Left atrial appendage (LAA) contrast filling defects are commonly found in patients undergoing multidetector cardiac computed tomography (CCT) before catheter ablation of atrial fibrillation. Delayed CCT allows quantification of the LAA delayed/initial attenuation ratio and improves accuracy for LAA thrombus detection, which may obviate routine transesophageal echocardiography (TEE) before ablation. CCT with contrast-enhanced scans (initial CCT) and with noncontrast-enhanced scans (delayed CCT) was performed in 176 patients. LAA was evaluated for filling defects. LAA apex, left atrial (LA) body, and ascending aorta (AA) attenuations (Hounsfield units) were measured on initial and delayed cardiac computed tomograms to calculate LAA, LA, LAA/LA, and LAA/AA attenuation ratios. LAA, initial LAA/LA, and initial LAA/AA attenuation ratios differed significantly in patients with versus without filling defects on cardiac computed tomogram, those with atrial fibrillation versus normal sinus rhythm, and those with abnormal left ventricular ejection fraction versus larger LA volumes (p <0.05). In 70 patients (40%) who underwent TEE, 13 LAA filling defects were seen on initial cardiac computed tomogram. Two defects persisted on delayed cardiac computed tomogram and thrombus was confirmed on transesophageal echocardiogram. Sensitivity, specificity, and positive and negative predictive values of initial CCT for LAA thrombi detection were 100%, 84%, 15%, and 100%, respectively. With delayed CCT these values increased to 100%. Intraobserver and interobserver reproducibilities for cardiac computed tomographic measurements were good (intraclass correlation 0.72 to 0.97, kappa coefficients 0.93 to 1.00). In conclusion, delayed CCT provided an increase in diagnostic accuracy of CCT for detection of LAA thrombus in patients with atrial fibrillation before ablation, which may decrease the need for routine TEE before the procedure.
Multidetector cardiac computed tomography (CCT) has been well validated for the evaluation of left atrial (LA) and pulmonary vein anatomy and cardiac computed tomographic images may be integrated with electrophysiologic mapping to guide radiofrequency catheter ablation for atrial fibrillation. CCT is a well-established, but until recently, rarely used technique for imaging the intracardiac thrombus. Intracardiac thrombus is depicted on cardiac computed tomogram as a filling defect. LA appendage (LAA) contrast-filling defects are commonly found during CCT before ablation but many of these patients have no thrombus confirmed by transesophageal echocardiography (TEE). Previous studies have reported experiences with CCT to detect LAA thrombus, demonstrating high sensitivity and negative predictive value but lower specificity and positive predictive value. Studies using delayed CCT for exclusion of LAA thrombus and differentiation between thrombus and slow flow or spontaneous echocardiographic contrast are limited. Our hypothesis was that delayed CCT would improve the accuracy for LAA thrombus detection, which raises the possibility of obviating routine TEE, thus decreasing procedure costs and the number of imaging tests and avoiding the significant discomfort and risks associated with TEE.
Methods
From April 2009 through July 2010, 176 consecutive studies consisting of initial and delayed cardiac computed tomograms obtained at the Mount Sinai Medical Center (New York, New York) and the Winthrop University Hospital (New York, New York) were examined. Seventy of these patients were evaluated with TEE at the same institutions and these results were examined. Baseline characteristics including a score for congestive heart failure, hypertension, age ≥75 years, diabetes, and stroke (doubled) were determined from medical chart documentation. Institutional review board approval and waiver of informed consent were obtained.
CCT was performed using 256-slice (Brilliance iCT, Philips Medical System, Highland Heights, Ohio), 128-slice (Somatom Definition AS 128, Siemens Medical Solutions, Forchheim, Germany), or 64-slice (Lightspeed VR 64-multislice computed tomograph, General Electric, Milwaukee, Wisconsin) scanner technologies with similar protocols in patients in a supine position during suspended end-expiration. Prospective electrocardiographic gating was employed, with tube current applied to 75% of the RR interval, for patients in normal sinus rhythm. For patients in atrial fibrillation a nongated protocol was employed. Dose parameters included 200 to 300 mA and 100 to 120 kVp for initial CCT and 80 to 100 kVp for 1-minute delayed CCT, respectively. Radiation exposure was estimated from the dose–length product. Nonionic iodinated contrast agent (60 to 80 ml; Isovue 300, Bracco, Milan, Italy) was injected through an antecubital vein at a rate of 4 ml/s with a power injector followed by normal saline 50 ml. A bolus tracking technique was used to appropriately trigger image acquisition once attenuation in the left atrium reached a preset threshold of 100 HU. For initial CCT images were acquired from the level of the aortic arch to the dome of the diaphragm. Approximately 1 minute after contrast injection a noncontrast delayed cardiac computed tomogram was acquired from below the aortic arch to the middle of the left ventricle. Other than contrast media, no medication was used before the CCT.
