Left ventricular (LV) scar identified by late gadolinium enhanced (LGE) cardiac magnetic resonance (CMR) is associated with adverse outcomes in coronary artery disease and cardiomyopathies. We sought to determine the prognostic significance of LV-LGE in atrial fibrillation (AF). We studied 778 consecutive patients referred for radiofrequency ablation of AF who underwent CMR. Patients with coronary artery disease, previous myocardial infarction, or hypertrophic or dilated cardiomyopathy were excluded. The end points of interest were major adverse cardiac and cerebrovascular events (MACCE), defined as a composite of cardiovascular death, myocardial infarction, and ischemic stroke/transient ischemic attack. Of the 754 patients who met the inclusion criteria, 60% were men with an average age of 64 years. Most (87%) had a normal LV ejection fraction of ≥55%. LV-LGE was found in 46 patients (6%). There were 32 MACCE over the mean follow-up period of 55 months. The MACCE rate was higher for patients with LV-LGE (13.0% vs 3.7%; p = 0.002). In multivariate analysis, CHA 2 DS 2 -VASc score (hazard ratio [HR] 1.36, 95% CI 1.05 to 1.76), the presence of LV-LGE (HR 3.21, 95% CI 1.31 to 7.88), and the LV-LGE extent (HR 1.43, 95% CI 1.15 to 1.78) were independent predictors of MACCE. In addition, the presence of LV-LGE was an independent predictor for ischemic stroke/transient ischemic attack (HR 3.61, 95% CI 1.18 to 11.01) after adjusting for CHA 2 DS 2 -VASc score. In conclusion, the presence and extent of LV scar identified by LGE-CMR were independent predictors of MACCE in patients with AF.
Atrial fibrillation (AF), the most common sustained cardiac arrhythmia, is a global health care burden with rising incidence and prevalence and is associated with increased morbidity and mortality. Given the side effects and limited efficacy of antiarrhythmic medications, radiofrequency catheter ablation (RFCA) is often used in symptomatic patients with AF in whom a rhythm control approach is pursued. This use of catheter ablation is supported by both American and European guidelines. In a growing number of centers, preablation cardiac magnetic resonance (CMR) is performed to provide detailed anatomical information about the left atrium and pulmonary veins and to enable 3-dimensional (D) image reconstruction and merging of these anatomical data with electrical mapping system data acquired during the procedure. CMR also offers additional data with late gadolinium enhanced (LGE) imaging. In the AF population, left atrial (LA) fibrosis identified by LGE-CMR has been shown to predict ablation outcome and to improve RFCA candidate selection. Moreover, left ventricular (LV) scar characterized by LGE imaging is associated with adverse clinical outcomes in a wide variety of cardiovascular (CV) patients. Recently, LV-LGE in AF was shown to be an independent predictor of all-cause mortality. However, data on LV-LGE in AF remain limited. Hence, we aimed to investigate the prognostic significance of LV-LGE in AF and hypothesized that the presence of LV scar would be associated with major cardiac and CV outcomes.
Methods
We retrospectively collected data on all consecutive patients with AF who were referred for RFCA and underwent CMR from June 2006 to January 2013. Of 778 patients, 16 with known coronary artery disease (CAD) and myocardial infarction (MI) and 8 with known hypertrophic and dilated cardiomyopathies were excluded. The primary imaging method for RFCA planning at our institution is CMR if there are no contraindications, such as severe renal impairment (glomerular filtration rate <30 ml/min/1.73 m 2 ), severe claustrophobia, and the presence of a pacemaker/defibrillator. Paroxysmal AF was defined as AF that terminates spontaneously within 7 days of onset, whereas persistent AF was defined as AF that fails to self-terminate within 7 days after onset. We defined previous MI by either clinical evidence of MI per electronic medical record review or electrocardiographic evidence per Minnesota codes 1.1.1 to 1.2.8. The patients’ CHA 2 DS 2 -VASc characteristics were determined by systematic chart review. The study protocol was approved by the Institutional Review Board at the University of Utah.
