Abstract 16387: The Contribution of Novel Genetic Markers to Atrial Fibrillation Risk Prediction Models
Background: The prevalence of atrial fibrillation (AF) is increasing even among populations thought to be at low risk. Whether risk prediction algorithms combined with knowledge of common genetic variation recently discovered to be associated with AF improves AF risk prediction in these populations is unknown.
Methods: In 20822 initially healthy women followed prospectively for a median of 14.5 years we used the Bayes Information Criterion to derive (n=13051) and then validate (n=6880) a novel and parsimonious clinical risk prediction algorithm for AF. We then evaluated whether a genetic risk score comprised of 5 recently published risk alleles in 3 loci (1q21, 4q25, and 16q22) improved the performance of this model and that of a previously published model from the Framingham Heart Study (FHS) in the entire cohort.
Results: After considering 31 potential covariates in the derivation cohort, the best fitting model (WHS AF risk algorithm) included terms for age, weight, height, systolic blood pressure and smoking (current and past). In the validation cohort, this model was well calibrated with good discrimination (C-statistic [SE]= 0.734 [0.03]). In the entire cohort (n=20822), the 10 allele genetic risk score was associated with incident AF (hazard ratio per allele 1.14, 95% CI 1.08–1.10, P<0.0001). The addition of the genetic score to the FHS and WGHS models offered statistically significant improvements in discrimination (Table). However, classification of participants into 10-year risk categories of <1, 1–5, and 5+% did not improve after adding the genetic score to the FHS, WHS, or age-alone models (Table).
Conclusion: A novel risk prediction model for AF derived and validated in a cohort of healthy women offers good discrimination and calibration. The addition of genetic information to this model, and to another derived in the FHS, improved discrimination but not clinical risk classification.
- © 2010 by American Heart Association, Inc.