Predicting atrial fibrillation using a combination of genetic risk score and clinical risk factors

doi:10.1016/j.hrthm.2020.01.006

Heart Rhythm

Volume 17, Issue 5, Part A, May 2020, Pages 699-705

https://doi.org/10.1016/j.hrthm.2020.01.006 Get rights and content

Background

Atrial fibrillation (AF) has a genetic basis, and environmental factors can modify its actual pathogenesis.

Objective

The purpose of this study was to construct a combined risk assessment method including both genetic and clinical factors in the Japanese population.

Methods

We screened a cohort of 540 AF patients and 520 non-AF controls for single nucleotide polymorphisms (SNPs) previously associated with AF by genome-wide association studies. The most strongly associated SNPs after propensity score analysis were then used to calculate a weighted genetic risk score (WGRS). We also enrolled 1018 non-AF Japanese subjects as a validation cohort and monitored AF emergence over several years. Finally, we constructed a logistic model for AF prediction combining WGRS and clinical risk factors.

Results

We identified 5 SNPs (in PRRX1, ZFHX3, PITX2, HAND2, and NEURL1) associated with AF after Bonferroni correction. There was a 4.92-fold difference in AF risk between the highest and lowest WGRS calculated using these 5 SNPs (P = 2.32 × 10⁻¹⁰). Receiver operating characteristic analysis of WGRS yielded an area under the curve (AUC) of 0.73 for the screening cohort and 0.72 for the validation cohort. The predictive logistic model constructed using a combination of WGRS and AF clinical risk factors (age, body mass index, sex, and hypertension) demonstrated better discrimination of AF than WGRS alone (AUC = 0.84; sensitivity 75.4%; specificity 80.2%).

Conclusion

This novel predictive model of combined AF-associated SNPs and known clinical risk factors can accurately stratify AF risk in the Japanese population.

Introduction

Atrial fibrillation (AF) is the most common arrhythmia and a major contributor to stroke and cardiovascular mortality.¹ The incidence of cerebral infarction from nonvalvular AF is approximately 5% per year, approximately 2–7 times higher than in the matched population without AF.²^,³ Early AF detection and therapeutic intervention are imperative because stroke prevention in high-risk AF patients is now possible with anticoagulant therapy.

Several studies have demonstrated a genetic basis for AF and modulation of AF pathogenesis by environmental factors.⁴^,⁵ Genome-wide association studies (GWASs) have identified several common single nucleotide polymorphisms (SNPs) that influence AF risk.⁶^,⁷ The most recent largest AF meta-analysis revealed 97 AF-associated loci in a mainly European population.⁸

Even in the Asian population including Japanese, 26-AF associated SNPs were demonstrated in previous GWASs.9, 10, 11 In addition to genetic factors, multiple clinical factors have been implicated in AF risk, such as hypertension, diabetes, aging, male sex, obesity, smoking, ischemic heart disease, valvular heart disease, and congestive heart failure.12, 13, 14

Stratification of AF risk is required for early AF detection and intervention to reduce the mortality caused by cerebral infarction or heart failure and the associated medical costs. We hypothesized that combining genetic and clinical risk factors can stratify AF risk more accurately than either alone. In this study, we constructed a novel risk model that includes both genetic and clinical risk factors to predict AF in the Japanese population.

Section snippets

Study participants

We retrospectively enrolled 565 Japanese patients with AF treated at Hiroshima University Hospital between November 2009 and April 2012. We excluded those with severe valvular disease (n = 1), congenital heart disease (n = 2), ischemic heart disease (n = 10), hypertrophic cardiomyopathy (n = 11), and dilated cardiomyopathy (n = 1). The remaining 540 Japanese AF patients were included as screening subjects. We also enrolled 520 Japanese non-AF controls from Hiroshima University as screening

Baseline patient characteristics and genotype distribution in the screening cohort

Baseline characteristics of the AF and the non-AF controls in the screening cohort are presented in Table 1. Compared to the non-AF group, the AF patients were older (59.1 ± 10.1 years vs 49.9 ± 14.7 years; P = 5.01 × 10⁻³⁰), more likely to be male (72.5% vs 48.0%; P = 2.27 × 10⁻¹⁶), more likely to have higher BMI (24.3 ± 3.4 vs 22.4 ± 3.4; P = 7.96 × 10⁻²⁰), more likely to have hypertension (56.4% vs 18.2%; P = 3.07 × 10⁻³⁵), and more likely to have diabetes (16.7% vs 6.7%; P = 2.73 × 10⁻⁷).

