Author + information
- Received May 6, 2002
- Revision received November 11, 2002
- Accepted November 27, 2002
- Published online April 16, 2003.
- ↵*Reprint requests and correspondence:
Dr. David Schwartzman, Cardiovascular Institute, UPMC Presbyterian, B535, 200 Lothrop Street, Pittsburgh, Pennsylvania 15213-2582, USA.
Objectives This study was designed to examine the dimensions and morphology of left atrial (LA) and distal pulmonary veins (PVs) using multidetector helical computed tomography (MDCT).
Background Detailed knowledge of LA and PV anatomy will assist in the development of techniques for ablative intervention. Multidetector helical computed tomography is a method for multidimensional imaging of cardiac anatomy.
Methods Multidetector helical computed tomography was used to image the LA and PVs in 70 subjects with and 47 subjects without atrial fibrillation (AF). Accuracy of the MDCT data was confirmed by correlation with echocardiography and endocardial electrogram recordings.
Results Left atrial and PV dimensions were significantly larger in AF versus non-AF subjects, men versus women, and subjects with persistent versus paroxysmal AF. There were no differences between groups in morphologic detail.
Conclusions Multidetector helical computed tomography images of the LA and PVs are accurate and provide detailed anatomic information. Significant differences in dimensions but not morphologic detail were apparent between groups.
The left atrium (LA) and distal pulmonary veins (PV) appear to play key roles in the initiation and sustenance of atrial fibrillation (AF). Techniques for AF ablation that focus on the LA are under development. Accurate, detailed information regarding LA/PV dimensions and morphology will promote this development. Using multidetector helical computed tomography (MDCT), multidimensional reconstruction of organ systems is possible (1). In the present report, we evaluate the use of MDCT for LA/PV imaging.
Multidetector helical computed tomography was performed in 117 subjects, who were divided into two groups: AF group, 70 subjects with paroxysmal or persistent AF; and no-AF group, 47 subjects without AF (Table 1).
1.2 MDCT technique
Image acquisition was performed using General Electric Lightspeed scanners (Milwaukee, Wisconsin). Iodinated contrast was given in a test dose (20 cc) to determine the moment of peak LA filling; subsequently, 125 cc was administered at a rate of 4 cc/s, after which scanning was performed during a single breath-hold, in 2.5-mm thickness steps: pitch = 6:1 high-speed mode, kVP = 120, mA = 230 to 350 (depending on body habitus), rotation time = 0.8 s. Duration of acquisition was 25 to 30 s. Scanning was not gated to the cardiac cycle.
Image reconstruction was performed using commercial General Electric software (Advantage version 3.1). By considering shape, continuity, and density criteria, sequential axial images were edited to remove all structures except LA and PV. Multidimensional reconstruction was then performed by coalescing these images, and presented as extra-atrial (Fig. 1) and intra-atrial (Fig. 2) vantages. Figure 3summarizes the MDCT measurements performed in this study.
To assess the accuracy of the MDCT technique, we correlated the measurements with echocardiographic measurements:
1. Transthoracic imaging, performed using commercial systems (Hewlett-Packard Inc., Palo Alto, California, or Acuson, Mountain View, California), was used to measure LA volume.
2. Intracardiac echocardiography (ICE) was performed using two systems: Boston Scientific (Natick, Massachusetts) (2): This system was used to image PV ostia. It is composed of a catheter incorporating a rotating transducer operating at 9 MHz. Acuson (AcuNAV) (3): This system was used to image the posterior LA, specifically the regions comprising the PV ostia. It is composed of a catheter incorporating a phased-array transducer programmed to operate at 7.5 MHz.
To further assess the accuracy of the MDCT technique, we correlated MDCT-designated endocardial locations with electrogram characteristics at those locations. Bipolar electrograms were recorded using a standard mapping catheter. The catheter was positioned at ostial (four per vein, equally spaced around the circumference) and nonostial (four per vein, each 1-cm proximal to vein ostium, equally spaced around the circumference), as defined by ICE. Electrograms were recorded during sinus rhythm (right veins) or during coronary sinus pacing (left veins).
Pulmonary vein ostium (atriovenous junction):the point of reflection of the parietal pericardium, as seen on the two-dimensional source images (Figs. 1 and 2). Corrected PV ostial circumference:ostial circumference/LA volume (Fig. 3). Ostial noncircularity:difference between maximum and minimum PV ostial diameters, expressed as (Diametermax− Diametermin/Diametermax(Figs. 2 to 4). ⇓Ostial branch:a PV branch joining within 5 mm of the atriovenous junction (Figs. 2, 5, and 6). Saddle:the region of tissue interposed between separate, ipsilateral veins (intervenous: Figs. 2 and 3) or between branches of the same vein (intravenous: Fig. 2). Common vein:coalescence of superior and inferior veins proximal to the junction with the LA body (Figs. 5 and 6). ⇓Supernumerary vein:an additional (neither superior nor inferior) vein(s), ⇓having an independent atriovenous junction (Figs. 2, 5, and 6).
Data are reported as mean ± SD, unless otherwise specified. A Student ttest was utilized to compare continuous variables, and a chi-square test to compare categorical variables. Correlations were performed using a Spearman test. For each test, a p value of <0.05 was considered significant.
Echocardiography and MDCT-derived LA volumes were significantly correlated (n = 70, r = 0.64, p < 0.001). Boston Scientific ICE and MDCT-derived PV ostial dimensions (35 subjects, 47 veins: left superior [LS] 15, right superior [RS] 12, left inferior [LI] 15, right inferior [RI] 5) were also significantly correlated: circumference (r = 0.78, p < 0.001); maximum diameter (r = 0.82; p < 0.001); minimum diameter (r = 0.85; p < 0.001). There was complete concordance between MDCT and Acuson ICE (26 subjects, 42 right and 42 left vein regions) in discerning common (n = 4) and supernumerary veins (n = 11), as well as ostial branches (n = 8).
