Introduction

Metabolic syndrome (MetS), a major risk factor in diabetes and cardiovascular disease, is rising in the adolescents due to the worldwide increased prevalence of obesity in this age population group [1]. Since MetS at young age is so closely related to metabolic disorders in adulthood, early recognition and treatment of at risk groups is very important [2]. Population-based studies show that higher anthropometric indices including weight, waist and hip circumferences as well as higher epicardial adipose tissue are associated with elevated risk of MetS in pediatrics [3, 4]. One of the possible methods for screening at-risk groups for MetS is insulin resistance measurements by the pancreatic suppression test or the hyperinsulinemic–euglycemic clamp technique [5]. However, these tests are invasive, expensive and complicated, so they are not suitable for clinical and epidemiological studies. Regarding the difficulties in fasting insulin measurement, an insulin-free surrogate for evaluating insulin resistance was sought and developed. Studies have shown that increased serum triglyceride (TG) can compromise muscle glucose metabolism and leads to a decrease in insulin sensitivity [6]. Therefore, fasting levels of triglycerides and glucose, the so-called triglycerides and glucose index (TyG index), seems to be an appropriate substitute marker of insulin resistance among adolescents [7]. The TyG index is a convenient, cost-effective and logical method for evaluating of insulin resistance, which is based on fasting TG and glucose instead of fasting insulin and glucose [8]. However, there are limited pediatric data regarding this index, and a valid cut-off value has not been established in Iran. Therefore, we conducted this study with the aim of determining of TyG index and finding valid cut-off values for metabolic abnormalities in a representative sample of Iranian children and adolescents.

Materials and methods

Study design and population

The data of current study were obtained from the fifth phase of a national school-based surveillance survey entitled “Childhood and Adolescence Surveillance and Prevention of Adult Non-communicable Disease” (CASPIAN) in 2015. The CASPIAN study is a large nationwide survey in Iran which was conducted on children and adolescents. There are vast data about this pediatrics study which are analyzed with different hypotheses, questions and approaches. Considering this article, the triglyceride–glucose index has not been analyzed in the CASPIAN study, so far. It is for the first time this index has been calculated and analyzed for predicting metabolic syndrome. However, the prevalence of metabolic syndrome and its association with other indices have been published in other articles. The study population was 14,400 individuals aged 7–18 years from urban and rural areas of 30 provinces in Iran. They were selected using multistage, stratified cluster sampling method. Sampling within each province was performed based on the student’s place of residence (urban or rural) and level of education (primary and secondary) using the proportional to size method and with equal sex ratio. This means that the number of boys and girls was the same in each province, and their ratio in urban and rural areas was proportional to the number of students in urban and rural areas. Similarly, the number of samples between different grades in urban and rural areas was divided according to the number of students in each grade. Achieving the desired number of samples was obtained using cluster sampling in each province with equal cluster sizes. Clusters were determined at school levels. The size of each cluster was 10 students, which means that a total of 10 statistical units (including 10 students and their parents) would be considered in each cluster. The sample size was 480 students in each province (48 clusters of 10 students), i.e., a total of 14,400 students at the national level. This was the maximum sample size that helped in achieving a good estimate of all risk factors of interest. In each province, 14 clusters out of 48 clusters were randomly selected for the biochemical test, i.e., a total of 4200 students. There were no significant differences in demographic data and physical measurements between randomly selected and non-selected pediatrics. Protocol details of the study have been explained previously [9].

Ethics

The study protocol was approved by the Research and Ethics Council of Isfahan University of Medical Sciences (code: 194049) and a written informed consent and a verbal consent were obtained from all the parents and students, respectively.

Anthropometric and biochemical measurements

Weight was measured on a scale placed on a flat ground to the nearest 0.1 kg while the subjects were minimally clothed. Height was measured in a standing position to the nearest 0.5 cm, without shoes [10]. Body mass index (BMI) was calculated as weight (kg) divided by square of height (m2). Waist circumference (WC) was measured using non-stretchable plastic tape between the uppermost lateral border of right ilium and that of left ilium to the nearest 0.1 cm. Hip circumference (HC) was measured using non-stretchable plastic tape at the widest part of the hip at the level of the greater trochanter to the nearest 0.1 cm [11]. Neck circumference was measured in the midway of the neck, between mid-cervical spine and mid anterior neck, to the nearest 0.1 cm, using a non-stretchable plastic tape with the subjects standing upright [12]. Wrist circumference was measured to the nearest 0.1 cm with subjects in a seated position using a non-stretchable plastic tape over the Lister tubercle of the distal radius and over the distal ulna [13]. World Health Organization (WHO) growth curves were used for categorizing of BMI; overweight as sex-specific BMI for age of 85th–95th, and obesity as sex-specific BMI for > 95th. Abdominal obesity was defined as waist-to-height ratio (WHtR) equal to or more than 0.5 [10].

