Premature ventricular contraction is associated with increased risk of atrial fibrillation: a nationwide population-based study

Premature ventricular contraction (PVC) and atrial fibrillation (AF) are common arrhythmias affecting 1–2% of the general population. During PVC, retrograde ventriculo-atrial activation can occur and act like an atrial ectopy. However, the clinical significance of this phenomenon is not fully understood. We aimed to elucidate whether the clinical diagnosis of PVC can increase the risk of new-onset AF. We performed a nationwide population-based analysis using the Korean National Health Insurance Service database. A total of 9,537,713 people without prior history of PVC and AF were identified. Among these people, 4135 developed PVC in 2009, and people with and without PVC were followed until 2018. People who had PVC showed an increased risk of new-onset AF as compared with people without PVC (hazard ratio [HR] = 2.705; 95% confidence interval [CI] = 2.428–3.013; p < 0.001). The risk of ischemic stroke was also significantly increased in people with PVC (HR 1.160; 95% CI 1.048–1.284; p = 0.0041). New-onset AF developed in 72 people (19.3%) among 374 people with PVC who had ischemic stroke during their follow-up. A significant interaction was observed between PVC and age with people < 65 years at greater risk of new-onset AF for having PVC. In this observational analysis, the risk of new-onset AF and ischemic stroke was increased in people with PVC. Additional evaluation to identify AF in people with PVC can be helpful.

Primary outcome endpoint. People who had a national health check-up in 2009 were classified into one of two groups: (1) those who were diagnosed with PVC in 2009 and (2) those who were not diagnosed with PVC in 2009. The incidence of new-onset AF and ischemic stroke was compared. The incidence of new-onset AF was defined as the number of new-onset AF cases calculated for 1000 patient * years of follow-up. The impact of the aforementioned non-genetic risk factors was evaluated in various age groups.
Definitions. The diagnosis of PVC was based on two criteria: (1) one inpatient or outpatient record (PVC 1) or (2) two or more outpatient records or one inpatient record (PVC 2) of the International Classification of Disease, Tenth Revision (ICD-10; I49.3) codes in the K-NHIS database. People with more symptomatic PVC will likely to have subsequent outpatient clinic visit or hospital admission. Therefore, PVC diagnosed by repeat outpatient or inpatient record might represent more symptomatic form of PVC.
Diagnosis of AF required two outpatient records or one inpatient record (I48). The robustness of this definition has been validated in previous studies 16,17 . One or more outpatient or inpatient records was required for the diagnosis of ischemic stroke (I63, I64). In the current study, transient ischemic attack was not included in ischemic stroke. The exact diagnosis codes for PVC, AF, and other diseases are described in Supplementary  Table S1.
Body weight status was classified into five groups according to body mass index (BMI): BMI < 18.5 kg/m 2 , 18.5 ≤ BMI < 23.0, 23.0 ≤ BMI < 25.0, 25.0 ≤ BMI < 30.0, and BMI ≥ 30.0. Smoker was defined as those who smoked at least 100 cigarettes in their lifetime and continued smoking within 1 month of the 2009 national health checkup. Ex-smokers were those who smoked at least 100 cigarettes in their lifetime but had not smoked within 1 month of the 2009 national health check-up, and never-smokers were those who smoked < 100 cigarettes in their lifetime. Based on the total amount of alcohol intake per week, patients were classified into three groups: (1) nondrinker, 0 g of alcohol per week; (2) mild-to moderate-drinker, less than 210 g but more than 0 g of alcohol per week; and (3) heavy-drinker, 210 g or more of alcohol per week. Diagnosis of diabetes was made for participants of the 2009 national health check-up who had impaired fasting sugar > 126 mg/dl or prior history of physician-diagnosed diabetes. Hypertension was defined as those who were previously diagnosed with hypertension or had either systolic blood pressure ≥ 140 mmHg or diastolic blood pressure ≥ 90 mmHg during 2009 national health check-up. Dyslipidemia was defined as a prior history of dyslipidemia diagnosis by a physician. Estimated glomerular filtration rate < 60 ml/min/1.73 m 2 was the diagnostic criteria for chronic kidney disease.
Statistical analysis. For continuous variables, the Student's t-test was used; results are described as mean ± standard deviation. Categorical variables are presented as percentile values and were compared with a chi-square test or Fisher's exact test as appropriate. Cumulative incidence curve analysis with log-rank t test and death as a competing risk was used to compare the incidence of new-onset AF and ischemic stroke in people with and without PVC. Univariate and multivariate Cox regression analyses were performed to calculate the haz-

