A novel index for diagnosis of type 2 diabetes mellitus: Cholesterol, High density lipoprotein, and Glucose (CHG) index

Abstract

Background

Gold standard methods of type 2 diabetes mellitus are expensive and therefore not practical for large scale studies in low‐income countries. We have investigated the total cholesterol, high density lipoprotein (HDL), and glucose (CHG) index for diagnosis of type 2 diabetes mellitus index which is derived from fasting state. In this study we aimed to compare the accuracy of with CHG index and triglycerides (TG) and glucose levels (TyG) as surrogates of type 2 diabetes mellitus.

Methods

A total of 9,704 individuals between 35 and 65 years of age were recruited as part of the Mashhad stroke and heart atherosclerotic disorder (MASHAD) study. They were categorized into two groups, those with and without type 2 diabetes mellitus. The cut‐off in groups to detection of type 2 diabetes mellitus was fasting blood glucose ≥126 mg/dL in blood sample. Receiver operating characteristic (ROC) curve analysis was used to establish the cut‐off of indices to evaluate the sensitivity and specificity of them.

Results

The best cut‐off of CHG index for diagnosis of type 2 diabetes mellitus was 5.57 which was associated with a sensitivity of 70.38% and specificity of 89.82% values. This was in comparison to the TyG index. LR+ CHG index was 6.91 compared to 3.47 for the TyG index and the AUC of CHG index was 0.864 (0.857, 0.871) compared with 0.825 (0.818, 0.833) for the TyG index. This indicates that the CHG index has a higher efficiency value to diagnose of type 2 diabetes mellitus.

Conclusions

The CHG index could be useful for the detection of type 2 diabetes mellitus.

A novel index for diagnosis of type 2 diabetes mellitus. Comparing the new index with other well‐known indices. The presented index is useful for detecting type 2 diabetes mellitus compared to the TyG index.

INTRODUCTION

Diabetes mellitus is one of the most common chronic diseases and a major cause of death worldwide ¹ . According to the International Diabetes Federation (IDF), the number of people with diabetes in 2021 was more than half a billion people worldwide, and by 2045, the number of these people will increase by 46%. Moreover, in 2021, an estimated 6.7 million adults worldwide died by diabetes‐related complications ² . According to these studies, the quantity/quality of sleep, smoking, dyslipidemia, high blood pressure, ethnicity, family history of diabetes, obesity, and inactivity are most related to the development of diabetes, especially type 2 diabetes mellitus ³ , ⁴ , ⁵ . Insulin resistance (IR) is an insensitive state of peripheral tissues to the effects of insulin. An accumulation of fatty acids in non‐adipose tissues, such as the liver and skeletal muscle, results in alternation of cells function and IR consequently; for example triglyceride content in hepatic tissue characteristics of hepatic IR ⁶ . The tissues that primarily become resistant to insulin are skeletal muscles, adipose tissue, and the liver. Decreased insulin sensitivity is associated with hyperglycemia, high blood pressure, dyslipidemia, hyperuricemia, increased inflammatory markers, endothelial dysfunction, and prothrombotic state which can eventually lead to type 2 diabetes mellitus, metabolic syndrome, and non‐alcoholic fatty liver disease ⁷ , ⁸ , ⁹ , ¹⁰ . The homeostasis model assessment of insulin resistance (HOMA‐IR) index was introduced as a simple, reliable, and repeatable index for use in large epidemiological studies and clinical settings ¹¹ , but, HOMA‐IR test is expensive and less accessible in underdeveloped countries, so, Luis et al. ⁶ introduced a product of fasting triglycerides (TG) and glucose levels (TyG) as a cheap and readily available surrogate for diagnosing IR in apparently healthy individuals. According to the introduction of high accuracy indices to predict or diagnose diseases in order to reduce costs, we aimed to introduce the cholesterol, HDL, glucose index (CHG) as a surrogate for diagnosis of type 2 diabetes mellitus, and the accuracy of this index compared to TyG was investigated.

MATERIALS AND METHODS

Study population

This is a cross‐sectional study based on Mashhad stroke and heart atherosclerotic disorder (MASHAD) cohort study that started in 2010 and continued until 2020. Total of 9,704 men and women at the age of 35–65 year participated in MASHAD study using a cluster random sampling design. Subjects were registered from three districts of Mashhad. Each district was divided into nine areas centered at Mashhad Healthcare Center divisions. The baseline investigation was started in 2010 obtaining a response of 79% after stratified cluster random sampling. After identifying eligible subjects, we arranged a meeting for the physical examination. This cohort study occurred in two phases. In the primary phase, the cross‐sectional anthropometric data for all subjects enrolled in the study was reported. In the next phase in our cohort of healthy individuals, participants were followed to identify the incidence rates of type 2 diabetes mellitus. They were asked to complete the follow‐up questionnaire to identify any changes in their health status and lifestyle every 3 years to document any changes in their lifestyle and health status. Based on the primary analysis, there was no significant difference in terms of age and gender in the participating population. Demographic, anthropometric, and lifestyle data were collected and recorded by a nurse and two healthcare professionals. Health‐related questionnaires include (a) demographic information, physical exercise, tobacco and alcohol consumption, food frequency questionnaire (FFQ), and 24‐h food record (b) anthropometric measurements, cardiovascular risk questionnaire, and (c) anxiety and depression questionnaire. All participants informed consent ¹² .

