Abstracts

Introduction: Genetic predisposition to dementia has been strongly sought, but risk loci remain elusive. Atrial fibrillation (AF), a known risk factor for cerebrovascular accidents (CVA), was recently linked with dementia, especially in a younger population. Two chromosomal loci, 4q25 and 16q22, are associated with AF and CVA risk and may similarly predispose to dementia. Therefore, tested for an association between three known AF-risk variants, rs2200733, rs2634073 (both 4q25), and rs7193343 (16q22) and dementia. Also, tested epsilon 4 allele of the ApoE gene (comprising two variants, rs429358 and rs74120) which is associated with Alzheimer's disease.

Methods: This was a retrospective, observational study using DNA from consenting patients (pts) enrolled in the Intermountain Heart Collaborative Study. All AF diagnoses preceded dementia diagnoses (including Alzheimer's) and were defined by ICD-9 codes. AF pts with dementia (n=132) were matched 1:2 by sex, AF onset age (±5 years) and follow up period (age at dementia ≤ age at last follow-up) to AF pts without dementia (n=264). A second 1:1 matching schema was performed between AF pts with dementia and dementia pts without AF (n=132) by sex and age at dementia diagnosis (AF dementia onset age ≤ non-AF dementia onset age). Genotyping employed Taqman real-time PCR. Chi-square trend tests performed to verify associations between AF/dementia groups and SNPs.

Results: Cases were 59.1% male with average age of AF onset at 74 years and average age of dementia at 78 in AF pts and 80 in non-AF pts. In AF pts, dementia was associated with rs2200733 (p=0.038) and rs429358 (p=0.003). However, when pts with CVA prior to dementia diagnosis were excluded from AF with dementia group, rs2200733 showed no significance with dementia (p=0.3681), yet rs429358 remained significant (p=0.001) and rs7412 became significant (p=0.006). rs2200733 (p=0.016) and rs2614073 (p=0.043) were both associated with AF in pts with dementia.

Conclusions: This data supports prior studies of ApoE risk of non-CVA related dementia/Alzheimer's in the general population. Furthermore, these findings suggest that association between rs2200733 and dementia maybe due to CVA in AF populations.

Background: 1. Intermountain Medical Center Heart Institute, Murray, Utah; 2. University of Utah School of Medicine, Salt Lake City, Utah; 3. Janssen Research and Development, Raritan, New Jersey

Introduction: It is unclear what factors determine the use of anticoagulants among patients presenting with acute coronary syndrome (ACS), in which the use of single (SAPT) or dual (DAPT) antiplatelet therapy is required.

Methods: We retrospectively evaluated clinical factors associated with in-hospital and post-discharge anticoagulant use among 5378 ACS patients enrolled in the Intermountain Heart Collaborative Study from 2004–2009. Multivariable logistic regression analyses were used to determine predictors of anticoagulant use in patients receiving SAPT or DAPT.

Results: In hospital, 99% of patients received an anticoagulant (83% heparin, 16% low molecular weight heparin, 12% warfarin, 8% other), 80% received DAPT, and 18% SAPT. Post-discharge, only 9% received anticoagulation (warfarin); 77% received DAPT and 20% SAPT (Table). Warfarin was prescribed in 7% of DAPT and 17% of SAPT patients; <1% received warfarin alone. Warfarin was used more frequently in patients with myocardial infarction (MI) (ST-elevation MI [STEMI] 11%, non-STEMI 10%, unstable angina 7%).

Conclusion: Among >5000 ACS patients, in-hospital use of anticoagulants was nearly universal. Post-discharge, warfarin use was rare and usually associated with the traditional indications of venous thromboembolism or atrial fibrillation. It was more frequently used with SAPT and prescribed slightly more frequently to STEMI and non-STEMI patients. How anticoagulation decisions affect long-term outcome requires further study.

1. Intermountain Medical Center Heart Institute, Murray, Utah; 2. University of Utah School of Medicine, Salt Lake City, Utah; 3. Janssen Research and Development, Raritan, New Jersey

Introduction: Acute coronary syndrome (ACS) patients benefit from dual antiplatelet therapy (DAPT), and most patients with atrial fibrillation (AF) merit anticoagulation stroke prophylaxis (AcP). Real world reporting of overall antithrombotic therapy among patients with ACS + AF is limited. We describe antithrombotic therapy in a large cohort of ACS patients discharged with AF.

