Association Between Nonalcoholic Fatty Liver Disease and the Incidence of Asthma in Adults: Finding from a Nationwide Study in Korea

Asthma and nonalcoholic fatty liver disease (NAFLD) are chronic diseases and are known to be associated with metabolic abnormalities. We aimed to clarify the association between NAFLD and the incidence of asthma in a large population-based cohort. We selected 160,603 healthy individuals without comorbidities from the National Health Insurance Service-National Sample cohort during 2009–2014. . NAFLD was dened by a surrogate marker, the fatty liver index (FLI). During a median of 5.08 years’ follow-up, 16,377 subjects (10.2%) were newly diagnosed with asthma and categorized into three groups according to FLI. The cumulative incidence of asthma was higher in subjects with higher vs. lower FLIs (FLI < 30 10.1%, 30 ≤ FLI < 60 10.8%, FLI ≥ 60 10.5% ). Higher FLI associated with an increased incidence of asthma (Harzard ratios (HR)highest vs. lowest FLI, 1.25; 95% CI, 1.15-1.36). The results using another denition of NAFLD as measured by hepatic steatosis index (HIS) are similar to the primary results. This association was more pronounced in women than in men (HR 1.46; 95% CI, 1.13-1.64 vs. HR 1.07; 95% CI, 0.94-1.20). This study demonstrated that NAFLD as measured by FLI and HIS may inuence the incidence rates of asthma in adults, especially in women. glucose and low density lipoprotein cholesterol incorporated into the existing Model to obtain Model During subgroup analyses, FLI integrated into the statistical models as a continuous variable after log transformation. A two-sided P value of < 0.05 was considered to indicate statistical signicance. Statistical analyses were performed using R software, version 3.3.3 (R Foundation Statistical


Introduction
Nonalcoholic fatty liver disease (NAFLD) is the most common chronic liver disease worldwide, and its prevalence in Asian-Paci c countries is estimated at 20-30% 1,2 .
NAFLD is associated with an increase of multisystem diseases including diabetes, ischemic cardiovascular disease, and chronic kidney disease 3,4 . Recently, an association of NAFLD with atrial brillation, a non-ischemic cardiac disease, has also been reported 5 .
Asthma is the most common chronic airway disease, with approximately 350 million cases reported worldwide in 2015 6 . Recent studies have reported the epidemiological link between metabolic syndrome (MetS) and asthma, especially obesity is a well-known risk factor for asthma and may mediate its severity 7,8 . Several studies have reported the association of asthma with other components of MetS, such as dyslipidemia and insulin resistance, as well as obesity [9][10][11] . NAFLD and MetS also share many features, and there is growing evidence of a bidirectional relationship between the two diseases 12,13 . However, to date, no studies have investigated the direct association between NAFLD and the development of asthma.
We hypothesized that NAFLD is associated with the incidence of asthma. The objective of this study was to investigate the direct association between NAFLD and asthma incidence using a large, claim-based cohort consisting of healthy Korean adult population without traditional risk factors and comorbidities.

Database and study cohort
We conducted a retrospective population-based cohort study using data from the National Health Insurance Service-National Sample Cohort 2.0 (NHIS-NSC 2.0) 2002-2015, as previously reported 14,15 . This database contains data collected from cohort members on sociodemographic characteristics, all medical claims including diagnosis, medical treatment, and health care utilization; and is linked with their data from the National Health Screening database. The database also includes mortality data such as the date and cause of death based on the death registration database of Statistics Korea. This study was approved by the Institutional Review Board of Chungnam National University Hospital, Sejong, Korea (protocol no. 2020-09-023). Witten consents were waived by ethics committee due to the retrospective nature of the study.

