Fatty Liver Index and the Risk of Atrial Fibrillation in a General Japanese Population　― The Suita Study ―

Abstract

Background: Atrial fibrillation (AF) is the most diagnosed arrhythmia in clinical settings. The fatty liver index (FLI) is a marker of liver steatosis with potential cardiovascular implications. This study investigated whether FLI could predict the risk of AF.

Methods and Results: We used data from the Suita Study, a Japanese population-based prospective cohort study. A total of 2,346 men and 3,543 women, aged 30–84 years, without prevalent AF were included and followed up. The diagnosis of AF was established during follow-up using electrocardiograms, hospital records, and death certificates. FLI was assessed during a baseline health checkup. Hazard ratios (HRs) and 95% confidence intervals (CIs) were calculated for incident AF per FLI quintile and log-transformed FLI. Within a median 14.5 years of follow-up, 142 men and 105 women developed AF. Compared with women in the third (middle) FLI quintile, women in the first (lowest), fourth, and fifth (highest) quintiles showed a higher risk of AF, with multivariable-adjusted HRs of 2.37 (95% CI 1.06–5.31), 2.60 (95% CI 1.30–5.17), and 2.04 (95% CI 1.00–4.18), respectively. No corresponding associations were observed in men. The change in log-transformed FLI was not associated with the risk of AF in either sex.

Conclusions: A U-shaped association between FLI and AF risk was detected in Japanese women. FLI could be a screening tool to detect women at high risk of developing AF.

The prevalence of atrial fibrillation (AF) has been increasing in Japan and globally.^1,2 AF constitutes a major risk factor for stroke, coronary artery disease, and heart failure.^3–5 Detecting people at high risk of developing AF could allow for early prevention and management, making a significant proportion of AF events preventable.⁶

Non-alcoholic fatty liver disease (NAFLD) is a common liver disease characterized by the accumulation of intrahepatic fat of at least 5% of liver weight in the absence of significant alcohol use.⁷ Simple fatty liver or steatosis, non-alcoholic steatohepatitis (NASH), fibrosis, and cirrhosis are the 4 stages of NAFLD.^7,8 Several mechanisms are involved in the accumulation of intrahepatic fat, including increased flux of fatty acids to the liver, increased de novo lipogenesis, and/or reduced fat clearance.^7,8 NAFLD is typically diagnosed using ultrasound or liver biopsy.⁸ However, a considerable proportion of NAFLD cases are missed because ultrasound is not usually performed during routine health checkups and NAFLD could be asymptomatic, especially in the early stages.⁸ Therefore, more feasible diagnostic methods, such as the fatty liver index (FLI), are indicated.⁹ FLI is an algorithm derived from measurements that are routinely assessed during clinical and laboratory checkups: γ-glutamyl transferase (GGT), waist circumference (WC), body mass index (BMI), and triglycerides (TG).¹⁰ The FLI does not require any additional imaging or invasive procedures. In addition to its feasibility, FLI has shown a high sensitivity and specificity in detecting NAFLD.¹⁰ These criteria made the FLI a quick and cost-effective tool for population-level screening to identify people at risk of NAFLD and to monitor the progression or regression of their condition over time.¹⁰

Previous studies indicated that higher FLI levels could predict future cardiovascular disease (CVD) events.^11–13 However, little is known about the association between FLI and AF risk, with no consistent conclusions.^13–15 Further, evidence from Japan, where the prevalence of NAFLD has been increasing,¹⁶ is lacking. In this context, we used data from the Suita Study, a prospective cohort study investigating urban Japanese people, to explore the association between FLI and the risk of AF.

Methods

Study Design and Population

As described elsewhere,^17–19 the Suita Study is a population-based prospective cohort study aiming to investigate lifestyle and clinical determinants of CVD. The Suita population included 2 randomly selected cohorts stratified by age and sex from the urban city of Suita in southwest Japan and a volunteer group. The 2 cohorts included 7,814 eligible participants (recruited between 1989 and 1998), and the volunteer group included 546 eligible participants (recruited between 1992 and 2006). Participants attended a baseline health checkup that included collection of information about their lifestyle, clinical assessment, blood sampling, and an electrocardiogram (ECG). Participants were then asked to return every couple of years for follow-up. In this study, we excluded participants who did not attend the baseline assessment or any follow-up (n=1,382), had AF or atrial flutter at baseline (n=42), lacked baseline measurements of GGT, WC, weight, height, or TG (n=21), or reported heavy drinking (≥46.0 g ethanol/day in men and ≥34.5 g ethanol/day in women; n=1,026), which left 5,889 participants (2,346 men, 3,543 women) for analysis.

