Introduction

In 2024, the Hepatology Branch of the Chinese Medical Association officially proposed to rename non-alcoholic fatty liver disease (NAFLD) as metabolic dysfunction-associated fatty liver disease (MAFLD)1. MAFLD was the most prevalent liver-related disease globally, with various contributing factors such as insulin resistance, oxidative damage, genetic predisposition, and environmental influences2. Research indicated that MAFLD, unlike non-alcoholic fatty liver disease (NAFLD), encompasses both obesity and metabolic irregularities, making it a more accurate identifier of patients with complex fatty liver and liver fibrosis3,4,5,6,7. Additionally, MAFLD was viewed as the hepatic manifestation of metabolic syndrome (MS)8. The prevalence of MAFLD had risen notably in Asia due to lifestyle changes, reaching 30–40% in the general population9. Currently, the prevalence of MAFLD in adults was as high as 35.58%10, which has replaced viral hepatitis as the number one chronic liver disease in China and has become a significant public health issue. Therefore, the prevention and treatment of MAFLD was urgent.

Some studies suggested that fatty liver can be detected early using simple methods, such as the triglyceride and glucose index (TyG), which was cost-effective and suitable for large-scale screening. Crowd research had shown promising results11. At the same time, related indexes such as TyG-BMI, TyG-WC, and TyG-WHtR were also derived. In addition, various factors were related to the onset of MAFLD, and IR played a vital role in the onset of MAFLD. Pathological changes in the intestinal microbiota and increased IR in adipose tissue and skeletal muscle affect hepatic lipid metabolism to promote further hepatic fat accumulation and inflammation12. However, there were still inconsistent conclusions on the specific relationship between the TyG-related indexes and MAFLD, and the differences may be mainly reflected in aspects such as sex and population13. Some studies have reported a positive correlation between TyG index level and the risk of MAFLD14. Ru Zhang et al. found that there was a linear relationship between the TyG index and the risk of MAFLD in males and a non-linear relationship in females15. Other studies have found no correlation between the TyG index and the risk of MAFLD in females. Although there was a positive correlation between the TyG index and the risk of MAFLD in males, there was a saturation threshold effect. When the TyG index was ≥ 8.64, the male TyG index and MAFLD risk were irrelevant16. In addition, other TyG-related indexes also have similar differences regarding sex, population, and region17,18,19,20,21. Xinjiang is located on the northwest border of China. It was a multi-ethnic area with a unique natural environment and eating habits. The diet was mainly based on dairy products and high-calorie foods, which may differ from people in other areas. Previous research by this research found that the prevalence of fatty liver among rural people in Xinjiang reached 22.2%, and there were metabolic risk factors such as obesity, blood pressure, and blood lipid and blood glucose factors22. Moreover, research on the relationship between TyG-related indexes and the risk of MAFLD has not been reported in this region, so it was indispensable to use TyG-related indexes for early screening of MAFLD. Therefore, our study focused on the rural population in Xinjiang, collecting data from 2016 to 2022 using a prospective cohort study that included physical exams, blood tests, and questionnaires. The goal was to understand the relationship between TyG-related indexes and MAFLD, identify high-risk groups early, and provide valuable insights for developing comprehensive prevention strategies for MAFLD in Xinjiang’s rural areas.

Materials and methods

Study population

A cohort study in southern Xinjiang, China, used a typical sampling method to enroll 14,376 rural residents aged 18 or older from the 51 st Farm. From 2019 to 2022, follow-up surveys were conducted every year. After excluding 403 floating population and pregnant women, those who were unable to participate in this survey, 516 people with incomplete basic information, 239 people with missing ultrasound examination information, and 424 people with serious diseases, 12,794 participants were successfully followed up, with a follow-up rate of 88.9%. This study further excluded 2,091 individuals with baseline MAFLD and ultimately included 10,703 individuals (Fig. 1). The study followed the Declaration of Helsinki guidelines and was approved by the Institutional Ethics Review Board of the First Affiliated Hospital of Shihezi University School of Medicine (IERB no.: 2016–121-01).

Fig. 1
figure 1

Flowchart showing the final selection process.