Study subjects
All subjects aged 20 years and older who underwent national health screening at least once between January 2009 and December 2014 were enrolled in this study. The rst health check-up was considered as the index check-up, and the year of the index check-up was the index year. Subjects who met the following exclusion criteria were excluded from the analysis: i) diagnosed with asthma within two years before the index year; ii) diagnosed with comorbid conditions including other pulmonary disease, liver disease, heart failure, hypertension, diabetes, cerebrovascular disease, ischemic heart disease, valvular heart disease, or peripheral artery disease within two years before the index year; iii) had a prescription of oral hypoglycemic, antihypertensive, or lipid-lowering agents within two years before the index year; iv) had any smoking history (current or former smoker); and v) had missing data in the index check-up and follow-up period.
National health screenings included physical examination, medical history, blood test, and chest X-ray, and collection of data on healthrelated behavior such as exercise habits, smoking, and alcohol consumption using self-administered questionnaires. Each diagnosis was de ned using the 10 th revision of the International Classi cation of Diseases (ICD-10). Each diagnosis code used in the analysis is shown in a supplemental table (Table S1).
De nition of adult-onset asthma cases and asthma-related healthcare use The de nition of asthma was as follows: i) Individuals with at least two or more claims with asthma according to the ICD-10 code (J45.x-J46.x) for the primary or sub-diagnosis; ii) prescribed at least one of the asthma-related medications: inhaled corticosteroids (ICS), ICS and LABA combined into a single inhaler (ICSs/LABAs), systemic LABAs, oral leukotriene antagonists (LTRA), short-acting β 2-agonists (SABAs), xanthine derivatives, and systemic corticosteroids 16 . In terms of asthma-related healthcare use, outpatient visits were de ned as patients diagnosed with asthma related ICD-10 codes (J45-46) and prescribed at least one of the asthma-related medications (see the above). Asthma-related hospitalization or emergency department (ED) visits were de ned as admission and ED visit under asthma related ICD-10 codes and prescribed oral or injected systemic corticosteroids to exclude healthcare use for other principal purposes 15,17 .

Data collection and measure
Body mass index (BMI) was calculated as weight (kg) divided by the square of the height (m). Obesity was de ned as a BMI of ≥25 kg/m 2 according to the World Health Organization criteria for Asian populations 18 . Alcohol consumption and daily activity were measured using a self-reported questionnaire.
Fatty liver index (FLI) is a noninvasive surrogate marker of NAFLD and is reported to be reliable in previous epidemiological studies [19][20][21][22] . FLI consists of BMI, waist circumference, triglyceride, and gamma-glutamyltransferase (GGT) and is calculated using the following formula:

FLI =
The FLI score ranges from 0 to 100, and the original study suggested an FLI score of ≥60 as the cut-off value to diagnose hepatic steatosis. In this study, subjects were then categorized into three groups: 0 ≤ FLI < 30, 30 ≤ FLI <60 and FLI ≥ 60, based on previous studies 21 .
However, few studies have been conducted to validate this application of FLI in Asian populations, taking into consideration ethnic characteristics such as lower BMI and waist circumference compared to those of Western populations. We therefore conducted supplemental analysis using different FLI cut-off criteria: 0 ≤ FLI < 25, 25 ≤ FLI < 35 and FLI ≥ 35 for males and 0 ≤ FLI < 10, 10 ≤ FLI < 20 and FLI ≥ 20 for females, proposed by Yang et al in a study conducted among Taiwanese people 22 . Other surrogate marker of fatty liver, hepatic steatosis index (HIS) was also calculated as supplementary criteria. HIS consists of liver enzymes, such as alanine transaminase (ALT), aspartate aminotransferase (AST), and BMI, calculated using the following formula: HIS = 8x (ALT/ AST ratio) + BMI (+2, if female; +2, if diabetes mellitus). HIS values of ≥ 36 rule in hepatic steatosis 23 .