Outcome and Exposure

At the follow-up visits, AF or atrial flutter was diagnosed by 2 trained internists using a standard 12-lead rest ECG according to the Minnesota Codes (8-3-1 to 4). Hospital records, health checkup findings, and death certificates of participants were systematically reviewed to confirm the diagnosis of AF.¹⁷ FLI was calculated using variables collected during the baseline checkup using the following formula:¹⁰

FLI = (e^{0.953 × loge(TG) + 0.139 × BMI + 0.718 × loge(GGT) + 0.053 × WC − 15.745}) / (1 + e^{0.953 × loge(TG) + 0.139 × BMI + 0.718 × loge(GGT) + 0.053 × WC − 15.745}) × 100

where loge is the natural logarithm.

Statistical Analyses

Statistical analyses were conducted using SAS version 9.4 software (SAS Institute Inc., Cary, NC, USA). The Chi-squared test was used to test the significance of differences in the proportions of different variables across FLI quintiles (see Table 1). Cox proportional hazards models were used to calculate hazard ratios (HRs) and 95% confidence intervals (95% CIs) for AF. The associations were calculated per quintiles, clinical cutoffs, and log(FLI) change. The third FLI quintile was assigned as the reference group in women because women showed an increased risk of AF with low and high FLI. In addition, FLI ≥60 has been suggested in the literature as a reliable diagnostic of NAFLD;¹⁰ therefore, we used this cut-off in a separate analysis. We also investigated linear associations by assessing AF risk per log-transformed FLI change. The associations were further displayed using restricted cubic spline curves. All analyses were stratified by sex. To investigate the potential impact of BMI and other variables on the association between FLI and AF risk, we computed P values for interation. Person-years were calculated from the baseline assessment date until that of AF diagnosis, death, leaving the study, or the end of follow-up (December 31, 2015), whichever came first.

Table 1. Participant Characteristics According to Fatty Liver Index Quintile at Baseline

Characteristics	Men						Women
	Q1	Q2	Q3	Q4	Q5	P value	Q1	Q2	Q3	Q4	Q5	P value
No. participants	469	469	470	469	469	–	708	709	709	709	708	–
Fatty liver index
Range	0.9–7.9	8.0–17.2	17.3–29.6	29.7–48.0	48.1–98.8	–	0.4–3.4	3.5–6.8	6.9–13.1	13.2–28.0	28.1–99.2	–
Median	4.8	12.3	22.8	37.6	61.0	–	2.2	4.9	9.4	19.4	44.7	–
Age group (%)
<50 years	26.9	23.0	23.2	26.5	27.7	<0.001	61.7	44.9	31.6	19.5	14.1	<0.001
50–59 years	14.7	18.8	19.2	22.6	26.2		17.4	22.7	30.0	26.4	28.3
60–69 years	26.6	30.9	30.4	29.6	29.4		12.0	20.3	25.4	33.7	36.3
≥70 years	31.8	27.3	27.2	21.3	16.7		8.9	12.1	13.0	20.4	21.3
Body mass index (%)
<18.5 kg/m2	32.2	2.1	0.8	0.0	0.0	<0.001	34.2	10.5	2.7	0.5	0.0	<0.001
18.5–24.9 kg/m2	67.2	96.2	89.6	70.4	34.3		65.7	89.1	92.7	76.6	37.0
≥25 kg/m2	0.6	1.7	9.6	29.6	65.7		0.1	0.4	4.6	22.9	63.0
Current smokers (%)	45.4	40.1	39.6	40.5	40.3	0.117	10.0	8.9	10.7	10.0	11.2	0.680
Current drinkers (%)	55.7	64.4	63.4	65.0	65.7	0.009	29.7	28.6	29.5	30.6	27.4	0.737
Hypertension (%)	25.2	27.3	38.1	40.3	46.1	<0.001	10.5	19.6	27.9	37.9	52.0	<0.001
Diabetes (%)	2.8	5.3	8.3	7.9	10.0	<0.001	1.3	1.3	2.3	3.5	9.6	<0.001
HDL-C (%)
<40 mg/dL (men)/<50 mg/dL (women)	7.5	18.1	22.5	31.3	46.3	<0.001	10.4	16.2	24.4	39.4	53.2	<0.001
40–59 mg/dL (men)/50–59 mg/dL (women)	56.5	59.9	61.1	57.6	48.0		26.0	31.9	32.7	29.6	26.0
≥60 mg/dL	36.0	22.0	16.4	11.1	5.7		63.6	51.9	42.9	31.0	20.8
Total cholesterol (%)
<200 mg/dL	63.5	55.2	44.0	40.9	37.9	<0.001	55.9	44.0	30.6	25.2	20.9	<0.001
200–239 mg/dL	29.9	35.4	43.2	43.3	40.3		35.0	41.2	43.9	42.9	42.9
≥240 mg/dL	6.6	9.4	12.8	15.8	21.8		9.1	14.8	25.5	31.9	36.2
Chronic kidney disease (%)	6.2	10.0	9.8	8.5	11.5	0.063	7.8	9.0	8.7	9.6	10.6	0.445
Cardiovascular disease (%)	3.6	3.0	5.3	3.2	6.0	0.080	0.7	0.4	0.7	1.3	1.8	0.055