All studies were performed either on 1.5-T Avanto or 3-T Verio magnetic resonance scanners (Siemens Medical Solutions, Erlangen, Germany). The study protocol consisted of cine imaging for cardiac structure and function, 3D contrast enhanced magnetic resonance angiography for LA and pulmonary vein anatomy, 2D-LGE imaging for viability, and 3D-LGE imaging for assessment of LA wall fibrosis/scar. 2D-LGE imaging was performed approximately 12 minutes after contrast injection (0.1 mmol/kg, MultiHance; Bracco Diagnostics Inc., Princeton, New Jersey) using phase sensitive inversion recovery sequences in short-axis and horizontal and vertical long-axis orientations. Scan parameters for 2D-LGE imaging were as follows: 3 T—repetition time (TR) = 2.5 ms, echo time (TE) = 1.1 ms, flip angle (FA) = 35°, pixel size = 1.88 × 1.88 mm, slice thickness = 7 mm; 1.5 T—TR = 2.5 ms, TE = 1.1 ms, FA = 45°, pixel size = 1.85 × 1.85 mm, slice thickness = 6 mm.
High-resolution LGE images for assessment of LA fibrosis/scar were acquired approximately 15 minutes after contrast injection using a 3D respiratory-navigated, electrocardiographic triggered, inversion recovery–prepared gradient echo pulse sequence. Inversion times were set to null normal LV myocardium. Data acquisition was restricted to 15% of cardiac cycle and was performed during LA diastole. The other scan parameters for 3D-LGE imaging at 3 T were TR = 3.1 ms, TE = 1.4 ms, FA = 14°, axial imaging volume with field-of-view = 400 × 400 × 110 mm, voxel size = 1.25 × 1.25 × 2.5 mm. Scan parameters for 3D-LGE imaging at 1.5 T were TR = 5.2 ms, TE = 2.4 ms, FA = 20°, axial imaging volume with field-of-view = 360 × 360 × 100 mm, voxel size = 1.25 × 1.25 × 2.5 mm. LV-LGE was considered present only if it was visible in all corresponding myocardial locations on short-axis, horizontal long-axis and vertical long-axis images. LGE distribution was categorized as subendocardial, midmyocardial, epicardial, transmural, or adjacent to right ventricular insertion points. Analysis of LGE images for quantification of LA and LV fibrosis was performed with custom software (Corview, Marrek Inc., Salt Lake City, Utah) and with CVI 42 software, version 5.0 (Circle Cardiovascular Imaging, Inc., Calgary, Alberta, Canada). The amount of fibrosis was estimated using a threshold-based algorithm for LA and a full width at half maximum value algorithm for LV.
The primary end points of interest were major adverse cardiac and cerebrovascular events (MACCE), defined as a composite end point of CV death, MI, and ischemic stroke/transient ischemic attack (TIA). The secondary end points of interest were the individual components themselves. MACCE was determined by identification of International Classification of Disease-9 codes (410, 411 for MI; 433.x1, 434.x1, 436, 437.1, 437.9, 362.34, 435 for ischemic stroke/TIA) and review of electronic medical records. In addition, CV death was ascertained using the Utah Population Database. The end points of interest were scrutinized by personnel who were blinded to the clinical data.
Continuous variables are presented as mean/SDs or median/interquartile ranges (IQRs), as appropriate. Categorical data are presented as numbers and percentage. Comparisons between 2 groups were made using the Student t test or Mann–Whitney nonparametric test for continuous variables, the chi-square test or Fisher’s exact tests for categorical variables, as appropriate. Univariate and multivariate logistic regression models were used to determine predictors of LV-LGE. The best overall multivariate models for LV-LGE were sought by stepwise forward selection with a probability to enter set at p = 0.05. Univariate and multivariate Cox regression models were used to obtain hazard ratios (HRs) for MACCE and ischemic stroke/TIA. Survival curves for MACCE and ischemic stroke/TIA were determined according to Kaplan–Meier methods, and comparison of MACCE and ischemic stroke/TIA was performed using a log-rank test. A p value of <0.05 was considered statistically significant, and all reported p values are 2 tailed. All analyses were performed using STATA 13 (StataCorp, College Station, Texas).