Statistical analysis
Data are presented as mean ± standard deviation for continuous variables and number (percentage) for categorical variables. One-way analysis of variance and chi-square tests were used to compare the differences between the FLI quartile groups. The incidence rate of asthma was expressed as the total number of newly diagnosed asthma cases per 100,000 person-years. Cumulative incidence rates were calculated and compared between quartile-based groups using Kaplan-Meier estimates and log-rank tests. Cox proportional hazards regression was used to estimate adjusted hazard ratios (HRs) and 95% con dence intervals (CI) for asthma incidence. Covariates included in Model 1 were age and sex. Later, clinical characteristics of variables whose association with adult-onset asthma showed borderline statistical signi cance (P < .10), physical activity, alcohol drinking, systolic blood pressure, diastolic blood pressure, fast blood glucose and low density lipoprotein cholesterol were incorporated into the existing Model 1 to obtain Model 2. During subgroup analyses, FLI was integrated into the statistical models as a continuous variable after log transformation. A two-sided P value of < 0.05 was considered to indicate statistical signi cance. Statistical analyses were performed using R software, version 3.

Baseline characteristics
Of the total 556,884 NHIS-NSC 2.0 sample cohort population, the data of 160,603 individuals were analyzed after excluding subjects who met the exclusion criteria ( Figure 1). The study subjects were divided into three groups according to their FLI value (0 ≤ FLI < 30, 30 ≤ FLI <60 and FLI ≥ 60). Detailed baseline characteristics by FLI groups are shown in Table 1. Groups with higher FLI had older individuals and a greater proportion of obese and male subjects than groups with lower FLI. Mean blood pressure and the proportion of alcohol consumers also tended to increase with FLI. In terms of laboratory data, the lipid pro le progressively worsened with increasing FLI value, and GGT and fasting glucose levels were also higher in patients with higher FLI.
Relationship between FLI and the incidence of asthma During 5.08 years' median follow-up (interquartile range, 3.08-6.16), a total of 16,377 subjects (10.2%) developed asthma. The incidence of asthma was high in groups with higher FLI (FLI < 30 10 [647/6,136]); the cumulative incidence of asthma is shown in Figure 2. Table 2 shows hazard ratios and their 95% CI from Cox proportional hazards regression analysis for subjects with adult-onset asthma according to the FLI group. After adjusting for age and sex (Model 1), the HR (95% CI) was 1.23 (1.13-1.33) for the highest group of FLI compared with that of the lowest. When the model was further adjusted for physical activity, alcohol drinking, systolic blood pressure, diastolic blood pressure, fast blood glucose, and low-density lipoprotein cholesterol (Model 2), the HR (95% CI) was 1.25 (1.15-1.36) for the highest group of FLI compared with that of the lowest. In addition to the FLI groups, the analysis conducted after categorizing the study subjects according to the different FLI cut-off criteria suggested in previous studies (HR 1.18, 95% CI 1.14-1.22) and HIS (HR 1.12, 95% CI 1.07-1.17), showed that asthma incidence tended to increase as each value increased (Table 2).
Furthermore, we performed subgroup analysis the association between asthma incidence and FLI according to sex. As a result, the FLI was associated with asthma incidence for both sexes, but was more pronounced in women (HR 1.46, 95% CI, 1.13-1.64 vs. HR 1.07, 95% CI, 0.94-1.20, Table 3).

Subgroup analysis of clinical variable affecting asthma incidence
The incidence of asthma in each clinical subgroup is shown in Figure S1. In all subgroups except alcohol drinkers, asthma occurrence tended to increase as FLI increased. Table S3 presents the hazard ratios for asthma incidence according to each clinical variable. As the age and BMI of the study population increased, the HR tended to increase slightly, and the incidence of asthma was higher in women than in men (male, HR 0.69, 95% CI, 0.67-0.72). Total cholesterol and low-density lipoprotein cholesterol were also associated with increased HR. Daily physical activity had a protective effect against asthma incidence. It was observed that the difference in asthma incidence according to sex was very distinct, we performed subgroup analysis the association between asthma incidence and FLI according to sex. As a result, the FLI was associated with asthma incidence for both sexes, but the difference of HR was observed in men (HR 1.07, 95% CI, 0.94-1.20) and women (HR 1.46, 95% CI, 1.13-1.64, Table S4).