HDL-C, high-density lipoprotein cholesterol; Q, quintile.

The regression models were adjusted for baseline variables as follows. Model I was adjusted for age (<50, 50–59, 60–69, or ≥70 years). Model II was further adjusted for smoking status (non-current or current), alcohol intake (non-current or current), hypertension (defined as blood pressure >140/90 mmHg or receiving medications; yes or no), diabetes (defined as fasting blood glucose ≥126 mg/dL or receiving medication; yes or no), total cholesterol (<200, 200–239, or ≥ 240 mg/dL), high-density lipoprotein cholesterol (<40 mg/dL in men/<50 mg/dL in women, 40–59 mg/dL in men/50–59 mg/dL in women, or ≥60 mg/dL), chronic kidney disease (defined as estimated glomerular filtration rate <60 mL/min/1.73 m²; yes or no), and CVD history (defined as having a positive history of stroke or ischemic heart disease; yes or no).

Results

The highest FLI quintiles included higher proportions of AF risk factors than the lowest FLI quintiles. Among men, the proportion of those with hypertension, diabetes, and total cholesterol ≥240 mg/dL was 46.1%, 10.0%, and 21.8%, respectively, in the fifth (highest) quintile, compared with 38.1%, 8.3%, and 12.8%, respectively, in the third (middle) quintile and 25.2%, 2.8%, and 6.6%, respectively, in the first (lowest) quintile (P<0.001). Among women, the proportion of those with hypertension, diabetes, and total cholesterol ≥240 mg/dL was 52.0%, 9.6%, and 36.2%, respectively, in the fifth (highest) quintile, compared with 27.9%, 2.3%, and 25.5%, respectively, in the third (middle) quintile and 10.5%, 1.3%, and 9.1%, respectively, in the first (lowest) quintile (P<0.001; Table 1). Within 81,492 person-years (median 14.5 years of follow-up), 247 participants (142 men, 105 women) developed AF.

The restricted cubic spline curves showed a U-shaped association between FLI and AF risk in women (Figure). In men, FLI was not associated with the risk of AF (Table 2). In women, compared with the third quintile, the first, fourth, and fifth quintiles showed increased risks of AF, with age-adjusted HRs of 2.21 (95% CI 1.01–4.83), 2.72 (95% CI 1.38–5.38), and 2.36 (95% CI 1.18–4.71), respectively, and multivariable-adjusted HRs of 2.37 (95% CI 1.06–5.31), 2.60 (95% CI 1.30–5.17), and 2.04 (95% CI 1.00–4.18), respectively (Table 3). BMI, as well as the other variables, did not interact with the association between FLI and AF in either sex (P value for interations >0.10). FLI ≥60 (vs. FLI <60) and the log-transformed FLI change were not associated with the risk of AF in either sex (Tables 2,3).

Figure.

Association between the fatty liver index (FLI) and the risk of atrial fibrillation in men and women separately. Multivariable-adjusted hazard ratios (solid lines) and 95% confidence intervals (dashed lines) were calculated, and the FLI was modeled using restricted cubic splines with knots at the 5th, 25th, 50th (reference), 75th, and 95th percentiles.

