To our knowledge, this is the largest prospective cohort study illustrating the 10-year risk of AF that takes into account both the burden of risk factors and genetic predisposition. Thus, the overall results were consistent with our study hypothesis, more specifically, among participants aged 45, 55 and 65 years, the overall 10-year risks of AF were 0.67%, 2.05% and 6. 34%, respectively. The 10-year risk of AF at the index ages of 45, 55, and 65 years ranged from 0.26%, 0.43%, and 2.27% among participants in the lowest tertiles of risk factor burden and genetic predisposition, rising to almost 1.65%, 4.22%. and 11.10% among those in the highest tertiles of risk factor burden and genetic predisposition. The 10-year risk was highest in men with high PRS and high risk factor burden at each index age.
There are three main clinical implications to be drawn from this analysis. First, risk factor burden/profiles, genetic predisposition, and their interactions play essential roles in AF risk, especially among male participants and those of younger age. Especially, the risk of AF attributable to a risk factor burden/profile increased along with the increased risk due to genetic predisposition. Second, age was the most prominent risk factor for AF. Along with aging, the contributions of risk factors and genetic predisposition to AF risk have decreased, especially among men. Third, at younger ages, high burden/risk profiles with low/intermediate PRS may lead to earlier onset of AF compared to the joint effect of high PRS and ideal burden/risk profiles. Thus, an optimal risk burden is important for AF prevention, especially for younger individuals or those with a high genetic predisposition for AF.
Comparison with other studies
The Framingham Heart Study estimated the short-term and lifetime risk of AF among 9,764 participants without AF aged 55 to 75 years. . Although of instrumental value, cohort studies are needed to further refine the joint effect of polygenic risk and other risk factors. Our updated 10-year risks, estimated by the contribution of risk factor burden and genetic predisposition, developed from a substantially larger prospective cohort study, offer superior generalizability to contemporary populations and help identify high-risk populations .
Although previous literature has reported the 10-year risk of AF to be 2.0%, 4.5%, and 7.6% among participants at the index age of 55 years with low, intermediate, and high polygenic risk , the lower 10-year risk observed in our study may be attributed to a relatively healthier cohort in the UK Biobank. Furthermore, the incidence of AF increases rapidly after age 40, and the 10-year risk of AF varies with age. .
Previous studies have mainly focused on middle-aged and older adults (aged ≥ 55 years) to investigate the 10-year risk of AF, and the applicability to younger individuals (aged ≤ 55) is uncertain. Indeed, prospective identification of these relatively young individuals at risk for AF may be beneficial, allowing for early intervention measures. . Our study included younger participants (index age 45 years) and observed that the joint effect of high risk factor burden/profiles and high hereditary predisposition on AF incidence decreased with advancing age. Furthermore, among the younger age group, compared to genetic predisposition, risk factor burden/profiles have a greater effect on the development of AF. Our results suggest that a more favorable burden/risk profile may delay the onset of AF risk, especially among young people and men. Similar to a previous study participants with high polygenic risk had twice the 10-year risk as those with low PRS, highlighting the independent impact of a hereditary predisposition to AF risk.
Three multivariate AF predictive models showed moderate discrimination (C-index 67.1% to 73.5%) but suboptimal calibration. The performances of our models were comparable to other models reported in previous studies [20, 29]. In our predictive models, the strongest risk factors were sex, history of MI or HF, and RPS. Both predicted and 10-year AF risk were higher in men than in women, as previously noted . One possible explanation could be that women had more favorable risk loadings compared to men. Furthermore, considering the main outcomes of the present study, which include AF and flutter, the burden of risk factors and genetic predisposition may have an independent prognostic impact in terms of different types of AF. .
Compared to the performances of previously used models, we included ethnicity as an important covariate in AF prediction models due to accumulating evidence that white individuals have a higher risk of AF than those who are not white. . Furthermore, the preponderance of AF cases in our population (56%, 71% and 82% at the index age of 45, 55 and 65 years, respectively) developed in participants with a history of myocardial infarction, HF, hypertension or diabetes, suggesting that comorbidities contributed more to AF risk in older individuals. Furthermore, lifestyle risk factors, including BMI, alcohol consumption, and smoking, which are modifiable, also played important roles in the 10-year risk and our predictive models. Alcohol consumption led to the highest 10-year AF risk among lifestyle risk factors. There is increasing evidence regarding the adverse effects of alcohol on left atrial function, electrical remodeling, and structure. These effects would contribute to FA . Considering the close association of risk factors with AF risk, these modifiable factors may be better targets for AF prevention. .
Furthermore, we hypothesize that our prediction models could be used to better inform targeted screening of at-risk individuals, which may be more cost-effective than routine screening of all patients aged ≥65 years. . [35, 36]. AF risk factor loading is an easily interpretable and accessible structured management tool that can be easily measured with existing and future data. Furthermore, with the increasing ubiquity of genomic data, both in the clinical context and through direct-to-consumer testing, the current list of optimal variants of a PRS for AF could potentially provide a more accurate personalized risk assessment. [15, 21]. For individuals with an index age of 45, 55 and 65 years, with a high burden of risk factors and high PRS, the 10-year risks of AF were 1.65%, 4.22% and 11.10%, respectively. which could potentially be used as thresholds. Application of prediction models could be used to inform screening thresholds in individuals with predicted high risks in 10 years, potentially leading to early detection, guidance on preventative approaches, and individualized counseling.
Strengths and limitations
Strengths of this study include the exceptionally large sample size, including younger participants; long-term follow-up; detailed information about behavior, clinical profile and genetic data; the large number of AF cases included in the analysis; and estimation of 10-year risks incorporating risk burden/profiles and genetic predisposition with adjustment for competing risks. Predictive models and potential thresholds were developed in our study to help implement early and targeted interventions.
Our study also has some limitations. It is important to highlight that biomarkers, electrocardiographic data, anemia, electrolyte imbalance and other potential risk factors for AF were not included in our study . However, some studies have pointed out that electrocardiogram data may not improve the performance of predictive models. [8, 38]. Some AF events caused by acute events such as trauma, surgery, etc. may have gone unnoticed, as most events were determined using inpatient data, therefore the true risk of AF may be underestimated, since undiagnosed AF is common . Furthermore, measurements of risk factor burden/profiles were obtained at baseline and could not reflect changes over time. Participants recruited from the UK Biobank are healthier and have a higher socioeconomic status than the general UK population, which may result in healthy volunteering bias. . Given that an adequate number of participants with various levels of exposure were assessed with high internal validity, this may not impact valid estimates of relationships . The UK Biobank does not record close relationships or does not disclose the relatedness coefficients of all participants estimated from genetic data, which prevents us from implementing mixed effects models that take into account relatedness between samples. Finally, our study used only one cohort to predict 10-year risks and AF predictive models, and whether our findings can be generalized to other ethnic groups needs further investigation.