Patients.
A total of 10,601,284 people underwent a national health check-up in 2009. People who were (1) under 20 years of age or (2) had a prior history of heart failure, PVC, AF, ischemic stroke, ventricular tachycardia, or ventricular fibrillation were excluded. In the analysis, 9,537,713 people were included. In total, 4135 people were diagnosed with PVC with 2141 and 1994 people having PVC 1 (one record) and PVC 2 (two outpatient record or one inpatient record), respectively, in 2009. The study flow is shown in Fig. 1. Baseline clinical characteristics of the cohort are summarized in Table 1. Significant difference in demographics among people with and without PVC was observed. People who developed PVC were significantly older; more likely to be female/ non-drinker/non-smoker; and had a higher prevalence of hypertension, diabetes mellitus, and dyslipidemia.
New-onset AF. In people with no PVC, a total of 192,549 new-onset AF developed during the 87,256,944 person * year follow-up (incidence: 2.207). A total of 330 new-onset AF cases were observed in people with PVC during the 36,419 person * year follow-up (incidence: 9.061). The incidence of new-onset AF was even higher in people with PVC 2 with 205 new-onset AF cases during the 17,225 person * year follow up (incidence = 11.902) as compared with PVC 1 (incidence = 6.512; non-adjusted HR 1.848; 95% CI 1.480-2.308; p < 0.001). Kaplan-Meier curve analysis showed significantly higher incidence of new-onset AF in people with PVC as compared with people with no PVC (p < 0.001; Fig. 2a,b). Multivariate adjusted models also revealed a significantly increased risk of new-onset AF in people with PVC as compared with those without PVC (