Table 2. FLI and the Risk of AF Among Men

	Incident cases	Incidence per 10,000 person-years	HR (95% CI)
			Model I	Model II
FLI quintile
Q1	26	43.8	1 (Ref.)	1 (Ref.)
Q2	29	47.1	0.95 (0.56–1.62)	0.95 (0.55–1.64)
Q3	27	44.5	0.95 (0.55–1.63)	0.91 (0.52–1.60)
Q4	26	41.1	0.97 (0.56–1.67)	0.94 (0.53–1.69)
Q5	34	53.0	1.31 (0.78–2.20)	1.23 (0.69–2.21)
Cut-off (FLI ≥60)	15	44.9	1.23 (0.72–2.11)	1.17 (0.66–2.06)
Log(FLI)	–	–	1.04 (0.87–1.25)	1.00 (0.81–1.23)

Model I was adjusted for age. Model II was further adjusted for smoking, alcohol consumption, hypertension, diabetes, high-density lipoprotein cholesterol, total cholesterol, chronic kidney disease, and cardiovascular disease. AF, atrial fibrillation; CI, confidence interval; FLI, fatty liver index; HR, hazard ratio; Q, quintile.

Table 3. FLI and the Risk of AF Among Women

	Incident cases	Incidence per 10,000 person-years	HR (95% CI)
			Model I	Model II
FLI quintile
Q1	15	13.8	2.21 (1.01–4.83)	2.37 (1.06–5.31)
Q2	13	12.2	1.35 (0.60–3.02)	1.39 (0.62–3.12)
Q3	11	10.5	1 (Ref.)	1 (Ref.)
Q4	35	36.7	2.72 (1.38–5.38)	2.60 (1.30–5.17)
Q5	31	34.3	2.36 (1.18–4.71)	2.04 (1.00–4.18)
Cut-off (FLI ≥60)	6	31.2	1.13 (0.50–2.58)	0.92 (0.39–2.15)
Log(FLI)	–	–	1.16 (0.96–1.41)	1.08 (0.86–1.35)

Model I was adjusted for age. Model II was further adjusted for smoking, alcohol consumption, hypertension, diabetes, high-density lipoprotein cholesterol, total cholesterol, chronic kidney disease, and cardiovascular disease. *In the case of assigning Q1 as a reference, the HRs and 95% CIs for AF per Model II would be 0.59 (0.28–1.25) for Q2, 0.42 (0.19–0.95) for Q3, 1.09 (0.56–2.14) for Q4, and 0.86 (0.42–1.76) for Q5. Abbreviations as in Table 2.

Discussion

This study indicated a U-shaped association between FLI and the risk of AF in Japanese women, but not men. However, FLI ≥60, a surrogate marker of NAFLD, and the log-transformed FLI change were not associated with the risk of AF in either sex.

There have been few investigations into the association between FLI and AF risk. Using data from 196,128 participants in the UK Biobank, a 10-unit increase in FLI was associated with an 8% increase in AF risk.¹³ The association was stronger in the overweight/obese group than in the normal weight/underweight group: 10% vs. 4%, respectively.¹³ In a nationwide Korean study including 8,048,055 participants, FLI 30–59 and ≥60 (vs. FLI <30) showed increased AF risk (by 5% and 12%, respectively).¹⁴ Unlike the UK Biobank study, the associations in the Korean study were strengthened in the underweight group and disappeared among the obese groups.¹⁴ When the sample was restricted to 5,333,907 young adults, AF risk increased by 47% and 21% among participants with FLI 30–59 and ≥60 (vs. FLI <30), respectively, and the associations remained consistent across BMI groups.¹⁵ In line with our study, the eligibility criteria in the previous studies excluded heavy drinkers and the studies did not adjust their results for BMI. However, we could not stratify our results by BMI because of the limited sample size. Although no multiplicative interaction was detected (for BMI, P value for interations=0.683 in men and P value for interations=0.375 in women), suggesting that BMI did not modify the association between FLI and AF risk in our study.