Full size image

Data collection

Clinical examinations and questionnaires were administered by trained healthcare professionals at the hospital’s physical examination center. In order to ensure the accuracy of the findings, medical insurance and hospitalization records from the years 2016 to 2022 were also reviewed. The clinical examination encompassed measurements of height, weight, BMI (body mass index), WC (waist circumference), blood pressure, systolic blood pressure (SBP), diastolic blood pressure (DBP), as well as the presence of Diabetes mellitus type 2 (T2DM), hypertension, dyslipidemia, or fatty liver. The final blood pressure reading was calculated as the average of three measurements. Chronic kidney injury was assessed using estimated glomerular filtration rate (eGFR)23. Diagnosis of T2DM and hypertension followed established criteria24,25. Obesity (BMI ≥ 30 kg/m2) was defined in accordance with the World Health Organization’s diagnostic guidelines for the adult population26. Dyslipidemia was identified by the presence of elevated levels of total cholesterol (TC) ≥ 6.22 mmol/L, triglycerides (TG) ≥ 2.26 mmol/L, low levels of high-density lipoprotein cholesterol (HDL-C) < 1.04 mmol/L, or high levels of low-density lipoprotein cholesterol (LDL-C) ≥ 4.14 mmol/L27. Abdominal ultrasonography was performed by a professional clinician.

Biochemical analysis

Participants fasted overnight for blood tests. Baseline lab tests included triglycerides, total cholesterol, HDL-C, LDL-C, fasting glucose, and serum creatinine. Blood was drawn after an 8-h fast, centrifuged to separate serum, and stored at −20 °C. All tests were conducted using an Olympus AU 2700 analyzer (Olympus Diagnostics, Hamburg, Germany) at the First Affiliated Hospital of Shihezi University School of Medicine’s lab.

Questionnaire survey

In the baseline and follow-up surveys, participants were interviewed in person using standardized questionnaires to collect demographic information, lifestyle habits, medical history, family background, and current medication use. Smoking was defined as having smoked ≥ 100 cigarettes or smoking continuously for ≥ 6 months28. Drinking was defined as consuming alcohol at least twice a month29.

Key definitions

MAFLD was defined as the presence of hepatic steatosis along with at least one of the following criteria: overweight/obesity (BMI ≥ 23.0 kg/m2 in Asian populations), type 2 diabetes mellitus (T2DM), or metabolic dysregulation30. However, our study did not assess the Homeostatic Model Assessment of Insulin Resistance (HOMA-IR) and plasma high-sensitivity C-reactive protein levels. The WHO recommends that adults engage in at least 150 min of moderate to vigorous physical activity (MVPA) per week, with each session lasting at least 10 minutes31. This study was conducted in rural areas. Most of the participants were farmers. Each farming session lasted for more than one hour. Therefore, activities such as lifting heavy objects, digging, and doing farm work were regarded as vigorous activities, while activities such as lifting light objects and doing housework were regarded as moderate-intensity activities. After referring to the suggestions of the WHO and combining the data of this questionnaire, the Movement were divided into no exercise, 1–3 times per week, and more than 4 times per week.

Anthropometric index calculation

$${\text{BMI}}\, = \,{\text{Weight}}_{{({\text{kg}})}} /{\text{Height}}_{{({\text{m}})}}^{{2}} ,$$
$${\text{WHtR }}\left( {{\text{waist}} - {\text{to}} - {\text{Height Ratio}}} \right)\, = \,{\text{WC}}_{{({\text{cm}})}} /{\text{Height}}_{{({\text{cm}})}} ,$$
$${\text{TyG}}\, = \,{\text{Ln }}\left[ {{\text{TG}}_{{({\text{mg}}/{\text{dL}})}} *{\text{FPG}}_{{({\text{mg}}/{\text{dL}})}} /{2}} \right],$$
$${\text{TyG}} - {\text{BMI}}\, = \,{\text{Ln }}\left[ {{\text{TG}}_{{({\text{mg}}/{\text{dL}})}} *{\text{FPG}}_{{({\text{mg}}/{\text{dL}})}} /{2}} \right] \, *{\text{ BMI}},$$
$${\text{TyG}} - {\text{WC}}\, = \,{\text{Ln }}\left[ {{\text{TG}}_{{({\text{mg}}/{\text{dL}})}} *{\text{FPG}}_{{({\text{mg}}/{\text{dL}})}} /{2}} \right] \, *{\text{ WC}},$$
$${\text{TyG}} - {\text{WHtR}}\, = \,{\text{Ln }}\left[ {{\text{TG}}_{{({\text{mg}}/{\text{dL}})}} *{\text{FPG}}_{{({\text{mg}}/{\text{dL}})}} /{2}} \right] \, *{\text{ WHtR}}.$$