Methods: 911 patients with AF + ACS enrolled in the Intermountain Heart Collaborative Study registry from 2004–2009 were identified. Clinical variables and discharge antithrombotic therapy (single antiplatelet therapy [SAPT] ± AcP or DAPT ± AcP) were collected. Predictors of AcP at discharge were determined using multivariable logistic regression analyses.

Results: 462 AF + ACS patients (50.7%) received coronary stents; 590 (64.8%) had a CHADS2 score ≥2. Only 15.6% received DAPT + AcP, 12.4% received SAPT + AcP, 48.9% DAPT only, 21.3% SAPT only, and 1.9% AcP only. Use of AcP did not differ based on CHADS2 score (≥2 = 31.2%; <2 = 27.4%; P=0.256).

Conclusions: Among patients with ACS + AF, fewer than one in three (30%) received AcP regardless of CHADS2 score; very few received DAPT + AcP. Scant data are available regarding optimal antithrombotic therapy management, and clinical trials are needed to guide therapy in this common and challenging patient population.

Background: Low density lipoprotein cholesterol (LDL-C) has long been the cornerstone of lipid lowering therapy for atherosclerosis risk reduction. However, recent investigations have reported the importance of subfractions and other atherogenic lipoproteins in reducing risk. We evaluated the predictive ability of atherogenic lipoproteins and subfractions among a higher risk cohort for the occurrence of death, myocardial infarction (MI), and repeat revascularization (MACE).

Methods: A total of 2,414 patients of the registry of the Intermountain Heart Study were evaluated. At time of angiography, lipid and subfractions were measured using the Vertical Auto Profile method (Atherotech; Birmingham, AL). Cox hazard regression was utilized to determine the association between lipid parameters and MACE adjusted by cardiac risk factors, medications, and lipids and subparticles. Average length of follow-up was 4.0±2.3 years.

Results: Age averaged 62.6±12.6 years, 65.5% were male, 54.2% had hyperlipidemia, 20.3% were diabetic, and 53.9% had angiographic CAD. Average LDL-C and non-HDL were 102.2±35.5 mg/dL and 131.5±39.8 mg/dL, respectively. The Table displays univariable and levels of multivariable analysis for the association of lipids and subfractions to MACE.

Conclusion: After adjustment, only the denser LDL (LDL3 and LDL4) and apolipoprotein B were significant predictors of MACE. The role of lipoproteins and subfractions for adverse outcomes in higher risk patients needs further study.

BACKGROUND: Many, but not all, pts with coronary atherosclerosis develop significant coronary artery calcification (CAC). However, the mechanisms and the predictors of CAC are poorly known.

METHODS: We analyzed the coronary angiograms of 1,728 pts to assess angiographic severity of fluoroscopically-determined CAC (F-CAC). A novel scoring system, reporting extent (0-3) and severity (0-3) for each major coronary vessel was developed. A global score was determined by multiplying extent and severity scores and summing the numbers for the 4 vessels. Multivariate logistic regression was performed to determine predictors of F-CAC presence by comparing Score 0 to Scores >1.

RESULTS: Significant predictors of F-CAC were age, sex, hyperlipidemia, diabetes, smoking, prior CAD, CAD presence and severity, depression history, GFR categories, BMI and LVEF (Table). Multivariate analysis showed age (OR=1.05, p<0.0001), sex (OR=1.43, p=0.009), smoking (OR=1.47, p=0.01), CAD (OR=3.87, p<0.0001), CAD disease severity (1 vs. 0: OR=2.48, p<0.0001; 2 vs. 0: OR=4.78, p<0.0001; 3 vs. 0: OR=6.00, p<0.0001), and GFR categories (30-59 vs. >60: OR=1.04, p=0.82; <30 vs. >60: OR=1.85, p=0.05) to be independent predictors of F-CAC presence.

CONCLUSIONS: Age, sex, smoking, CAD presence and severity, and GFR were found to predict F-CAC severity as assessed by angiography. These clinical predictors can potentially help tailor clinical therapy and lead to a better understanding of coronary calcification pathophysiology.