Blood pressure (BP) was measured two times in the sitting position after 15 min of rest using a standardized mercury sphygmomanometer on the right arm. The first and fifth Korotkoff sounds were recorded as systolic blood pressure (SBP) and diastolic blood pressure (DBP), respectively. The average of the two recorded values was used in the analyses. Fasting blood samples were taken after 12–14 h of overnight fast. Fasting blood glucose (FBG), total cholesterol (TC), low density lipoprotein-cholesterol (LDL-C), high density lipoprotein-cholesterol (HDL-C), and triglycerides (TG) were measured enzymatically by Hitachi auto-analyzer (Tokyo, Japan) [14].

Definition

In this study, MetS was defined according to Adult Treatment Panel III (ATP III) criteria modified for the pediatric age group. Participants who had at least three of the following criteria were classified as having MetS: TG concentration of 150 mg/dL or greater; HDL-C concentration of 40 mg/dL or less; FBG concentration of 100 mg/dL or greater; abdominal obesity: waist-to-height ratio > 0.5; and either SBP or DBP greater than the 90th percentile for age, sex, and height [15]. The TyG index was calculated as follows: TyG index = Ln [triglyceride (mg/dL) × fasting glucose (mg/dL)/2] [16].

Statistical analysis

All variables were checked for normality and presented as the mean ± standard deviation, or number (percentage). The independent sample t test was used to compare continuous variables and the Chi-square test was used to compare proportions according to age and sex groups. TyG index is presented in deciles by age and sex groups. To estimate valid cut-off values of TyG index for predicting metabolic syndrome, the receiver operation characteristic (ROC) curve analysis was used with an estimation of the sensitivity and specificity. Data were analyzed separately for sex and age groups. The estimated cut-off values were determined using the minimum value of which represents the maximum sum of sensitivity and specificity. The area under curve (AUC) shows the ability of TyG index cut-off points to discriminate students with and without metabolic syndrome. Analyses were performed using STATA version 11.0 (STATA Statistical Software: Release 11. STATA Corp LP. Package, College Station, TX, USA); P values less than 0.05 were considered as statistically significant.

Results

Totally, 3843 students (52.3% boys) with mean (SD) age of 12.45 (3.04) years were assessed. According to WHO criteria, 9.4% of participants (8.7% of boys and 10.2% of girls) were overweight and 11.4% (12.5% of boys and 10.3% of girls) were obese. Abdominal obesity was documented in 21.1% of students (21.6% of boys and 20.5% of girls). Mean of TyG index in normal-weight, overweight and obese pediatrics was 8.19, 8.25, 8.21 in the 7–12 years’ age group and 8.20, 8.23, 8.20 in the 13–18 years’ age group, respectively, and there was no significant difference in the mean values of TyG index among different weight status (P > 0.05). The general characteristics of participants including anthropometric, clinical, and biochemical measurements according to age and gender are presented in Table 1. In the 7–12 years’ age group, no significant differences were observed between boys and girls except for the mean values of waist, wrist and neck circumferences which were significantly lower in girls (P < 0.01). In the 13–18 years’ age group, the mean values of waist, wrist and neck circumferences, weight, SBP, DBP, total cholesterol and LDL-C were significantly higher in boys compared to girls (P < 0.01). There were no significant differences in the mean values of TyG index and the prevalence of MetS between male and female participants in both age groups.

Table 1 General characteristics of the participants by gender and age group
Full size table

The distribution of the TyG index by age and gender is summarized in Table 2. In the 7–12 years’ age group, the 5th percentile of TyG index was 7.49 in both boys and girls. In the 13–18 years’ age group, the 5th percentile of TyG index was 7.49 and 7.48 in boys and girls, respectively. In the 7–12 years’ age group, the 95th percentile of TyG index was 9.03 and 8.99 in boys and girls, respectively, and in the 13–18 years’ age group, it was 9.02 and 9.04 in boys and girls, respectively.

Table 2 The distribution of the triglyceride–glucose index by age and sex group
Full size table

The ROC curves of the TyG index for MetS according to ATP III criteria are displayed in Fig. 1, stratified by age groups. The area under the curve (AUC) revealed values of 0.83 for the 7–18 years’ age group, 0.82 for the 7–12 years’ age group and 0.83 for the 13–18 years’ age group (Fig. 1).

