Comparison of triglyceride/HDL-C ratio and triglyceride glucose index in identifying NAFLD in Chinese population: a cross-sectional study

Triglyceride (TG) to high-density lipoprotein cholesterol (HDL-C) ratio and triglyceride glucose index (TyG) have been recommended as surrogate markers for nonalcoholic fatty liver disease (NAFLD). In the present study, we aimed to compare the ability of TG/HDL-C, TyG and TyG-HDL-C, a new indicator deriving from TyG and HDL-C, in identifying NAFLD in Chinese population.


Abstract
Background Triglyceride (TG) to high-density lipoprotein cholesterol (HDL-C) ratio and triglyceride glucose index (TyG) have been recommended as surrogate markers for nonalcoholic fatty liver disease (NAFLD). In the present study, we aimed to compare the ability of TG/HDL-C, TyG and TyG-HDL-C, a new indicator deriving from TyG and HDL-C, in identifying NAFLD in Chinese population.

Methods
A total of 11,524 subjects who participated in a health checkup program were included. NAFLD was diagnosed by ultrasonography.

Results
The prevalence rates of NAFLD progressively increased across the quartiles of TG/HDL-C, TyG and TyG-HDL-C (P for tend < 0.0001). After adjustment for confounding factors, the three indicators were all independently associated with the risk of NAFLD. The odds ratios (ORs) for NAFLD of the second to fourth TyG-HDL-C quartiles compared to the rst quartile were 2.24 (1.92-2.62), 4.24 (3.64-4.94) and 8.96 (7.64-10.50), respectively. There was no difference in areas under the ROC curve (AUCs) of the three indicators in men, regardless of body weight. AUCs of TyG were signi cantly higher than thats of TG/HDL-C in overweight and obese women (P < 0.0001), with the optimal cut-off point of 8.56 (sensitivity = 74.4 %, speci city = 60.9 %). In normal or low body weight women, TyG-HDL-C, with a AUC of 0.827 (0.810-0.843), was better than the other two indictors, with AUCs of 0.813 (0.796-0.830), 0.815 (0.797-0.831), respectively (P = 0.0015). The optimal cut-off point of TyG-HDL-C in normal or low body weight women was 6.8 (sensitivity = 78.1%, speci city = 72.1 %).
Conclusions TG/HDL-C is as effective as TyG in predicting NAFLD in Chinese men. TyG is superior to TG/HDL-C in identifying NFAFLD in Chinese overweight and obese women. While in Chinese normal and low weight women, TyG-HDL-C is a better predictor for NAFLD.

Background
Nonalcoholic fatty liver disease (NAFLD) ranges from simple hepatic steatosis to non-alcoholic steatohepatitis (NASH) and cirrhosis. NAFLD is the most common chronic liver disease worldwide, with an estimated global prevalence of 15-40% that continues to increase rapidly [1]. And in mainland of shown that NAFLD was also related to type 2 diabetes mellitus, abdominal obesity, dyslipidemia, hypertension and cardiovascular disease. Furthermore, the prediction for cardiovascular events was often independent of traditional cardiovascular risk factors [7]. Therefore, the early identi cation of NAFLD may be important for public health. There is a strong medical need of simple, noninvasive, cost-effective and reliable tools for predicting NAFLD.
Therefore, in the present study we sought to compare the ability of the TyG index, TG/HDL ratio and TyG-HDL-C in detectig NAFLD in Chinese adults with different sex and weight.

Subjects
The study population was recruited from adults who underwent health checkups at the A liated Hospital of Qingdao University between January 2018 and December 2018. Subjects with an alcohol intake > 140g/week for men and 70g/week for women, a history of viral hepatitis, auto-immune hepatitis or other forms of chronic liver disease were excluded from the study. Finally, a total of 11,524 subjects were included in the nal analysis, consisting of 6,326 men and 5,198 women.
Our study was approved by institutional review board of the A liated Hospital of Qingdao University, and the informed consent requirement was exempted because of our retrospective estimation of unidenti ed database.