Results
From June 2006 to January 2013, a total of 778 patients with AF underwent CMR for RFCA planning at the University of Utah. After excluding patients with a history of CAD, MI, or hypertrophic or dilated cardiomyopathy, the final cohort included 754 patients with an average age of 64 and CHA 2 DS 2 -VASc score of 2.1 (IQR 1 to 3), respectively. In this cohort, 61% were men, 55% had persistent AF, and the median AF duration was 24 months. Warfarin and novel anticoagulants were prescribed in 435 patients (57.7%) and 16 patients (2.1%), respectively. There were 97 patients (12.9%) with LV ejection fraction (EF) <55%. There was greater proportion of persistent AF in patients with LVEF <55% than that in those with LVEF ≥55% (77.3% [75 of 97] vs 51.8% [340 of 656]; p <0.001). Compared with patients without LV-LGE, patients with LV-LGE were more likely to have diabetes mellitus (23.9% vs 11.6%; p = 0.014), dyslipidemia (45.7% vs 27.0%; p = 0.006), heart failure (19.6% vs 7.8%; p = 0.005), smoking history (39.1% vs 25.6%; p = 0.043), enlarged LA size (31.1 cm 2 vs 28.6 cm 2 ; p = 0.049), and higher CHA 2 DS 2 -VASc score (2.7 vs 2.1; p = 0.009). There were no significant differences between patients with and without LV-LGE with regard to age, gender, AF type and duration, previous AF ablation history, other medical co-morbidities, body mass index, LV systolic dysfunction, renal function, or medications used, as listed in Table 1 .
| Variable | Overall Cohort (n = 754) | Left Ventricular Late Gadolinium Enhancement | p Value ∗ | |
|---|---|---|---|---|
| Present (n = 46) | Absent (n = 708) | |||
| Age (yrs) | 64.2±12.1 | 66.2±11.0 | 64.1±12.2 | 0.241 |
| Male | 459 (61%) | 29 (63%) | 430 (61%) | 0.096 |
| Duration of AF (months) (median, IQR) | 24.1 (4.2-72.3) | 21.7 (5.3-70.5) | 24.1 (4.1-72.5) | 0.625 |
| Persistent AF | 416 (55%) | 30 (65%) | 386 (55%) | 0.157 |
| Prior AF ablation | 46 (6%) | 5 (11%) | 41 (6%) | 0.163 |
| CHA 2 DS 2 -VASc Score † | ||||
| Mean (SD) | 2.1±1.6 | 2.7±1.6 | 2.1±1.6 | 0.009 |
| Score | 0.007 | |||
| 0-1 | 325 (43%) | 11 (24%) | 314 (44%) | |
| ≥2 | 429 (57%) | 35 (76%) | 394 (56%) | |
| Diabetes mellitus | 93 (12%) | 11 (24%) | 82 (12%) | 0.014 |
| Hypertension | 437 (58%) | 31 (67%) | 406 (57%) | 0.181 |
| Dyslipidemia | 212 (28%) | 21 (46%) | 191 (27%) | 0.006 |
| Ischemic Stroke/TIA | 73 (10%) | 8 (17%) | 65 (9%) | 0.068 |
| Heart failure | 64 (9%) | 9 (20%) | 55 (8%) | 0.005 |
| Peripheral arterial disease | 10 (1%) | 2 (4%) | 8 (1%) | 0.065 |
| Smoker | 199 (26%) | 18 (39%) | 181 (26%) | 0.043 |
| Thyroid disease | 110 (15%) | 9 (20%) | 101 (14%) | 0.324 |
| Obstructive sleep apnea | 139 (18%) | 9 (20%) | 130 (18%) | 0.838 |
| Body Mass Index (kg/m 2 ) | 29.7±6.9 | 29.7±6.4 | 29.7±6.9 | 0.973 |
| Glomerular filtration rate (ml/min/1.73m 2 ) | 74.6±19.2 | 75.0±18.9 | 74.6±19.2 | 0.892 |
| LVEF <55% | 97 (13%) | 9 (20%) | 88 (13%) | 0.163 |
| Left atrial area (cm 2 ) | 28.7±8.2 | 31.1±7.6 | 28.6±8.3 | 0.049 |
| Left atrial fibrosis (%) | 16.0±9.8 | 17.1±10.1 | 15.9±9.8 | 0.486 |
| Medications | ||||
| Class I antiarrhythmics | 101 (13%) | 6 (13%) | 95 (13%) | 0.987 |
| Class III antiarrhythmics | 115 (15%) | 6 (13%) | 109 (15%) | 0.667 |
| Aspirin | 256 (34%) | 19 (41%) | 237 (34%) | 0.277 |
| Statins | 233 (31%) | 15 (33%) | 218 (31%) | 0.796 |
| Beta-blockers | 319 (42%) | 23 (50%) | 296 (42%) | 0.276 |
| Calcium channel blockers | 225 (30%) | 9 (20%) | 216 (31%) | 0.116 |
| ACEI/ARB | 249 (33%) | 17 (37%) | 232 (33%) | 0.558 |
| Diuretics | 204 (27%) | 18 (39%) | 186 (26%) | 0.058 |
| Novel oral anticoagulants | 16 (2%) | 0 (0%) | 16 (2%) | 0.303 |
| Warfarin | 435 (58%) | 27 (59%) | 408 (58%) | 0.887 |
∗ p value = left ventricular late gadolinium enhancement present versus absent in patients without a clinical history of myocardial infarction.
