Association between atherosclerosis-related indicators and fatty liver: a population-based observational study Authors: Xiaoping Liu, Ping Yuan, Ying Huang Published: 2025 ABSTRACT AbstractBackgroundFatty liver has been proven to be closely related to cardiovascular diseases (CVDs), but there are not sufficient studies on its correlation with atherosclerosis-related index.AimThe objective of this study was to determine the potential association of atherosclerosis-related indicators with fatty liver among a Japanese population by a cross-sectional analysis.MethodsMedical data of 909 Japanese individuals age range 24–84 years from the Dryad database were enrolled into our analysis. These participants have received a physical examination with a standardized questionnaire, blood test and imaging examination. We obtained atherosclerosis-related indicators including lipid profile, triglyceride glucose index (TyG), atherogenic index (AI) and brachial-–ankle pulse wave velocity (baPWV)-max. Fatty liver was confirmed by standardized abdominal ultrasound.ResultsAmong these 909 participants, the prevalence of fatty liver was 28.93%. Our curve analysis supported the significant associations of AI, TyG and baPWV-max with fatty liver. Our multivariate logistic regression consistently demonstrated that AI [1.76 (1.49, 2.07), < 0.0001], TyG [2.71 (1.96, 3.75), < 0.0001] and baPWV-max [2.71 (1.96, 3.75), < 0.0001] were significantly associated with fatty liver when age, sex, BMI, systolic BP, diastolic BP, current smoker and exercise were adjusted for. Furthermore, ROC curve indicated that TyG had a higher AUC value of 0.7660 than AI (AUC = 0.7547) and baPWV-max (AUC = 0.6223) on detecting the presence of fatty liver. TyG index exceeding 8.2611 indicated the occurrence of fatty liver with a high sensitivity (80.99%) and specificity (60.84%).ConclusionOur results suggested that TyG value was significantly and positively correlated with fatty liver in the Japanese population, as well as AI and baPWV-max. When TyG value was more than 8.2611, this indicator showed a better predictive value for fatty liver than the other three traditional biomarkers.Keywords: Fatty liver, Metabolic dysfunction-associated steatotic liver disease, Atherosclerosis, TyG, Cardiovascular risk Introduction Fatty liver has become the most common chronic liver disease worldwide [1, 2]. Based on existing statistics, 38% of the global population has been affected by metabolic dysfunction-associated steatotic liver disease in the past 20 years [3–5]. It has been believed to be closely related to metabolic abnormalities including obesity, insulin resistance, diabetes and lipid metabolism, which have been considered as independent factors for adverse cardiovascular events [6, 7]. In fact, existing studies have confirmed that in most patients, fatty liver is often accompanied by cardiovascular diseases (CVDs)-related risk factors [4–7]. Fatty liver can result in an inflammatory state in the human body, leading to vascular endothelial dysfunction and oxidative stress response and ultimately promoting the occurrence and development of cardiovascular events [8, 9]. Previous studies have also shown that the typical characteristics of dyslipidemia in patients with fatty liver are a high level of triglyceride (TG) and low-density lipoprotein cholesterol (LDL-C) and a low level of high-density lipoprotein -cholesterol (HDL-C) [10, 11]. This changed lipoprotein profile is a common manifestation of atherosclerotic dyslipidemia [11–13]. In patients with fatty liver, subclinical atherosclerosis indicators might even be found to be significantly elevated, including a higher carotid intima–media thickness, greater coronary artery calcification and higher coronary high-risk obstructive plaque and non-calcified plaque [14–16]. In addition, patients with fatty liver could also suffer from neuroendocrine dysfunction, intestinal microbiota imbalance and genetic abnormalities that can lead to liver necrotizing inflammation and fibrosis, which were also