Aortic Root Measurement by Cardiovascular Magnetic Resonance
Specification of Planes and Lines of Measurement and Corresponding Normal Values
Background— Cardiovascular magnetic resonance is widely used for aortic root visualization and measurement, but methods still need to be standardized. Our aim was to identify appropriate planes of acquisition and lines of measurement and record corresponding normal values.
Methods and Results— We studied 120 healthy volunteers, 10 of each gender in each decile between 20 and 80 years, by using a 1.5-T cardiovascular magnetic resonance system. Steady-state free precession cine acquisitions aligned with the left ventricular outflow tract in oblique sagittal and coronal orientations were used to locate 2 sinus planes that transected the root at its widest point in its maximally expanded systolic and at its end diastolic positions. We measured the cusp-cusp and the cusp-commissure dimensions in these cine planes, each as the average of 3. Diastolic cusp-commissure dimensions were smaller than diastolic cusp-cusp dimensions (32.0±3.5 mm versus 34.6±4.0 mm in men, 28.4±2.8 mm versus 30.7±3.3 mm in women, P<0.001 for both). The diastolic cusp-commissure dimensions increased by 0.9 mm per decade in men and 0.7 mm per decade in women (P<0.001 for both) and gave higher R2 values with respect to age and body surface area (0.40 for men, 0.27 for women) than diastolic cusp-cusp, systolic cusp-commissure, or sinus measurements made in the left ventricular outflow tract planes.
Conclusions— The results indicate the importance of consistent methods for measurement of the aortic root by cardiovascular magnetic resonance. We recommend diastolic cusp-commissure measurements, which yielded favorable R2 values with respect to age and body surface area and were found to correspond closely with reference echocardiographic root measurements recorded in the Framingham cohort. We recorded reference values for these and other possible aortic root measurements by cardiovascular magnetic resonance.
Received February 14, 2008; accepted July 9, 2008.
Accurate and reproducible measurements of the aortic root are needed for informed decision-making on the timing and nature of surgical replacement of the aortic valve and root. Aortic root size also has prognostic implications in patients with pathology such as Marfan disease,1 and clear visualization is important in relation to surgery that preserves the aortic valve and root. Measurements of aortic root dimensions may also be clinically relevant in patients with aortic regurgitation.2 A recent study has shown that in patients older than 65 years without known cardiovascular disease, aortic root dilatation is associated with risk factors for coronary heart disease and was predictive of congestive heart failure, stroke, cardiovascular disease mortality, and all-cause mortality.3
Clinical Perspective p 113
Cardiovascular magnetic resonance (CMR) is becoming increasingly available. It has the advantages of clear visualization of the aorta with unrestricted access and freedom from ionizing radiation. It allows repeated studies to monitor the progress of aortic root dilatation, potentially providing guidance on the timing and nature of surgical intervention, but data are lacking on normal dimensions as measured by this modality. Values for normal ranges have been established by echocardiography,4,5 using methods of acquisition that differ from those available to CMR. The acoustic windows through which the heart can be imaged by transthoracic echocardiography are limited,6 thus restricting lines of measurement, whereas CMR allows cine imaging in any plane, including planes that transect the axis of the aortic root to display all 3 sinuses. This seems advantageous but raises the previously unaddressed question of whether the line of measurement (which can either be from cusp-cusp or from cusp-commissure) makes a difference and if so, by how much. This plane also allows measurement of cross-sectional area of the aortic root, which could reasonably be indexed to body surface area (BSA).
The purpose of this study was to measure aortic root dimensions by CMR using planes transecting, as well as aligned with, the axis of the aortic root and to establish appropriate methods and the corresponding reference values.
One hundred twenty normotensive volunteers were recruited, 10 of each gender in each decile between 20 and 80 years. Each volunteer was interviewed and the project was explained. Measurements of height, weight, and blood pressure were recorded (Table 1). BSA was calculated according to the Du Bois formula.7 Anyone with known coronary heart disease, valvular disease, heart failure, hypertension (systolic blood pressure >150 mm Hg, diastolic blood pressure of >90 mm Hg as measured by mercury sphygmomanometer), or atrial fibrillation was excluded. Ethical approval was obtained from the local regional ethics committee, and informed consent was obtained from all subjects.