*University of Utah, Salt Lake City, UT; †Intermountain Medical Center, Murray, UT

Background: Background: Vitamin (Vit) D is a critical factor related to calcium (CA) absorption and metabolism, and Vit D levels are associated with serum CA levels, coronary artery disease (CAD) and major adverse CV events. Additionally, severity of coronary calcification has been used as a surrogate marker for CAD severity. However, whether either Vit D or serum CA levels are associated with severity of coronary calcification is unknown.

Methods: A total of 1,728 consecutive pts (age=63±12 yrs; males=61%, severe [≥70% stenosis in at least one major vessel] CAD=61%) undergoing clinically indicated coronary arteriography with available baseline demographics, Vit D and CA levels were included in the study. To assess fluoroscopic severity of coronary calcification, we developed a novel scoring system, reporting both extent (0-3) and severity (0-3) for each major coronary vessel. A global score for each individual pt was determined by multiplying extent and severity score for each vessel scored and adding all the scored vessels (range=0-36; categorized into 4 levels of severity: 0 [n=1225], 1-2 [n=163], 3-4 [n=118], ≥5 [n=222]). Vit D levels (ng/dL) were stratified into normal (>30; n=273), low (16-30; n=1266), and very low (≤15; n=189) categories. Associations between Vit D and coronary calcium severity categories were then made and correlations between serum CA levels and coronary calcium severity were determined.

Results: No significant association between Vit D categories and mean coronary calcium scores (normal=1.41±3.4; low=1.93±4.3; very low=1.58±3.8; p=0.13) was found, as also between coronary calcium severity categories and mean Vit D levels (ng/dL) (0 = 24.4±9.0; 1-2 = 23.9±6.2; 3-4 = 24.5±10.1; >5 = 24.5±8.1; p=0.89). Additionally, no significant correlations were found between serum CA levels and global coronary severity score (r=0.004, p=0.88) or calcium global score categories (r=0.02, p=0.44).

Conclusions: Despite the known relationship between Vit D and CA metabolism, no association between either serum levels or severity of coronary calcification could be found. The underlying mechanism of differential calcification of coronary atherosclerotic plaque remains to be discovered.

Control/Tracking Number: 13-A-14499-ACC

Background: Innovative strategies and tools for stratification of 30-day readmission (30dR) risk are needed due to penalties legislated in the Affordable Care Act (ACA). The Intermountain Risk Score (IMRS) is a sex-specific tool derived to predict mortality using the complete blood count (CBC), basic metabolic profile (BMP), and age. IMRS also predicts incident heart failure (HF), but its ability to predict 30dR is unknown.

Methods: Patients discharged with a primary ICD-9 diagnosis of HF from an Intermountain Healthcare hospital (4/1999-4/2011) were studied to test IMRS for all-cause 30dR (N=6616) and to derive a new risk score (IMRS-HF) for 30dR. Recent patients (hospitalized 4/2011-10/2012) validated IMRS-HF (N=459). Age, sex, length of stay, and components of the CBC, BMP, and Charlson Index were evaluated by sex-specific Cox regression for use in IMRS-HF, with beta-coefficients used to assign risk values.

Results: IMRS predicted mortality (c=0.70, p<0.001), but was weak for 30dR (c=0.54, p=0.001). In derivation patients, IMRS-HF predicted 30dR for females (n=3461, IMRS-HF range: 0-9) with hazard ratio (HR)=1.22 per +1 score (95% CI=1.16, 1.29; p<0.001) and for males (n=3155, IMRS-HF range: 1-19) with HR=1.25 per +1 score (CI=1.19, 1.31; p<0.001). Highest vs. lowest IMRS-HF groups were substantially stratified (females: HR=5.15 for scores 7-9 [24% 30dR] vs. 0 [5% 30dR]; males: HR=4.47 for 15-19 [22% 30dR] vs. 1-8 [6% 30dR]). C-statistics were c=0.604 (p<0.001) for females and 0.636 (p<0.001) for males. In validation patients, females (n=169) had HR=2.13 (p=0.14) for scores 5-9 (16% 30dR) vs. 0-3 (8% 30dR) and HR=3.37 (p=0.019) for 4 (25% 30dR) vs. 0-3 with c=0.566 (p=0.31), and males (n=290) had HR=8.58 (p=0.041) for scores 15-19 (28% 30dR) vs. 1-8 (4% 30dR) with step-wise risk (HR=1.23 per +1 score, p=0.004; c=0.649, p=0.005).

