particularly be higher in smaller subjects. Also, different classes of blood pressure lowering drugs may induce alterations in heart rate (beta-blockers) and the location of reflection sites (vasoactive drugs) and hence differentially affect the impact of wave reflection independent of the level of blood pressure decrease(7). In this study, input impedance was calculated as an intermediate step in the linear wave separation analysis. The data confirm that, when studied in a global manner, the arterial system in patients with Marfan disease is not drastically different from normal subjects, an observation that confirms the findings of Yin et al. that were based on invasive data (25). We also wish to draw some attention on a finding concerning P b /P f. When plotting P b /P f as a function of body length (Figure 6), it immediately becomes clear that this factor is not independent of body size, as one might expect of a true reflection coefficient (in the control population, there was a significant inverse association between P b /P f and body length). We believe that the influence of length on P b /P f is explained by the fact that in taller subjects, the forward and backward wave travel longer distances. So, upon arrival at the reflection site, the amplitude of the forward wave is smaller due to damping. In addition, the reflected wave needs to travel a longer distance back up the aorta, and is more damped as well. As a result, the taller the subject is, the smaller the amplitude of the reflected component, P b, and thus of P b /P f.

16 In our opinion, one of the strong aspects of the study is its fully non-invasive character that allows transferring fundamental hemodynamic research from the experimental laboratory to the clinical setting. At the same time, it is acknowledged that this brings along methodological considerations and limitations that deserve some attention and comments. First, we scaled carotid diameter distension waveforms to assess carotid systolic pressure. Although the methodology of scaling diameter distension waveforms was found adequate (20), the relation between diameter and pressure is nonlinear (11) and we may have underestimated carotid systolic pressure, especially in the older subjects or in subjects with high blood pressure. Nevertheless, carotid arteries appear less affected by Marfan disease, and Marfan patients and controls were matched for age and blood pressure, so that the eventual impact on the data should be the same for both studied groups. Second, the local aortic reflection coefficients are derived from data measured at four distant, discrete locations, so that the aorta is implicitly approximated as a tube consisting of 4 segments, with discrete changes in mechanical properties. The computation of Z 0-xx along the aorta is also based on carotid blood pressure, which was assumed to adequately represent blood pressure along the entire aorta. Pressure amplification is present along the aorta (14) and we may have underestimated pulse pressure (and underestimated Z 0-xx ) especially at the most distal locations (AB). It is an open question how much this assumption has affected our findings. For the control group, our findings on the variation of local reflection coefficients along the aorta are in line with the data from Ting et al.(19) and Gillesen et al.(2) (see above), suggesting that the effect is not important enough to affect these general findings. Our data do not allow us to directly make a similar statement for the Marfan patients. Nevertheless, using estimated regional pulse wave velocity in the abdominal aorta segment and the relation between Z 0-xx, cross-sectional area A, and blood density (Z 0-xx = PWV/A), the different evolution in Z 0-AB between patients and controls in the 3 rd tertile was confirmed (data are not shown) using a method independent of blood pressure. Third, as evident from our data and the discussion above, body size is an important confounding

17 factor in the analysis of wave reflection. While the different stature allowed us to enlarge the range of physiological parameters affecting wave reflection, ill-matched populations in terms of body stature - may pose an important limitation in clinical studies. Fourth, by inclusion of older Marfan patients, it cannot be excluded that the population is biased in the sense that older patients would have a milder manifestation of the disease, as they have reached a higher age without surgery despite the presence of the disease. Finally, for the wave separation analysis, we combined central flow with a surrogate for central pressure (and not the true central pressure). Although this methodology is quite commonly applied, it is acknowledged that this assumption may affect the accuracy of the wave separation, and hence the value of P b /P f. In conclusion, we have demonstrated that stiffening and dilatation of the proximal aorta in patients with Marfan disease does not lead to an increase in local aortic characteristic impedance of the aorta. In both healthy subjects and patients with Marfan disease, local reflection coefficients are postive in the proximal and distal aorta, and virtually zero in the mid aortic region. In healthy subjects, the proximal reflection coefficient diminshes with age, while it increases in the distal region. We could not demonstrate an association between local reflection coefficients along the aorta and the augmentation index, which is primarily determined by pulse wave velocity and the apparent length of the aterial system.