Clinical measurements
Physical examination was performed and anthropometry was obtained comprised of weight, height, systolic and diastolic blood pressure (SBP, DBP). Body mass index (BMI) was calculated as body weight (kg) divided by body height squared (m 2 ). According to the criteria for Chinese people, normal and low weight was de ned as BMI < 24 kg/m 2 , overweight as BMI ≥ 24 and < 28 kg/m 2 , and obesity as BMI ≥ 28 kg/m 2 . Fasting blood samples were collected after at least 10 h overnight. Triglycerides (TG), total cholesterol (TC), low-density lipoprotein cholesterol (LDL-C), high-density lipoprotein cholesterol (HDL-C), FPG, serum uric acid (SUA) and homocysteine (HCY) were measured using an autoanalyzer (Beckman, Palo Alto, CA). The TyG index was calculated as established formulas, TyG = Ln [TG (mg/dL) ×FPG (mg/dL)/2] [15]. TyG-HDL-C was calculated by subtracting HDL-C(mmol/L) from TyG.

Diagnosis of NAFLD
NAFLD was diagnosed according to the guidelines of NAFLD issued by the Chinese National Consensus Workshop on Nonalcoholic Fatty Liver Disease [16]. The diagnosis of NAFLD was based on ultrasonography ndings of hepatic steatosis associated with characteristic echo patterns, including ultrasound beam attenuation, diffuse hyperechogenicity of liver, and poor visualization of intrahepatic structures, using a Toshiba Nemio 20 sonography machine with a 3.5-MHz probe (Toshiba, Tokyo, Japan).

Statistical analysis
Continuous variables were presented as median (interquartile range) because of their skew distribution, and categorical variables were displayed as percentages (%). The signi cance of differences between groups was evaluated by Mann-Whitney U test for continuous variables and chi-square test for categorical variables. Logistic regression was used to evaluate the association between TG/HDL-C, TyG, TyG-HDL-C and NAFLD. Four models were applied: model 1 was unadjusted. Model 2 was adjusted for age, sex and BMI. Model 3 was adjusted for all variables in model 2 plus SBP, SUA and Hcy. Model 4 was adjusted for all variables in model 3 plus HDL-C for TyG quartiles or plus FPG for TG/HDL-C quartiles. Receiver operator characteristic (ROC) curve analyses were performed to assess the abilities of the three indicators to detect NAFLD. Z test was used for comparing the differences between the areas under the curve (AUCs). P < 0.05 was considered statistically signi cant. SPSS for Windows version 23.0 (IBM Co., Armonk, NY, USA) and MedCalc (MedCalc software, Olstead, Belgium) were used for statistical analyses in the present study.

Results
Clinical characteristics of the study population A total of 11,524 subjects were included in the present study. The mean age of the population was 55.7 (± 11.5) years and mean BMI 25.8 (± 3.5) kg/m 2 . There were 4,843 participants diagnosed as NAFLD by liver ultrasonic examination, with a prevalence of 42%. In comparison with subjects without NAFLD, subjects with NAFLD were more likely to be male, and had higher levels of BMI, SBP, DBP, FPG, TG, TC, LDL-C, UA, HCY, TG/HDL-C ratio, TyG and TyG-HDL-C (P < 0.0001). In contrast, HDL-C was signi cantly reduced in the NAFLD group. Notably, the mean age of women with NAFLD was older than that of without NAFLD, while the mean age of men with NAFLD was younger than that of without NAFLD (Table 1). Comparison of the associations of TG/HDL-C, TyG and TyG-HDL-C with NAFLD risk The prevalence of NAFLD was all signi cantly increased along with the TG/HDL-C, TyG and TyG-HDL-C quartiles (P for trend < 0.0001, Fig. 1). The phenomenon is more obvious in TyG-HDL-C quartiles. The prevalence of the highest TyG-HDL-C quartile was 73.7%, which showed a 6.8-fold increased with the ones in the lowest quartile. While the prevalence of the fourth quartile of TG/HDL-C and TyG rose 6.2 times and 6.3 times, respectively, compared with that of ones in lowest quartile.
We also observed a signi cant trend of increasing odds ratio for NAFLD with increasing levels of TG/HDL-C, TyG and TyG-HDL-C in