Conclusions: IMRS only weakly predicted 30dR in HF patients. In contrast, 30dR was significantly stratified by IMRS-HF, a new clinical decision tool. IMRS-HF was validated in an independent population, with especially strong findings among males. IMRS-HF may be useful for in-hospital care aimed at preventing readmission and avoiding ACA reimbursement penalties.

Background: Stroke prevention in AF is guided by clinical factors with inadequate predictive power. Most thrombi observed in AF are observed in the left atrial appendage (LAA).

This study was designed to 1) determine the LAA volume among patients with AF and a history of stroke compared to AF patients without a history of stroke, 2) identify a LAA volume threshold that significantly increases the risk of stroke in AF patients, and 3) determine the relative predictive value of LAA size compared with the CHADS and CHADS-VASC models when applied to patients without a history of prior stroke.

Methods: Patients (N=48) with a history of atrial fibrillation and stroke were compared with controls (N=48) with a history of atrial fibrillation but no history of stroke. MRA images from case and control populations were segmented to determine LAA volume.

Results: Patients with a history of stroke had larger LAA mean volumes than controls (28.8 ± 13.5 cm3 vs. 21.7 ± 8.27 cm3, P = 0.002). LAA volume > 34 cm3 was identified as a threshold that conveys a significantly increased risk of stroke (Multivariable OR 7.11, P = 0.003). LAA volume increases of 10 cm3conveys a greater risk of stroke (Multivariable OR 1.88 per +10 cm3, p=0.007) than a 1 point increase in the CHADS2 score (Multivariable OR 1.52 per +1 point, p=0.06) when applied to a primary prevention population.

Conclusions: Larger LAA volume is associated with stroke in the setting of atrial fibrillation. LAA volume analysis by MRA can potentially improve risk stratification for stroke in AF patients.

Background: Methadone-associated sudden death is a leading public health issue. High plasma methadone concentration (PMC) and QTc interval prolongation increase risk for sudden death. We examined clinical, genetic and pharmacokinetic variables and methadone dose for associations with plasma methadone concentrations and QTc prolongation.

Methods: Consenting participants (n=31) initiating methadone maintenance therapy were enrolled in the MEMORIES Trial (Clinical Trials.gov NCT01191242). Peak (4 hr post-oral dose) and trough (24 hr) plasma samples obtained on days 1, 7 and 21 were analyzed for methadone and EDDP (the principal methadone metabolite) by liquid chromatographic tandem mass spectrometry. Pre-treatment and d21 rate-corrected QT interval (QTc) was measured by ECG.

Results: PMC was associated with dose and body mass index, although 60.4% of PMC variability was unexplained by dose. Variants in ABCB1, CYP3A4 and CYP2B6 genes were not predictive of PMC; however, stereo-selectivity of CYP2B6 was reduced by the variant rs3745274. Baseline QTc did not predict day (d) 21 QTc (r2=0.056, p=0.25); however d21 trough plasma EDDP was highly correlated with QTc (r2=0.386, p=0.001). For subjects with d21 QTc ≥447msec (n=6), d21 trough EDDP was 43.8±15.4 ng/mL vs 22.9±10.7ng/mL for QTC<447msec, p=0.004). Findings were similar on d 8 (r2=0.197, p=0.026; 104.4 ±36.7 ng/mL and 70.2 ±26.1 ng/mL for QTc ≥ and <447msec, respectively, p<0.02). Trough EDDP ≥25 ng/mL predicted day 21 QTc≥447 msec, with sensitivity of 100% and specificity of 78.9%.

Conclusions: Individual response to oral methadone is highly variable and incompletely explained by dose or currently known clinical and genetic variables. Variables, presently unknown, may account for the remainder of inter-individual variability. Methadone-increased QTc interval directly correlated with plasma trough level EDDP as early as d8 and, at d21, identified with 100% sensitivity those with QTc ≥447msec. The use of EDDP appears promising as a risk marker for QTc prolongation and deserves further study.

Background: Statin therapy is well documented to lower cholesterol and prevent cardiovascular adverse events. However, 10-15% of patients cannot tolerate many statins due to the development of myalgias. Pitavastatin, a newer statin, is minimally affected by the cytochrome P450 system, water soluble so it doesn't enter the brain, and does not lower Co-enzyme Q10. However, whether these special biochemical characteristics make pitavastatin both usable and efficacious among patients who have not been able to tolerate other statins is not known.