Postprocessing of cardiac computed tomographic images was performed on a dedicated advanced image processing workstation (Extended Brilliance Workspace 3.5, EBV, Philips Medical System). Postprocessing included multiplanar, volume-rendered, and shaded-surface display reconstructions. Reconstructed cardiac computed tomographic images were reviewed and interpreted by 2 experienced independent investigators blinded to clinical data. The LAA was qualitatively evaluated in multiple axial planes for a filling defect, defined as incomplete visualization or opacification of the entire LAA with contrast. With postcontrast-delayed CCT it is possible to differentiate thrombus from pseudothrombus. Persistence of an LAA filling defect on delayed cardiac computed tomogram suggests the presence of LAA thrombus, whereas resolution suggests slow flow. For quantitative evaluation attenuation ratios were determined based on techniques previously described. LAA apex, LA body, and ascending aorta (AA) mean attenuations (Hounsfield units) were measured on initial and delayed cardiac computed tomograms using 1-cm 2 regions of interest to calculate (1) LAA attenuation ratio (ratio of LAA attenuation on delayed cardiac computed tomogram to initial cardiac computed tomogram), (2) LA attenuation ratio (ratio of LA attenuation on delayed cardiac computed tomogram to initial cardiac computed tomogram), (3) initial LAA/LA attenuation ratio (ratio of LAA to LA attenuation on initial cardiac computed tomogram), (4) delayed LAA/LA attenuation ratio (ratio of LAA to LA attenuation on delayed cardiac computed tomogram), (5) initial LAA/AA attenuation ratio (ratio of LAA to AA attenuation on initial cardiac computed tomogram), and (6) delayed LAA/AA attenuation ratio (ratio of LAA to AA attenuation on delayed cardiac computed tomogram). Using dedicated commercially available software (Cardiac Review, EPS, Philips Medical System) shaded-surface display of the left atrium excluding pulmonary veins was used to calculate LA volumes and maximum LA anteroposterior dimensions as previously described.
Transesophageal echocardiographic images acquired after standard clinical preparation using commercially available equipment (iE33, Phillips, Andover, Massachusetts; or Acuson Sequoia, Siemens, Mountain View, California) were interpreted by 2 experienced cardiologists blinded to clinical data. The LAA was graded as normal (no spontaneous echocardiographic contrast or thrombus), spontaneous echocardiographic contrast, or thrombus based on previous descriptions and the measured left ventricular ejection fraction was recorded.
SPSS 17 (SPSS, Inc., Chicago, Illinois) was used for statistical analysis. Continuous variables were described as mean ± SD. Categorical variables were presented as frequency and percentage. Continuous variables were compared between groups using t test. Categorical variables were compared using chi-square or Fisher’s test as appropriate. Multivariate logistic regression was used to evaluate independent predictors of contrast-filling defect on the first scan. Sensitivity, specificity, and positive and negative predictive values of initial and delayed CCT to detect LAA thrombus were determined assuming TEE as the reference standard. Intraclass correlation coefficients and Bland-Altman analysis were used to assess intraobserver and interobserver concordances in quantitative cardiac computed tomographic measurements in 10 randomly selected studies. Kappa coefficient was calculated to assess intraobserver and interobserver concordances on presence of a contrast-filling defect on initial and delayed cardiac computed tomograms. A p value <0.05 was considered statistically significant.
