Long-term mortality due to infection associated with elevated liver enzymes: a population-based cohort study

We aimed to investigate whether elevated liver enzymes in the adult population were associated with mortality due to infection. As a population-based cohort study, data from the National Health Insurance Service Health Screening Cohort were used. Adult individuals (aged ≥ 40 years) who underwent standardized medical examination between 2002 and 2003 were included, and infectious mortality was defined as mortality due to infection between 2004 and 2015. Aspartate transaminase (AST), alanine aminotransferase (ALT), γ-glutamyl transpeptidase (γ-GTP), AST/ALT ratio, and dynamic AST/ALT ratio (dAAR) were included in multivariable Cox modeling. A total of 512,746 individuals were included in this study. Infectious mortality occurred in 2444 individuals (0.5%). In the multivariable model, moderate and severe elevation in AST was associated with 1.94-fold [hazard ratio (HR):1.94, 95% confidence interval (CI) 1.71–2.19; P < 0.001] and 3.93-fold (HR: 3.93, 95% CI 3.05–5.07; P < 0.001) higher infectious mortality respectively, compared with the normal AST group. Similar results were observed for moderate and severe elevation in ALT and mild, moderate, and severe elevation in γ-GTP. Additionally, a 1-point increase in the AST/ALT ratio and dAAR was associated with higher infection mortality. Elevated liver enzymes (AST, ALT, AST/ALT ratio, γ-GTP, and dAAR) were associated with increased infectious mortality.

Infectious disease is one of the most important causes of mortality and morbidity in human history 1 . Mortality and morbidity due to infectious diseases are common despite advances in medicine. Every year, the influenza virus infects 10-40% of the world population 2 . During the 2009-2010 H1N1 influenza pandemic in the United States, 59 million people were infected, 265,000 were hospitalized, and there were 12,000 deaths 3 . Currently, the world population is facing the coronavirus disease-2019 pandemic as a global public health crisis 4 .
The liver plays a critical role in host defense mechanisms during infection 5 . It contains many innate immune cells, such as macrophages, natural killer cells, natural killer T cells, and gamma-delta T cells, which contribute to the immunologic elimination of microorganisms by the endothelial cells and Kupffer cells of the liver 6 . The antigen-rich blood from the gastrointestinal tract passes through the liver, and the lymphocytes and natural killer T cells play critical roles as the first immune defense mechanism against invading pathogens 7 . Thus, impaired immunity due to liver dysfunction might lead to chronic infection and cancer 8 . Among common liver diseases, liver cirrhosis is considered an immunocompromised condition that increases the susceptibility to bacterial infection, resulting in a higher rate of mortality 9 . Similarly, abnormal liver function due to any liver disease can cause innate immune dysfunction 10 , resulting in a higher risk of infection and mortality due to infection. Liver enzymes have been widely used to evaluate and screen for abnormal liver function in the adult population [11][12][13] , and elevated liver enzymes might be related to a higher risk of mortality. However, the relationship between the elevation of liver enzymes and the risk of infection or infectious mortality has not been identified.
Therefore, this study aimed to investigate whether elevated liver enzymes were associated with a higher risk of infectious mortality among the adult population. Specifically, liver enzymes, such as aspartate transaminase (AST), alanine aminotransferase (ALT), and γ-glutamyl transpeptidase (γ-GTP) are evaluated during a standardized medical examination in South Korea. Data source: NHIS-HEALS database and the study population. The NHIS-National Health Screening Cohort (NHIS-HEALS) was used for this study 14 . As the sole public insurance system in South Korea, the NHIS collects information regarding demographics, socioeconomic status, diagnosis of diseases using the International Classification of Diseases, tenth revision (ICD)-10 codes, and treatment of the diseases. All disease diagnoses and prescription information of drugs and/or procedures should be registered in the NHIS database to receive financial support from the government, suggesting that the reliability of the NHIS database is sufficient. Subscribers to the NHIS who are ≥ 40 years of age are recommended to undergo a standardized medical examination every 2 years 15  Infectious mortality as primary endpoint. Infectious mortality was defined as death due to infection.