Discussion
This study revealed an increased risk of new-onset AF and ischemic stroke in people with PVC. Various risk factors for AF have been reported in previous studies, but whether PVC could increase the risk of new-onset AF was unknown 17,18 . Our study is the first to report such an association. Due to sufficient sample size, we were able to perform various subgroup analyses and demonstrated that the risk of new-onset AF and ischemic stroke is more pronounced in people < 65 years. Since the identification of PVC was based on diagnostic codes, patients with PVC in our nationwide cohort are highly likely to be symptomatic patients seeking specific medical care for PVC. Therefore, our data can have practical implications for patients with PVC having outpatient or inpatient medical follow-ups. Heterogeneous duration of PVC originating from ICD-10 code based diagnosis can be problematic.
To resolve this limitation, we excluded people with prior diagnosis of PVC during 2002 to 2008 and only those with newly diagnosed PVC in 2009 were included. Due to sufficient follow-up duration and large sample size, we were able to observe the influence of newly diagnosed PVC on AF and ischemic stroke. www.nature.com/scientificreports/ PVC and new-onset AF. Common risk factors exist between PVC and AF, e.g., age, hypertension, and diabetes mellitus. In our cohort, people with PVC were older and had higher prevalence of hypertension and diabetes, which are all risk factors associated with AF. However, increased risk of new-onset AF in people with PVC was statistically significant even after adjusting for these common risk factors. Conversely, people with PVC were less likely to be male in contrast with AF, which had male predominance. Body weight status, another important risk factor for AF, also did not differ between people with and without PVC. Our prior study showed no reliable association between dyslipidemia and AF, but the prevalence of dyslipidemia was higher in people with PVC than in those without PVC in this study. People with PVC 2 had greater risk of new-onset AF compared with people with PVC 1. Diagnosis of PVC 2 was based on intensified criteria compared with PVC 1; therefore, PVC 2 might represent a more advanced form of PVC. Higher risk of new-onset AF in people with PVC 2 also support the association between PVC and new-onset AF.
The underlying mechanism linking PVC and increased risk of new-onset AF is not clear. Retrograde ventriculo-atrial conduction can occur with PVC. These retrograde atrial activations can act like an atrial ectopies originating from pulmonary veins. Reduced left ventricular function and subsequent increase in left atrial pressure can also contribute to increased risk of new-onset AF in people with. We have previously reported the impact of hypertension and diabetes on new-onset AF is more profound in young people as compared to the old 20 . Similarly, pronounced impact of PVC on new-onset AF and ischemic stroke in young people is another important finding of our study. The underlying mechanism of such interaction is not clear but may involve different characteristics of PVC in young vs. old age. Common genetic predisposition of PVC and AF in young people can also be the cause of such interaction. www.nature.com/scientificreports/ Ischemic stroke. Prior studies reported an association between PVC and ischemic stroke 14,15 . However, the underlying pathophysiology linking PVC and ischemic stroke is not fully elucidated. Not only AF but also premature atrial contraction has previously been reported to increase the risk of ischemic stroke 21 . Retrograde ventriculo-atrial conduction by PVC can act like a premature atrial contraction, which may be the underlying cause of the increased risk of ischemic stroke. Increased risk of AF in people with PVC can also contribute to the occurrence of ischemic stroke. Among the 4135 people with PVC, 330, 374, and 72 people developed newonset AF, ischemic stroke, or both, respectively, suggesting a possible causal relationship between new-onset AF and ischemic stroke. Another hypothesis is concealed cardiomyopathy. Although we excluded people with prior diagnosis of heart failure, people with subclinical cardiomyopathy would not have been excluded. Premature ventricular contraction can be the first symptom of underlying subclinical cardiomyopathy in those people. When subclinical cardiomyopathy becomes clinical heart failure it can increase the risk of new-onset AF. If the underlying cardiomyopathy also affects atrial myocardium, it can also increase the risk of new-onset AF and ischemic stroke 22 . Adjusting the confounder effect of subclinical cardiomyopathy will be an important issue in future trials.
Our study raises the hypothesis that PVC can increase the risk of new-onset AF and ischemic stroke. Although people with PVC were older and had higher prevalence of hypertension, diabetes mellitus, and dyslipidemia, an association between PVC and ischemic stroke was shown after adjusting multiple parameters. Furthermore, when PVC were diagnosed with more intensified criteria (PVC 2), the incidence of ischemic stroke was higher ( Table 2). Whether suppression of PVC with drugs or radiofrequency catheter ablation can reduce the risk of new-onset AF and ischemic stroke should be tested in future trials.
Limitations. This study has several limitations. First, we were not able to obtain the burden of PVC measured by Holter monitoring. Currently, the K-NHIS database does not have any data regarding PVC burden. However, based on the number of inpatient or outpatient records for PVC, we classified PVC into PVC 1 or PVC 2. Diagnosis of PVC 2 was based on more intensified criteria and risk of new-onset AF, and ischemic stroke was higher in PVC 2 as compared with PVC 1. It is possible that that advanced form of PVC (PVC 2) are associated with more adverse events. Second, this study was based on a nationwide administrative database, and the results  www.nature.com/scientificreports/ might have missing data due to coding inaccuracies. It is possible that people with asymptomatic PVC were not included in the PVC group. Although identification of PVC with ICD-10 codes (I49.3) using nationwide medical database is a novel approach, it needs further validation. Differences in baseline demographics originating from the retrospective nature of the study is another limitation. Third, our database was based only on East Asian people, and caution is required when applying our results to other ethnic groups. Fourth, we were not able to distinguish between paroxysmal (trigger type) and non-paroxysmal (substrate type) AF, which might have different pathophysiology. If retrograde ventriculo-atrial activation is responsible for triggering AF in people with PVC, trigger type AF rather than substrate type AF will be the predominant form of AF observed in people with PVC.

Conclusions
Premature ventricular contractions might increase the risk of new-onset AF and ischemic stroke. Whether suppression of PVC with drugs or radiofrequency catheter ablation can reduce the risk of new-onset AF and ischemic stroke should be tested in future trials.