related to the occurrence and development of adverse cardiovascular events [17–19]. Here, we have obtained the raw data from the Dryad database (10.1136/bmjopen-2014-005413) where a cross-sectional study including 912 individuals aged 24–84 years with a health check in Japan was performed. In this thesis, we conducted a secondary study on the raw data to further determine a potential association between atherosclerosis-related indicators and fatty liver among a Japanese population. Materials and methods Study design We conducted a secondary analysis of a previous study containing a total of 912 individuals who had undergone a medical health check including atherosclerosis-related indicators such as lipid profile, atherogenic index (AI), triglyceride glucose index (TyG) and brachial–ankle pulse wave velocity (baPWV)-max, at the Medical Health Checkup Center at Murakami Memorial Hospital, Gifu, Japan [20, 21]. Briefly, the objective of this original study was to promote public health by detecting early chronic diseases and evaluating their underlying risk factors in an adult Japanese population. All included participants underwent health check-up programs at the Murakami Memorial Hospital from March 2004 to December 2012. According to the Declaration of Helsinki, informed consent from each participant was obtained. The raw medical data from these Japanese individuals were stored in the Dryad database, and we obtained the data for free by following the database guidelines without the need for additional ethical proof. For the purpose of this study, those participants who ever received medication, tested positive for hepatitis B antigen and/or hepatitis C antibody, had hormone replacement therapy and oral contraceptives and pregnant women were excluded [22]. Atherosclerosis-related index Measurement of baPWV-max by an automatic waveform analyzer (Colin Medical Technology, Komaki, Japan) has been reported by a method previously [23]. baPWV was measured in a supine position in a room at a temperature around 25 °C. baPWV was calculated as the pulse wave propagation distance divided by the pulse wave propagation time and expressed in cm/s [23]. The intraobserver and interobserver coefficients of variation were reported to be 10% (r = 0.87, p < 0.01) and 8.4% (r = 0.98, p < 0.01), respectively [23]. The TyG index has been considered as an indicator based on fasting TG and blood glucose, and is mainly used to evaluate insulin resistance and cardiovascular metabolic risk [24]. The formula for TyG index is as follows: ln [TG (mg/dL) × blood glucose (mg/dL)/2] [24]. AI is a clinical index developed by the international medical community to quantify the degree of atherosclerosis. Its calculation formula is as follows: [Total cholesterol (mmol/L)—HDL-C (mmol/L)]/HDL-C (mmol/L) [25]. Fatty liver Abdominal ultrasonography was performed to confirm the diagnosis of fatty liver by trained technicians using Aloka SSD-650CL (Aloka Co, Ltd, Tokyo, Japan). Then, one gastroenterologist made a diagnosis of fatty liver without any reference according to these stored ultrasonographic images [26]. Of the four known criteria (hepatorenal echo contrast, liver brightness, deep attenuation and vascular blurring), these participants were required to have hepatorenal contrast and liver brightness to be given a diagnosis of fatty liver [27]. Blood measurements As briefly described in a previous study [20], blood samples were obtained from the fasting venous blood of each participant. Then, plasma and/or serum samples were immediately centrifuged and stored at − 80 °C in a fridge until biochemical analysis. Blood examinations were performed for biochemical analysis in Japan by MODULAR ANALYTICS (Hitachi High-Technologies Corp. Ltd, Tokyo, Japan). The analytical coefficients of variation for blood glucose, alanine aminotransferase (ALT), aspartate transaminase (AST), TC, HDL-C, LDL-C, TG and uric acid were less than 5%. The estimated glomerular filtration rate (eGFR) was calculated by using the Japanese Society of