Subjects

Fifty-one consecutive patients with AAD, diagnosed by contrast-enhanced CT, who underwent intraoperative echocardiography during emergency surgery at our institution between February 2009 and June 2011 were studied ( Figure 1 ). Ethical approval for this study was granted by the Sakakibara Heart Institute ethics committee, and patients gave informed consent before participation in the study.

Flowchart from diagnosis to surgery. Patients had chest or back pain and were transferred to the emergency room (ER) at our institution or other hospitals. Fifteen patients underwent contrast-enhanced CT (noncardiac mode) because of suspicion of aortic dissection and were diagnosed as having AAD at our institution. The others were diagnosed at other hospitals and transferred to our institution for emergency surgery (ES). All underwent ES with intraoperative TEE (IOE), and their coronary involvement was evaluated.

CT

All patients underwent contrast-enhanced CT (noncardiac mode) because of suspicion of aortic dissection and were diagnosed as having AAD. Fifteen underwent both early-phase and late-phase contrast-enhanced CT (SOMATOM Sensation 16; Siemens Healthcare, Forchheim, Germany) according to our routine protocol; the image acquisition of early-phase CT began 30 sec after the intravenous administration of contrast material at an injection rate of 3 mL/sec, and late-phase CT began 90 sec after contrast material injection. Computed tomographic images had a slice thickness of 5 mm and included not only transverse sections but also coronal and sagittal sections.

Thirty-six patients from other hospitals underwent contrast-enhanced CT according to unknown protocols. Some of these patients’ computed tomographic images were firm images, and coronal and sagittal sections were not always included. However, their images were also sections with slice thickness of 5 mm, and the image quality was sufficient to diagnose AAD.

From these computed tomographic images, we detected the intimal flap and both coronary ostia and evaluated whether there was coronary involvement.

TEE

Both 2D TEE and 3D TEE were performed using an iE33 ultrasound system (Philips Medical Systems, Bothell, WA) with a 3D matrix-array transesophageal transducer (X7-2t).

The transesophageal probe was introduced after the patient was anesthetized and an endotracheal tube was placed. On 2D TEE, after the orienting landmark of the aortic valve had been demonstrated at a distance of approximately 30 cm from the patient’s teeth, the coronary arteries were identified as two parallel lines originating from the aortic lumen. The intimal flap of dissection is visualized as a mobile linear echo separating the true lumen from the false lumen in the aortic root.

We depicted the aortic valve and proximal ascending aorta at approximately 135° (long-axis view) to see the coronary arteries and intimal flap by rotating the probe from medially to laterally and by moving up and down. Next, we depicted the aortic valve at approximately 45° (short-axis view) to see the coronary arteries and intimal flap. Then, by manipulating the probe to identify the proximal extension of the intimal flap, we attempted to depict both the coronary ostium and the flap on the same planes of both short-axis and long-axis views in diastole, during which coronary flow signals were mainly observed. Finally, we evaluated the spatial relationship between the intimal flap and the coronary ostium to diagnose whether there was coronary involvement. In a series of these procedures, we acquired approximately 10 images. If we suspected coronary involvement, several images were additionally acquired. We also assessed whether patients had aortic regurgitation, other valvular heart diseases, and concomitant diseases while acquiring additional images.

Next, a 3D data set was acquired using 3D zoom mode or full-volume mode with 3D TEE. Three-dimensional zoom mode was a magnified pyramidal volume approximately 60° × 60° in size, which included the aortic valve and both the left and the right coronary ostia. This pyramidal volume data set of two consecutive cardiac cycles was acquired and stored digitally. In full-volume mode, a wide-angle pyramidal volume containing the aortic valve and both coronary ostia was acquired by electrocardiographic gating to merge narrow pyramidal volumes into a sector of approximately 60° × 60° to 100° × 100°, obtained over four to seven consecutive heartbeats. Three-dimensional zoom imaging was useful in patients with arrhythmias because it can be acquired without dependence on electrocardiographic gating.