Methods: A total of 40 consecutive patients with documented intolerance to any dose of at least two different statins were prospectively enrolled into a protocol whereby they received a trial of pitavastatin therapy at a dose of 2 mg per day. Initially pitavastatin samples were provided and if they could tolerate the regimen, a long-term prescription was given. Baseline clinical characteristics in all patients and fasting baseline and post-pitavastatin treatment fasting LDL cholesterol levels were obtained in all patients found to tolerate pitavastatin therapy.

Results: A total of 40 patients (mean age = 65 years; males = 40%, documented CAD = 50%; hypertension = 80%, diabetes = 18%, positive family history of CAD = 58%, smoker = 10%) were enrolled in the protocol. Of these 27 (68%) were able to tolerate the pitavastatin samples and continued on maintenance pitavastatin therapy. Among those able to tolerate pitavastatin, baseline fasting LDL cholesterol was reduced from 147±27 mg/dL to 93±25 mg/dL, resulting in an average LDL-cholesterol reduction of 34%. Patients who could best tolerate pitavastatin tended to be males and those with no history of coronary artery disease or diabetes.

Conclusion: Low dose pitavastatin is an acceptable alternative to abandoning statin therapy among two thirds of patients documented to be intolerant to other statins, providing an average LDL-cholesterol reduction of 34%.

BACKGROUND: Telomere length (TL) is a correlate of biological age and has been associated with risk for coronary artery disease (CAD) and longevity. We tested for associations between TL and cardiovascular outcomes and all-cause mortality among patients (pts) referred for coronary angiography.

METHODS: Peripheral blood DNA was obtained from consenting pts (n=3569) at angiography, and TL measured in triplicate by monochrome (SYBR Green I) multiplex quantitative PCR (Bio-Rad CFX384 Detection System) and normalized to a quantitatively-measured, single-copy gene (albumin) in the same reaction. Pt information was extracted from Intermountain Healthcare's electronic records database, and survival status was verified by a national death index.

RESULTS: Pts were a mean of 62.9 yrs (±13.48); 91.3% were white and 63.8% were male, 2257 pts (63.2%) had CAD. During follow-up (median 9.2 yrs, IQR 2.5 yrs), 1122 pts (32%) died, 530 (15%) had MI, and 232 (6.5%) had subsequent CVA. TL was correlated with age (r=-0.244, p<0.0001). In univariate analysis longer TL was associated with a decrease risk of CAD (OR= 0.54, p<0.0001) and decreased risk for death (HR 0.51, p<0.0001) but only the association with death remained after adjustment for age and other risk factors (HR= 0.78, p=0.01). TL was not associated with MI or CVA (Table).

CONCLUSION: TL is a strong univariable predictor of survival that is not eliminated by adjustment for age and other risk factors. The relationship of TL to survival deserves further investigation.

*†Intermountain Medical Center, Murray, UT, *University of Utah, Salt Lake City, UT

Background: Accurate evaluation of biomarkers indicative of acute myocardial injury. Both troponin-I and troponin-T are validated markers, but often only one or the other is available on any specific blood analysis platform. It is important to know the performance of each before choosing an institutional standard.

Methods: At Intermountain Medical Center, simultaneous results of the Roche platform troponin-T assay were compared to the existing standard Abbott platform troponin-I assay on 3,306 consecutively ordered clinical samples in 1,860 individual patients. Sensitivity and specificity for both troponin-I and troponin-T were calculated using the other assay as the imputed gold standard, based on the recommended normal range of ≤0.04 ng/mL and <0.03 ng/mL for troponin-I and troponin-T respectively. Additionally, chart review was performed on all patients in which the initial results of troponin-T were normal but troponin-I was elevated, or vise versa, Results were categorized into four categories: 0=unassociated diagnosis, 1 = definite/high probability acute coronary syndrome (ACS), 2 = possible ACS; 3 = non-ACS diagnosis associated with myocardial injury.