Discussion

Almost half of AF patients are asymptomatic, and these patients die of associated cardiovascular events at a rate 3 times higher than symptomatic AF patients.¹⁶ Thus, early AF detection is critical for reducing mortality. Various new technologies, such as wearable electrocardiographic patches, Apple watch/smartphones, and irregular beats-detecting blood pressure machines are being applied with increasing frequency in general practice. Nonetheless, early AF detection still is challenging,

Conclusion

The combination risk model using AF-associated SNPs (rs3903239, rs2106261, rs6817105, rs7698692, and rs6057225) and clinical risk factors (age, hypertension, BMI, and sex) can stratify AF risk in the Japanese population more accurately than either WGRS or clinical factors alone.

Acknowledgments

We thank the clerical and medical staff at Hiroshima University Hospital for their assistance.

References (25)

J.D. Roberts et al.
Impact of genetic discoveries on the classification of lone atrial fibrillation
J Am Coll Cardiol
(2010)
R.B. Schnabel et al.
Development of a risk score for atrial fibrillation (Framingham Heart Study): a community-based cohort study
Lancet
(2009)
A.M. Chamberlain et al.
A clinical risk score for atrial fibrillation in a biracial prospective cohort (from the Atherosclerosis Risk in Communities [ARIC] study)
Am J Cardiol
(2011)
K.C. Siontis et al.
Typical, atypical, and asymptomatic presentations of new-onset atrial fibrillation in the community: characteristics and prognostic implications
Heart Rhythm
(2016)
H. Lin et al.
Gene expression and genetic variation in human atria
Heart Rhythm
(2014)
S. Garnier et al.
Involvement of BAG3 and HSPB7 loci in various etiologies of systolic heart failure: results of a European collaboration assembling more than 2,000 patients
Int J Cardiol
(2015)
L. Liu et al.
Genetic variants associated with susceptibility to atrial fibrillation in a Japanese population
Can J Cardiol
(2017)
E.J. Benjamin et al.
Impact of atrial fibrillation on the risk of death: the Framingham Heart Study
Circulation
(1998)
P.A. Wolf et al.
Atrial fibrillation as an independent risk factor for stroke: the Framingham Study
Stroke
(1991)
S. Levy et al.
Characterization of different subsets of atrial fibrillation in general practice in France: the ALFA study
Circulation
(1999)

M.S. Olesen et al.

Atrial fibrillation: the role of common and rare genetic variants

Eur J Hum Genet

(2014)

P.T. Ellinor et al.

Large scale replication and meta-analysis of variants on chromosome 4q25 associated with atrial fibrillation

Eur Heart J

(2008)

Cited by (18)