The acquired 3D data set was digitally stored for subsequent analysis using QLAB software (Philips Medical Systems) supplied with the iE33. The multiplanar reconstruction mode of QLAB-3DQ allows the extraction of any plane from the 3D data set. In diastole, two planes (green-edged and red-edged planes) showing the short-axis and long-axis views, respectively, of the aortic root were set through the proximal part of the left coronary artery (LCA) or the right coronary artery (RCA). By this adjustment, we were able to obtain optimally aligned cross-sectional short-axis and long-axis views of the aortic root depicting both the coronary ostium and the flap ( Figure 2 , Video 1 [available at www.onlinejase.com ]). We then evaluated the spatial relationship between the coronary ostium and the medial flap to diagnose whether there was coronary involvement.

Quad-screen display of a full-volume data set extracted by QLAB-3DQ. Red, green, and blue lines show the different cross-sectional planes. In this case, two orthogonal planes ( green-edged and red-edged planes ), which show the short-axis and long-axis views, respectively, of the aortic root, are set through the proximal part of the RCA ( arrows ) in diastole and adequately adjusted to identify also the proximal extension of the intimal flap ( asterisk ). These optimally aligned cross-sectional short-axis and long-axis views of the aortic root allow evaluation of coronary involvement in acute type A aortic dissection by 3D TEE (see Video 1 ).

We compared the results of computed tomographic and 2D and 3D transesophageal echocardiographic evaluations, which had been confirmed by two or more reviewers, with the operative findings.

Diagnostic Criteria for Coronary Involvement

From the 2D transesophageal echocardiographic images acquired in diastole, we classified the spatial relationship of the ostium with the flap into four types: (1) true lumen (the ostium branching from the true lumen; Figures 3 A-1 and 3 A-2, Videos 2–4 [available at www.onlinejase.com ]), (2) false lumen (the ostium branching from false lumen; Figure 3 B, Videos 5 and 6 [available at www.onlinejase.com ]), (3) dissection (the flap extending through the ostium into the coronary artery; Figure 3 C, Video 7 [available at www.onlinejase.com ]), and (4) unclear (relationship not assessable (Figures 3 D-1 and 3 D-2, Videos 8–10 [available at www.onlinejase.com ]). From these 2D transesophageal echocardiographic evaluations, types 2 and 3 fulfill the diagnostic criteria for coronary involvement. Likewise, we classified the spatial relationship into four types in the same manner from the 3D transesophageal echocardiographic evaluations, the extracted optimal images from the 3D data set in diastole, and types 2 and 3 were diagnosed as coronary involvement.

Two-dimensional transesophageal echocardiographic images and schematic diagrams illustrating four patterns of spatial relationship between the coronary ostium and intimal flap ( asterisk ): (1) true lumen (T), (2) false lumen (F), (3) dissection, and (4) unclear. Types 2 and 3 fulfill the diagnostic criteria of coronary involvement. (A-1) Short-axis view of the aortic root showing the intimal flap stopping at the edge of left coronary ostium, corresponding to type 1 (see Video 2 ). (A-2) Short-axis and long-axis views showing the intimal flap stopping at the edge of right coronary ostium, corresponding to type 1 (see Videos 3 and 4 ). (B) Short-axis and long-axis views showing the thrombosed false lumen (Th) and RCA (R) ( arrows ) branching from the false lumen, corresponding to type 2 (see Videos 5 and 6 ). (C) Short-axis view showing the intimal flap extending into the LCA (L), corresponding to type 3 (see Video 7 ). (D-1) Short-axis view showing the flap appearing to overlie the left coronary ostium, corresponding to type 4 (see Video 8 ). (D-2) Short-axis and long-axis views showing the flap appearing to overlie the right coronary ostium, corresponding to type 4 (see Videos 9 and 10 ). Ao , Aorta; LA , left atrium; LV , left ventricle; PA , pulmonary artery.

Reproducibility

Intraobserver variability was determined by having one observer repeat the evaluation of the status of the coronary ostia 2 weeks later using computed tomographic and 2D and 3D transesophageal echocardiographic images in 15 randomly selected patients. Interobserver variability was determined by having a second observer perform these evaluations in the same 15 patients. Intraobserver and interobserver variability values were calculated using κ statistics.

Statistical Analysis

Standard methods were used to calculate sensitivity, specificity, positive predictive value, and negative predictive value. Continuous data are expressed as mean ± SD. Categorical data are presented as numbers or percentages. Continuous variables were compared using Student’s t tests. Categorical variables were compared using Fisher’s exact tests or χ ² tests as appropriate. SPSS version 17.0 (SPSS, Inc., Chicago, IL) was used for statistical analyses.