Diagnostic accuracy of TG/HDL, TyG and TyG-HDL-C for NAFLD
To compare the ability of TG/HDL-C, TyG and TyG-HDL-C for detecting NAFLD, subjects were further grouped by sex and weight, and the ROC curves were analyzed. As shown in Table 3, with the decrease of BMI, the areas under the ROC curve (AUCs) of the three indicators became larger, except for TyG in men.
There was no difference in AUCs of the three indicators in men, regardless of body weight. Furthermore, the optimal cut-off point of TG/HDL-C was 1.02 in men, with a sensitivity of 70.9% and speci city of 66.4 %. In contrast, the AUCs of TyG were signi cantly higher than thats of TG/HDL-C in overweight and obese women, while similar to thats of TyG-HDL-C. The optimal cut-off point of TyG for NAFLD of overweight and obese women was 8.56, with a sensitivity of 74.4% and speci city of 60.9 %. In normal or low body weight women, the AUCs of TG/HDL-C and TyG had no difference, while TyG-HDL-C was better than the former two. The optimal cut-off point of TyG-HDL-C in normal or low body weight women was 6.8, with a sensitivity of 78.1% and speci city of 72.1 % (Fig. 2).

Discussion
In the present study,we observed TG/HDL-C, TyG and TyG-HDL-C were all closely and positively related to risk of NAFLD, after adjustment for potential confounders. We also demonstrated that, in men, there was no difference in the effectiveness of TG/HDL-C, TyG and TyG-HDL-C for identifying NAFLD, regardless of body weight. In contrast, TyG was superior to TG/HDL-C for predicting NAFLD in overweight and obese women, while not different from TyG-HDL-C. In women with normal or low body weight, TyG and TG/HDL-C had no difference in identi cation for NAFLD, while TyG-HDL-C was better than the former two. The optimal cut-off point of TyG-HDL-C for NAFLD in normal and low body weight women was 6.8. The optimal cut-off point of TyG for NAFLD in overweight and obese women was 8.56. Compared with the other two indicators, TG/HDL-C is more suitable for predicting NAFLD in Chinese men for its simple calculation, with the cut-off point of 1.02, although there was no difference in effectiveness.
In recent years, a growing body of evidence has shown an association between TG/HDL-C and NAFLD, as well as TyG and NAFLD. Two cohort studies [12,14] showed that higher TG/HDL-C was strongly associated with increased risk of NAFLD. Similarly, in a cross-sectional study [11] involving 18,061 participants, TG/HDL-C was independently related with NAFLD in apparently healthy individuals and was recommended as a surrogate for NAFLD. Several cross-sectional studies [8][9][10] aslo revealed a strong and positive association between TyG and NAFLD, and TyG was effective to screen NAFLD. Similarly, in a nine-year cohort study [17], the researchers found that TyG could predict the risk of NAFLD in the next few years. Consistently, our present study also revealed a close relationship between TG/HDL-C and NAFLD, as well as TyG and NAFLD. Furthermore, both TG/HDL-C and TyG were effective in detecting NAFLD.
The mechanisms associated with the onset of NAFLD remain poorly understood. The current evidence [18] suggests a "multiple-hit" hypothesis including insulin resistance (IR), hormones secreted from the adipose tissue, nutritional factors, gut microbiota and genetic and epigenetic factors, among which IR plays an important role. Moreover, IR has been found to be associated with the development of NAFLD [19,20]. Besides,the severity of IR was also positively related to the progress and prognosis of NAFLD [21]. On the other hand, TG/HDL-C [22][23][24][25] and TyG [10,[26][27][28][29] were both found to be closely associated with insulin resistance in different populations and have been proposed as markers of insulin resistance. Thus, IR may be the mechanism underlying the link between TG/HDL-C, TyG and NAFLD.