Results
One hundred seventy-six consecutive patients were included. Baseline characteristics of the overall population and according to presence (n = 40, 23%) or absence (n = 136, 77%) of an LAA filling defect on initial cardiac computed tomogram are listed in Table 1 . Mean radiation dose for delayed CCT was 1.25 ± 0.5 mSv. Patients with an LAA filling defect on initial cardiac computed tomogram had a significantly higher prevalence of hypertension, atrial fibrillation during CCT, and lower left ventricular ejection fraction. LAA and LA volumes and LA diameter were also significantly larger in patients with a filling defect on initial cardiac computed tomogram. In multivariate analysis history of hypertension and LA volume remained independently associated with presence of a contrast-filling defect on initial cardiac computed tomogram.
| Variable | Overall (n = 176) | Without Defect (n = 136) | With Defect (n = 40) | p Value |
|---|---|---|---|---|
| Age (years) | 59.5 ± 12.4 | 58.9 ± 12.6 | 61.4 ± 11.7 | 0.72 |
| Age >75 years | 17 (10%) | 11 (8%) | 6 (15%) | 0.22 |
| Men | 132 (75%) | 99 (73%) | 33 (83%) | 0.21 |
| Body mass index (kg/m 2 ) | 30.2 ± 14.6 | 30.8 ± 16.5 | 28.2 ± 4.7 | 0.37 |
| Caucasian ethnicity | 141 (80%) | 107 (79%) | 34 (85%) | 0.38 |
| Chronic heart failure | 19 (11%) | 13 (10%) | 6 (15%) | 0.38 |
| Hypertension ⁎ | 107 (61%) | 75 (55%) | 32 (80%) | <0.01 |
| Hypercholesterolemia ⁎ | 68 (39%) | 49 (36%) | 19 (48%) | 0.19 |
| Diabetes mellitus | 24 (14%) | 21 (15%) | 3 (8%) | 0.20 |
| Stroke/transient ischemic event | 4 (2%) | 4 (3%) | 0 (0%) | 0.57 |
| Anticoagulation therapy | 147 (84%) | 111 (82%) | 36 (90%) | 0.21 |
| Antiplatelet therapy | 43 (24%) | 33 (24%) | 10 (25%) | 0.92 |
| Left ventricular ejection fraction (%) | 54.4 ± 7.7 | 51.6 ± 9.1 | 41.4 ± 12.7 | <0.01 |
| CHADS2 score | 0.07 | |||
| 0 | 57 (32%) | 51 (38%) | 6 (15%) | |
| 1 | 78 (44%) | 55 (40%) | 23 (57%) | |
| 2 | 32 (18%) | 23 (17%) | 9 (22%) | |
| ≥3 | 9 (5%) | 7 (5%) | 2 (5%) | |
| Atrial fibrillation during multidetector cardiac computed tomography | 85 (48%) | 55 (40%) | 30 (75%) | <0.01 |
| Gated multidetector cardiac computed tomography | 92 (52%) | 82 (60%) | 10 (25%) | <0.01 |
| 256-slice scanner | 135 (77%) | 102 (75%) | 33 (83%) | 0.32 |
| 120-kVp initial multidetector cardiac computed tomography | 161 (92%) | 124 (91%) | 37 (93%) | 0.79 |
| 80-kVp delayed multidetector cardiac computed tomography | 131 (74%) | 99 (73%) | 32 (80%) | 0.36 |
| Radiation dose (mSv) | 6.7 ± 3.0 | 6.3 ± 5.9 | 8.0 ± 3.3 | <0.01 |
| Left atrial volume (ml) | 147.0 ± 46.1 | 137.7 ± 41.3 | 179.0 ± 47.8 | <0.01 |
| Left atrial diameter (mm) | 46.7 ± 9.2 | 44.9 ± 8.6 | 52.2 ± 8.7 | <0.01 |
| Left atrial appendage volume (ml) | 12.8 ± 6.1 | 11.9 ± 4.6 | 16.0 ± 9.1 | <0.01 |
| Left atrial attenuation ratio † | 0.45 ± 0.14 | 0.46 ± 0.16 | 0.44 ± 0.13 | 0.38 |
| Left atrial appendage attenuation ratio † | 0.66 ± 0.50 | 0.52 ± 0.17 | 0.90 ± 0.79 | <0.01 |
| Initial left atrial appendage/left atrial attenuation ratio † | 0.77 ± 0.25 | 0.86 ± 0.17 | 0.45 ± 0.24 | <0.01 |
| Delayed left atrial appendage/left atrial attenuation ratio † | 0.95 ± 0.17 | 0.95 ± 0.14 | 0.93 ± 0.26 | 0.46 |
| Initial left atrial appendage/ascending aorta attenuation ratio † | 0.80 ± 0.24 | 0.89 ± 0.14 | 0.48 ± 0.25 | <0.01 |
| Delayed left atrial appendage/ascending aorta attenuation ratio † | 0.92 ± 0.18 | 0.93 ± 0.16 | 0.88 ± 0.24 | 0.21 |
⁎ Self-reported, on medication, or medical chart documentation.