Explaining the U-shaped association between FLI and AF among women in our study is quite challenging. However, FLI is a marker of liver steatosis that induces systemic inflammation, atherosclerosis, and insulin resistance, factors that can create a proarrhythmic state.^20,21 In addition, previous studies linked elevated FLI to clinical conditions closely associated with AF risk, such as hypertension and diabetes.^22,23 A high-fat diet, a major risk factor for fatty liver,²⁴ has led to a profound transformation of atrial energy metabolism with fat accumulation in mouse models, inducing electrical remodeling of their atrial myocardium that becomes vulnerable to AF.²⁵ Both liver steatosis and AF are associated with local and systemic inflammation that could lead to liver and atrial fibrosis, respectively.^26,27 Conversely, a low FLI could reflect underweight, a major risk factor for AF among Asian populations.²⁸ BMI is inversely associated with concentrations of B-type natriuretic peptide (BNP),²⁹ a marker of AF and heart failure.³⁰ Further, frailty associated with being underweight is associated with elevated BNP concentrations and AF risk.^31,32 However, when we repeated the analysis after excluding women with BMI <18.5 kg/m², who were mostly concentrated in the lowest FLI quintiles, the increased risk of AF among the first (lowest) FLI quintile remained consistent (data not shown), suggesting the involvement of other mechanisms.

Conversely, the absence of a significant U-shaped association between FLI and AF risk among men in this study could be explained by 2 main reasons. First, we excluded heavy drinkers from the analysis, and heavy drinking is positively associated with both AF and FLI. Thus, when heavy drinkers were included in the analysis, elevated FLI became significantly associated with AF risk in men (data not shown). In addition, low FLI in men, unlike in women, was not associated with AF risk. Median FLI values in men per quintile were more than double the corresponding values in women, suggesting that the number of men with low FLI was too limited to influence AF risk.

One of the main strengths of the Suita Study is using a prospective cohort design with a long follow-up period to assign temporality. In addition, we investigated randomly selected participants representing urban Japanese people who constitute 80% of the general population. Further, we used standardized methods for AF diagnosis, including ECG, health records, and death certificates, to minimize AF underreporting.

However, some limitations should be addressed. First, assessing FLI at baseline only may have diluted the association because high-risk participants may have received close monitoring and medical education during the follow-up period. Second, weight change is expected to influence FLI values because BMI and WC are involved in the FLI algorithm. Weight loss in a previous Suita Study report³³ and a meta-analysis of Japanese cohort studies³⁴ was associated with increased CVD mortality. Third, we had no data about liver infections, especially hepatitis B and C; therefore, we cannot confirm whether the association between FLI and AF risk was independent of liver infections. However, a study including 3.74 million first-time Japanese blood donors between 2001 and 2005 put the prevalence of hepatitis B s antigen (HBsAg) and anti-hepatitis C virus antibodies at 0.31% and 0.26%, respectively, suggesting that participants with liver infections were unlikely to have affected the results.³⁵ Fourth, although we adjusted for several potential confounders and the P values for their interactions with the association between FLI and AF were not statistically significant, we cannot entirely exclude their impacts on the associations investigated. Further, because of the observational nature of this study, the presence of unmeasured or undetected confounders is likely.

Conclusions

We detected a U-shaped association between FLI and AF risk in Japanese women. FLI could be a useful screening tool to detect women at high risk of AF. From the preventive perspective, women with high or low FLI should be considered for AF screening.

Acknowledgments

The authors thank the Suita Medical Association, the Suita City Health Center, the Preventive Cardiology and Preventive Healthcare Departments, and all cohort members.

We would like to thank Dr. Kawanishi and Dr. Misaki, the former and current presidents of the Suita City Medical Association.

Sources of Funding

This study was supported by the Intramural Research Fund for the Cardiovascular Diseases of the National Cerebral and Cardiovascular Center (20-4-9), Japan Health Research Promotion Bureau (2019-(1)-1), the Japan Science and Technology Agency (JPMJPF2018), Health and Labour Sciences Research Grants from the Ministry of Health, Labour and Welfare of Japan (20FA1002), and the Meiji Yasuda Health Insurance Company and Research Institute.

Disclosures

The authors have no competing interests to declare.

Conflict of Interest

We have no competing interest to declare.

Author Contributions

A.A. drafted the manuscript and conducted the statistical analysis. Y.K. supervised the work and provided the resources. All authors contributed to data interpretation and revision of the manuscript.

IRB Information

The Suita Study protocol was approved by the Institutional Review Board of the National Cerebral and Cardiovascular Center (M25-043-4). The study was conducted in accordance with the Declaration of Helsinki. All participants provided written informed consent before participation.

Data Availability

Data will be available upon reasonable request.

