Aortic root measurement on CT: linear dimensions, aortic root area and comparison with echocardiography. A retrospective cross sectional study

Abstract

Objective:

We sought to assess the different CT aortic root measurements and determine their relationship to transthoracic echocardiography (TTE).

Methods:

TTE and ECG-gated CT images were reviewed from 70 consecutive patients (mean age 54 ± 18 years; 67% male) with tricuspid aortic roots (trileaflet aortic valves) between Nov 2009 and Dec 2013. Three CT planes (coronal, short axis en face and three-chamber) were used for measurement of nine linear dimensions. TTE aortic root dimension was measured as per guidelines from the parasternal long axis view.

Results:

All CT short axis measurements of the aortic root had excellent reproducibility (intraclass correlation coefficient, ICC 0.96–0.99), while coronal and three-chamber planes had lower reproducibility with ICC 0.90 (95% CI 0.84–0.94) and ICC 0.92 (0.87–0.95) respectively. CT coronal and short axis maximal dimensions were systematically larger than TTE (mean 2 mm larger, p < 0.001), while CT cusp to commissure measurements were systematically smaller (CT RCC-comm mean 2 mm smaller than TTE, p < 0.001). All CT short axis measurements had excellent correlation with aortic root area with CT short axis maximal dimension marginally better than the rest (Pearson’s R 0.97).

Conclusion:

Systematic differences exist between CT and TTE dependent on the CT plane of measurement. All CT short axis measurements of the aortic root had excellent reproducibility and correlation with aortic root area with maximal dimension appearing marginally better than the rest. Our findings highlight the importance of specifying the chosen plane of aortic root measurement on CT.

Advances in knowledge:

Systematic differences in aortic root dimension exist between TTE and the various CT measurement planes. CT coronal and short axis maximal dimensions were systematically larger than TTE, while CT cusp to commissure measurements were smaller. CT readers should indicate the plane of measurement and the specific linear dimension to avoid ambiguity in follow-up and comparison.

Introduction

Progressive dilatation of the aortic root confers risk for rupture and the indication for prophylactic surgery is partly dependent on size. Multiple imaging modalities exist for measurement of the aortic root. Normative data and thresholds for surgery have been defined by transthoracic echocardiography (TTE) at the mid sinus level^1–3 which is typically used for initial assessment and surveillance.^4–6

Multidetector ECG-gated CT is increasingly used for measurement of the aortic root and offers a high spatial resolution three-dimensional data set from which the complex geometry of the aortic root can be appreciated. There are several CT planes from which multiple measurements of the aortic root at its largest dimension, typically at the mid-sinus level, can be made with no clear consensus on the optimal method^4,7–13 (Figure 1). Additional measurement variability is introduced by timing of assessment (systole vs diastole) with guidelines suggesting diastolic measurement to ensure reproducibility.⁵ A short axis (en face) multiplanar reformatted plane of the aortic root is commonly created on CT allowing definition of cusp to commissure and cusp to cusp dimensions (Figure 2). However, neither of these parameters is necessarily the maximal dimension that can be obtained nor directly equivalent to echocardiography. When reconstructed on CT, the standard parasternal long axis TTE measurement of the aortic root typically passes somewhere between the right and non-coronary cusps with poorly defined landmarks.

Figure 1.

Different CT planes of aortic root measurement with corresponding TTE view. Leading edge to leading edge measurement convention utilized for TTE; anterior aortic wall thickness included with care taken to exclude right ventricular outflow tract wall thickness. TTE, transthoracic echocardiography.

Figure 2.

CT planes of aortic root measurement. From the short axis aortic root plane, anatomically landmarked (mid-cusp to mid-cusp, and mid-cusp to opposite commissure) and non-anatomically landmarked (maximal) dimensions can be measured. White arrows denote cusp mid-point.

A two-dimensional (2D) area taken from the CT short axis aortic root plane would appear to be a better reflection of total root size, and there is recent evidence for its use in prognosis.¹⁴ However, the relationship between aortic root area and (more commonly used and expeditiously obtained) linear measurements is unknown.

Unlike CT with multiple measurement planes, clinical TTE assessment of aortic root dimension is typically only undertaken from the parasternal long axis view. TTE aortic root measurement is not ordinarily performed from the parasternal short axis view due to the absence of a simultaneous corresponding orthogonal imaging plane, and inability to confirm the height of measurement.

