Background
Velocity vector imaging (VVI) is a novel two-dimensional speckle-based imaging technique for evaluation of tissue deformation. The aim of this study was to determine the feasibility and variability of VVI for the assessment of aortic strain, distensibility, and stiffness in patients with aortic valve disease.
Method
Eighty-five patients (mean age 66 ± 11 years) with aortic stenosis (AS) or aortic regurgitation (AR) were examined in the operating room before the operation using transesophageal echocardiography (TEE). The two-dimensional short-axis images and M-mode recordings of the descending aorta were acquired simultaneously with the invasive blood pressure measurement in the radial artery. The TEE images were analyzed off-line using VVI software.
Results
In comparison with patients with AS, patients with AR displayed significantly higher circumferential strain (7.6% ± 4.5% vs. 3.7% ± 1.9%, P < . 001) and distensibility (27.1 ± 12.8 kPa −1 10 −3 vs. 17.2 ± 7.2 kPa −1 10 −3 , P < . 001) by VVI and distensibility (32.8 ± 16.7 kPa −1 10 −3 vs. 21.7 ± 10.6 kPa −1 10 −3 , P < .004) by M-mode. Stiffness was higher in AS than AR, as measured by VVI (13.3 ± 6.0 vs. 10.5 ± 6.0, P < . 01) and M-mode (11.2 ± 6.1 vs. 10.4 ± 9.1, P < . 048). The correlations between VVI and M-mode distensibility ( r = 0.84) and stiffness ( r = 0.84) were both highly significant ( P < .0001). The VVI strain measurements showed low inter- and intraobserver variability with intraclass correlations greater than 0.95 and coefficients of variation less than 10%.
Conclusion
VVI-derived strain, distensibility, and stiffness differ significantly between AR and AS and correlate strongly with the corresponding M-mode–derived parameters. VVI is a feasible method for the assessment of the elastic properties of the descending aorta with low variability and has the advantage of incorporating the entire aortic wall circumference in the analysis, consequently accounting for local variations in the elastic properties of the aorta.
Increased arterial stiffness is associated with atherosclerosis and has been identified as an important predictor of cardiovascular events and morbidity. When determining regional arterial stiffness, the aorta is a vessel of major interest because of its substantial contribution to the arterial buffering function (Windkessel effect) and because aortic pulse-wave velocity is an independent predictor of outcome in different patient populations. Aortic valve disease is not an isolated disease of the aortic valve leaflets. Both structure and function of the thoracic aorta are also affected. Therefore, determination of aortic stiffness is of interest in aortic valve disease, and abnormalities in aortic stiffness may be a manifestation of aortic involvement in the atherosclerotic process or may reflect connective tissue disorders.
To determine local aortic stiffness, noninvasive imaging techniques have been used, such as transthoracic echocardiography (TTE) or transesophageal echocardiography (TEE) using two-dimensional images or M-mode recordings, magnetic resonance imaging, and computed tomography. Velocity vector imaging (VVI), a novel method based on ultrasonic two-dimensional images, is a recent development shown to be feasible for the assessment of cardiac mechanics. VVI involves speckle tracking combined with tissue-blood border detection and the periodicity of the cardiac cycle using RR-intervals. The tissue velocity is displayed as a vector projected on a two-dimensional echocardiographic image where the vector shows the direction of the movement and the length of the vector is related to the velocity of the tissue ( Figure 1 A). VVI permits angle-independent measurements of tissue velocity and deformation, strain, as well as rotational displacement. Because VVI tracks moving tissue, the area change can be calculated automatically frame by frame along the entire circumference of the cavity.
The aim of the present study was to explore the possibility of applying VVI on TEE images of the descending aorta to quantify its elastic properties expressed as circumferential strain, rotational displacement, distensibility, and stiffness in patients with aortic stenosis (AS) or aortic regurgitation (AR). The VVI measures were compared with the M-mode–derived variables.
Materials and Methods
Study Population
The study population consisted of 85 patients included in the Advanced Study of Aortic Pathology, which is a prospective single-center study performed at the Karolinska University Hospital, Stockholm, Sweden. Consecutive patients with aortic valve disease or aneurysm of the ascending aorta requiring surgery were recruited if they had no significant coronary artery disease according to coronary angiography. The 85 patients were divided according to the predominant lesion into AS or AR groups on the basis of preoperative TTE and confirmed by the intraoperative TEE and surgical findings. A combination of equal severity of AS and AR was described in two patients, and they were excluded from subgroup analysis because that group was too small for statistical analysis. Additional analyses of subgroups of patients with “pure” AS and “pure” AR were performed. Eighty-two patients (97%) had sinus rhythm, whereas the remaining three patients had well-treated atrial fibrillation, with heart rate within normal range.