† Validated risk score model to establish the stroke risk in patients with nonrheumatic atrial fibrillation.
In our patients with AF without a known history of CAD, MI, or hypertrophic or dilated cardiomyopathy, we identified 46 patients (6.1%) with LV-LGE on CMR. The median LV-LGE extent was 1.2% of LV mass (IQR 0.9 to 2.1). Of these patients, 80.4% had normal LV systolic function. LV-LGE ischemic patterns were found in 23 patients (23 of 46 [50%]). Of these patients with ischemic LV-LGE patterns, 13 of 23 (57%) had subendocardial LGE and 10 of 23 (43%) had transmural LGE. In patients with nonischemic LGE patterns, midmyocardial, epicardial, and right ventricular insertion locations were present in 20 patients (20 of 23 [87%]), 1 patient (1 of 23 [4%]), and 2 patients (2 of 23 [9%]), respectively. Representative images of LV-LGE are shown in Figure 1 . None of these patients with LV-LGE underwent stress testing or invasive coronary angiography based on the CMR findings, and the origin of LV-LGE remains inconclusive. LV systolic dysfunction (LVEF <55%) was found in 4 patients (17.4%) with ischemic LGE patterns versus 5 patients (21.7%) with nonischemic LGE patterns (p = 0.710).
We sought to identify risk factors for LV-LGE using a logistic regression model. On univariate analysis, AF duration, CHA 2 DS 2 -VASc ≥2, diabetes mellitus, dyslipidemia, heart failure, and smoking history were associated with the presence of LV-LGE. In a multivariate model, AF duration (odds ratio [OR] 1.003, 95% CI 1.001 to 1.004), CHA 2 DS 2 -VASc ≥2 (OR 2.16, 95% CI 1.05 to 4.44), heart failure (OR 2.55, 95% CI 1.13 to 5.75), and smoking history (OR 1.94, 95% CI 1.03 to 3.65) remained significantly associated with the presence of LV-LGE.
There were 32 instances of MACCE (4.2%) over the mean follow-up period of 55.5 ± 18.5 months. CV death, MI, and ischemic stroke/TIA events occurred in 10 (1.3%), 7 (0.9%), and 19 (2.5%) patients, respectively. Patients with LV-LGE had significantly higher MACCE (13.0% [6 of 46] vs 3.7% [26 of 708]; p = 0.002) and ischemic stroke/TIA (8.7% [4 of 46] vs 2.1% [15 of 708]; p = 0.006) compared with patients without LV-LGE ( Figure 2 ). There was no significant difference in CV death (4.4% [2 of 46] vs 1.1% [8 of 708]; p = 0.065) and MI (2.2% [1 of 46] vs 0.9% [6 of 708]; p = 0.363) in patients with and without LV-LGE ( Figure 2 ).
