Cumulative Triglyceride-Glucose Index and Risk of Metabolic Dysfunction–Associated Fatty Liver Disease: Findings from the Kailuan Cohort

Introduction

Metabolic dysfunction–associated fatty liver disease (MAFLD) is characterized by the presence of hepatic steatosis in individuals who have at least one metabolic risk factor (ie., obesity or dyslipidemia), and minimal or no alcohol consumption. It was previously known as fatty liver or nonalcoholic fatty liver disease (NAFLD).¹ Recently, MAFLD has been suggested as a replacement for NAFLD to highlight the pathogenic link between fatty liver disease and metabolic dysfunction.² Research has shown that insulin resistance (IR) plays a crucial role in the development of MAFLD. IR hinders glucose disposal, leading to a compensatory increase in insulin production by beta cells and resulting in hyperinsulinemia, which creates a detrimental cycle of worsening IR and its metabolic consequences.³ In this context, recent studies have focused on the relationship between a cost-effective surrogate marker of IR - the triglyceride-glucose (TyG) index⁴ and the development of MAFLD, suggesting that higher TyG index levels are positively correlated with an increased risk of MAFLD.⁵ Nevertheless, the TyG index is a dynamic condition that changes over time, previous cohort studies focus only on the levels of TyG index at baseline,^6–9 the impact of the cumulative TyG index (cum-TyG index) on the risk of new-onset MAFLD is still not fully understood. Whether the accumulation of time-weighted cum-TyG index influences the risk of new-onset MAFLD is still uncertain. To this end, this study aims to assess the relationship between the cum-TyG index and the risk of new-onset MAFLD in a large cohort study.

Materials and Methods

Study Participants

The Kailuan Study is a prospective research initiative carried out in the Kailuan Community located in Tangshan, China. The comprehensive design and methodologies of the study were detailed in earlier publications.¹⁰ Since 2006, a total of 101,510 employees from the Kailuan Group taken part in the Kailuan Study. In this research, individuals with a prior history of MAFLD (n = 10,025), those with a previous cancer diagnosis (n = 349), and participants with missing data (n = 1305) were excluded. Ultimately, 89,831 subjects were included in the final analysis.

Data Collection

Data regarding demographic variables were gathered using questionnaires. Blood samples from the elbow vein were collected on the morning of the physical examination day to assess serum biochemical indices, all of which were analyzed using an automated biochemical analyzer (Hitachi 747; Hitachi, Tokyo, Japan). Smoking habits were divided into three categories: never smokers, former smokers, and current smokers. Educational attainment was classified into three distinct levels: low education level, medium education level, and high education level. The frequency of physical activity was categorized as none, occasional, and frequent exercise.

Cum-TyG Index and High TyG Index Exposure Duration

The cum-TyG index was assessed by calculating the weighted average of the TyG index recorded at each visit. In detail, cum-TyG = (TyG index₂₀₀₆ + TyG index₂₀₀₈)/2 × TI_2006-2008 + (TyG index₂₀₀₈ + TyG index₂₀₁₀)/2 × TI_2008-2010. In this formula, TyG index₂₀₀₆, TyG index₂₀₀₈, and TyG index₂₀₁₀ represented the TyG index values recorded during the year of 2006, 2008, and 2010, respectively, while TI_2006-2008, TI_2008-2010 indicated the average values of time intervals for all participants between 2006 and 2008, and between 2008 and 2010. Subsequently, participants were divided into four groups based on the quartiles of cum-TyG index: quartile 1: 20.83 < cum-TyG index ≤ 32.27, quartile 2: 32.27 < cum-TyG index ≤ 34.83, quartile 3: 34.83 < cum-TyG index ≤ 37.60, and quartile 4: 37.60 < cum-TyG index ≤ 55.42.

The appropriate cutoff value for the TyG index predicting MAFLD was determined using a time-dependent receiver operating characteristic curve. The duration of high TyG index exposure was calculated. 0-year high TyG index exposure indicated that the TyG index levels were below all three critical values; 2-year high TyG index exposure indicated that the TyG index levels were above one of three critical values; 4-year high TyG index exposure indicated that the TyG index levels were above two of three critical values; 6-year high TyG index exposure indicated that the TyG index levels were above all three critical values. Subsequently, participants were divided into four groups based on the duration of high TyG index exposure.