Each subject gave written consent to participation in the study, which was approved by the Ethics Committee of the Karolinska Institute Stockholm, Sweden.
Transthoracic Echocardiography
All patients were examined according to the study protocol by TTE before surgery. For the TTE, a Philips iE33 ultrasound scanner (Philips Medical Systems, Bothell, WA) was used. Two-dimensional echocardiography and Doppler measurements were performed according to standards outlined by the American Society of Echocardiography. To grade the AS peak and mean, transvalvular gradients were calculated using the Bernoulli equation and aortic valve area, according to the continuity equation. The AR color flow jet area, vena contracta, pressure half-time, jet density, and diastolic flow reversal in the descending aorta were used for classification of AR into grades 1 to 3 (1 = mild; 2 = moderate; 3 = severe). Grade 0.5 was used for minimal AR. “Pure” AS was defined as a mean gradient > 40 mm Hg and AR grade < 1, and “pure” AR was defined as AR grade 3 and a mean gradient < 20 mm Hg. Patients with significant pathology of other valves were excluded.
Transesophageal Echocardiography
All patients were examined according to the study protocol by TEE on the operating table under general anesthesia before surgery. A Siemens Sequoia c 512 ultrasound scanner (Siemens Medical Systems, Mountain View, CA) with a transducer frequency of 6 or 7 MHz was used in all patients. Two-dimensional echocardiography and Doppler measurements were performed according to standards outlined by the American Society of Echocardiography. The gain, depth, and sector width were individually adjusted. Two-dimensional image sequences for VVI analysis were recorded with the highest possible frame rate (mean 71 Hz) using the acoustic capture function. Short-axis views of the descending aorta were acquired at three predefined distances from the teeth (35 ± 5 cm). We acquired three to four loops and registered M-mode at the same levels. Blood pressure was measured invasively in the radial artery simultaneously with the acquisition of the TEE images. Electrocardiogram was recorded and displayed on the ultrasound images.
Velocity Vector Imaging and M-Mode Analyses
All analyses were performed off-line with Syngo US WP 3.0 VVI (Siemens Medical Systems). The best loops (by visual assessment), preferably at the 35-cm level, were chosen, and the blood–intima border was traced manually. If tracking of the blood–intima border was unsatisfactory according to software indication or visual assessment, a new tracing was performed or another loop (at the same level) was chosen. Maximal systolic area (As), minimal diastolic area (Ad), area change over time (dA/dt), maximal systolic circumferential strain (VVI strain), and minimal systolic rotational displacement (VVI rot) were automatically calculated, and the relevant values of each variable were depicted from the respective curves ( Figure 1 B–D). The results shown are means of two to four cardiac cycles from the two levels with best image quality (except in three patients in whom only one level had adequate image quality). Strain values between the levels were compared.
The M-mode was recorded in short-axis view of the descending aorta, paying attention to the placement of the cursor in the center of the vessel, perpendicular to the aortic wall. The maximal systolic diameter (Ds) and minimal diastolic diameter (Dd) were measured along the M-line, from the leading edge of the near wall intima–lumen echo to the leading edge of the echo from the far wall lumen–intima interface as previously described, at the same levels as in the VVI analysis ( Figure 1 E). All measurements were averaged from three to four cardiac cycles. Five patients were excluded from the analysis because of poor image quality: VVI analysis was not possible in two patients, M-mode images could not be analyzed in two patients, and none of the methods could be used in one patient. In 80 patients, constituting 96% of the original sample, M-mode and VVI recordings could be compared.
Reproducibility
The intra- and interobserver variability of measurements were determined for VVI variables (As, Ad, dA/dt, VVI strain, and VVI rot) in 32 patients. For intraobserver variability, the same observer (J.P.) repeated the VVI measurements after more than 1 week, whereas interobserver variability was assessed from the measurements by two independent observers (J.P. and M.Y.). Interobserver variability of M-mode variables was assessed in 40 patients.
Calculation of Aortic Elasticity
From M-mode Ds and Dd, the following indices of aortic elasticity were calculated:
distensibility = ((Ds 2 ∗π/4) − (Dd 2 ∗π/4)∗10 7 )/(Dd 2 ∗π/4)∗PP∗1333 kPa −1 10 −3 ;
stiffness = ln (SBP/DBP)/((Ds-Dd)/Dd)
where SBP and DBP refer to radial systolic and diastolic blood pressures (mm Hg) and pulse pressure (PP) = SBP-DBP.
VVI distensibility was calculated as = As-Ad∗10 7/ Ad∗PP∗1333 kPa −1 10 −3 .