18 Acknowledgements This study was supported by a research grant from the Ghent University (BOF 011D4701) (J. De Backer) and by a research grant from the Fund for Scientific Research Belgium (FWO G029002) (A. De Paepe). Johan De Sutter is a senior clinical investigator of the Fund for Scientific Research Flanders (Belgium) (FWO Vlaanderen).

19 References 1. Boudoulas H, Toutouzas P, and Wooley C. Functional abnormalities of the aorta. Armonk, NY: Futura Publishing, Gillessen T, Gillessen F, Sieberth H, Hanrath P, and Heintz B. Age-related changes in the elastic properties of the aortic tree in normotensive patients: investigation by intravascular ultrasound. Eur J Med Res 1: , Greenwald SE, Carter AC, and Berry CL. Effect of age on the in vitro reflection coefficient of the aortoiliac bifurcation in humans. Circulation 82: , Groenink M, de Roos A, Mulder BJ, Spaan JA, and van der Wall EE. Changes in aortic distensibility and pulse wave velocity assessed with magnetic resonance imaging following beta-blocker therapy in the Marfan syndrome. Am J Cardiol 82: , Groenink M, de Roos A, Mulder BJ, Verbeeten B, Jr., Timmermans J, Zwinderman AH, Spaan JA, and van der Wall EE. Biophysical properties of the normal-sized aorta in patients with Marfan syndrome: evaluation with MR flow mapping. Radiology 219: , Hirata K, Triposkiadis F, Sparks E, Bowen J, Wooley CF, and Boudoulas H. The Marfan syndrome: abnormal aortic elastic properties. J Am Coll Cardiol 18: 57-63, Hirata K, Vlachopoulos C, Adji A, and O'Rourke MF. Benefits from angiotensin-converting enzyme inhibitor 'beyond blood pressure lowering': beyond blood pressure or beyond the brachial artery? J Hypertens 23: , Jeremy RW, Huang H, Hwa J, McCarron H, Hughes CF, and Richards JG. Relation between age, arterial distensibility, and aortic dilatation in the Marfan syndrome. Am J Cardiol 74: , Kelly R, Hayward C, Avolio A, and O'Rourke M. Noninvasive determination of age-related changes in the human arterial pulse. Circulation 80: , Meijboom LJ, Westerhof BE, Nollen GJ, Spaan JA, de Mol BA, Jacobs MJ, and Mulder BJ. Beta-blocking therapy in patients with the Marfan syndrome and entire aortic replacement. Eur J Cardiothorac Surg 26: , Meinders JM and Hoeks AP. Simultaneous assessment of diameter and pressure waveforms in the carotid artery. Ultrasound Med Biol 30: , Milnor WR. Hemodynamics. Baltimore, Maryland, USA: Williams&Wilkins, Murgo JP, Westerhof N, Giolma JP, and Altobelli SA. Aortic input impedance in normal man: relationship to pressure wave forms. Circulation 62: , 1980.

