POSTER SESSION 3
P817
Phenotypic profiling to determine individuals at risk for atrial fibrillation: A latent class clustering analysis of the PREVEND cohort
M. Rienstra
M. Rienstra
B. Geelhoed
B. Geelhoed
J.E. Siland
J.E. Siland
R.A. Vermond
R.A. Vermond
P. Van Der Harst
P. Van Der Harst
H.L. Hillege
H.L. Hillege
I.C. Van Gelder
I.C. Van Gelder
University Medical Center Groningen, Cardiology
,
Groningen
,
Netherlands
Abstract
Introduction: The clinical and pathophysiological heterogeneity variety of risk factors predisposing to AF is large. Better understanding of the phenotypic heterogeneity of those at risk for AF may allow better risk prediction. Latent class clustering analyses may allow the identification of clusters of individuals that share similar risk factors (specific profile), instead of the traditional analysis of specific risk factors of AF.
Methods: We determined the risk of AF using a hypothesis-free, latent class clustering approach, and compared the performance with a traditional risk-factor-based AF prediction model in the community-based PREVEND cohort.
Results: The mean age of the 8,265 individuals was 49±13 years, and 49.8% were men. During 10-year follow-up, 250 (3%) individuals developed AF. We build a latent class model based of uncorrelated risk factors (Figure). The model with 6 distinct classes gave the optimum tradeoff between a high statistical likelihood and a low number of model parameters. The smallest class included 347, the largest 2590 individuals. All clusters had a specific profile (Figure). The incidence of AF varied; being 0.7% (class 1), 1.9% (class 2), 1.4% (class 3), 7.0% (class 4), 4.6% (class 5), and 15.0% (class 6; logrank p<0.001). Internal validation was done with parametric bootstrapping. The discrimination (C-statistic 0.805 vs 0.817, delta-C -0.012, p=0.18) and reclassification (IDI -0.013, p=0.008, NRI -0.013, p=0.73, and categoryless NRI -0.081, p=0.22) performance of the cluster-based model was comparable to the traditional risk-factor-based AF model.
Conclusion: Despite the heterogeneity between individuals at risk for AF, latent class clustering produces mutually exclusive groups of individuals with AF risk factors, and is feasible and clinically valid.
Figure
P818
P wave duration/morphology and left atrial volume at computed tomography in patients without and with (paroxysmal or persistent) atrial fibrillation
A. Cima
A. Cima
M. Brunelli
M. Brunelli
S. Raffa
S. Raffa
A. Grosse
A. Grosse
M. Frommhold
M. Frommhold
T. Daehne
T. Daehne
A. Madaffari
A. Madaffari
C. Lindemann
C. Lindemann
J.C. Geller
J.C. Geller
Zentralklinik Bad Berka, Arrhythmia and Electrophysiology Section, Division of Cardiology
,
Bad Berka
,
Germany
Abstract
Introduction: P wave duration of ≥120msec (left atrial enlargement, LAE) and/or advanced degree of interatrial block (IAB, biphasic +/- P waves in the inferior leads) are ECG parameters used to describe LA anatomical and electrical remodeling. Aim of this study was to describe the incidence of LAE and IAB in a large series of patients (pts) without, with paroxysmal or persistent atrial fibrillation (AF) and compare ECG findings with LA volume measured with computed tomography (CT).
Methods: ECG (50mm/s) P wave duration (msec) and morphology (bimodal P wave, identified by 2 clearly separated peaks, biphasic p wave ≈ 50% of second component with opposite polarity) were analyzed within 3 days of a 64-slice dual source CT of the heart (Siemens, acquisition in sinus rhythm during ventricular diastole) in pts without, with paroxysmal or persistent AF, and LA volume was calculated (in ml, excluding the veins) and normalized for body surface area.
Results: 305 pts [83 (27%) without (no) AF, 112 (37%) paroxysmal and 112 (36%) persistent AF, median age 60 (52, 68) years, 122 (40%) females, 40 (15, 79) months since first AF detection] were enrolled. P wave duration positively correlated with LA volume (Pearson 0.4961; P<.0001), and P wave duration [100 (100, 110) vs. 120 (110, 140) vs. 140 (130, 160) msec] and LA volume [41 (35, 52) vs. 54 (46, 63) vs. 67 (60, 79) ml/m2] were significantly different between no, paroxysmal and persistent AF (P<.0001). The No AF group had a significantly higher rate of positive P waves in lead I and II (82% vs. 37% vs. 32%, P<.0001) whereas the rate of +/- or +/flat P wave in lead III and aVF (12% vs. 25% vs. 33%, P=.0020) and of LAE (37% vs. 72% vs. 97%, P<.0001) was higher in both AF groups. A +/- P wave in V1 was more frequently seen in no and paroxysmal AF groups (76% vs. 85% vs. 69%, P=.0212).
Conclusion: P wave duration positively correlates with LA volume. A progressive increase in LA Volume and rate of LAE at ECG is observed when pts without, with paroxysmal and persistent AF are compared. Presence of AF (particularly of its per (...truncated)