The NHIS database provided the date of death and the main cause of death of all individuals. Infectious mortality was evaluated from January 1, 2004 to December 31, 2015, over 12 years. The specific diagnoses of infectious mortality are presented as ICD-10 codes in Supplementary Table 3.
Measurements as confounders for adjustment. The following information was collected as covariates for this study: demographic information (age, sex, and BMI), socioeconomic status (residence and annual income level), presence of comorbidities (underlying disability, Charlson comorbidity index, mild to moderate and severe liver disease), and lifestyle information (smoking status, alcohol consumption, and exercise frequency). Place of residence was classified into three groups (Seoul, other metropolitan cities, and other areas), and BMI was categorized into four groups (< 18.5, 18.5-24.9, 25.0-29.9, and > 30 kg/m 2 ). The annual income level was categorized into five groups (0-20%, 20-40%, 40-60%, 60-80%, and 80-100%), and the underlying disability was divided into two groups (mild to moderate and severe disability groups). In South Korea, all physical disabilities must be registered in the NHIS to receive various benefits. The disabilities were categorized into six levels based on severity. The 1st (most severe) to 3rd level of disability was considered as the severe disability group, while the 4th to 6th (mildest) level was considered as the moderate disability group. Smoking status was divided into four groups: non-smoker, previous smoker, current smoker, and unknown [no-response group]). Alcohol consumption was divided into four groups (non-drinker, mild drinker, heavy drinker, and unknown [no-response group]). The mild drinker group was defined as alcohol consumption ≤ 70 g per week in males and ≤ 20 g per week in females, while the heavy drinker group was defined as alcohol consumption > 70 g per week in males and > 20 g per week in females. Exercise frequency was divided into six groups (no exercise, 1-2 times per week, 3-4 times per week, 5-6 times per week, exercise almost every day, and unknown [no-response group]). The Charlson comorbidity index was calculated using registered ICD-10 codes between 2002 and 2003, as shown in Supplementary Table 4 18 .
Statistical analysis. The clinico-epidemiological characteristics of the participants in this study are presented as means with standard deviations for continuous variables and numbers with percentages for categorical variables. We constructed multivariable Cox regression models to investigate whether the AST, ALT, AST/ALT ratio, γ-GTP, and dAAR were associated with infectious mortality from 2004 to 2015. In this time to event analysis, infectious death was set as an event, and the survival time from January 1, 2004 to the date of death was set as duration. The four multivariable Cox models were fitted to avoid multicollinearity between the liver enzymes (AST, ALT, AST/ALT ratio, γ-GTP, and dAAR), and all covariates were included in the multivariable models www.nature.com/scientificreports/ for adjustment. In addition, as a sensitivity analysis, we investigated the association of elevated liver enzymes with infectious mortality from 2005-2015, not 2004-2015, to avoid reverse causation bias. This is because there was a short period of latency between elevated liver enzymes and infectious mortality in 2004 19 . Further, we performed subgroup analyses according to comorbidities of liver disease and alcohol consumption because both liver disease and alcohol consumption affect the elevation of liver enzymes 13 . The multivariable Cox regression models were also used for the subgroup analyses using the same methods as those for the main analyses. It was confirmed that there was no multicollinearity in all multivariable models involving the entire cohort, with a variance inflation factor of < 2.0. The Cox regression results are presented as hazard ratios (HRs) with 95% confidence intervals (CIs). All statistical analyses were performed using R software (version 4.0.3 with R packages, the R Project for Statistical Computing, Vienna, Austria), and P < 0.05 was considered statistically significant.