20 14. Nichols W and O Rourke M. McDonald s Blood Flow in Arteries. Theoretical, Experimental and Clinical Principles. London: Edward Arnold, O'Rourke MF. Mechanical principles. Arterial stiffness and wave reflection. Pathol Biol (Paris) 47: , O'Rourke MF, Avolio A, and Qasem A. Clinical assessment of wave reflection. Hypertension 42: e15-16; author reply e Epub 2003 Sep 2022., Segers P, Rabben SI, De Backer J, De Sutter J, Gillebert TC, Van Bortel L, and Verdonck P. Functional analysis of the common carotid artery: relative distension differences over the vessel wall measured in vivo. J Hypertens 22: , Segers P and Verdonck P. Role of tapering in aortic wave reflection: hydraulic and mathematical model study. J Biomech 33: , Ting CT, Chang MS, Wang SP, Chiang BN, and Yin FC. Regional pulse wave velocities in hypertensive and normotensive humans. Cardiovasc Res 24: , Van Bortel LM, Balkestein EJ, van der Heijden-Spek JJ, Vanmolkot FH, Staessen JA, Kragten JA, Vredeveld JW, Safar ME, Struijker Boudier HA, and Hoeks AP. Non-invasive assessment of local arterial pulse pressure: comparison of applanation tonometry and echo-tracking. J Hypertens 19: , Verbeke F, Segers P, Heireman S, Vanholder R, Verdonck P, and Van Bortel LM. Noninvasive Assessment of Local Pulse Pressure: Importance of Brachial-to-Radial Pressure Amplification. Hypertension 46: Epub 2005 May 2023., Westerhof N and O'Rourke MF. Haemodynamic basis for the development of left ventricular failure in systolic hypertension and for its logical therapy. J Hypertens 13: , Westerhof N, Sipkema P, van den Bos CG, and Elzinga G. Forward and backward waves in the arterial system. Cardiovascular Research 6: , Westerhof N, Stergiopulos N, and Noble M. Snapshots of Hemodynamics. An aid for clinical research and graduate education. New York: Springer Science + Business Media, Yin FC, Brin KP, Ting CT, and Pyeritz RE. Arterial hemodynamic indexes in Marfan's syndrome. Circulation 79: , 1989.

21 Tables Table 1. Population characteristics and general hemodynamic data. Data are mean values ± standard deviation. P-values are results from t-test analysis of the control group versus the patient population. The data between brackets designate the range of the parameter in the population. Control (n=26) Marfan (n=26) P-value M/F 12/14 12/ Age (years) 35.5 ± 11.8 (14 60) 32.7 ± 11.5 (13 54) 0.37 Length (m) 1.73 ± 0.11 ( ) 1.83 ± 0.10 ( ) Weight (kg) 67.0 ± 13.0 (50 95) 75.4 ± 14.3 (47 105) 0.03 BSA (m 2 ) 1.80 ± 0.21 ( ) 1.98 ± 0.20 ( ) DBP (mmhg) 62.6 ± 8.2 (48 80) 61.6 ± 8.6 (47 80) 0.68 MAP (mmhg) 84.4 ± 9.4 ( ) 84.8 ± 9.35 ( ) 0.87 SBP ca (mmhg) ± 11.8 ( ) ± 11.7 ( ) 0.64 HR (beats/min) 67.0 ± 10.0 ( ) 61.8 ± 10.0 ( ) 0.07 SV (ml) 78.9 ± 22.4 ( ) 82.0 ± 27.4 ( ) 0.67 CO (l/min) 5.2 ± 1.2 ( ) 4.9 ± 1.4 ( ) 0.51

22 Table 2 A xxs A xxd Z 0-xx xx-yy cm 2 cm 2 mmhg.ml -1.s AA DA DIA AB Control 6.34 ± ± ± Marfan 7.79 ± 2.24 ** 6.30 ± 2.13 ** ± Control 0.31 ± 0.08 Control 3.44 ± ± ± AA-DA Marfan 0.27 ± 0.11 Marfan 4.42 ± 0.90 ** 3.57 ± 0.85 * ± * Control 0.00 ± 0.11 Control 3.11 ± ± ± DA-DIA Marfan 0.05 ± 0.08 Marfan 3.62 ± ± ± Control 0.31 ± 0.16 Control 1.84 ± ± ± DIA-AB Marfan 0.30 ± 0.16 Marfan 2.19 ± ± ± Aortic cross-sectional area in systole (A xxs ) and diastole (A xxd ), local characteristic impedance Z 0-xx at four different levels along the aorta and local reflection coefficients xx-yy. Data are reported as mean values ± standard deviation. * P<0.05; ** P<0.01: difference between Marfan patients and controls in t-test analysis.