Fig. 1
figure 1

ROC curves for metabolic syndrome by age. a ROC curves for metabolic syndrome in the 7–18 years’ age group. b ROC curves for metabolic syndrome in the 7–12 years’ age group. c ROC curves for metabolic syndrome in the 13–18 years’ age group. ROC receiver operating characteristic

Full size image

The cut-off values of the TyG index with their corresponding sensitivity, specificity and AUC for metabolic syndrome by gender and age groups were tabulated in Table 3. The cut-off values of the TyG index were 8.33 (sensitivity: 87% and specificity: 68%) in total participants, 8.47 (sensitivity: 77% and specificity: 78%) in boys, and 8.33 (sensitivity: 90% and specificity: 67%) in girls. In the 7–12 and 13–18 years’ age groups, these values were 8.47 (sensitivity: 75% and specificity: 78%) and 8.34 (sensitivity: 89% and specificity: 68%) in total, 8.39 (sensitivity: 80% and specificity: 73%) and 8.47 (sensitivity: 81% and specificity: 78%) in boys, 8.33 (sensitivity: 91% and specificity: 68%) and 8.35 (sensitivity: 88% and specificity: 68%) in girls, respectively.

Table 3 The cut-off values of the triglyceride–glucose index for metabolic syndrome by gender and age group
Full size table

Discussion

In this study we determined the cut-off values of the TyG index, 8.33 in total participants, 8.47 in boys, and 8.33 in girls. The area under the ROC curve (AUC) of TyG index for MetS was 0.83.

MetS is growing in the pediatric age group especially in Middle East countries [17]. The prevalence of MetS in Iranian children and adolescents has been increased [18]. However, no study has examined the cut-off values of surrogate measures of MetS in Iranian pediatrics. Several population-based studies have demonstrated the clinical advantages of the TyG index as a surrogate measure for insulin resistance [19,20,21,22], but since the range of triglyceride and MetS is diverse in different populations, it is essential to validate the new indices in each population. Therefore, in this study, for the first time we tried to indicate the percentile distribution of TyG index, as one of the indirect indices for MetS, and its cut-off values in Iranian children and adolescents with the aim of screening at-risk groups of MetS.

Insulin resistance is strongly associated with MetS, so several population-based studies have used MetS as the indicator of insulin resistance and proposed cut-off values using ROC curves for MetS [23,24,25,26,27]. In this study we conducted ROC curve analysis of the TyG index for MetS according to ATP III criteria for children and adolescents to determine a valid cut-off value. The ROC curve is an appropriate method for determination of an ideal cut-off value by considering both sensitivity (true positives) and specificity (true negatives). It propose a graphical illustration of each cut-off value for any diagnostic test that uses a continuous variable  [28]. Therefore, it is a suitable tool for MetS diagnosis. The area under the ROC curve (AUC) is widely realized as the measure of a diagnostic test’s discriminative power. The maximum value for the AUC is 1.0, but, in general, ROC curves with an AUC more than 0.75 are clinically useful [29]. In current study, the AUC of TyG index for MetS was 0.83 which demonstrating a suitable discriminative power and potential utility for diagnosis of MetS.

The cut-off value of TyG index was 8.33 in total participants which was somehow similar to previous studies [30, 31]. In a study performed on different ethnics of American and Korean adolescents, the cut-off values for TyG index was evaluated according to three metabolic syndrome definitions commonly used in adolescents and ranged from 8.45 to 8.65 in Mexican American, Non-Hispanic White, from 8.35 to 8.55 in Korean adolescents and from 8.15 to 8.35 in Non-Hispanic Black adolescents [32]. Kim et al. compared the cut-off values of two methods of insulin resistance measurements for MetS in 10- to 18-year-old Korean subjects. The cut-off values of the homeostasis model assessment (HOMA-IR) and TyG index were 3.29 and 8.48, respectively, according to the criteria of Cook et al.; 2.96 and 8.41, respectively, according to the criteria of de Ferranti et al.; and 3.54 and 8.66, respectively, according to the criteria of the IDF [33].

It has been demonstrated that TyG index is associated with several parameters of insulin resistance and totally contributes better performance in comparison with the HOMA-IR for screening of insulin resistance [34,35,36]. These results can be justified by the fact that the TyG index consists of two main parts of MetS including fasting level of TG and glucose in its equations; however, HOMA-IR consists of only one component of fasting glucose in MetS.

In the present study for the first time we contribute cut-off values of TyG index for MetS in Iranian children and adolescents by age and gender groups. However, this study has some limitations. First of all, we used only one definition for MetS, but since there is no outcome about the determining of MetS in children and adolescents, it was better to use different definitions for MetS. Moreover, in our survey TyG index was not compared with other direct or indirect methods of insulin resistance measurements.

Conclusion

The area under the ROC curve (AUC) of TyG index for MetS was 0.83 in total participants. The cut-off values of the TyG index according to ATP III criteria were 8.33 (8.21–8.45) in total participants, 8.47 (8.36–8.58) in boys, and 8.33 (8.18–8.48) in girls. These outcomes are clinically helpful for diagnosing MetS in children and adolescents. However, for obtaining more valid cut-off values, further studies with stratification of at-risk groups using the different criteria for MetS and also correlation studies between the TyG index and the gold standard methods for insulin resistance are suggested.