Population.
A total of 514,795 individuals underwent the standardized medical examination from 2002 to 2003. Among them, 1320 individuals who died during this period and 729 individuals whose information about liver enzymes were missing were excluded. A total of 512,746 individuals were included in this study. Among them, infectious mortality occurred in 2444 individuals (0.5%), as shown in Fig. 1. The clinico-epidemiological characteristics of the study's participants are presented in Table 1.  Tables 3 and 4. The HRs of elevation in AST, ALT, γ-GTP, and dAAR for infectious mortality tended to be higher in the liver disease group than in the no liver disease group. The HRs of elevation in AST, ALT, γ-GTP, and dAAR for infectious mortality tended to be higher in the alcohol user group than in the non-alcohol user group.

Discussion
This population-based cohort study showed that elevated liver enzymes (AST, ALT, AST/ALT ratio, γ-GTP, and dAAR) were associated with a higher risk of infectious mortality among the South Korean adult population. This association was most evident in the elevation of AST. Specifically, our results suggested that the elevation of AST, ALT, and γ-GTP above 40 U was associated with a higher risk of infectious mortality. In subgroup analyses, this association was more evident in the existing liver disease and alcohol user groups. This is the first study to report an increased risk of infectious mortality due to abnormal liver function among the adult population. We found that approximately 25% of the adult population had elevated liver enzyme levels between 2002 and 2003. It was estimated that about 9% of asymptomatic patients had elevated enzyme levels on standard liver function tests in the United States 12 . Our cohort included 16.8% of the patients who were diagnosed with mild, moderate, and severe liver disease; thus, 25% of the cohort showing elevated liver enzyme in our study can be considered acceptable.
A previous study reported that advanced liver disease generally increases the risk of infection, and this phenomenon is known as cirrhosis-associated immune dysfunction 20 . Cirrhosis-associated immune dysfunction was defined as syndromic abnormalities of the immune function, characterized by immunodeficiency and systemic inflammation 20 . As an immune defense mechanism, the liver prevents the systemic spread of microbial and dietary antigens entering through the gastrointestinal tract 21 . This function is offset by immune tolerance to non-pathogenic exogenous material 22 . Another mechanism is the synthesis of soluble molecules that are essential for an effective immune response 5 . However, cirrhosis disrupts the cellular organization of the liver and reduces the ability of hepatocytes to synthesize proteins 20 . These cirrhosis-induced liver injuries compromise the immune function of the liver by damaging the reticuloendothelial system and synthesis of innate immune proteins. These damaged reticuloendothelial systems due to liver cirrhosis have been identified as risk factors for bacterial infection and lower survival rates 23 . In addition to the previous reports regarding cirrhosis-associated immune dysfunction and risk of infection, our study reported that infectious mortality might be increased in the adult population with abnormal liver function. In other words, not only advanced liver diseases, such as liver cirrhosis but also adult populations with mild liver disease might have a higher risk for infectious mortality.
The results regarding elevated γ-GTP and increased risk of infectious mortality are also important. γ-GTP is found in hepatocytes and biliary epithelial cells and is a very sensitive marker of hepatobiliary disease 24 , although it has lower specificity than AST and ALT for diagnosing liver disease 24 . The γ-GTP level is known to increase more consistently, with high sensitivity than AST and ALT among patients with all types of anicteric non-alcoholic chronic liver disease 25 . Moreover, the serum γ-GTP level for assessment of liver fibrosis or cirrhosis combined with platelet count is an important biomarker in patients with chronic hepatitis B infection 26 . Although we did not evaluate the γ-GTP to platelet count ratio due to the lack of platelet count data in our database, our results suggest that γ-GTP elevation might significantly reflect abnormal liver function and can be a predictive factor for higher infectious mortality among individuals with abnormal liver function.
In the subgroup analyses, our study showed that the impact of elevation in liver enzymes on increased infectious mortality was more evident in individuals with pre-existing liver disease and alcohol users. In general, the severity of liver disease correlates with increased liver enzyme levels 11 . In the population with chronic liver disease or alcohol users, the liver enzymes might directly reflect liver function. Elevation of liver enzymes is also seen in non-hepatic diseases, such as congestive heart failure, celiac disease, thyroid disorders, or striated muscle disease 13 . However, there are few reports on the relationship between the risk of infection and elevated liver enzyme levels in those without liver disease among the adult population; therefore, further research is necessary.
Interestingly, we used dAAR to evaluate the impact of subclinical advanced liver fibrosis using age, AST, and ALT according to a recent study 17 . In the recent study 17 , the dAAR score provides a method to stratify easily the adult population regarding risk for incident severe liver disease. Furthermore, the dAAR score can be used as a screening tool for the unselected general population and as a trigger for further liver evaluations in the future. However, currently, it is a new concept in screening for liver function; therefore, further research is required to evaluate the clinical usefulness of dAAR.
Our study had some limitations. First, some important variables, such as platelet count and bilirubin, were not included in the study because the NHIS database did not contain these data. Therefore, the impact of the AST to platelet ratio index or albumin-bilirubin score on infectious mortality could not be analyzed. Second, multivariable adjustment only adjusted for known confounders, and there might be unmeasured confounders  Table 2. Multivariable Cox regression models for infectious mortality. CI, confidence interval; AST, aspartate transaminase; ALT, alanine aminotransferase; γ-GTP, γ-glutamyl transpeptidase. www.nature.com/scientificreports/ that might affect the results. Third, we used ICD-10 codes to define liver disease and calculate the Charlson comorbidity index; however, the actual underlying diseases might differ with registered ICD-10 codes in this study. For instance, if individuals with liver disease did not visit outpatient clinics due to mild symptoms or poor access to healthcare utilization, they were not registered as having liver disease in the NHIS database in this study. Lastly, the lifestyle information was collected using a questionnaire, and there might be a selection bias due to non-response to the surveys in standardized medical examination 27 .
In conclusion, this population-based cohort study showed that elevated liver enzymes (AST, ALT, AST/ALT ratio, γ-GTP, and dAAR) were associated with increased infectious mortality among South Korean adults. Specifically, our results suggest that the elevation of AST, ALT, and γ-GTP above 40 U were significantly associated with infectious mortality. Therefore, screening and optimal management of adult populations with elevated enzymes are clinically important for preventing infectious mortality. www.nature.com/scientificreports/

Data availability
The datasets used and/or analyzed during the current study are available from the corresponding author upon reasonable request.