Cardiovascular Magnetic Resonance
All subjects were imaged using a 1.5-T system (Siemens Sonata, Erlangen, Germany). Transaxial and coronal half-fourier acquisition single-shot turbo spin echo (HASTE) multislice pilot images were acquired in diastole during held expiration. Steady-state free precession cine acquisitions were then acquired, also during held expiration, using the following parameters: echo time of 1.57 ms, 15 segments, repetition time of 46 ms without view sharing, slice thickness of 7 mm, field of view of 350 mm × 263 mm, and pixel size of 1.4 mm × 2.2 mm. To acquire cine planes, an oblique sagittal cine was aligned orthogonal to the coronal scouts in the axis of the left ventricular outflow tract (LVOT) and proximal ascending aorta (Figure 1A). An oblique coronal cine acquisition was then located orthogonal to the oblique sagittal cine, aligned with the axis of the LVOT (Figure 1B). Both of these LVOT cines showed movement of the aortic root through the cardiac cycle. The appropriate frames from these were then used to align 2 cine planes transecting the aortic root at sinus level, 1 aligned relative to the phase of maximal systolic distension (Figure 1C) and the other at end diastole (Figure 1D). These 4 cine planes will be referred to as the sagittal LVOT, coronal LVOT, systolic sinus, and diastolic sinus planes, respectively.
The measurements were performed by a single observer (E.D.B.) using CMRtools (Cardiovascular Imaging Solutions, London, United Kingdom) to manually define lines and areas. Images were loaded in DICOM 3.0 format to the software in which pixel dimensions provided the basis for calculations of length and area, the latter using the rasterization of contours onto a subpixel grid, the technique having been verified during the development of the software by using magnetic resonance images of a calibration phantom.
Measurements in Sagittal and Coronal LVOT Planes
In the sagittal and coronal LVOT planes (Figure 1A and 1B), maximum systolic and end diastolic measurements were made at 3 levels: the level of the aortic annulus, the level of the maximum diameter across the sinuses, and at the sinotubular junction, as illustrated in Figure 1A and 1B.
Measurements in Sinus Planes
In each of the systolic and diastolic sinus planes, 2 sets of 3 aortic root dimensions were measured, 3 cusp-cusp and 3 cusp-commissure (Figure 1C and 1D). These linear measurements were made in the relevant sinus planes at maximum systolic distension and at end diastole, and the average of each set of 3 measurements was taken. Each separate measurement was also recorded as right coronary cusp (RCC) to left coronary cusp (LCC), LCC to noncoronary cusp (NCC), and NCC to RCC, and from each named cusp to its opposite commissure.
Aortic Root Area Measurement
The systolic and diastolic cross-sectional areas of the aortic root were measured by manual planimetry from the relevant sinus plane at maximum systolic distension and at end diastole.
Indexed Aortic Root Measurements
The aortic root linear and area measurements were indexed to BSA and to height, the latter being a measure of body size that is relatively independent of obesity.
The means, standard deviations, and 95% confidence intervals of all linear dimensional measurements were calculated for each gender and for each decile. Aortic root dimensions were plotted with respect to age (Figure 2), and the data were analyzed for normality and the presence of outliers. The measurements of any outliers were checked to exclude input errors.
Paired t tests were performed to compare all aortic root dimensions in systole with those in diastole in both male and female cohorts. Unpaired t tests were performed to compare all aortic root dimensions between men and women.
Repeated measures analysis of variance was used to determine whether there were significant differences between each set of 3 cusp-cusp or cusp-commissure measurements. If significant differences were found, paired t tests with Bonferroni correction for multiple comparisons were performed to assess the differences between pairs of measurements.
Multiple linear regression analysis was performed by using the Statistical Package for the Social Sciences (SPSS), version 10 for Windows (SPSS Inc, Chicago, Ill), for all aortic root dimensions for both sexes, with age and BSA as independent variables. This was repeated using age and height as the independent variables. In each analysis, R2 values were determined to give the proportion of the variability in the aortic root measurements attributable to age and BSA and to age and height, respectively.
All variables analyzed were found to be approximately normally distributed. The means, standard deviations, and 95% confidence intervals of linear measurements in the systolic and diastolic sinus planes are shown in Table 2 together with the sinus measurements in the sagittal LVOT plane. Tables 3 and 4⇓ show the same sets of measurements indexed to BSA and height, respectively. Table 5 shows means, standard deviations, and 95% confidence intervals of cross-sectional aortic root areas.