With current absence of agreement and inconsistency in choice of dimension reported by CT readers, we sought to determine the optimal CT plane for aortic root assessment on basis of reproducibility and magnitude of correlation with CT measured root area.

With threshold for surgery based on TTE data, we also aimed to assess the presence/absence of systematic bias between TTE and the various CT planes of aortic root measurement.

Methods and materials

Patients

Consecutive patients who had undergone TTE and ECG-gated cardiac CT within 30 days of each other from November 2009 to December 2013 at a single tertiary institution were retrospectively identified from Radiology Information Systems and echocardiography databases. Institutional ethics approval was obtained, and the nature of this study was deemed to not require patient consent.

All examinations were clinically indicated. Exclusion criteria were congenital heart disease affecting the aortic valve (e.g. bicuspid), previous aortic root surgery including aortic root repair and prosthetic valve replacement, aortic root dissection, and technical limitations precluding accurate aortic root measurement on CT due to artifact.

CT

Contrast-enhanced ECG-gated CT of the heart and aortic root was performed on a dual-source multidetector CT scanner (Siemens Somatom Definition FlashTM, Erlangen, Germany) with iodixanol contrast (VisipaqueTM, GE Healthcare, Milwaukee, WI). Syngo.via VB10B (Siemens, Erlangen, Germany) was used for CT image analysis.

Three multiplanar reformatted planes were created in diastole (Figure 1) – (1) coronal (coronal plane of the chest automatically reconstructed by the workstation perpendicular to axial scan plane); (2) short axis aortic root (manually reconstructed true cross-sectional en face view of the aortic root perpendicular to the two orthogonal long axes of the aorta at the mid-sinus level); and (3) three-chamber (manually reconstructed double oblique reformats centered on the mitral valve so that the left atrium, left ventricle, and proximal thoracic aorta are visible in one image with the MPR long axis reference line passing through the left ventricular apex and mid mitral valve, analogous to the parasternal long axis view on TTE). The best diastolic phase as generated by scanner was used for measurement and both readers used the same phase.

The aortic cusps are named by coronary artery origin; no variant coronary artery anatomy was encountered during the study. The commissures are named by the cusps that comprise them. All measurements were made at the largest dimension, typically at the mid-sinus level.

Nine linear measurements of the aortic root at mid-sinus level were taken from the three reformatted planes using the inner edge to inner edge (I-I) convention⁴ (Figure 2), defined as below:

• Coronal plane (coronal).

• Short axis aortic root plane: midpoint of one cusp to another (mid cusp to mid cusp).

  • Right cusp to left cusp (RCC-LCC).

  • Left cusp to non-coronary cusp (LCC-NCC).

  • Non-coronary cusp to right cusp (NCC-RCC).

• Short axis aortic root plane: mid cusp to opposite commissure.

  • Right cusp to left/non-coronary commissure (RCC-comm).

  • Left cusp to non-coronary/right commissure (LCC-comm).

  • Non-coronary cusp to right/left commissure (NCC-comm).

• Short axis aortic root plane: maximal possible dimension assessed visually regardless of cusp/commissure relation (maximal)

• Three-chamber plane (3Ch)

Aortic root area was also measured from the reconstructed short axis plane by manual tracing. Aortic wall thickness was excluded with tracing of the lumen only.

Two CT radiologists (HK, HP) independently reconstructed the cardiac CT views and made their measurements on separate occasions, blinded to the other’s results.

TTE

All patients underwent comprehensive TTE assessment on commercially available echocardiography systems (Vivid E9 and Vivid S70, General Electric Company, Chicago IL). All studies were performed in accordance with current American Society of Echocardiography guidelines.¹⁵

Studies were retrieved and analyzed using Prosolv (Fujifilm, Tokyo, Japan). As per current chamber quantification guidelines,¹⁵ aortic root measurements were taken from 2D imaging in the zoomed parasternal aortic long axis view at end-diastole, using the leading edge to leading edge (L-L) convention at mid sinus level (Figure 1).

Two echocardiologists (KL, SB) made their aortic root measurements on separate occasions blinded to each other’s results.

Statistics

MedCalc (v. 19.0.5) was used to perform statistical analysis. Continuous data are presented as mean ± one standard deviation. Intraclass correlation coefficient (ICC) estimates and their 95% confidence intervals were calculated to assess TTE and CT interobserver reproducibility. Comparison between CT and TTE measurements was performed using Bland–Altman analysis. Pearson’s correlation coefficient was used to assess the relationship between CT aortic root area and linear measurements from TTE and all CT planes. Two-tailed Student’s t-test was performed to assess the significance of observed differences between TTE and CT measurements. A p-value of <0.05 was considered significant.