Age, CHA 2 DS 2 -VASc, dyslipidemia, history of ischemic stroke/TIA, angiotensin-converting enzyme inhibitor/angiotensin receptor blocker use, warfarin use, LA area, presence of LV-LGE, and extent of LV-LGE provided unadjusted associations with MACCE as listed in Table 2 . In a multivariate model, CHA 2 DS 2 -VASc (HR 1.36, 95% CI 1.05 to 1.76), the presence of LV-LGE (HR 3.21, 95% CI 1.31 to 7.88), and the extent of LV-LGE (HR 1.43, 95% CI 1.15 to 1.78) were significantly associated with MACCE ( Table 3 ). Kaplan–Meier estimates highlighting the significant difference in MACCE occurrence according to the presence or absence of LV-LGE are shown in Figure 3 .
| Variable | Hazard Ratio | 95% Confidence Interval | p Value |
|---|---|---|---|
| Age | 1.05 | 1.02-1.09 | 0.004 |
| Male | 0.56 | 0.28-1.11 | 0.098 |
| Duration of AF | 1.00 | 0.99-1.00 | 0.166 |
| Persistent AF | 0.57 | 0.27-1.21 | 0.144 |
| Prior AF ablation ∗ | |||
| CHA 2 DS 2 -VASc Score | 1.50 | 1.23-1.84 | <0.001 |
| Diabetes mellitus | 1.69 | 0.69-4.10 | 0.247 |
| Hypertension | 2.13 | 0.96-4.74 | 0.065 |
| Dyslipidemia | 2.03 | 1.01-4.08 | 0.047 |
| Ischemic stroke/TIA | 1.63 | 1.08-2.48 | 0.022 |
| Heart failure | 1.91 | 0.74-4.97 | 0.183 |
| Peripheral arterial disease | 2.39 | 0.33-17.51 | 0.391 |
| Smoker | 1.49 | 0.72-3.09 | 0.283 |
| Thyroid disease | 1.40 | 0.58-3.41 | 0.456 |
| Obstructive sleep apnea | 0.30 | 0.71-1.24 | 0.096 |
| Body mass index | 0.96 | 0.90-1.02 | 0.168 |
| Glomerular filtration rate | 0.98 | 0.97-1.00 | 0.123 |
| Class I antiarrhythmics | 0.64 | 0.19-2.09 | 0.456 |
| Class III antiarrhythmics | 0.79 | 0.28-2.26 | 0.663 |
| Aspirin | 0.92 | 0.43-1.94 | 0.821 |
| Statins | 1.19 | 0.57-2.47 | 0.637 |
| Beta-blockers | 0.92 | 0.46-1.87 | 0.821 |
| Calcium channel blockers | 1.42 | 0.69-2.91 | 0.335 |
| ACEI/ARB | 2.28 | 1.13-4.56 | 0.020 |
| Diuretics | 1.84 | 0.91-3.73 | 0.090 |
| Novel oral anticoagulants ∗ | |||
| Warfarin | 0.47 | 0.23-0.95 | 0.036 |
| LVEF <55% | 0.95 | 0.33-2.70 | 0.919 |
| Left atrial area | 1.04 | 1.00-1.07 | 0.028 |
| Left atrial fibrosis | 1.01 | 0.97-1.05 | 0.587 |
| LV-LGE Present | 4.09 | 1.68-9.96 | 0.002 |
| LV-LGE Extent † | 1.36 | 1.10-1.67 | 0.003 |
∗ Hazard ratios could not be calculated for previous AF ablation and novel oral anticoagulants due to no major adverse cardiac/cerebrovascular event during the follow-up period in patients with a history of previous AF ablation or novel oral anticoagulants.
† LV-LGE extent hazard ratio is for each 1% increase in LGE volume.
| Variable | Model 1 | Model 2 | ||||
|---|---|---|---|---|---|---|
| Hazard Ratio | 95% CI | p Value | Hazard Ratio | 95% CI | p Value | |
| Age | 1.02 | 0.98-1.06 | 0.390 | 1.02 | 0.98-1.06 | 0.347 |
| CHA 2 DS 2 -VASc | 1.36 | 1.05-1.76 | 0.019 | 1.41 | 1.09-1.83 | 0.009 |
| LV-LGE Present | 3.21 | 1.31-7.88 | 0.011 | – | – | – |
| LV-LGE Extent ∗ | – | – | – | 1.43 | 1.15-1.78 | 0.001 |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