References

• 1.   Kodani E, Kaneko T, Fujii H, Nakamura H, Sasabe H, Tamura Y, et al. Prevalence and incidence of atrial fibrillation in the general population based on national health insurance special health checkups: TAMA MED Project-AF. Circ J 2019; 83: 524–531.
• 2.   Kornej J, Börschel CS, Benjamin EJ, Schnabel RB. Epidemiology of atrial fibrillation in the 21st century: Novel methods and new insights. Circ Res 2020; 127: 4–20.
• 3.   Staerk L, Sherer JA, Ko D, Benjamin EJ, Helm RH. Atrial fibrillation: Epidemiology, pathophysiology, and clinical outcomes. Circ Res 2017; 120: 1501–1517.
• 4.   Odutayo A, Wong CX, Hsiao AJ, Hopewell S, Altman DG, Emdin CA. Atrial fibrillation and risks of cardiovascular disease, renal disease, and death: Systematic review and meta-analysis. BMJ 2016; 354: i4482.
• 5.   Arafa A, Kokubo Y, Sheerah HA, Sakai Y, Watanabe E, Li J, et al. Developing a stroke risk prediction model using cardiovascular risk factors: The Suita Study. Cerebrovasc Dis 2022; 51: 323–330.
• 6.   Du X, Dong J, Ma C. Is atrial fibrillation a preventable disease? J Am Coll Cardiol 2017; 69: 1968–1982.
• 7.   Nassir F, Rector RS, Hammoud GM, Ibdah JA. Pathogenesis and prevention of hepatic steatosis. Gastroenterol Hepatol (N Y) 2015; 11: 167–175.
• 8.   Bellentani S. The epidemiology of non-alcoholic fatty liver disease. Liver Int 2017; 37: 81–84.
• 9.   Kahl S, Straßburger K, Nowotny B, Livingstone R, Klüppelholz B, Keßel K, et al. Comparison of liver fat indices for the diagnosis of hepatic steatosis and insulin resistance. PLoS One 2014; 9: e94059.
• 10.   Bedogni G, Bellentani S, Miglioli L, Masutti F, Passalacqua M, Castiglione A, et al. The fatty liver index: A simple and accurate predictor of hepatic steatosis in the general population. BMC Gastroenterol 2006; 6: 33.
• 11.   Olubamwo OO, Virtanen JK, Voutilainen A, Kauhanen J, Pihlajamäki J, Tuomainen TP. Association of fatty liver index with the risk of incident cardiovascular disease and acute myocardial infarction. Eur J Gastroenterol Hepatol 2018; 30: 1047–1054.
• 12.   Kim JH, Moon JS, Byun SJ, Lee JH, Kang DR, Sung KC, et al. Fatty liver index and development of cardiovascular disease in Koreans without pre-existing myocardial infarction and ischemic stroke: A large population-based study. Cardiovasc Diabetol 2020; 19: 51.
• 13.   Zou B, Yeo YH, Cheung R, Ingelsson E, Nguyen MH. Fatty liver index and development of cardiovascular disease: Findings from the UK Biobank. Dig Dis Sci 2021; 66: 2092–2100.
• 14.   Lee SR, Han KD, Choi EK, Oh S, Lip GYH. Nonalcoholic fatty liver disease and the risk of atrial fibrillation stratified by body mass index: A nationwide population-based study. Sci Rep 2021; 11: 3737.
• 15.   Lee SR, Han KD, Choi EK, Oh S, Lip GYH. Non-alcoholic fatty liver disease and the risk of incident atrial fibrillation in young adults: A nationwide population-based cohort study. Front Cardiovasc Med 2022; 9: 832023.
• 16.   Eguchi Y, Wong G, Lee EI, Akhtar O, Lopes R, Sumida Y. Epidemiology of non-alcoholic fatty liver disease and non-alcoholic steatohepatitis in Japan: A focused literature review. JGH Open 2020; 4: 808–817.
• 17.   Kokubo Y, Watanabe M, Higashiyama A, Nakao YM, Kusano K, Miyamoto Y. Development of a basic risk score for incident atrial fibrillation in a Japanese general population: The Suita Study. Circ J 2017; 81: 1580–1588.
• 18.   Arafa A, Kokubo Y, Kashima R, Teramoto M, Sakai Y, Nosaka S, et al. Association between white blood cell count and atrial fibrillation risk: A population-based prospective cohort study. Circ J 2022; 87: 41–49.