Nephrology equation [28]. Lifestyle factors All participants proactively completed a standardized questionnaire assisted by a trained team of interviewers. Habits regarding smoking status were classified into two subgroups (current smoker or not). Alcohol consumption was not included in our study due to missing data for most individuals. The duration, sport type and frequency of exercise of participants were recorded in the questionnaire [29]. Participating in any sport regularly for at least once a week was defined as regular exercise [30]. Body mass index (BMI) was calculated by dividing body weight (in kilograms) by the square of height (in meters). Statistical analysis The Empower Stats software 5.2 is a statistical tool based on R language, which was used for all statistical analyses in our study. A P < 0.05 was considered statistically significant. The Kolmogorov–Smirnov test (K–S) was used to analyze the normality distribution of continuous data. Continuous variables were presented as mean ± standard deviation (SD), and categorical variables were presented as the frequency (%). The real-world curves of association between fatty liver and atherosclerosis-related indicators including TyG, AI and baPWV-max were established by univariate and multivariate analyses. Then, the relationship of fatty liver with these atherogenic indicators was examined by multivariate logistic regression. We applied multivariable logistic regression model to explore the potential association between these atherosclerosis-related indicators and fatty liver after adjusting for some confounding factors as follows: Adjusted I: age and sex; Adjusted II: age, sex, BMI, systolic BP, diastolic BP, current smoker and exercise. All logistic regression ORs were calculated by per unit increase for TyG, AI or baPWV-max. The presence of fatty liver was defined as the primary outcome. Also, the saturation effect curve was also plotted to explore a turning point in the relationship. Finally, receiver operating characteristic curve (ROC) was plotted to evaluate the diagnostic value of the atherosclerosis-related indicators for detecting the existence of fatty liver. Results Individual characteristics Excluding 3 individuals with missing data, the remaining 909 participants were enrolled into our analysis, as shown in Fig. 1. Their mean age was 51.13 ± 9.57 years, with a mean BMI value of 23.12 ± 3.13 kg/m2 and 64.91% male participants was (Table 1). The systolic and diastolic BP were 120.22 ± 14.92 and 76.12 ± 9.96 mmHg, respectively. The prevalence of fatty liver, current smoker and exercise among these patients were 263 (28.93%), 197 (21.67%) and 177 (19.47%), respectively. Importantly, atherosclerosis-related indicators including AI, TyG and baPWV-max were 3.17 ± 1.23, 8.29 ± 0.67 and 1415.45 ± 246.47 cm/s, respectively. As for blood measurements, AST, ALT, fasting blood glucose, HDL-C, TC, LDL-C, TG and eGFR were 20.84 ± 8.08 IU/L, 22.66 ± 14.27 IU/L, 98.04 ± 14.08 mg/dL, 53.55 ± 14.61 mg/dL, 209.82 ± 35.97 mg/dL, 128.00 ± 31.71 mg/dL, 99.85 ± 75.05 mg/dL and 70.41 ± 12.05 mL/min/1.73 m2, respectively. Final participants for analysis Table 1General characteristics (N = 909)VariablesMean ± SDAge (years)51.13 ± 9.57BMI (kg/m2)23.12 ± 3.13Gender (male), n (%)590 (64.91%)Systolic BP (mmHg)120.22 ± 14.92Diastolic BP (mmHg)76.12 ± 9.96Current smoker, n (%)197 (21.67%)Exercise, n (%)177 (19.47%)Fatty liver, n (%)263 (28.93%)Atherosclerosis-related index baPWV-max (cm/s)1415.45 ± 246.47 TyG8.29 ± 0.67 AI3.17 ± 1.23Blood measurements AST (IU/L)20.84 ± 8.08 ALT (IU/L)22.66 ± 14.27 Fasting glucose (mg/dL)98.04 ± 14.08 HDL-C (mg/dL)53.55 ± 14.61 TC (mg/dL)209.82 ± 35.97 LDL-C (mg/dL)128.00 ± 31.71 TG (mg/dL)99.85 ± 75.05 eGFR (mL/min/1.73 m2)70.41 ± 12.05 Uric acid (mg/dL)5.25 ± 1.38BMI body mass index, BP blood pressure, baPWV brachial–ankle pulse wave velocity, TyG triglyceride glucose index, AI atherosclerosis index, ALT alanine aminotransferase, AST aspartate transaminase, eGFR estimated glomerular filtration rate, HDL-C