Methods

This study is a retrospective observational study conducted from 2009 to 2013 in our institution. Subjects were patients with AF aged >18 years without chest pain who underwent pulmonary vein isolation who had preoperative cardiovascular multidetector computed tomography (MDCT) imaging performed for procedural planning. Exclusion criteria were subjects with known history of CAD based on previous coronary angiography or previous coronary revascularization (n = 19), severe valvular diseases or with prosthetic heart valve (n = 3), known history of systolic dysfunction with left ventricular ejection fraction <40% (n = 22), and patients without noncontrast MDCT studies (patients had other preoperative imaging modalities or did not have one; n = 113). Type of AF was determined by patients’ electrophysiologist according to the 2001 American College of Cardiology/American Heart Association/European Society of Cardiology guidelines for the management of patients with AF. Paroxysmal AF was defined as episodes lasting for <7 days. Persistent AF was defined as episodes lasting ≥7 days.

We randomly selected the same number of patients with NSR (n = 430) with matching age and gender from the same period to be the control group for this study. These control patients were selected from a series of patients without known CAD who presented with acute chest pain suspicious for cardiac chest pain to our emergency department. Myocardial infarction was ruled out in these patients on a basis of electrocardiogram and cardiac enzymes findings. Controls underwent MDCT imaging for CACS assessment and were observed in a chest pain unit. The same exclusion criteria were applied to the control group. Institutional review board approval was obtained for this study.

A 64-slice MDCT (Philips Precedence; Philips Healthcare, Eindhoven, The Netherlands) was used throughout the study period. In patients with AF, non–contrast-enhanced CACS assessment was followed by contrast-enhanced studies. The scans were performed when ventricular rates were <80 beats/min. Metoprolol was used as needed to control ventricular rate. In the control group with NSR, only non–contrast-enhanced CACS assessments were performed. Images were acquired during a single breath-hold, using prospective electrocardiogram gating with imaging triggered at 75% of the R-R interval. CACS was calculated as previously described by Agatston et al.

For patients with AF, demographics and cardiovascular risk factors were collected at the date of the CACS testing. Framingham coronary artery disease 10-year risk score (FRS) was calculated for each patient using relevant clinical information nearest to the CACS testing date. In the control group, all clinical information was collected during the observation period in a chest pain unit.

In patients with AF with CAD (CACS >0), we also investigated difference between the actual age of the particular patient and expected arterial age calculated from an epidemiologic cohort. This was performed by calculating the predicted arterial age of each particular patient by entering their CACS into the arterial age calculator from the MESA cohort. Then, mean actual age was compared to mean predicted arterial age by paired sample t test.

In our study, we investigated potential impacts of CACS on utilization of statin therapy in patients with AF using CACS ≥300 according to the 2013 American College of Cardiology/American Heart Association Guideline on the Treatment of Blood Cholesterol to Reduce Atherosclerotic Cardiovascular Risk in Adults. Patients with AF who did or did not use statins at the time of CACS assessment were identified in our cohort. Then, the prevalence of patients with AF with and without CACS ≥300 was examined. The number of patients with AF who may and may not benefit from statin therapy for subclinical CAD was subsequently analyzed. Second, as it has been shown that CACS >100 is a predictor of stroke, we added CACS >100 to our inclusion criteria for vascular diseases in the CHA ₂ DS ₂ -VASc model. Using CACS data with the original CHA ₂ DS ₂ -VASc model, we calculated the number of patients who would have had an increase in CHA ₂ DS ₂ -VASc score to ≥1 because such patients would potentially be candidates for oral anticoagulation for stroke prevention.

Descriptive statistics for studied variables are presented as mean with standard deviation, median with interquartile range (IQR), and frequency with percentage as appropriate. Independent Students’ t test and Wilcoxon-Mann-Whitney U test were used to compare trends of normally and nonnormally distributed continuous variables, respectively. For comparison of trends between categorical variables, chi-square test and Fisher’s exact test (when necessary) were performed. To assess for associations of the studied covariates with CAD, logistic regression analyses were used. Presence of AF and type of AF were entered into multivariate models, adjusted by FRS, body mass index, and warfarin use. The regression analysis results are presented as odd ratio (OR) and 95% confidence interval (CI). Performance of prediction models for presence of subclinical CAD (CACS >0) was evaluated by comparison of global chi-square score of each predictive model using analysis of variance F tests. The baseline model had FRS alone, whereas the second model added the presence of AF into the baseline model. The second model additionally included AF type. All statistical analyses were performed with IBM SPSS/PASW Statistics 20 (SPSS Inc., Chicago, Illinois). A 2-tailed p value of <0.05 was considered statistically significant.