Figure 2A shows diastolic linear aortic root dimensions at sinus level for males and females plotted against age. Figure 2B and 2C shows the same measurements indexed to BSA and height, respectively, plotted against age. Figure 3A shows diastolic aortic area measurements at sinus level for males and females plotted against age. Figure 3B and 3C shows the same measurements indexed to BSA and height, respectively, plotted against age.
Multiple Linear Regression Analysis
Table 6 shows the results of multiple linear regression analyses of linear and cross-sectional area measurements with age and BSA as independent variables. Diastolic results gave better R2 values with respect to age and BSA than systolic results in both male and female cohorts, and R2 values were better in males than in females. Table 7 shows the results of the multiple linear regression analyses of the same measurements but with age and height as the independent variables. Again, diastolic results gave better R2 values with respect to age and height than systolic results in both cohorts, with the male cohort giving better R2 values as before. The regression against height gave generally poorer R2 values than the regression against BSA.
Comparison of Male and Female Cohorts
The aortic root dimensions (all measured in millimeters) were larger in men than in women in all planes, as expected. Systolic measurements of cusp-cusp, cusp-commissure, and sagittal sinus LVOT dimensions were 36.0±4.1 versus 31.9±3.2, 33.6±3.4 versus 29.9±2.7, and 33.8±3.8 versus 30.7±3.2, respectively (P<0.001 for all). Diastolic measurements in the same planes were 34.6±4.0 versus 30.7±3.3, 32.0±3.5 versus 28.4±2.8, and 32.4±4.2 versus 29.0±3.3, respectively (P<0.001 for all).
Comparison of Systolic and Diastolic Measurements
In men, systolic cusp-cusp, cusp-commissure, and sagittal sinus LVOT measurements were larger than the corresponding mean diastolic values (36.0±4.1 versus 34.6±4.0, 33.6±3.4 versus 32.0±3.5, and 33.8±3.8 versus 32.4±4.2, respectively, P<0.001 for all). The same was true for the female cohort (31.9±3.2 versus 30.7±3.3, 29.9±2.7 versus 28.4±2.8, and 30.7±3.2 versus 29.0±3.3, P<0.001 for all).
Comparison of Cusp-Cusp With Cusp-Commissure Measurements
There were significant differences between cusp-cusp and cusp-commissure measurements. Systolic cusp-cusp measurements were larger than systolic cusp-commissure measurements (36±4.1 versus 33.6±3.4 in men, 31.9±3.2 versus 29.9±2.7 in women, P<0.001 for both), and diastolic cusp-cusp measurements were larger than diastolic cusp-commissure measurements (34.6±4.0 versus 32.0±3.5 in men, 30.7±3.3 versus 28.4±2.8 in women, P<0.001 for both).
Comparison of Individual Cusp-Cusp and Cusp-Commissure Measurements
Repeated measures analysis of variance showed that there was slight asymmetry between the nonaveraged cusp-cusp and cusp-commissure measurements of the aortic root in both systole and diastole (P<0.001 for all analyses), suggesting anteroposterior constraint (Table 8). For both male and female cusp-cusp measurements, the NCC-RCC was smaller than the LCC-NCC and the RCC-LCC measurements in both systole and diastole (male systolic: 34.8±4.3 versus 36.3±3.9, P<0.001, and 34.8±4.3 versus 36.7±4.7, P<0.001, respectively; male diastolic: 33.6±4.2 versus 34.9±3.8, P<0.001, and 33.6±4.2 versus 35.2±4.5, P<0.001, respectively; female systolic: 31.0±3.1 versus 32.1±3.0, P<0.001, and 31.0±3.1 versus 32.5±3.8, P<0.001, respectively; female diastolic: 29.7±3.2 versus 31.1±3.1, P<0.001, and 29.7±3.2 versus 31.3±3.9, P<0.001, respectively).