Diagnosis of MAFLD

The evaluation of the severity of liver steatosis was determined through abdominal ultrasonography, utilizing a high-resolution B-mode topographical ultrasound system equipped with a 3.5 MHz probe (ACUSON X300, Siemens, Germany). In this study, abdominal ultrasonography was consistently performed by experienced radiologists who were unaware of the baseline data and laboratory findings of the participants. The diagnosis of MAFLD was established based on the most recent consensus criteria:¹¹ MAFLD is characterized by liver steatosis identified through ultrasonography, in conjunction with one of the following three criteria: type 2 diabetes mellitus, overweight or obesity (body mass index (BMI) ≥ 23.0 kg/m²), or lean/normal weight (BMI < 23.0 kg/m²) accompanied by two metabolic risk abnormalities. In our study, metabolic risk abnormalities were defined as follows: 1. Waist circumference (WC) ≥90 cm for men or 85 cm for women. 2. Blood pressure ≥ 130/85 mmHg. 3. Plasma triglycerides (TG) ≥1.70 mmol/L. 4. Plasma high-density lipoprotein cholesterol <1.0 mmol/L for men or <1.3 mmol/L for women. 5. Prediabetes (ie., fasting glucose levels between 5.6 and 6.9 mmol/L). 6. Plasma high-sensitivity C-reactive protein (hs-CRP) level exceeding 2 mg/L.

Statistical Analyses

All analyses were conducted utilizing SAS, version 9.4 (SAS Institute, Inc, Cary, NC). Continuous variables that followed a normal distribution were represented as mean and standard deviation, whereas those exhibiting a skewed distribution were represented as medians along with the interquartile range. Categorical variables were presented in terms of proportions. The cumulative incidence of MAFLD across various groups was determined using the Kaplan–Meier method, and the Log rank test was applied for inter-group comparisons. Three Cox proportional hazards regression models were performed. Model 1 was a univariate model, model 2 was adjusted for age, sex, model 3 was adjusted for all variables in baseline characteristics. Sensitivity analyses were carried out to eliminate the potential influence of reverse causation. Subgroup analyses were conducted following stratification based on age, sex, and BMI. Additionally, restricted cubic spline (RCS) was utilized to test non-linear relationships between the cum-TyG index levels and the risk of new-onset MAFLD. Two-sided p-values of less than 0.05 were considered statistically significant.

Results

Characteristics of the Study Participants

The baseline characteristics, according to the quartiles of cum-TyG index, are shown in Table 1. A total of 89,831 participants were included in the present study. Their mean age was 52.16 ± 12.79 years, and 79.83% were men. Compared with quartile 1, participants in the quartile 4 had higher WC, BMI, systolic blood pressure (SBP), diastolic blood pressure (DBP), fasting blood glucose (FBG), total cholesterol (TC), TG, hs-CRP, and with higher prevalence of hypertension, diabetes and hyperlipemia. The baseline characteristics, according to the duration of high TyG index exposure, are shown in Table 2. Compared with 0-year high TyG index exposure, participants in 6-year high TyG index exposure had higher WC, BMI, SBP, DBP, FBG, TC, TG, hs-CRP and with higher prevalence of hypertension, diabetes and hyperlipemia.

Table 1 Baseline Characteristics According to the Quartiles of Cum-TyG Index

Table 2 Baseline Characteristics According to the Duration of High TyG Index Exposure

Cum-TyG Index and the Risk of MAFLD

A total of 16,387 cases of newly diagnosed MAFLD were recorded over a median follow-up duration of 10 years. The Kaplan–Meier curves illustrated that the cumulative incidence of new-onset MAFLD was greatest among participants in quartile 4 (Figure 1A). The Cox proportional hazards regression analysis indicated that the hazard ratio (HR) and 95% confidence interval (CI) for quartile 4 were 1.648 (1.571–1.730) compared to quartile 1 (Table 3). Moreover, the results from the sensitivity analyses were consistent with those of the primary analyses (Table 4). RCS analysis indicated that the cum-TyG index and new-onset MAFLD followed a non-linear relationship (P for overall<0.001, P for non-linearity<0.001) (Figure 2A). Subgroup analysis showed that quartile 4 was significantly associated with the risk of MAFLD in the elderly (HR: 1.873, 95% CI: 1.590–2.207), women (HR: 2.056, 95% CI: 1.860–2.274), and participants with lower BMI (<28) (HR: 1.775, 95% CI: 1.677–1.879) (Table 5).

Table 3 Association Between the Cum-TyG Index and the Risk of MAFLD

Table 4 Sensitivity Analysis According to the Quartiles of Cum-TyG Index

Table 5 Subgroup Analysis According to the Quartiles of Cum-TyG Index

Figure 1 (A) Cumulative incidence of MAFLD according to the quartiles of cum-TyG index, (B) Cumulative incidence of MAFLD according to the duration of high TyG index exposure.

Figure 2 (A) Restricted cubic splines according to the quartiles of cum-TyG index, (B) Restricted cubic splines according to the duration of high TyG Index exposure.

High TyG Index Exposure Duration and the Risk of MAFLD

The Kaplan–Meier curves illustrated that the cumulative incidence of new-onset MAFLD was greatest among participants in 6-year high TyG index exposure (Figure 1B). The Cox proportional hazards regression analysis indicated that the HR and 95% CI for 6-year high TyG index exposure were 2.122 (2.010–2.239) compared to 0-year high TyG index exposure (Table 6). Moreover, the results from the sensitivity analyses were consistent with those of the primary analyses (Table 7). RCS analysis indicated that the duration of high TyG index exposure and new-onset MAFLD followed a non-linear relationship (P for overall<0.001, P for non-linearity<0.001) (Figure 2B). Subgroup analysis showed that 6-year high TyG index exposure was significantly associated with the risk of MAFLD in the middle-aged and young (HR: 2.122, 95% CI: 2.004–2.247), women (HR: 2.477, 95% CI: 2.226–2.757), participants with lower BMI (HR: 2.510, 95% CI: 2.359–2.671) (Table 8).