Results

Demographics

A total of 70 patients were included with a mean age of 54 years (±18 years) and 67% were male (47/70). The median time between CT and echocardiographic examinations was 4 days (IQR 13 days). The main indications for CT were ‘exclusion of coronary artery disease’ (36/70; 51%) and ‘pre-cardiac surgery’ (14/70, 20%).

Reproducibility

Mean aortic root measurements of this patient cohort by modality and measurement plane are summarized in Table 1. Aortic root area measured on CT had the highest interobserver reproducibility (ICC 0.99, 95% CI 0.9–1.0) compared to all linear measurements including TTE. Linear measurement reproducibility remained excellent (all ICC >0.9), and was highest with the CT short axis plane (ICC 0.98 for maximal dimension). The reconstructed three-chamber plane had poorer reproducibility (ICC 0.92, 95% CI 0.87–0.95) and the lowest was for the coronal plane (ICC 0.90, 95% CI 0.84–0.94) (Table 1).

Table 1.

TTE and CT aortic root measurements at different planes, and respective interobserver reproducibility

Measurement	Mean (mm)	Mean difference (cardiac CT - TTE) in mm	P value	Standard deviation of mean CT - TTE difference (mm)	ICC	95% CI
TTE	32.9	N/A	N/A	N/A	0.95	0.88–0.97
CT coronal plane
Coronal	34.1	1.2	<0.0001	2.12	0.90	0.84–0.94
CT short axis plane
RCC-LCC	31.4	−1.5	<0.0001	1.88	0.96	0.89–0.98
LCC-NCC	31.2	−1.7	<0.0001	2.43	0.98	0.90–0.99
NCC-RCC	30.6	−2.3	<0.0001	1.84	0.97	0.91–0.99
RCC-comm	30.9	−2.0	<0.0001	1.63	0.98	0.96–0.99
LCC-comm	32.3	−0.6	0.012	1.77	0.97	0.93–0.99
NCC-comm	31.8	−1.1	<0.0001	1.90	0.98	0.97–0.99
Maximal	35.3	2.4	<0.0001	2.13	0.98	0.97–0.99
CT 3Ch plane
3Ch	32.0	−0.9	0.0007	2.04	0.92	0.87–0.95
CT short axis plane
Area	N/A	N/A	N/A	N/A	0.99	0.90–1.00

ICC, Intraclass correlation coefficient;TTE, transthoracic echocardiography.

p-values represent difference between CT and TTE using paired, two tailed t-test.

Systematic differences/Bland–Altman analyses

When comparing TTE and various linear CT measurements of the aortic root, small but statistically significant systematic biases were noted (all p < 0.05). Of the linear CT measurements, the mid-cusp to opposite commissure and mid-cusp to mid-cusp dimensions were systematically smaller than TTE (mean ranging from −2.3 to −0.6 mm) (Figure 3). The maximal dimension (without cusp/commissure relation) obtained from the CT short axis plane was on average 2.4 mm larger than TTE (Figure 3a).

Figure 3.

Bland–Altman analysis of different CT plane and mean CT/TTE measurements. All CT measurements (panels c to f) except for short axis maximal dimension and coronal plane (panels a and b) underestimated TTE aortic root size. Red lines indicate mean CT-TTE difference. TTE, transthoracic echocardiography.

Root measurement from the CT coronal plane tended to be 1.2 mm larger than TTE (Figure 3b). The CT short axis RCC-comm measurement that most closely approximates TTE was on average 2.0 mm smaller than TTE (Figure 3c). Whilst the CT three-chamber (−0.9 mm; Figure 3d) measurement also underestimated TTE aortic root dimensions, a large spread of measurement difference was noted, with CT underestimation of up to 7.7 mm likely due to variability in creation of this arbitrary plane on CT.

Correlation analysis with aortic root area

All linear measurements from CT and TTE had excellent correlation with area (all Pearson’s

R > 0.9, all p < 0.001) (Table 2). Notably, correlation was strongest with CT short axis plane maximal dimension (Pearson’s R 0.97) and weakest (though still excellent) with TTE measurements (Pearson’s R 0.92; Figure 4).

Table 2.

