Population-based prediction of atrial fibrillation

Sammanfattning: Background: Atrial fibrillation (AF) is the most common sustained cardiac arrhythmia and increases in prevalence. As AF confers increased risk of morbidity, stroke, dementia and mortality, a major objective of cardiovascular epidemiology today is the development of tools for prediction and prevention of AF and its consequences. Although AF has traditionally been considered a non-heritable disease, a familial component to AF has recently been established and genetic polymorphisms associated with AF risk. The aim of this thesis was therefore to evaluate the ability of genetic polymorphisms, conventional cardiovascular risk factors and blood biomarkers reflecting diverse pathophysiological pathways to predict onset of AF in the general population. Methods: Individuals with incident AF were identified from a large, population-based cohort study (Malmö Diet and Cancer, n=30 447) using national registers during up to 17.8 years follow-up. Genetic polymorphisms on chromosomes 4q25, 16q22 and in the KCNH2 gene reproducibly associated with AF were genotyped in the entire cohort. Six blood biomarkers (MR-proANP, MR-proADM, NT-proBNP, copeptin, CRP and cystatin c) were measured in a random subcohort (n=6104). Results: Case validity of AF diagnoses in national registers was estimated to be high, 95-97%. A substantial proportion of population risk was conferred by conventional risk factors (hypertension, 34-38%; obesity, 10-11%). Conventional risk factors predicted AF with reasonable accuracy (C-statistic 0.732), although age and sex considered alone were only modestly less accurate. Addition of blood MR-proANP and CRP improved predictive accuracy modestly (C-statistic 0.753). Polymorphisms on chromosomes 4q25 and 16q22 but not in KCNH2 were associated with AF independently of conventional risk factors, but did not significantly improve predictive accuracy. In a meta-analysis of published studies including up to 150 000 individuals, the association of polymorphisms with AF was robust across diverse study designs, but with widely varying risk estimates. Genetic polymorphisms but not conventional risk factors were associated with AF in heart failure (HF) patients. A SNP on chromosome 16q22 was more strongly associated with AF in this context (p for interaction = 7x10-4), suggestive of a pathophysiological interaction of the gene affected by this SNP and HF. Conclusion: The results of this thesis have implications for population-based prediction of AF and provides novel information on risk factors for AF in HF patients.