Table 6 Association Between the Duration of High TyG Index Exposure and the Risk of MAFLD

Table 7 Sensitivity Analysis According to the Duration of High TyG Index Exposure

Table 8 Subgroup Analysis According to the Duration of High TyG Index Exposure

Discussion

To the best of our knowledge, this is the first cohort study evaluating the association of the cum-TyG index with the risk of new-onset MAFLD. Our findings suggested that elevated cum-TyG index is linked to a heightened risk of developing new-onset MAFLD. Therefore, the cum-TyG index can be integrated into risk prediction tools for the prevention of MAFLD in clinical settings.

In accordance with our findings, Zhang et al conducted a cohort study that included 2056 participants, and demonstrated that the incidence of MAFLD appeared to rise with increasing quartiles of the TyG index.⁷ Yang et al carried out a study involving 71,299 participants. Their multivariate logistic regression analysis revealed a positive relationship between the TyG index and the risk of MAFLD.⁹ Su et al analyzed NHANES data in their study, which showed a significant positive linear relationship between the TyG index and the risk of MAFLD.⁸ Taheri et al conducted a case-control study with 4241 participants, demonstrating that a higher TyG index level was significantly positively correlated with an increased risk of MAFLD in the Iranian population.⁶

Nonetheless, previous cohort studies have primarily concentrated on the baseline levels of the TyG index (6–9). The TyG index represents a dynamic condition that evolves over time. However, the accumulative effect of dyslipidemia and IR on the risk of developing new-onset MAFLD remains inadequately comprehended. In terms of specific pathological mechanism, a defining characteristic of MAFLD is the increase in hepatocyte lipid droplets (LDs) that store triglycerides, cholesterol esters, and various other lipid types.¹² At the tissue level, the accumulation of fat can be harmless, occurring without inflammation. However, it may also advance to a condition of chronic inflammation that is associated with steatohepatitis. While steatohepatitis is not immediately life-threatening, patients face a heightened risk of developing severe liver diseases.¹³ On the other hand, a significant factor in the progression of MAFLD is lipotoxicity, which arises from the hepatic buildup of lipids that can initiate cellular stress responses. In the context of metabolic-associated fatty liver, lipid accumulation is typically non-threatening as inert lipid types like TG and cholesterol esters dominate at this phase. In contrast, steatohepatitis is often marked by increased levels of free cholesterol, diacylglycerols, and/or ceramides.¹⁴ Additionally, ceramides and diacylglycerol species can lead to IR.¹⁵ Therefore, there is compelling evidence supporting that an excess of lipids and their derivatives, when not safely stored as neutral lipids in LDs, can play a causal role in the progression of MAFLD.¹²

Our research has some limitations. Firstly, the homeostasis model assessment IR (HOMA-IR) score is included in the latest diagnosis criteria of MAFLD.¹¹ The HOMA-IR score is a straightforward approach to illustrate the interaction between glucose and insulin dynamics during fasting to estimate IR. HOMA-IR is simplified by the formula: fasting insulin (μU/dL) × FBG (mmol/L)/22.5,¹⁶ however, fasting insulin was not measured in the Kailuan study. Secondly, Ultrasonography is a commonly available imaging method used for identifying fatty liver; however, the accuracy and reliability reported have varied significantly.¹⁷ Research has shown that ultrasound has a diagnostic accuracy of approximately 90 to 95% for moderate and severe hepatic steatosis, while its accuracy for diagnosing mild liver steatosis is merely 60%,¹⁸ indicating ultrasonography for MAFLD diagnosis may under detect milder liver steatosis. Although a number of covariates have been included in Cox regression models, the possibility of residual confounding in this study may diminish the reliability of the evidence. In this regard, it has been demonstrated that dietary and genetic factors play important roles in the development and progress of MAFLD,^19,20 however, they are unavailable in the Kailuan study. Thirdly, personal data and socioeconomic status were gathered through self-reported questionnaires, which may be influenced by recall bias. Additionally, considerably more males than females were enrolled into the Kailuan study, the results may not be applicable to other populations.

Conclusions

Our study filled the knowledge gap by demonstrating that the cum-TyG index and the length of time exposed to a high TyG index correlate with a heightened risk of developing new-onset MAFLD, independent of other conventional risk factors. Cum-TyG index may be incorporated into the risk prediction tools for MAFLD prevention. It is crucial to control blood lipid levels within the recommended range to effectively prevent the development of MAFLD.