Pearson’s R correlation of all measurements (TTE and CT linear dimensions, and CT measured aortic root area)

	CT aortic root area (95% CI)	TTE	CT short axis (maximal dimension)	CT short axis (NCC-RCC)	CT short axis (RCC-comm)	CT 3Ch	CT coronal
CT aortic root area		0.924	0.969	0.959	0.965	0.941	0.965
TTE	0.924 (0.88–0.95)		0.882	0.905	0.926	0.896	0.878
CT short axis (maximal dimension)	0.969 (0.95–0.98)	0.882		0.946	0.935	0.913	0.955
CT short axis (NCC-RCC)	0.959 (0.94–0.97)	0.905	0.946		0.956	0.921	0.925
CT short axis (RCC-comm)	0.965 (0.94–0.98)	0.926	0.935	0.956		0.933	0.922
CT 3Ch	0.941 (0.91–0.96)	0.896	0.913	0.921	0.933		0.899
CT coronal	0.965 (0.94–0.98)	0.878	0.955	0.925	0.922	0.899

TTE, transthoracic echocardiography..

All p < 0.001.

Figure 4.

Pearson’s correlation between linear dimensions and CT measured aortic root area. Strongest correlation between CT short axis plane maximal dimension and area, beyond that of TTE. TTE, transthoracic echocardiography.

Discussion

In our cohort of patients with tricuspid aortic roots, we found that CT short axis plane measurements of the aortic root had greater reproducibility than those taken from the coronal and reconstructed three-chamber planes. Furthermore, of the linear CT measurements, coronal and short axis maximal dimensions were on average systematically larger than TTE, while CT mid-cusp to opposite commissure and mid-cusp to mid-cusp dimensions were systematically smaller.

Our findings of CT-TTE systematic bias are consistent with other data. Park and colleagues¹⁶ assessed 53 bicuspid aortopathy patients and found that TTE root measurement was smaller by a mean of 3.1 ± 2.6 mm compared to the CT short axis plane maximal dimension. In a mixed cohort (n = 112) of patients with bicuspid (n = 40) and tricuspid roots, Plonek et al¹⁷ similarly showed that standard TTE measurements (mean 44.8 ± 8.4 mm) were smaller compared to the CT short axis plane maximal dimension (mean 49.1 ± 9.0 mm).

Comparable to our data, Plonek’s group also demonstrated that dimensions obtained from the various CT planes differed significantly (sagittal, coronal and axial root dimensions all smaller than short axis maximal; all p < 0.05) but systematic comparison of CT dimensions to standard TTE measurement was not undertaken. They also assessed CT aortic root area by perimeter tracing (17.3 ± 6.3 cm²) and as a circle (19.2 ± 7.1 cm²) but area correlation with linear dimensions was not performed.

In this study, the CT short axis plane mid-cusp to opposite commissure and mid-cusp to mid-cusp measurements were on average 2 mm smaller than TTE. Conversely, the short axis plane maximal and coronal dimensions were on average approximately 2 mm larger. These findings were statistically significant. As depicted in Figure 5, the CT coronal measurement may occasionally be equivalent to the short axis maximal dimension and is significantly larger than the sagittal-like orientation of TTE measurement. While it is tempting to suggest a CT correction factor for echocardiographic reference ranges, caution is necessary due to the large spread of measurement difference in individual subjects in our study.

Figure 5.

Comparison of TTE with the various possible CT measurement dimensions. In the CT short axis plane (panels d and e), the typical PLAX TTE measurement (panel a) transecting both the right and non-coronary cusps is depicted by the white line. The pink line in panel E represents the CT three-chamber plane (panel b); the purple line in panel E represents the CT coronal plane (panel c). TTE, transthoracic echocardiography.

The reconstructed three-chamber plane on CT is intended to mirror the parasternal long axis view on TTE, and therefore one would expect good agreement between the two. While we found little systematic error between them, the random error was large (Figure 3d) which is attributable to the arbitrary nature of the plane of measurement. On TTE, this is dependent on the rib space from which an optimal acoustic window is found. When viewed en face, the TTE measurement usually relates to some dimension between the right and non-coronary cusps with no clear defining landmarks on CT, limiting precise reproducibility with the corresponding CT three-chamber measurement.

Technical differences in echocardiography may explain some (but not all) of the observed systematic measurement differences between the two modalities. The absence of simultaneous orthogonal biplane in TTE imaging is one of these apparent technical differences.