Previous study by Er et al. [30] showed that TyG was more e cient than TG/HDL-C for identifying insulin resistance. Another study by Babic et al. [31] showed TG/HDL-C might be a useful predictor of glycemic control in normal weight, and TyG index in overweight and obese patients with diabetes mellitus type 2.
The comparison of TG/HDL-C and TyG in predicting NAFLD was not investigated in previous studies. We created a new indicator, TyG-HDL-C, deriving from three variables highly correlated with IR. Besides, BMI is an important factor affected TyG e cacy for identifying NAFLD [8]. Thus, our present study compared the ability of TG/HDL-C, TyG and TyG-HDL-C in predicting NAFLD in different sex and weight groups. Our study found with the decrease of BMI, all of the three indicators were more closely related with NAFLD. There was no difference in the ability of TG/HDL-C, TyG and TyG-HDL-C for identifying NAFLD, regardless of body weight. Compared with the other two indicators, TG/HDL-C, with a AUC of 0.746 (0.735-0.757), is more suitable for predicting NAFLD in Chinese men for its simple calculation, although there was no difference in effectiveness. And the optimal cut-off point was 1.02, with a sensitivity of 70.9% and speci city of 66.4 %. In contrast, TyG was superior to TG/HDL-C for predicting NAFLD in overweight and obese women, while not different from TyG-HDL-C. The optimal cut-off point of TyG for diagnosing NAFLD in overweight and obese women was 8.56, with a AUC of 0.748 (0.732-0.763), a sensitivity of 74.4% and speci city of 60.9 %. In women with normal or low body weight, TyG and TG/HDL-C had no difference in identi cation for NAFLD, while TyG-HDL-C was better than the former two. The optimal cutoff point of TyG-HDL-C for NAFLD in normal or low body weight women was 6.8, with a AUC of 0.827 (0.810-0.843), a sensitivity of 78.1% and speci city of 72.1 %.
The differences of TyG and TG/HDL-C in diagnostic e cacy between men and women may be caused by the different distribution of HDL-C in gender, which had dose-response relationship with IR and the risk of NAFLD [32][33][34]. In our study, the level of HDL-C in men was lower than that in women (men, 1.36 ± 0.32mmol/L vs women, 1.58 ± 0.35mmol/L, P <0.0001), and the prevalence of low HDL-C in men was signi cantly higher than that in women (men,11.8% vs women, 3.9%, P <0.0001). A nationally representative survey [35] of 163,641 Chinese adults also found low HDL-C and high TG were the primary forms of dyslipidemia, and the prevalence of low HDL-C in men was much higher than that in women. In the Chinese population, low HDL-C mainly affects men. This may be the reason why TG/HDL-C is weaker than TyG in identifying NAFLD, while in men, TG/HDL-C is as effective as TyG. We also found a phenomenon that the mean age of women with NAFLD was older than that of without NAFLD, while the mean age of men with NAFLD was younger than that of without NAFLD. This may be related to different exercise and eating habits of men in different ages.
This research has some advantages, for instance, the relative large sample size and the subgroups analysis. However, there are several limitations in the present study. First of all, we diagnosed NAFLD using ultrasonography, which has limited in detecting mild liver steatosis. However, this noninvasive method is still widely used in clinical practice. Secondly, the study was cross-sectional in design, therefore, a causal relationship cannot be obtained. Thirdly, information about IR was unavailable in our study,which may be potential link.

Conclusions
TG/HDL-C is as effective as TyG in predicting NAFLD in Chinese men. TyG is superior to TG/HDL-C in identifying NFAFLD in Chinese overweight and obese women. While in Chinese normal and low weight women, TyG-HDL-C is a better predictor for NAFLD.