For both male and female cusp-commissure measurements, the RCC-commissure was smaller than the LCC-commissure and the NCC-commissure measurements in both systole and diastole (male systolic: 32.9±3.6 versus 34.1±3.4, P<0.001, and 32.9±3.6 versus 33.8±3.4, P<0.001, respectively; male diastolic: 31.4±3.8 versus 32.3±3.5, P<0.001, and 31.4±3.8 versus 32.3±3.5, P<0.001, respectively; female systolic: 29.3±3.0 versus 30.2±2.9, P<0.001, and 29.3±3.0 versus 30.2±2.8, P<0.001, respectively; female diastolic: 27.6±2.8 versus 28.9±3.0, P<0.001, and 27.6±2.8 versus 28.8±2.9, P<0.001, respectively).
Additional Linear Measurements
The additional linear measurements from the sagittal LVOT plane (annular and sinotubular junction) and all measurements from the coronal LVOT plane are shown in Table 9.
Effect of Age on Aortic Root Size
Diastolic cusp-commissure dimensions showed evidence of increase by 0.9 mm per decade in men (P<0.001) and 0.7 mm per decade in women (P<0.001), as shown in Figure 2. No statistically significant differences were found in the slopes of these plots between men and women.
This study shows that differences between serial measurements of the aortic root by CMR could arise from the inadvertent use of different lines of measurement, particularly at sinus level. It underlines the need for standardization of CMR measurements of the aortic root, an issue that does not seem to have been adequately addressed previously despite the widespread use of CMR for aortic visualization and measurement. Measurements at sinus level are important because this is normally the region of greatest aortic width where further dilatation can be associated with aortic regurgitation. Furthermore, the trefoil shape has particular implications for dimensional measurements. We made our measurements using steady-state free precession cine images, which are widely used because of the good blood-tissue contrast they provide, even when local blood velocities are low.
The cusp-commissure dimensions in the sinus planes were typically 2 to 3 mm smaller than the corresponding cusp-cusp measurements. Of the 2, the cusp-commissure measurements showed the more favorable R2 values with respect to age and BSA and also corresponded more closely with the reference echocardiographic aortic root measurements recorded in the Framingham study.4,5 The ultrasonic method used, recommended by the American Society of Echocardiography,8 measured the distance from the leading edge of the anterior aortic root wall to the leading edge of the posterior aortic root wall as identified in the end diastolic part of an M-mode trace, which can be located with the help of 2D echocardiography. The exact orientation of the line of insonation relative to the 3 sinuses may depend on the parasternal acoustic windows available and the orientation of the root in an individual but must generally pass through the anterior wall of the right coronary cusp and the posterior wall of the root at or close to the posterior commissure. It is not surprising, therefore, that the CMR end diastolic cusp-commissure root measurements and those at sinus level in the sagittal LVOT plane corresponded most closely with the echocardiographic measurements. It might have been expected, however, that the leading edge to leading edge method, by including the thickness of the anterior wall, would have given slightly higher measurements than those by CMR, which measures from one blood-wall boundary to another. In practice, the results were remarkably close (CMR diastolic cusp-commissure dimensions: 32.0±3.5 and 28.4±2.8, versus Framingham echo: 32±3 and 28±3, for men and women, respectively4), although the lines and methods of measurement by the 2 modalities are not identical. A contributory factor could be the slight asymmetry of the aortic root. Our separate, nonaveraged cusp-cusp and cusp-commissure measurements showed evidence of slight anteroposterior constraint, with the right coronary cusp to posterior commissure measurement being ≈1 mm less than the other 2 cusp-commissure measurements in both systole and diastole (Table 8).
Of our linear aortic root measurements, the average of 3 end diastolic cusp-commissure measurements gave the highest R2 values by multiple linear regression analysis with respect to age and BSA (Table 6). This indicates that the proportion of variability in aortic root size explained by age and BSA is highest for this measurement and is higher than corresponding Framingham echocardiographic values (CMR diastolic cusp-commissure: n=60, R2=0.40, and n=60, R2=0.27, versus Framingham echo: n=1848, R2=0.21, and n=2152, R2=0.22, for men and women, respectively5). The mean ages for men and women in our volunteers were similar to those of the Framingham cohort (49.3±17.2 and 49.2±16.6 in our volunteers versus 48±13 and 49±14 in the Framingham cohort, for men and women, respectively). Multiple regression analyses were also performed with age and height as independent variables (Table 7). Unlike BSA, height is relatively independent of obesity9 and is potentially more closely related to genetic predisposition. However, as in the Framingham echocardiographic study,4 we found poorer R2 values than when using age and BSA as independent variables (Table 7).