VVI stiffness was calculated with the same formula as for M-mode measurements, using diameters derived from the aortic area: systolic diameter = √ (4As/π) and diastolic diameter = √ (4Ad/π). The systolic diameter obtained from VVI was compared with M-mode Ds.
Statistical Analysis
Analyses were performed using the commercially available software Statistica 8.0 (StatSoft, Inc., Tulsa, OK). Normally distributed data are presented as arithmetic mean and standard deviation. For comparisons between groups, t test or Mann–Whitney U test was used for independent samples according to the distribution. The chi-square test was used to analyze variables on a nominal scale. Inter- and intraobserver reliability were analyzed by the intraclass correlation coefficient (ICC), which represents the portion of the total variance caused by the variance between subjects. An ICC greater than 0.75 represents excellent reproducibility, whereas values between 0.4 and 0.75 represent fair to good reproducibility. The standard error of measurement was also calculated, as well as the coefficient of variation (CV%). To compare the VVI and M-mode methods, a one-way repeated-measures analysis of variance was performed. Correlations between the methods were estimated using Pearson’s product-moment correlation coefficients. Limits of agreement, according to Bland–Altman, were also calculated and graphically presented. Univariate and forward stepwise multiple linear regression analyses were used to evaluate predictors of different VVI measures. P < .05 was considered statistically significant.
Results
The clinical characteristics and echocardiographic data of the study population are shown in Table 1 . The majority of the patients had AS (62%). The patients with AR were significantly younger, taller, and more often of male gender compared with those with AS.
| All n = 85 ∗ | AS n = 54 | AR n = 29 | P value | |
|---|---|---|---|---|
| Age (y) | 65.7 ± 11.4 | 68.8 ± 10.2 | 60.6 ± 11.7 | <.002 |
| Gender (male/female) | 53/32 | 29/25 | 22/7 | <.015 |
| Height (m) | 1.73 ± 0.11 | 1.71 ± 0.10 | 1.77 ± 0.10 | <.010 |
| Weight (kg) | 80.3 ± 15.7 | 79.2 ± 17.5 | 82.8 ± 11.9 | <.272 |
| BSA (m 2 ) | 1.94 ± 0.22 | 1.91 ± 0.23 | 2.00 ± 0.18 | <.113 |
| Systolic blood pressure (mm Hg) | 108 ± 15 | 111 ± 14 | 103 ± 15 | .027 |
| Diastolic blood pressure (mm Hg) | 57 ± 11 | 62 ± 9 | 49 ± 11 | <.0001 |
| Pulse pressure (mm Hg) | 42 ± 22 | 49 ± 12 | 54 ± 14 | .081 |
| LVEDD (mm) | 51.5 ± 9.2 | 46.7 ± 5.1 | 60.2 ± 8.8 | <.0001 |
| LVESD (mm) | 35.9 ± 8,5 | 31.9 ± 5.3 | 43.3 ± 8.8 | <.0001 |
| EF (%) | 57.7 ± 7.5 | 59.4 ± 6.0 | 54.7 ± 9.1 | .005 |
| FS (%) | 30.7 ± 7.3 | 32.0 ± 6.8 | 28.4 ± 8.0 | .035 |
| AR grad (1-3/3) | 1.4 ± 1.2 | 0.6 ± 0.5 | 2.5 ± 0.7 | <.0001 |
| AV PG (mm Hg) | 59.9 ± 37.9 | 83.6 ± 24.7 | 16.1 ± 9.1 | <.0001 |
| AV MnG (mm Hg) | 37.4 ± 24.7 | 52.9 ± 16.4 | 9.0 ± 4.7 | <.0001 |
| AVA (cm 2 ) | 1.5 ± 1.1 | 0.8 ± 0.3 | 3.0 ± 0.8 | <.0001 |
∗ Two patients with a combination of equal severity of AS and AR were excluded from subgroup analysis.
Velocity Vector Imaging Data
The results of the VVI analysis are shown in Table 2 . Patients with AR had a significantly greater systolic and diastolic aortic area, a higher dA/dt, and a higher VVI strain than patients with AS. VVI rotational displacement differed significantly between the groups. However, the standard deviation of this variable was wide, in comparison with other variables. The VVI distensibility was higher and the VVI stiffness was lower in patients with AR than in AS ( Table 2 ). There was no significant difference in VVI strain between the proximal and distal levels of the descending aorta (5.3% ± 3.8% vs. 5.0% ± 3.5%, P = .58).