Discussion
In this study, we investigated the association between NAFLD as measured using FLI, a validated surrogate marker, and the incidence of asthma in a large, representative sample of the Korean population. NAFLD was associated with an increase in the incidence of asthma and in most clinical subgroups except alcohol drinkers, an increase in the incidence of asthma was observed with an increase in the FLI score. Furthermore, as the FLI score increased, the utilization of asthma-related health care services such as hospitalization days and outpatient visits also increased.
In current study, the incidence of asthma was slightly higher in the middle group (30 ≤ FLI < 60) compared to the highest FLI group.
However, as a result of con rming the hazard ratios, it was found that asthma incidence increased as FLI increased, and the same result was obtained in the analysis using other types of FLI categories and HIS, another surrogate marker of NAFLD. Therefore, our ndings support the possibility of NAFLD in uencing the occurrence of asthma.
NAFLD is highly prevalent worldwide, and it is known that NAFLD occurs along the lines of MetS progression 24 . Similar to MetS, the association of NAFLD with extrahepatic complications such as cardiovascular disease, chronic kidney disease and insulin resistance are well known 3,4,24 . Because potential associations with other diseases including atrial brillation and periodontitis have recently been suggested, there is increased interest in the effect of NAFLD on multiple organs 5,25 .
Numerous studies have reported a correlation between MetS and impairment of lung function. Of the MetS components, robust epidemiological data linking abdominal obesity with the incidence and exacerbation of asthma are available 26 . Other factors such as dyslipidemia, hyperglycemia, and hypertension have also shown to be independently associated with asthma. Waist circumference and insulin resistance were revealed independent risk factors for asthma in a large Norwegian adult asthma cohort, and high triglyceride (TG) level and insulin resistance were independently associated with asthma incidence in a cross-sectional study of children aged 4-12 years in West Virginia 9,27 .
Given the above results, a relationship between asthma and NAFLD may be suspected. Therefore, we evaluated the association between these two diseases using a large population after excluding known risk factors such as diabetes, hyperlipidemia, hypertension, and smoking exposure and con rmed the independent association of NAFLD with asthma incidence. To the best of our knowledge, this is the rst study to con rm the association between asthma incidence and NAFLD. However, the magnitude of association was not very high, and the following explanation of our results may be considered.
Because the study used health insurance data, asthma patients of all severity levels were included because it was not possible to determine the control status or severity level. Given that less than 10% of all asthma patients were diagnosed with severe asthma, it is believed that a number of mild asthmatic patients were included, which could have diluted the association.
The association between NAFLD and asthma incidence identi ed in this study may be explained by several pathophysiological mechanisms underlying the two diseases. First, in patients with NAFLD, elevated levels of circulatory TG and free fatty acids have been observed. Excess fat increases the production of proin ammatory cytokines such as tumor necrosis factor-alpha and interleukin-6 through the activation of innate immune responses such as pattern recognition receptor 28, 29 . Excess fatty acid is also involved in the production of reactive oxygen species by inducing endoplasmic reticulum stress 30 . Increased proin ammatory cytokines and oxidative stress may cause airway epithelial damage and airway in ammation, which may affect asthma development 31,32 . Second, insulin resistance, which plays a key role in the development of fatty liver disease, leads to excessive insulin secretion in the blood. Insulin is thought to contribute to airway hypersensitivity and airway remodeling by causing epithelial damage and smooth muscle proliferation in the airway 33 . In addition, hyperinsulinemia was reported to have been involved in enhancing airway responsiveness through parasympathetic stimulation in an asthma mouse model 34 .
Similar to the above-mentioned observations, our study also con rmed that an increase in age, weight, or poor lipid pro le affects the incidence of asthma. Daily physical activity tended to have a positive effect on the suppression of asthma.