Mean attenuation values in the AA, left atrium, and LAA were 323.0 ± 93.7, 338.3 ± 104.4, and 258.5 ± 108.3 HU, respectively, for initial CCT and 152.4 ± 39.8, 148.4 ± 55.2, and 139.2 ± 47.6 HU, respectively, for delayed CCT. LAA, initial LAA/LA, and initial LAA/AA attenuation ratios differed significantly in patients with versus without filling defects ( Table 1 ). LAA, initial LAA/LA, and initial LAA/AA attenuation ratios differed in patients with versus without persistent atrial fibrillation, decreased (<50%) versus normal left ventricular ejection fraction, and larger LA volume >140 ml (median value in study population) versus ≤140 ml ( Table 2 ).
| Variable | LAA Ratio | LAA/LA Initial Scan | LAA/AA Initial Scan |
|---|---|---|---|
| Rhythm | |||
| Atrial fibrillation | 0.75 ± 0.60 | 0.67 ± 0.25 | 0.71 ± 0.26 |
| Sinus rhythm | 0.58 ± 0.36 | 0.86 ± 0.22 | 0.88 ± 0.20 |
| p value | 0.031 | <0.001 | <0.001 |
| Left ventricular ejection fraction (%) | |||
| <50 | 0.70 ± 0.41 | 0.67 ± 0.25 | 0.72 ± 0.27 |
| ≥50 | 0.52 ± 0.25 | 0.81 ± 0.22 | 0.85 ± 0.21 |
| p value | 0.023 | 0.014 | 0.021 |
| Left atrial volume (ml) | |||
| ≥140 | 0.81 ± 0.64 | 0.87 ± 0.19 | 0.89 ± 0.15 |
| <140 | 0.52 ± 0.21 | 0.67 ± 0.27 | 0.70 ± 0.28 |
| p value | <0.001 | <0.001 | <0.001 |
Seventy patients (11 with LAA filling defect on initial but none on delayed cardiac computed tomogram, 2 with LAA filling defect on initial and delayed cardiac computed tomograms, and 57 without filling defects on initial and delayed cardiac computed tomograms) subsequently underwent TEE. Median time from CCT to TEE was 2.5 days (interquartile range 7.3) and 57 patients (81%) underwent the 2 tests on the same day. On transesophageal echocardiogram 23 patients had spontaneous echocardiographic contrast but no thrombus, 2 patients had thrombus, and 45 studies were normal (no spontaneous echocardiographic contrast or thrombus). Baseline characteristics and cardiac computed tomographic data of the subcohort according to transesophageal echocardiographic result are listed in Table 3 . Compared to the overall population baseline characteristics of the subcohort were not significantly different (data not shown). In addition, presence of a filling defect on initial cardiac computed tomogram (n = 13, 19%) was comparable to the overall population (n = 40, 23%). Patients with abnormal transesophageal echocardiogram (spontaneous echocardiographic contrast or thrombus) had a higher prevalence of chronic heart failure and worse score for congestive heart failure, hypertension, age ≥75 years, diabetes, and stroke (doubled) and a larger proportion had been on antiplatelet therapy compared to patients with normal transesophageal echocardiogram. LA, initial LAA/LA, and initial LAA/AA attenuation ratios were significantly different between the 2 groups. Figure 1 shows median LAA attenuation values on initial and delayed cardiac computed tomograms according to transesophageal echocardiographic result. As expected, attenuation values were lower in patients with thrombus. LAA attenuation on delayed cardiac computed tomogram was lower in patients with spontaneous echocardiographic contrast versus normal transesophageal echocardiogram (112.2 ± 24.6 vs 130.6 ± 31.6).