• 19.   Arafa A, Kokubo Y, Shimamoto K, Kashima R, Watanabe E, Sakai Y, et al. Stair climbing and incident atrial fibrillation: A prospective cohort study. Environ Health Prev Med 2022; 27: 10.
• 20.   Ding YH, Ma Y, Qian LY, Xu Q, Wang LH, Huang DS, et al. Linking atrial fibrillation with non-alcoholic fatty liver disease: Potential common therapeutic targets. Oncotarget 2017; 8: 60673–60683.
• 21.   Haghbin H, Gangwani MK, Ravi SJK, Perisetti A, Aziz M, Goyal H, et al. Nonalcoholic fatty liver disease and atrial fibrillation: Possible pathophysiological links and therapeutic interventions. Ann Gastroenterol 2020; 33: 603–614.
• 22.   Huh JH, Ahn SV, Koh SB, Choi E, Kim JY, Sung KC, et al. A prospective study of fatty liver index and incident hypertension: The KoGES-ARIRANG Study. PLoS One 2015; 10: e0143560.
• 23.   Kitazawa A, Maeda S, Fukuda Y. Fatty liver index as a predictive marker for the development of diabetes: A retrospective cohort study using Japanese health check-up data. PLoS One 2021; 16: e0257352.
• 24.   Lian CY, Zhai ZZ, Li ZF, Wang L. High fat diet-triggered non-alcoholic fatty liver disease: A review of proposed mechanisms. Chem Biol Interact 2020; 330: 109199.
• 25.   Suffee N, Baptista E, Piquereau J, Ponnaiah M, Doisne N, Ichou F, et al. Impacts of a high-fat diet on the metabolic profile and the phenotype of atrial myocardium in mice. Cardiovasc Res 2022; 118: 3126–3139.
• 26.   Heyens LJM, Busschots D, Koek GH, Robaeys G, Francque S. Liver fibrosis in non-alcoholic fatty liver disease: From liver biopsy to non-invasive biomarkers in diagnosis and treatment. Front Med (Lausanne) 2021; 8: 615978.
• 27.   Li CY, Zhang JR, Hu WN, Li SN. Atrial fibrosis underlying atrial fibrillation [Review]. Int J Mol Med 2021; 47: 9.
• 28.   Hsu JC, Yang YY, Chuang SL, Chung YW, Wang CH, Lin LY. Underweight is a major risk factor for atrial fibrillation in Asian people with type 2 diabetes mellitus. Cardiovasc Diabetol 2021; 20: 226.
• 29.   Rivera M, Cortés R, Salvador A, Bertomeu V, de Burgos FG, Payá R, et al. Obese subjects with heart failure have lower N-terminal pro-brain natriuretic peptide plasma levels irrespective of aetiology. Eur J Heart Fail 2005; 7: 1168–1170.
• 30.   Werhahn SM, Becker C, Mende M, Haarmann H, Nolte K, Laufs U, et al. NT-proBNP as a marker for atrial fibrillation and heart failure in four observational outpatient trials. ESC Heart Fail 2022; 9: 100–109.
• 31.   Yao S, Guo J, Shi G, Zhu Y, Wang Y, Chu X, et al. Association of BNP with frailty in elderly population: Rugao Longevity and Ageing Study. J Nutr Health Aging 2019; 23: 73–78.
• 32.   Villani ER, Tummolo AM, Palmer K, Gravina EM, Vetrano DL, Bernabei R, et al. Frailty and atrial fibrillation: A systematic review. Eur J Intern Med 2018; 56: 33–38.
• 33.   Arafa A, Kokubo Y, Sheerah HA, Sakai Y, Watanabe E, Li J, et al. Weight change since age 20 and the risk of cardiovascular disease mortality: A prospective cohort study. J Atheroscler Thromb 2022; 29: 1511–1521.
• 34.   Arafa A, Kashima R, Kokubo Y. Impact of weight change since a young age on cardiovascular mortality risk: A pooled analysis of Japanese epidemiological evidence. Environ Health Prev Med 2022; 27: 9.
• 35.   Tanaka J, Akita T, Ohisa M, Sakamune K, Ko K, Uchida S, et al. Trends in the total numbers of HBV and HCV carriers in Japan from 2000 to 2011. J Viral Hepat 2018; 25: 363–372.