Statistical analysis

The descriptive analysis presented continuous variables as mean ± SD or median (IQR), and categorical variables as proportions. The t-test and chi-square test were used to compare continuous and categorical variables, respectively. The TyG-related index was evaluated both as a continuous variable and in quartiles. The association between TyG-related indexes and MAFLD was examined using Cox proportional hazards models, quantified by HR and 95% CI. The predictive value of TyG-related indexes for MAFLD risk was assessed using ROC curves and AUC values, with AUC > 0.7 indicating acceptable prediction performance. Subgroup and sensitivity analyses were conducted to assess the stability of the association and explore interactions. RCS analysis was performed to investigate the dose–response relationship between TyG-related indexes and MAFLD risk, adjusting for covariates such as age, sex, TC, HDL-C, LDL, SCr, marital status, movement, smoking, and drinking. Statistical analyses were carried out using SPSS (version 25.0), MedCalc (Version 20.1.0), and R software (version 4.3.3), with statistical significance defined as a two-sided P-value < 0.05.

Results

Baseline characteristics

Baseline data indicated a MAFLD prevalence of 16.3% (2091/12,794 participants), with male at 15.7% and female at 17%. After excluding those with existing or past MAFLD, the cohort study comprised 10,703 participants, whose baseline characteristics were outlined in Table 1. Over a mean follow-up of 5.70 ± 0.44 years, 5.6% (596/10,703) developed MAFLD, higher among females at 6.7% (357/5332) vs. males. Participants’ mean age was 36.78 ± 14.08 years, with average TyG, TyG-BMI, TyG-WC, and TyG-WHtR values of 8.50 ± 0.70, 213.47 ± 43.20, 757.87 ± 144.72, and 4.66 ± 0.88, respectively. MAFLD patients had higher mean values for age, BMI, WC, WHtR, SBP, DBP, TC, LDL-C, and TyG-related indexes compared to non-MAFLD patients (all P < 0.05). The cumulative incidence of MAFLD was notably higher among those over 60 years of age, individuals with a primary school education or lower, widowed individuals, and those with dyslipidemia (all P < 0.05) (Table 1). Furthermore, females showed higher values for BMI, WHtR, DBP, HDL-C, and TyG-WHtR (all P < 0.05) (Table S1).

Table 1 Baseline characteristics.
Full size table

Association between TyG relevant index and the risk of incident MAFLD

After adjusting the complete model (Model 2), a significant increase in MAFLD risk was observed with elevated levels of TyG, TyG-BMI, TyG-WC, and TyG-WHtR (all P < 0.05). Notably, TyG-WHtR exhibited the highest risk, with a 43.8% increase in MAFLD risk per unit increase. Comparing the highest and lowest quartiles of TyG-related indexes, the risk of developing MAFLD was elevated by 1.363, 6.993, 2.540, and 2.541 times, respectively (Table 2). Sex-specific analysis revealed a stronger association between TyG-related indexes and MAFLD in males than in females. The risk of MAFLD associated with TyG-WHtR remained the highest for both sexes, with a 33.4% increase in males and a 44.6% increase in females per unit increase. Interestingly, a saturation threshold effect was observed in females, where TyG levels ≥ 8.86 showed a positive correlation with MAFLD risk, while levels < 8.86 did not. Notably, no significant association was found between male TyG levels and MAFLD risk (Table S2).

Table 2 Cox regression model of TyG-related indexes and risk of MAFLD.
Full size table

RCS analysis revealed a nonlinear relationship between TyG-related indexes and the risk of MAFLD in the general population after adjusting for the full model (all P-non-linear < 0.0001). However, upon sex stratification, the nonlinear relationship between male TyG and MAFLD was no longer evident (P-non-linear = 0.5856) (Fig. 2).