23 Table 3 P Standardized r 2 PWV (cm/s) x (cm) HR (beats/min) P b /P f Contribution of pulse wave velocity (PWV), effective length of the arterial tree (x), heart rate (HR) and global reflection coefficient (P b /P f ) to the augmentation index, as assessed with linear regression analysis. P-value is the statistical significance of the parameter in the model, the standardized coefficient, indicating the relative importance of the parameter. r 2 indicates the predictive value of the model after additional inclusion of the parameter.

24 Figure captions Figure 1. Panel A: Magnetic resonance imaging of the aorta for assessment of the distance between the four aortic measuring locations: ascending aorta (AA), descending aorta (DA), diaphragm (DIA) and lower abdominal aorta (AB). Panel B displays measured normalized flows at these locations, with indication of the moment when half-peak-flow is reached (AU: arbitrary units). Plotting the distance traveled by the propagating flow front as a function of this time, the slope of the regression line yields aortic pulse wave velocity (PWV, cm/s; panel C). Figure 2: Assessing central waveforms. (i) Applanation tonometry yields the brachial artery pressure waveform, which is calibrated using sphygmomanometer brachial systolic (SBP BA ) and diastolic blood pressure (DBP). Averaging this curve gives mean arterial pressure (MAP). (ii) Diameter distension curves, measured using ultrasound, yield carotid diameter distension waveform, which is used as a surrogate for carotid pressure waveform. The waveform is calibrated suing DBP and MAP. The peak value of the calibrated waveform yields carotid systolic pressure (SBP CA ). The calibrated curve clearly shows an inflection point, which is used to assess the time delay (T f-b ) between the onset of pressure rise and the inflection point. Figure 3. Evolution in time (tertile T 1 to T 3 ) and along the aorta (AA: ascending aorta; DA: descending aorta; DIA: diaphragm; AB: lower abdominal aorta) of aortic cross section (panel A), characteristic impedance (panel B) and local reflection coefficients (panel C) in controls (open circles) and in patients with Marfan disease (closed circles). Figure 4. Evolution with age (tertile 1 to 3) in Controls and Marfan patients of the augmentation index (AIx; panel A), the effective length of the arterial tree (x; panel B), aortic pulse wave

25 velocity (PWV; panel C) and the global reflection coefficient (P b /P f ; panel D). Data are mean values; error bars are standard deviation. Figure 5. Input impedance modulus (panel A) and phase angle (panel B) measured in control subjects (open circles) and patients with Marfan disease (closed circles). Error bars are standard deviations. Figure 6. Association between body length and global wave reflection coefficient, P b /P f in control subjects (open symbols) and patients with Marfan disease (closed symbols). The association was significant in control subjects only (dashed line), the correlation coefficient being -0.64, P=0.001.

26 A AA AB DIA DA normalized flow (AU) distance travelled (cm) time (s) B AA DA C AB DIA PWV time (s) Figure 1

27 Applanation Tonometry Ultrasound Diameter Distension Diameter (mm) Pressure (mmhg) SBP BA MAP DBP Time (s) Time (s) Pressure (mmhg) inflection point T f-b DBP Time (s) SBP CA MAP Figure 2

28 Aortic cross-sectional area (cm 2 ) Aortic characteristic impedance (mmhg.ml -1.s) Local Reflection Coefficient Control Marfan T1 T2 age T3 0 AA-DA DA-DIA DIA-AB DA DIA AB AA valve bifurcation Control Marfan AA DA DIA AB Figure 3. A valve B C 0.0 age bifurcation T3 T2 T1-0.1 valve bifurcation T1 Control Marfan T2 age T3

29 A Control Marfan B AIx (%) x (cm) age tertile age tertile 800 C 0.65 D PWV (cm/s) P b /P f (-) age tertile age tertile Figure 4

30 A 60 B modulus Z in (mmhg.ml -1.s) Control Marfan phase Z in ( ) frequency (Hz) frequency (Hz) Figure 5

31 0.8 Control Marfan 0.7 P b /P f (-) length (m) Figure 6