| Variable | Overall (n = 70) | Normal TEE (n = 45) | SEC or Thrombus (n = 25) | p Value |
|---|---|---|---|---|
| Age (years) | 57.8 ± 9.8 | 57.4 ± 9.9 | 58.4 ± 9.8 | 0.68 |
| Age >75 years | 2 (3%) | 0 (0%) | 2 (8%) | 0.12 |
| Men | 51 (73%) | 32 (71%) | 19 (76%) | 0.66 |
| Body mass index (kg/m 2 ) | 30.3 ± 6.0 | 30.2 ± 6.2 | 30.6 ± 5.8 | 0.77 |
| White | 61 (87%) | 39 (87%) | 22 (88%) | 1.00 |
| Chronic heart failure | 11 (16%) | 4 (9%) | 7 (28%) | 0.046 |
| Hypertension ⁎ | 40 (57%) | 25 (56%) | 16 (60%) | 0.72 |
| Hypercholesterolemia ⁎ | 26 (37%) | 17 (38%) | 9 (36%) | 0.88 |
| Diabetes mellitus | 16 (23%) | 9 (20%) | 7 (28%) | 0.56 |
| Stroke/transient ischemic event | 1 (1%) | 1 (2%) | 0 (0%) | 1.00 |
| Anticoagulation therapy | 58 (83%) | 38 (84%) | 20 (80%) | 0.74 |
| Antiplatelet therapy | 20 (29%) | 8 (18%) | 12 (48%) | <0.01 |
| Left ventricular ejection fraction (%) | 54.6 ± 11.6 | 50.9 ± 9.0 | 47.6 ± 12.4 | 0.25 |
| CHADS2 score | 0.048 | |||
| 0 | 23 (33%) | 15 (33%) | 8 (32%) | |
| 1 | 29 (41%) | 23 (51%) | 6 (24%) | |
| 2 | 13 (19%) | 5 (11%) | 8 (32%) | |
| ≥3 | 5 (7%) | 2 (4%) | 3 (12%) | |
| Atrial fibrillation during multidetector cardiac computed tomography | 37 (53%) | 20 (44%) | 17 (68%) | 0.059 |
| Filling defect on initial multidetector cardiac computed tomogram | 13 (19%) | 5 (11%) | 8 (32%) | 0.052 |
| Gated multidetector cardiac computed tomography | 34 (49%) | 26 (58%) | 8 (32%) | 0.04 |
| 256-slice scanner | 35 (50%) | 30 (67%) | 5 (20%) | <0.01 |
| 120-kVp initial multidetector cardiac computed tomography | 67 (96%) | 43 (96%) | 24 (96%) | 1.00 |
| 80-kVp delayed multidetector cardiac computed tomography | 35 (50%) | 30 (67%) | 5 (20%) | <0.01 |
| Radiation dose (mSv) | 6.4 ± 2.8 | 6.3 ± 2.1 | 6.7 ± 3.7 | 0.61 |
| Left atrial volume (ml) | 150.2 ± 45.8 | 140.1 ± 46.3 | 170.0 ± 38.7 | 0.01 |
| Left atrial diameter (mm) | 46.0 ± 9.6 | 45.8 ± 8.2 | 46.7 ± 15.4 | 0.83 |
| Left atrial appendage volume (ml) | 14.2 ± 7.4 | 12.6 ± 6.0 | 17.3 ± 9.0 | 0.01 |
| Left atrial attenuation ratio † | 0.43 ± 0.12 | 0.45 ± 0.11 | 0.39 ± 0.12 | 0.05 |
| Left atrial appendage attenuation ratio † | 0.57 ± 0.30 | 0.55 ± 0.23 | 0.61 ± 0.40 | 0.44 |
| Initial left atrial appendage/left atrial attenuation ratio † | 0.77 ± 0.23 | 0.82 ± 0.19 | 0.69 ± 0.29 | 0.05 |
| Delayed left atrial appendage/left atrial attentuation ratio † | 0.93 ± 0.18 | 0.95 ± 0.12 | 0.90 ± 0.25 | 0.36 |
| Initial left atrial appendage/ascending aorta attentuation ratio † | 0.81 ± 0.24 | 0.87 ± 0.18 | 0.71 ± 0.29 | <0.01 |
| Delayed left atrial appendage/ascending aorta attenuation ratio † | 0.89 ± 0.19 | 0.92 ± 0.17 | 0.85 ± 0.23 | 0.15 |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