Fig. 2
figure 2

RCS analysis of TyG-related indexes and MAFLD risk. Adjusted age group, sex, TC, HDL-C, LDL-C, SCr, education, marital status, smoking, drinking, and movement. Male and female groups were not adjusted for sex. TyG, triglyceride and glucose index; BMI, body mass index; WC, waist circumference; WHtR, waist-to-Height Ratio.

Full size image

Subgroup analysis was conducted based on various factors including age, education, marital status, and lifestyle habits. The TyG-related indexes were found to be associated with the risk of MAFLD, especially in unmarried individuals aged 18–30. Significant interactions were observed between TyG-related indexes and marital status, obesity, diabetes, and dyslipidemia (P < 0.001) (Fig. 3, Table S3). Sensitivity analyses were performed after handling individuals with a history of medication or family history of diabetes, hypertension, or hyperlipidemia, and the results were consistent (Table S4).

Fig. 3
figure 3

Subgroup analysis. A, Subgroup analysis of TyG; B, Subgroup analysis of TyG-BMI; C, Subgroup analysis of TyG-WC; D, Subgroup analysis of TyG-WHtR. Adjusted by age group, sex, TC, HDL-C, LDL-C, SCr, education, marital status, smoking, drinking, and movement. Variables related to subgroups were not adjusted during the analysis. For example, TC, HDL-C, and LDL-C were not adjusted for the dyslipidemia subgroup. TyG, triglyceride and glucose index; BMI, body mass index; WC, waist circumference; WHtR, waist-to-Height Ratio; T2DM, Diabetes mellitus type 2.

Full size image

Diagnostic value of TyG relevant index for MAFLD

The results of ROC analysis demonstrated that the TyG-BMI index exhibited a robust predictive capability for detecting MAFLD across various subgroups within the total population, with statistically significant findings (P < 0.05). The Area Under the Curve (AUC) and Youden index reached optimal levels. The identified critical threshold for the TyG-BMI index was 218.91 (Table S5).

Discussion

MAFLD has emerged as a prevalent chronic liver disease worldwide, posing a significant threat to public health32. Numerous studies focused on identifying reliable, practical, and cost-effective predictors of MAFLD, addressing a critical need in clinical practice. Against this backdrop, we hypothesized a potential relationship between TyG-related indices and MAFLD, prompting our investigative analysis. Notably, this study represented a novel endeavor in rural Xinjiang, China. Our findings revealed a significant association between TyG-related indices and the risk of MAFLD across the general population and various subgroups. Furthermore, we observed a stronger correlation between TyG-related indices and MAFLD in females than males. TyG showed no significant association with MAFLD in the male cohort—a departure from previous research outcomes. Interestingly, we identified a non-linear relationship between TyG-related indices and MAFLD in the overall population, which was not observed in the male TyG-BMI subgroup, highlighting a unique disparity from prior studies.

In this study, the average values of TyG, TyG-BMI, TyG-WC, and TyG-WHtR were 8.50 ± 0.70, 213.47 ± 43.20, 757.87 ± 144.72, and 4.66 ± 0.88 respectively, which were higher than those in other studies19,21. Ru Zhang et al. found that the risk of occurrence of TyG in the highest quartile was twice that of the lowest quartile, which was higher than that in this study. The RCS analysis results showed that the relationship between TyG and the risk of MAFLD was linear in males (P-overall < 0.001, P-non-linear = 0.746) but non-linear in females (P-non-linear = 0.040)15. Males in rural Xinjiang, China, tend to engage in more high-intensity physical activities than females. On the other hand, females typically spend more time at home caring for children and doing household chores. This difference in physical activity levels may contribute to females having lower metabolic rates than males. Consequently, females may have a lower baseline BMI and higher WHtR than males, along with lower eGFR levels. These factors can increase the risk of conditions such as diabetic kidney disease (DKD)33 and obesity, among others. As a result, the relationship between TyG and MAFLD may vary significantly between different sexes. The results of another cohort study from Hebei, China, also showed that male TyG was positively correlated with the risk of MAFLD. However, the relationship was non-linear and had a saturation threshold effect. The inflection point was 8.64. When the TyG index was < 8.64, TyG was associated with the risk of MAFLD, and the risk was positively correlated. When the TyG index was ≥ 8.64, the two had no correlation (HR = 1.28, 95%CI: 0.81–2.03). In females, there was no correlation between TyG and the risk of MAFLD16. Our results were opposite and echo the RCS results of female TyG, which suggests that in our population, there may also be a saturation threshold effect on the risk of female TyG and MAFLD, and the inflection point is 8.86. This inflection point can be considered a female Monitoring point for MAFLD high-risk groups and can be verified through further studies.