high-density lipoprotein cholesterol, LDL-C low-density lipoprotein cholesterol, TC total cholesterol, TG triglyceride General characteristics (N = 909) BMI body mass index, BP blood pressure, baPWV brachial–ankle pulse wave velocity, TyG triglyceride glucose index, AI atherosclerosis index, ALT alanine aminotransferase, AST aspartate transaminase, eGFR estimated glomerular filtration rate, HDL-C high-density lipoprotein cholesterol, LDL-C low-density lipoprotein cholesterol, TC total cholesterol, TG triglyceride Univariate and multivariate curve analysis Our univariate curve analysis supported the significant associations of AI, TyG and baPWV-max with fatty liver (all P < 0.05, Fig. 2A, B, C), respectively. Furthermore, the results from multivariate curve analysis still displayed that AI, TyG and baPWV-max were independently associated with fatty liver, respectively, after age, sex, BMI, systolic BP, diastolic BP, current smoker and exercise were adjusted for (all P < 0.05, Fig. 3A, B, C). Univariate smooth curve analysis Multivariable smooth curve analysis Multivariate logistic regression analysis To further ensure the association of the atherosclerosis-related indicators with fatty liver, the results from multivariate logistic regression analysis consistently demonstrated that AI [1.76 (1.49, 2.07), < 0.0001], TyG [2.71 (1.96, 3.75), < 0.0001] and baPWV-max [1.00 (1.00, 1.00) < 0.0001] were significantly associated with fatty liver, respectively, when age, sex, BMI, systolic BP, diastolic BP, current smoker and exercise were adjusted for (Table 2). Also, sensitivity analysis from Table 3 showed similar results on the association between these indicators and fatty liver by considering “eGFR” as a covariate. Table 2The correlation between atherosclerosis-related index and fatty liver by multivariate logistic regressionExposureAdjusted IAdjusted IIAI2.04 (1.77, 2.36), < 0.00011.76 (1.49, 2.07), < 0.0001TyG4.41 (3.30, 5.90), < 0.00012.71 (1.96, 3.75), < 0.0001baPWV-max1.00 (1.00, 1.00), < 0.00011.00 (1.00, 1.00), < 0.0001Adjusted I for age and sexAdjusted II for age, sex, BMI, systolic BP, diastolic BP, current smoker and exerciseAI atherosclerosis index, TyG triglyceride glucose index, baPWV brachial–ankle pulse wave velocity, BMI body mass index, BP blood pressure The correlation between atherosclerosis-related index and fatty liver by multivariate logistic regression Adjusted I for age and sex Adjusted II for age, sex, BMI, systolic BP, diastolic BP, current smoker and exercise AI atherosclerosis index, TyG triglyceride glucose index, baPWV brachial–ankle pulse wave velocity, BMI body mass index, BP blood pressure Table 3The correlation between atherosclerosis-related index and fatty liver by sensitivity analysisExposureAdjusted IAdjusted IIAI2.04 (1.77, 2.36) < 0.00011.78 (1.51, 2.09) < 0.0001TyG4.42 (3.29, 5.94) < 0.00012.75 (1.99, 3.82) < 0.0001baPWV-max1.00 (1.00, 1.00) 0.51111.00 (1.00, 1.00) 0.9261Adjusted I for age, sex and eGFRAdjusted II for age, sex, BMI, systolic BP, diastolic BP, current smoker, exercise and eGFRAI atherosclerosis index, TyG triglyceride glucose index, baPWV brachial–ankle pulse wave velocity, BMI body mass index, BP blood pressure The correlation between atherosclerosis-related index and fatty liver by sensitivity analysis Adjusted I for age, sex and eGFR Adjusted II for age, sex, BMI, systolic BP, diastolic BP, current smoker, exercise and eGFR AI atherosclerosis index, TyG triglyceride glucose index, baPWV brachial–ankle pulse wave velocity, BMI body mass index, BP blood pressure Diagnostic analysis The saturation effect curve was plotted to explore a turning point in the real-world relationship between the atherosclerosis-related indicators and fatty liver. Our results did not show any obvious turning point on the associations of AI, TyG and baPWV-max with fatty liver, respectively (Table 4). Therefore, we further performed a ROC analysis for these indicators (Figs. 4). Our results indicated that TyG had a higher AUC value of 0.7660 than AI (AUC = 0.7547) and baPWV-max (AUC = 0.6223), suggesting that TyG might be a relatively valuable biomarker for diagnosing fatty liver, compared with other indicators (Table 5). The TyG index exceeding 8.2611 indicated the occurrence of fatty liver with a high sensitivity (80.99%) and specificity (60.84%). Table 4Saturation effect analysis for fatty liverExposureAITyGbaPWV-maxModel 1 Linear effect1.78 (1.51, 2.09) < 0.00012.75 (1.99, 3.82) < 0.00011.00 (1.00, 1.00) < 0.0001Model 2 Turning point (K)1.57.751266 < KInf (0.00, Inf) 0.98520.90 (0.21, 3.82) 0.88901.01 (1.01, 1.02) 0.0002 > K1.71 (1.45, 2.03) < 0.00013.13 (2.16, 4.53) < 0.00011.00 (1.00, 1.00) 0.0008 Log likelihood ratio tests0.0210.1510.001Adjusted for age, sex, BMI, systolic BP, diastolic BP, current smoker and exerciseAI atherosclerosis index, TyG triglyceride glucose index, baPWV brachial–ankle pulse wave velocity, BMI body mass index, BP blood pressure Saturation effect analysis for fatty liver Adjusted for age, sex, BMI, systolic BP, diastolic BP, current smoker and exercise AI atherosclerosis index, TyG triglyceride glucose index, baPWV brachial–ankle pulse wave velocity, BMI body mass index, BP blood pressure ROC analysis Table 5ROC analysis for continuous predictorsExposureROC area (AUC)95% CI low95% CI upBest thresholdSpecificitySensitivityAI0.75470.72070.78873.20380.69660.7110TyG0.76600.73210.79998.26110.60840.8099baPWV-max0.62230.58360.66101327.50000.45360.7414AI atherosclerosis index, TyG triglyceride glucose index, baPWV brachial–ankle pulse wave velocity ROC analysis for continuous predictors AI atherosclerosis index, TyG triglyceride glucose index, baPWV brachial–ankle pulse wave velocity Discussion Fatty liver has become an important predictor for CVD risk, which was independent of other traditional factors such as hypertension, diabetes and hyperlipidemia [1–4]. Early risk stratification for patients with fatty liver is particularly crucial for reducing adverse cardiovascular risk [5–7]. Our results showed that atherosclerosis-related indicators including TyG and AI had a significant correlation with fatty liver, and TyG had a better predictive performance for fatty liver with a higher sensitivity (80.99%) and specificity (60.84%), compared with AI and baPWV-max. As the indicator originally designed for insulin resistance, more and more high-quality clinical evidence have found that TyG has a significant correlation with the risk of cardiovascular outcomes, exhibiting various linear or non-linear characteristics [24, 31]. For instance, previous studies suggested that a high TyG index was not only associated with adverse composite event, but also significantly associated with cardiovascular risk such as coronary heart disease (CHD), angina and heart attack [31–33]. A recent analysis involving a Chinese population from Fuwai Hospital published in the Lancet suggested a reverse L-shaped relationship between TyG index and the risk of all-cause and cardiovascular mortality, with critical value of 9.75 and 9.85, respectively [34]. Specifically, when below the critical value, for every 1 unit increase in TyG value, there was no association with all-cause mortality risk, but only a slight increase in cardiovascular mortality risk (9%). On the contrary, every 1 unit increase in TyG was associated with the risk of all-cause mortality and cardiovascular mortality increases by 110% and 99%, respectively, when the critical value was exceeded [34]. A higher TyG index was also associated with the prevalence of symptomatic CHD and the risk of major adverse cardiovascular and cerebrovascular events in patients with ST-segment elevation myocardial infarction (STEMI) [34]. Furthermore, TyG index has been found to be significantly correlated with the occurrence of new-onset hypertension, and this association remains consistent across different subgroups confirmed by a sensitivity analysis [35]. The TyG index even served as an independent predictor of the likelihood of developing hypertension in the general population [35]. These previous studies were almost consistent with our results that a high TyG value was significantly and positively correlated with a risk of fatty liver, which served as an independent predictor for cardiovascular-related comorbidities. We observed that TyG index with a value more than 8.2611 indicated the occurrence of fatty liver with a sensitivity 80.99% and specificity 60.84%. Our results further compared TyG with other atherosclerosis-related indicators such as AI and baPWV-max, suggesting a greater advantage on detecting fatty liver. In recent evidence, studies have also shown that the TyG index was positively correlated with the risk of non-alcoholic fatty liver disease (NAFLD) [36]. After adjusting for other potential influencing factors, the TyG index remained an independent risk for NAFLD [36]. Other evidence even showed that the combination of TyG index and ALT level had a higher diagnostic value for NAFLD, suggesting that TyG index ≥ 6.9 had a high diagnostic value for it [37]. Among American adults with NAFLD/MASLD (metabolic fatty liver disease), the TyG–BMI and TyG–waist circumference (WC) index were significantly associated with all-cause mortality and cardiovascular mortality risk, respectively [38]. In patients without progressive liver fibrosis, the non-linear positive correlation between TyG–BMI and TyG–WC and all-cause mortality was more pronounced [38]. Therefore, based on the existing evidence and the findings from our study, it suggested that TyG might have an unexpected value on detecting the early presence of fatty liver. Our studies had several limitations. Firstly, this was a cross-sectional design and the causal association could not be clearly defined. Secondly, only Japanese participants with a small sample size from a single center were enrolled into our study that could not be generalizable to other Asian populations or biogeographic ethnic groups. But existing clinical evidence has shown that fatty liver and cardiovascular diseases often coexist in various populations (white, yellow, black or others), suggesting a clear correlation between the two. Thirdly, although we have used ROC curve for diagnostic analysis, TyG has been confirmed to be related to various diseases, such as CVDs, severe acute pancreatitis and chronic kidney disease, which may still interfere with the accuracy of our results. Finally, the pathological type of fatty liver and alcohol consumption were not included in our analysis due to the retrospective design. Further evidence from prospective cohort study are needed in the future to support our findings. Additionally, ultrasound diagnosis of fatty liver is not the gold standard, potentially affecting our results. Pathological diagnosis might be better to confirm our conclusion. Conclusion Briefly, our results demonstrated that atherosclerosis-related index such as TyG, AI and baPWV-max, had a significant correlation with fatty liver in Japanese population. When we observed the TyG value exceeding 8.2611, this indicator showed a best value for detecting the presence of fatty liver than other three traditional biomarkers. Author contributions Xiaoping liu completed the original manuscript text; Ping Yuan completed data validation; Ying Huang revised the final manuscript. All authors reviewed the manuscript. Funding The study was not supported by any fund. Data availability Our study data were obtained from the Dryad database (10.5061/dryad.m484p). Declarations Ethics approval and consent to participate According to the Declaration of Helsinki, informed consent from each participant was obtained. The raw medical data from these Japanese individuals were stored in the Dryad database, and we obtained the data for free by following the database guidelines, without the need for additional ethical proof. Competing interests The authors declare no competing interests. Footnotes Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Data Availability Statement Our study data were obtained from the Dryad database (10.5061/dryad.m484p).