As the aortic root is displaced by ventricular contraction, it must be decided whether to locate a sinus plane with respect to the root’s systolic or diastolic position. A plane of cine acquisition that moves with the movement of the aortic root10 would have the advantage of allowing both systolic and diastolic measurements to be made from a single acquisition. If, as in this study, a fixed plane of cine acquisition and the end diastolic phase are to be used for measurement, then the plane must be located to pass through the sinuses at their end diastolic position.
Regarding the additional measurements of the aortic annulus and sinotubular junction made in the sagittal and coronal LVOT planes, the former showed relatively low R2 values with respect to age and BSA, and the latter relatively high. This may in part reflect the different structural and geometric characteristics of the 2 levels. What is widely referred to as the annulus tends to be ovoid in cross section and is bounded by a combination of muscular and fibrous components, none of which is strictly ring-like.11,12 The sinotubular junction, however, is nearly circular in cross section and more uniform in structure, which may facilitate consistent measurement.
Limitations of the Study
The healthy volunteers imaged in this study may have been able to comply better with the breath-holding requirements of CMR than some potential patients. Parallel imaging techniques were not used but could be implemented either to reduce breath-hold duration or to improve spatial or temporal resolution, albeit at the expense of signal-to-noise ratio in the resulting images.13 In this study of healthy volunteers, we acquired cine images with a temporal resolution of 46 ms. However, in future studies, particularly those including patients with tachycardia, view-sharing may be beneficial because it would effectively increase the temporal resolution to allow more precise identification of the end systolic and end diastolic frames.
Although linear regression analysis of measurements in our cohort with respect to age and BSA showed favorable R2 values relative to those of the large Framingham cohort, the 10 subjects in each decile group for each gender were not sufficient to determine the 5th and 95th percentile limits with respect to age and BSA.
The acquisition order of cines, from oblique sagittal to oblique coronal to the systolic and diastolic root planes, may have resulted in sequential increase in the precision of slice location relative to the mobile anatomy. Subsequent realignment of the sagittal LVOT cine relative to the coronal LVOT and sinus plane cines might have improved the accuracy of measurements in the sagittal LVOT plane and given higher R2 values for the corresponding annulus and sinotubular junction measurements.
On the grounds of favorable R2 values with respect to age and BSA (Table 6) and closeness of agreement with established reference echocardiographic aortic root measurements,4 we recommend that end diastolic cusp-commissure measurements are made in the diastolic sinus plane of cine acquisition. This plane allows all 3 sinuses to be visualized, and we recommend that the cusp-commissure measurements are recorded as the average of 3, unless the sinuses are found to be unusually asymmetrical, when it may be preferable to name and record the lines of measurement individually. We report normal reference values for these and other possible measurements of the aortic root from steady-state free precession cine acquisitions, including dimensions at annulus and sinotubular junction levels and cross-sectional area at sinus level.
Padial LR, Oliver A, Sagie A, Weyman AE, King ME, Levine RA. Two-dimensional echocardiographic assessment of the progression of aortic root size in 127 patients with chronic aortic regurgitation: role of the supraaortic ridge and relation to the progression of the lesion. Am Heart J. 1997; 134: 814–821.
Gardin JM, Arnold AM, Polak J, Jackson S, Smith V, Gottdiener J. Usefulness of aortic root dimension in persons ≥65 years of age in predicting heart failure, stroke, cardiovascular mortality, all-cause mortality and acute myocardial infarction (from the Cardiovascular Health Study). Am J Cardiol. 2006; 97: 270–275.
Vasan RS, Larson MG, Levy D. Determinants of echocardiographic aortic root size: the Framingham heart study. Circulation. 1995; 91: 734–739.
Anderson RH, Ho SY, Brecker SJ. Anatomical basis of cross-sectional echocardiography. Heart. 2001; 85: 716–720.
Sahn DJ, DeMaria A, Kisslo J, Weyman A. Recommendations regarding quantitation in M-mode echocardiography: results of a survey of echocardiographic measurements. Circulation. 1978; 58: 1072–1083.
Anderson RH. Clinical anatomy of the aortic root. Heart. 2000; 84: 670–673.