A total of 38 veins in 32 subjects were mapped: LS 14, RS 9, LI 10, RI 5. At ostial sites, the incidence with which a typical fractionated or bifid PV ostium potential (Fig. 2) was recorded (86% of all sites in all subjects) was significantly greater than at nonostial sites (13%; p < 0.001).
2.1.3 Within-group (AF) MDCT analysis
In the AF group, superior vein ostial circumferences were significantly greater than inferior vein circumferences (Table 2) . There were no differences between corresponding right and left veins. Pulmonary vein circumference (aggregate for all veins: r = 0.43, p < 0.001) and maximal diameter (aggregate for all veins: r = 0.48, p < 0.001) were both significantly correlated with LA volume. Pulmonary vein circumference (aggregate for all veins: r = 0.23, p = 0.01) and maximal diameter (aggregate for all veins: r = 0.29, p = 0.001) were also significantly correlated with LA maximal planar diameter (X in Fig. 3). After “correction” for LA volume, there were no significant differences between PV ostial circumferences. The following morphologic observations, made in both AF and no-AF groups, require emphasis: 1) Supernumerary veins were common on the right, principally draining the middle lobe and less frequently the superior segment of the lower lobe (Figs. 2, 5, and 6). Conversely, supernumerary veins were rarely observed on the left. 2) The incidence of PV ostial branching was significantly higher on the right, and significantly more common in the inferior than superior vein (Figs. 1, 2, 5, and 6). 3) The average ostium-to-first branch distance in left veins was significantly greater than in right veins (Fig. 1). 4) A common PV occurred frequently on the left, but almost never on the right (Figs. 5 and 6). 5) Veins could be sharply angulated (Fig. 1). 6) Spatial overlap between proximal branches of separate (ipsilateral) veins was common (Fig. 1). 7) Generally, vein ostia were neither circular nor planar (Figs. 2 and 4). The magnitude of ostial noncircularity was significantly less in the RI than other veins.
2.2 Between-group MDCT analyses (Tables 2 to 4)
2.2.1 AF and no-AF (Table 2)
Left atrial and PV-ostial dimensions were significantly greater in the AF group. There were no differences between groups in vein multiplicity, ostial branching, ostium-to-first-branch distance, saddle length, or ostial noncircularity. Once “corrected” for LA volume, ostial circumferences were significantly greater in the no-AF group.
2.2.2 Men and women (Table 3)
Volume and most dimensions of the LA were significantly greater in men. Pulmonary vein ostial dimensions (before and after “correction” for LA volume) were similar. There were no significant differences between groups in vein multiplicity, ostial branching, ostium-to-first-branch distance, saddle length, or ostial noncircularity.
2.2.3 Paroxysmal and persistent AF (Table 4)
Volume and most dimensions of the LA were significantly greater in subjects with persistent AF. Pulmonary vein ostial dimensions (before and after “correction” for LA volume) were similar. There were no significant differences between groups in vein multiplicity, ostial branching, ostium-to-first-branch distance, saddle length, or ostial noncircularity.
Utilizing MDCT images, we have characterized multidimensional LA and PV anatomy. The accuracy of the MDCT images was supported by correlation with echocardiographic and electrographic indices. Although the differences in LA volume and dimensions in the between-group analyses were expected, it is important to detail such differences, particularly in regard to the development of interventional technologies. It is equally important to recognize the complex and heterogeneous nature of LA and PV anatomy.
We and others have previously reported on the use of multidimensional imaging of LA and PV using CT (4,5). The use of other imaging techniques, including cineangiography and magnetic resonance, has also been reported (6,7). Our data do not permit comment as to the relative merits of these techniques.
Our finding that, after correction for LA volume, all PV diameters were similar suggests that PV ostial dilation is a process associated with enlargement of the LA body. This is inconsistent with a pair of studies from the same group utilizing cineangiography (6)and magnetic resonance imaging (7), which concluded that the magnitude of dilation in superior veins in patients with AF was not proportionate to the degree of LA enlargement, a phenomenon not seen in the inferior veins. These authors speculated that this phenomenon could be related to the innate material properties of the superior veins. To better parallel these studies, we also performed a correlation between maximal LA diameter and PV ostial dimension. This was significantly less robust than the correlation with volume, suggesting that a single-diameter criterion is inadequate. This should not be surprising for a geometrically complex chamber such as the LA.
3.1 Study limitations
Subjects were clearly selected. Intergroup matching was performed retrospectively. Artifacts (“smudging”) were commonly observed on both extra-atrial (Figs. 1 and 5) and intra-atrial (Fig. 2) images. We believe that these artifacts were in part due to the lack of cardiac cycle gating (such as cardiac motion artifact). These artifacts may have limited the accuracy of our data.
Subsequent to the performance of this study, we have observed three patients with AF who had PV anatomical features that were not typical of the study cohort (Fig. 7). The accuracy of the CT image was corroborated in each patient by ICE imaging.
- atrial fibrillation
- intracardiac echocardiography
- left atrium
- left inferior pulmonary vein
- left superior pulmonary vein
- multidetector helical computed tomography
- pulmonary vein
- right inferior pulmonary vein
- right superior pulmonary vein
- Received May 6, 2002.
- Revision received November 11, 2002.
- Accepted November 27, 2002.
- American College of Cardiology Foundation
- Packer D.L.,
- Stevens C.L.,
- Curley M.G.,
- et al.
- Lin W.-S.,
- Prakash V.S.,
- Tai C.-T.,
- et al.