Utilization of the L-L convention for TTE measurement in contrast to I-I for CT would also contribute to minor but systematic differences in results, reflecting TTE inclusion of anterior aortic wall thickness with root luminal dimension. Pathological data have shown normal aortic root wall thickness to approximate 1.6 mm,¹⁸ and a similar measurement difference has been demonstrated between TTE I-I and L-L root measurements.¹⁹

Most importantly, single plane/linear measurements may fail to reflect the geometric complexities of the aortic root.^20,21 On basis of LaPlace’s Law, aortic area would the most reliable determinant of wall stress and consequent risk of rupture in situations of root dilatation.

Our data have shown very high reproducibility of the area metric, possibly beyond all linear dimensions. Masri et al¹⁴ recently showed that use of indexed CT short axis root area was superior to CT cusp to commissure dimension in the prediction of mortality and surgical intervention. Whilst all in their cohort had tricuspid roots, an abnormal cut point of 10 cm²/m² was empirically chosen from previous bicuspid and Marfan aortopathy literature.^22,23

Notably, a significant proportion (327 of 771, 42%) had aortic root dilatation (4.5–5.5 cm) below the current threshold for surgical intervention (>5.5 cm). Within this cohort of intermediate dilatation, 44% had an abnormal indexed area of which 78% died.

TTE is currently still the central imaging technique in a multimodality assessment paradigm pre-surgery. With further data, CT measured root area may become the gold-standard metric for triggering surgical intervention. However, an expeditious 2D measurement would still be useful. While all short axis CT measurements demonstrate good characteristics, our data suggest that the CT short axis plane maximal dimension may be the best in view of its excellent reproducibility and strong correlation with root area.

In our study, interobserver reproducibility was high for TTE root measurement (ICC 0.95, 95% CI 0.88–0.97). It should be noted that although measurements were made by two echocardiologists separately, these were performed on the same set of TTE clips for each patient. In contrast, CT measurement was performed by two radiologists using independently reconstructed views from axial data sets. It is likely that interobserver agreement of TTE root measurement would be poorer in actual clinical practice compared to our observations, with additional variability introduced by significance operator/sonographer dependence in TTE image acquisition. Despite this ostensibly higher TTE ICC in our study, we have shown that the reproducibility of the majority of CT linear dimensions still surpasses TTE.

Lastly, it is important to note that our CT measurements were performed in diastole. It is recognized that significant variability in aortic root size may arise across the cardiac cycle²⁴ and relates to aortic pulsatility, but is not clearly predictable. In accordance with current guidance,⁵ a diastolic phase of measurement was used to ensure reproducibility.

Study limitations

There are limitations to our data. This was a retrospective cross-sectional study of asymptomatic patients with mostly normal size tricuspid aortic roots. Only 17 patients (24%) had a dilated aortic root based on TTE (>37 mm). Patients with prior aortic root or valve surgery, congenital heart disease and aortic root dissection were excluded to ensure data consistency. It would be important to test our observations in other aortic root patient subsets (e.g. bicuspid roots) to determine generalizability. With regard to cusp nomenclature, we acknowledge that alternate terminology exists which relates the sinus to its spatial relationship with the pulmonary valve (i.e. right, left and non-facing sinus) and is particularly pertinent in congenital heart disease cohorts. However, we have retained use of common nomenclature relating to coronary artery origin to avoid confusion for non-congenital heart disease clinicians in our non-congenital disease cohort. An additional limitation is the relative age of our data set (2009–2013). However, it was still acquired with a second-generation dual source scanner with 64 detector rows. This scanner is still currently in clinical use in our service and satisfies contemporary UK NICE guidance for cardiac CT scanner performance. Lastly, data on body surface area/body mass index and co-morbidities (such as AF and chronic lung disease) were not collected, and these may bias towards poor TTE imaging.

Conclusion

There are systematic differences between CT and TTE measurements of the aortic root depending on the CT plane and specific dimension chosen. This is particularly relevant when CT is being used for surveillance intermittently along with TTE as the values may not be interchangeable. The often-used CT mid-cusp to opposite commissure measurement is systematically smaller than TTE, while the maximal dimension on CT is systematically larger. Additionally, with differences between the various CT planes even in the same patient, it would be important for CT readers to indicate the plane of measurement and the specific linear dimension to avoid ambiguity in follow-up and comparison. This may be also be efficiently achieved by appending a key image capture to CT reports, depicting the measured dimension. Of the CT linear aortic root measurements, the maximal dimension in the short axis plane has excellent reproducibility and correlates well with aortic root area.