ROC analysis showed that TyG-BMI showed good predictive value for MAFLD, which was consistent with the study of Mingxing Chang et al.20. However, Zhi Liu et al.’s study in Sichuan reported that the predictive value of TyG for MAFLD was higher than that of TyG-BMI17. After comparing Sichuan MAFLD patients and non-MAFLD patients, the BMI was lower than that in this study, which may be caused by differences in BMI levels. The reason may be that the living and eating habits of rural people in Xinjiang differ from those of other ethnic groups. Xinjiang people eat mostly dairy products and high-calorie foods. It may also be related to Xinjiang’s climate and geographical location. Some studies have shown that arid climate and geographical conditions may lead to the occurrence of hypertension-related diseases34, and high BMI itself is closely related to BP. Moreover, Lin Ning’s study also found that TC and TG levels were positively correlated with indexes such as latitude and temperature difference and negatively correlated with indexes such as annual average temperature, annual average relative humidity, and annual precipitation35. Rural areas in Xinjiang, China, have lower humidity, dry climate, frequent sandstorms, lower annual rainfall, and higher latitudes than Sichuan, China. This may lead to higher TC and TG levels in residents, thereby increasing the level of BMI. Secondly, some studies have reported that TyG-WC and TyG-WHtR in Americans had better predictive value for MAFLD than TyG-BMI19,21. This may be due to heterogeneity caused by country and sample size13. Some studies have shown that the genetic background of the Uighurs in Xinjiang was relatively complex, with European and Asian ancestry36, which may lead to differences in the identical indexes between different countries and ethnic groups. In addition, subgroup analysis results showed that TyG-BMI had the highest AUC95%CI in the 18–30-year-old and unmarried subgroups, reaching 0.786 (0.774, 0.799) and 0.862 (0.775, 0.949) respectively. On the one hand, it may be because the average age of our population is relatively low; on the other hand, it may be because getting married will reduce the possibility of depression and anxiety37, thereby reducing the BMI level and the risk of metabolic syndrome38,39. Therefore, it may cause unmarried people to compare with married and older people. Other marital statuses are associated with higher risk of MAFLD. It showed that TyG had a crucial predictive value for MAFLD in this subgroup of people. It was also rarely reported in previous studies, showing the differences in the risk association between TyG-related indexes and MAFLD at different ages. However, expanding the sample size or conducting multi-center studies was still necessary to verify this inference. In summary, maintaining good health awareness, maintaining a healthy diet, and improving the living environment were of great significance to the prevention and control of MAFLD, and it was necessary to select appropriate MAFLD prediction indexes according to different regions and groups.

Strengths and limitations

The study examined the link between TyG-related indexes and MAFLD risk, focusing on rural residents in Xinjiang, China. It provided epidemiological insights for early MAFLD screening in rural areas. However, the results may not apply to other populations, and the diagnosis was based on imaging, not histology. Ultrasound, though, is a sensitive and practical tool for large-scale fatty liver diagnosis40. The study’s findings were reliable, but potential confounders were not controlled for, and the impact of time on TyG-related indexes was not considered. Future research will explore these aspects. TyG, TyG-BMI, TyG-WC, and TyG-WHtR were independently associated with MAFLD, indicating the need for early detection and intervention to manage high-risk groups effectively.

Conclusion

TyG, TyG-BMI, TyG-WC, and TyG-WHtR were independently linked to MAFLD, with a stronger link in females. TyG did not associate with MAFLD in males. TyG-related indexes predict MAFLD, especially TyG-BMI in unmarried people aged 18–30. These findings are important for early management and prevention of MAFLD in rural China.