Polygenic risk scores in atrial fibrillation: Associations and clinical utility in disease prediction
2024, Heart Rhythm
Atrial fibrillation (AF) is a common heart arrhythmia and a major cause of cardioembolic stroke. Therefore, accurate prediction is desirable to allow high-risk individuals to be identified early and their risk lowered before complications arise. Polygenic risk scores (PRSs) have become a popular method of quantifying aggregated genetic risk from common variants, but their clinical value in AF remains uncertain. This literature review summarizes the associations between PRS and AF risk and discusses the evidence for the clinical utility of PRS for AF prediction. Stroke risk in patients with AF is also considered. Despite consistent associations between PRS and AF risk, the performance of PRS as a stand-alone tool for AF prediction was poor. However, addition of PRS to the existing AF prediction models commonly improved the predictive performance above that of the clinical models alone, including in cohorts with comorbid cardiovascular disease. Associations between PRS and cardioembolic stroke risk in patients with AF have also been reported, but improvements in stroke prediction models from PRS have been minimal. PRS are likely to add value to the existing clinical AF prediction models; however, standardization of PRS across studies and populations will likely be required before they can be meaningfully adopted into routine clinical practice.
Relation of Antecedent Symptoms to the Likelihood of Detecting Subclinical Atrial Fibrillation With Inserted Cardiac Monitors
2021, American Journal of Cardiology
Citation Excerpt :
Briefly, compared to Holter monitoring, the ICM correctly identified AF in 97.4% (Reveal LINQ) and 96.1% (Reveal XT) of patients (diagnostic sensitivity) and correctly excluded AF in 97.0% (Reveal LINQ) and 85.4% (Reveal XT) of patients (diagnostic specificity). The reported overall accuracy to detect durations of AF was 99.4% and 98.5% for Reveal LINQ and Reveal XT, respectively.19 Participants were ≥18 years old with no AF history but deemed at risk for AF based on demographic characteristics, with or without symptoms.
Atrial fibrillation (AF) comes to attention clinically during an evaluation of symptoms, an evaluation of its adverse outcomes, or because of incidental detection during a routine examination or electrocardiogram. However, a notable number of additional individuals have AF that has not yet been clinically apparent or suspect—subclinical AF (SCAF). SCAF has been recognized during interrogation of pacemakers and defibrillators. More recently, SCAF has been demonstrated in prospective studies with long-term monitors—both external and implanted. The REVEAL AF trial enrolled a demographically “enriched” population that underwent monitoring for up to 3 years with an insertable cardiac monitor. SCAF was noted in 40% by 30 months. None of these patients had AF known before the study; however, some had nonspecific symptoms common to patients with known AF. The current study assessed whether patients with versus without such symptoms were more likely to have SCAF detected. We found that only palpitations had an association with AF detection when controlling for other baseline symptoms (hazard ratio 1.61 (95% confidence interval 1.12 to 2.32; p = 0.011). No other prescreening symptoms evaluated were associated with an increased likelihood of SCAF detection although patients without detected SCAF had an even higher frequency of symptoms than those with detected SCAF. Thus, REVEAL AF demonstrated that the presence of palpitations is associated with an increased likelihood of SCAF whereas other common symptoms are not; and, symptoms, per se, may more likely be consequent to associated disorders than they are a direct consequence of SCAF.
A novel multi-scale convolutional network with attention-based bidirectional gated recurrent unit for atrial fibrillation discrimination
2021, Biocybernetics and Biomedical Engineering
Citation Excerpt :
As a major cardiovascular disease, atrial fibrillation (AF) is a type of supraventricular tachyarrhythmia and resulted from the abnormal conduction, affecting over 33.5 million individuals worldwide [1]. The timely detection of this disease will be critical to reducing more harm to patients and various related complications such as coronary heart disease and thrombus [2–4]. Therefore, the development of the proper method strategy for AF detection has received considerable attention in recent years.
Atrial fibrillation (AF) is a major cardiovascular disease that has affected thousands of individuals worldwide. The electrocardiogram (ECG) is the most extensively applied approach to detect AF at present, while the traditional detection strategy based on the visual observation of ECG data is often laborious and inefficient. In this work, we specially designed an intelligent recognition system based on a novel convolutional neural network that utilizes the multi-scale convolution kernel and bidirectional gated recurrent unit with attention mechanism for AF detection. Also, two standard control groups using 10-fold cross-validation were performed to assess the validity of the proposed model. The empirical results not only demonstrate the high efficiency of multi-scale convolution kernel, but also show that the model has a more superior classification performance to several state-of-the-art methods with an accuracy of 98.3% and 97.7% on two public databases, respectively. Due to its high performance, we plan to develop the model into portable devices to benefit more individuals such as the elderly and athletes.
Genetic testing in cardiovascular disease
2024, Medical Journal of Australia
Impact of rs1805127 and rs55742440 Variants on Atrial Remodeling in Hypertrophic Cardiomyopathy Patients with Atrial Fibrillation: A Romanian Cohort Study
2023, International Journal of Molecular Sciences
Risk stratification of atrial fibrillation and stroke using single nucleotide polymorphism and circulating biomarkers
2023, PLoS ONE

View all citing articles on Scopus

: Yukiko Nakano was supported by JSPS KAKENHI Grant Number 17K09501.

View full text

ClinicalAtrial FibrillationPredicting atrial fibrillation using a combination of genetic risk score and clinical risk factors

Background

Objective

Methods

Results

Conclusion

Introduction

Section snippets

Study participants

Baseline patient characteristics and genotype distribution in the screening cohort

Discussion

Conclusion

Acknowledgments

J Am Coll Cardiol

Lancet

Am J Cardiol

Heart Rhythm

Heart Rhythm

Int J Cardiol

Can J Cardiol

Impact of atrial fibrillation on the risk of death: the Framingham Heart Study

Circulation

Atrial fibrillation as an independent risk factor for stroke: the Framingham Study

Stroke

Characterization of different subsets of atrial fibrillation in general practice in France: the ALFA study

Circulation

Atrial fibrillation: the role of common and rare genetic variants

Eur J Hum Genet

Large scale replication and meta-analysis of variants on chromosome 4q25 associated with atrial fibrillation

Eur Heart J

Clinical
Atrial Fibrillation
Predicting atrial fibrillation using a combination of genetic risk score and clinical risk factors