Furthermore, we analyzed the correlation between FLI and asthma-related healthcare service utilization, and the number of patients who experienced hospitalization was higher in the group with low FLI, but the number of hospitalizations per patient, number of days of hospitalization, and outpatient frequency increased with higher FLI. The discrepancy between the incidence of asthma and asthma-related hospitalization may have been in uenced by the fact that the severity of asthma was not re ected in this study. In fact, the increase in the number of hospitalizations per patient, the number of days of hospitalization, and intensive care unit admission in the high FLI group in this study may also suggest a possible association between FLI and asthma severity. In a study of more than 70,000 adolescents in Brazil published by Kuchinir et al., a signi cant association was reported between MetS and asthma control 35 . In a similar vein, progression of NAFLD may also affect the severity of asthma. Further research is needed that re ects the severity of asthma.
As another important point, in this study, women showed a stronger correlation between FLI and asthma incidence than men. Several previous studies reported a higher correlation between obesity and asthma in women 36,37 , but others provided con icting results 38 . Some studies have reported the correlation between metabolic abnormality and the occurrence of asthma 9,39 , but few studies have investigated gender differences. There are several possible mechanisms for differences in the association between asthma incidence and FLI by gender. It is thought that the female sex hormone estrogen may be involved in Th2 in ammation and airway hypersensitivity 40 . It has also been suggested that the level of leptin, an adipokine with pro-in ammatory effect, is high in women independently of body fat mass 41 .
Overlapping genetic foci associated with asthma and obesity have been reported, and genetic differences are thought to have an effect 42,43 , but it is still unclear.
Currently, NAFLD is a disease of increasing global concern, several guidelines recommend regular screening in patients with chronic conditions such as obesity, diabetes mellitus, and MetS 44 . In addition to weight reduction, oral hypoglycemic agents are currently being used in an attempt to treat NAFLD 44 . Based on the results of the current study, improvement of NAFLD through such interventions is expected to have a positive effect on asthma. Further studies are needed in the future.
This study has some limitations. First, the incidence of asthma may be overestimated because asthma patients may asymptomatic or have symptoms but do not need medical care for years. To minimize this error, all patients who were diagnosed with asthma or were prescribed asthma-related medication within two year before the index year were excluded based on previous studies 45 . However, the duration of two year may not have been su cient. However, since the focus of current study was not on the asthma incidence rate itself, but on the identi cation of the correlation between asthma incidence and fatty liver, it seems that the effect on the outcome was not signi cant. Second, in this study, determination of NAFLD status did not involve imaging tests such as abdominal ultrasonography or magnetic resonance imaging, which are currently rst-line diagnostic methods. However, the FLI used in the current study was validated as a predictor of NAFLD in several existing studies in both Asian and Western populations 19,21,22,46 . In addition, health checkups in various countries currently do not include imaging tests. Therefore, we think that FLI, which can be easily obtained from data obtained from health checkups, may rather comprehensively re ect the presence of hepatic steatosis in the general population. Third, we were unable to include family history of allergic disease, allergen exposure, and atopic status known as asthma risk factors in this study due to the nature of health insurance data. Finally, the generalizability of our study ndings is limited to the Korean population. Further studies will be needed for other ethnic groups.
In conclusion, the results of this study suggest that NAFLD, which was measured using the FLI, was associated with an increase in asthma incidence in healthy Korean population, especially in women. It was also con rmed that progression of NAFLD may affect the severity of asthma. These ndings suggest that clinicians should be aware of the higher risk of asthma among patients with NAFLD. Further research on the impact of NAFLD management on the incidence and severity of asthma is needed.  Table 3. sex speci c associations between fatty liver index and incidence of asthma  Overview of the study population Figure 2 Cumulative incidence of adult-onset asthma by FLI groups

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download. 20210601AsthmaandFLIsupplementaryFinal.docx