| AS n = 54 | AR n = 29 | P value | |
|---|---|---|---|
| VVI | |||
| Aortic area, systole (cm 2 ) | 4.7 ± 1.1 | 5.9 ± 1.4 | <.001 |
| Aortic area, diastole (cm 2 ) | 4.3 ± 1.1 | 5.0 ± 1.4 | <.014 |
| dA/dt (cm 2 /s) | 3.1 ± 1.0 | 7.5 ± 3.5 | <.001 |
| VVI strain (%) | 3.7 ± 1.9 | 7.6 ± 4.5 | <.001 |
| Rotational displacement (degrees) | −0.85 ± 0.77 | −1.19 ± 0.77 | <.05 |
| M-mode | |||
| Aortic diameter, systole (cm) | 2.40 ± 0.32 | 2.70 ± 0.38 | <.01 |
| Aortic diameter, diastole (cm) | 2.30 ± 0.33 | 2.40 ± 0.39 | .084 |
| Aortic diameter difference (cm) | 0.14 ± 0.06 | 0.25 ± 0.13 | <.001 |
| Distensibility and stiffness from VVI and M-mode | |||
| VVI distensibility (kPa −1 10 −3 ) | 17.2 ± 7.2 | 27.1 ± 12.8 | <.001 |
| M-mode distensibility (kPa −1 10 −3 ) | 21.7 ± 10.6 | 32.8 ± 16.7 | <.004 |
| VVI stiffness (index) | 13.3 ± 6.0 | 10.5 ± 6.0 | <.010 |
| M-mode stiffness (index) | 11.2 ± 6.1 | 10.4 ± 9.1 | <.05 |
M-Mode Data
Patients with AR had wider systolic diameters than patients with AS. However, the diastolic diameters did not differ significantly. The M-mode strain and distensibility were higher in patients with AR in comparison with those with AS, whereas M-mode stiffness was higher in AS than in AR ( Table 2 ).
Comparison between M-Mode and Velocity Vector Imaging
There was no significant difference between Ds measured from M-mode and systolic dimension calculated from the VVI As. The aortic diameters measured by M-mode and VVI did not differ significantly in the AS (2.43 ± 0.29 cm vs. 2.41 ± 0.33 cm, P = .115) or AR group (2.72 ± 0.34 vs. 2.71 ± 0.38, P = .544), and the methods correlated significantly in the AS ( r = 0.94, P < . 0001) and AR ( r = 0.96, P < . 0001) groups.
Distensibility
Distensibility calculated by VVI was lower compared with distensibility by M-mode ( P < . 001), but there was a strong correlation ( r = 0.84) between the two methods with an ICC of 0.75. The Bland–Altman plots showed that the differences were independent of the mean values; however, there was a systematic bias of 5.22 kPa −1 10 −3 between the methods ( Figure 2 A).
Stiffness
The VVI stiffness was higher compared with M-mode stiffness ( P < . 003). There was a strong correlation ( r = 0.84) between the two methods with a high ICC of 0.79. The Bland–Altman plot showed a small systematic bias of −0.18 in the log-transformed differences between the methods ( Figure 2 B). There was a significant negative correlation between circumferential VVI strain and VVI-derived stiffness ( r = −0.64) and M-mode–derived stiffness ( r = −0.51). The dA/dt correlated moderately with VVI-derived stiffness ( r = −0.43).
In a subgroup analysis, patients with “pure” AR (n = 18) demonstrated significantly higher distensibility 38.8 ± 15.4 kPa −1 10 −3 compared with “pure” AS (n = 23) 21.4 ± 9.2 kPa −1 10 −3 ( P < . 0001) by VVI and M-mode (47.6 ± 19.0 kPa −1 10 −3 vs. 28.8 ± 16.2 kPa −1 10 −3 , P = .013). Stiffness was higher in “pure” AS than in “pure” AR, as measured by VVI (14.1 ± 7.5 vs. 8.8 ± 4.1, P = .008) and tended to be so by M-mode (11.7 ± 7.7 vs. 7.6 ± 4.5, P = .057).
Multiple Regression Analysis
Significant univariate relationships with clinical variables for VVI-derived strain and dA/dt are shown in Table 3 . In the stepwise multiple regression analysis, valve lesion and age were the only independent determinants (adjusted R 2 = 0.42, P < . 05) of the VVI strain. Furthermore, a significant interaction term consisting of age and valve lesion was found for the VVI strain, showing that the effect of the valve lesion on the measured variable was not the same in younger and older subjects ( Figure 3 ). When diastolic pressure was added to the model, it seemed to be a stronger determinant than the valve lesion and the R 2 value increased to 0.53 ( P < . 05). The independent determinants of dA/dt were age, valve lesion, and diastolic pressure ( R 2 = 0.61, P < . 05). Even for dA/dt, a significant interaction term was found, consisting of gender and valve lesion.