Results: Over-all 1,005 (30.4%) and 854 (25.8%) of samples for troponin-I and troponin-T, respectively, were elevated. Sensitivity and specificity of troponin-T, using troponin-I as the gold standard were 0.76 and 0.96, respectively, and conversely, sensitivity and specificity of troponin-I, using troponin-T as the gold standard were 0.89 and 0.90, respectively. Categorization of the 139 (7.5%) patients in which troponin-T was normal but troponin-I was elevated revealed category 0: n=20 (14.4%); category 1: n=33 (23.7%); category 2: n=31 (22.3%); category 3: n=55 (39.6%). For the 67 (3.6%) patients with troponin-I normal and troponin-T elevated, categories were: 0: n=51(76.1%); 1: n=2(3.0%); 2: n=6(9.0%); 3: n=8(11.9%).

Conclusions: In this comparison, troponin-I was significantly more sensitive and nearly as specific as troponin-T. Clinical chart review of patients with troponin-T/troponin-I discordance generally supported the results of troponin-I. These findings may limit the utility of this troponin-T assay.

Objective: To determine whether the use of an electronic discharge orders tool was correlated with higher adherence to core measures and lower hospital readmissions.

Background: Despite increased used of standardized discharge orders and evidence suggesting the effectiveness of these tools, little is known about adherence to core measures and 30-day all-cause hospital readmission among patients discharged with these electronic systems.

Methods: A retrospective study was conducted on heart failure (HF) patients discharged from Intermountain Healthcare hospitals between January 2011 and September 2012. The primary outcome was 30-day all-cause readmissions as defined by The Centers for Medicare & Medicaid Services. Adherence to HF core measures was evaluated on three inpatient quality measures. Adult smoking cessation advice/counseling was excluded as it is a non‐accountability measure. Conditional logistic regression was used to estimate odds ratios and 95% confidence intervals.

Results: A total of 2,409 encounters were found eligible to adhere to at least one core measure. The discharge orders tool was used in slightly over half (55%) of these encounters. Full adherence to HF quality measures was higher among patients discharged with the electronic tool (OR = 9.89; 95% CI = 6.08 - 16.11). The observed readmission rate among patients discharged with and without the electronic discharge orders tool was 15.5% and 18.0%, respectively. After adjusting for length of stay, age, sex, and severity of illness we found that the likelihood of readmission was lower (OR = 0.77; CI = 0.62 - 0.97) among patients discharged with the electronic tool.

Conclusions: The use of an electronic discharge orders tool enhanced with decision support algorithms aimed at increasing the prescription of evidence-based discharge medications and the provision of HF instructions was associated with increased compliance with HF core measures and with lower readmission rates among heart failure patients.

Background: Many patients who develop atrial fibrillation (AF) will experience a worsening of their arrhythmia over time. Catheter ablation is an effective rhythm control strategy with favorable long-term outcomes. The optimal time to proceed with ablation during the disease course is unknown. Further, it is unknown if delays in treatment will negatively influence outcomes.

Methods: A total of 684 consecutive patients who underwent an AF ablation procedure that had long-term established care within an integrated health care system were evaluated. Recursive partitioning was used to determine categories associated with changes in risk from the time of first AF diagnosis (ICD-9 code: 427.31) to first AF ablation: 1: 30-180 (n=187), 2: 181-545 (n=116), 3: 546-1825 (n=186), 4: >1825 (n=195) (in days). Outcomes evaluated include one year AF recurrence, stroke, heart failure hospitalization, and death.

Results: With increasing time to treatment, patients were older (1: 63.7±11.1, 2: 62.6±11.8, 3: 66.4±10.2, 4: 67.6±9.70, p<0.0001) and had more hypertension (1: 64.7%, 2: 82.8%, 3: 74.7%, 4: 80.0%, p=0.001). There was no difference in paroxysmal/persistent AF subtype ratio between groups. For each increase strata of time increase, there was a direct increase of one year AF recurrence and/or need for long-term antiarrhythmics (Figure, p-trend=0.03). After adjustment, clinically significant differences in risk of recurrent AF were found when compared to the 30-180 day time category: 181-545: odd ratio (OR)=1.38, p=0.25; 546-1825: OR=1.62, p=0.05; and >1825: OR=1.58, p=0.06. No differences were observed in stroke or heart failure hospitalization among the groups. Death rates were higher in the most delayed group (1: 3.2%, 2: 4.3%, 3: 4.3%, 4: 11.8%, p-trend=0.001)

Conclusions: Delays in treatment with catheter ablation impact procedural success rates independent of temporal changes to the AF subtype at ablation. These data suggest catheter ablation should be considered early in the AF disease process.