Laboratory evaluation

After 14 h of overnight fasting, blood samples were taken by venipuncture of the peripheral vein. Serum TG, low‐density lipoprotein cholesterol (LDL‐C), high density lipoprotein cholesterol (HDL‐C), and total cholesterol (TC), were measured as described previously ¹² ; dyslipidemia was defined as TC ≥200 mg/dL, or TG ≥150 mg/dL, or LDL‐C ≥130 mg/dL, or HDL‐C <40 mg/dL (for men) and HDL‐C <50 (for women) ¹² . Type 2 diabetes mellitus was defined as fasting blood glucose (FBG) ≥126 mg/dL or diabetic medication treatments ¹² .

Anthropometric assessments

Height, weight, body mass index (BMI), waist circumference (WC), hip circumference (HC), waist to hip ratio (WHR), and mid‐upper arm circumference (MAC) were measured according to the standardized protocols in all participants, as described previously ¹² . Height (cm), waist circumference (cm), hip circumference (cm), and mid‐upper arm circumference (cm) were measured to the nearest millimeter using a tape measure. Weight (kg) was measured to the nearest 0.1 kg using electronic scales. The BMI was calculated by dividing weight (kg) to height squared (m²) ¹² . According to the report of WHO organization overweight is 25 ≤ BMI < 30 kg/m² and obesity is BMI ≥30 kg/m². Waist‐to‐hip ratio (WHR) was calculated by dividing waist circumference to hip circumference ¹³ . According to the International Diabetes Federation (IDF) cut‐off point of WC ≥80 for women and WC ≥94 for men defined as a high WC ¹² .

Definition

Formula of the indices definition:

• The TyG index was calculated as the Ln [(FBG (mg/dL)*triglyceride (mg/dL))/2] ⁶ ,

• TyG‐BMI = TYG*BMI,

• TYG‐WC = TYG*WC,

• CHG index Ln [TC (mg/dL)*FBG (mg/dL)/2*HDL (mg/dL)],

• CHG‐BMI = CHG*BMI,

• CHG‐WC = CHG*WC.

Statistical analysis

All statistical analyses were performed using SPSS version 20 (SPSS, Chicago, IL, USA). Mean and standard deviation (SD) were reported for quantitative normal parameters. Chi‐square was used to comparison qualitative parameters. Receiver operating characteristic (ROC) curve analysis was performed to set the cut‐off level of TyG, TyG‐BMI, TyG‐WC, CHG, CHG‐ BMI, and CHG‐WC to evaluate sensitivity and specificity of indicators to diagnose of type 2 diabetes mellitus. This issue was done by Yoden index. A P < 0.05 was considered as significant.

RESULTS

A total of 9,704 subjects entered and following data refinement, 9,564 individual records were analyzed. The demographic and clinical characteristics of study population are summarized in Table 1. The mean age was 51.77 ± 7.73 and 47.45 ± 8.17 years in the individuals with and without type 2 diabetes mellitus groups respectively (P < 0.05); of these, there were 3,815 (39.89%) men and 5,749 (60.11%) women. The number of 1,369 (14.32%) and 8,195 (85.68%) were in the individuals with and without type 2 diabetes mellitus groups, respectively. In all variables and parameters were observed significant differences between two groups except sex, smoking status, and HDL (Table 1).

Table 1.

Population characteristics

Variable	With type 2 diabetes mellitus	Without type 2 diabetes mellitus	P‐value
Age (year)	51.77 ± 7.73	47.45 ± 8.17	<0.001
Sex
Male	521 (38.05%)	3,294 (40.20%)	0.135
Female	848 (61.94%)	4,901 (59.80%)
Smoking status
Non‐smoker	916 (66.91%)	5,647 (68.91%)	0.898
Ex‐smoker	178 (13.00%)	763 (9.31%)
Current‐smoker	275 (20.09%)	1785 (21.78%)
WC (cm)	99.27 ± 11.40	94.55 ± 12.02	<0.001
BMI (kg/m2)	28.93 ± 4.62	27.71 ± 4.73	<0.001
FBG (mg/dL)	160.23 ± 68.73	81.56 ± 12.46	<0.001
HDL (mg/dL)	42.21 ± 10.04	42.91 ± 9.90	0.017
TG (mg/dL)	179.94 ± 119.67	136.54 ± 86.09	<0.001
LDL (mg/dL)	120.06 ± 40.24	116.14 ± 34.26	<0.001
TC (mg/dL)	201.42 ± 45.99	189.68 ± 37.67	<0.001
SBP (mmHg)	128.90 ± 19.82	120.54 ± 17.99	<0.001
DBP (mmHg)	81.43 ± 11.20	78.68 ± 11.18	<0.001
TyG	9.33 ± 0.73	8.46 ± 0.56	<0.001
TyG‐BMI	270.07 ± 48.78	235.40 ± 46.63	<0.001
TyG‐WC	927.15 ± 135.29	802.74 ± 127.09	<0.001
CHG	5.86 ± 0.52	5.18 ± 0.30	<0.001
CHG‐BMI	169.42 ± 30.23	144.12 ± 27.64	<0.001
CHG‐WC	491.52 ± 74.39	581.89 ± 84.92	<0.001
METS‐IR	48.09 ± 8.88	42.34 ± 8.61	<0.001

BMI, body mass index; CHG, TC‐HDL‐glucose index; DBP, diastolic blood pressure; FBG, fasting blood glucose; HDL, high density lipoprotein; LDL, low‐density lipoprotein; METS‐IR, metabolic score for insulin resistance; SBP, systolic blood pressure; TC, total cholesterol; TG, triglycerides; TyG, triglyceride‐glucose index; WC, waist circumference.

The sensitivity and specificity of indexes cut‐offs is shown in Table 2. As shown in Table 2 the TyG, TyG‐BMI, TyG‐WC, CHG, CHG‐BMI, and CHG‐WC cut‐offs were significant statistically (P < 0.001). The best cut‐off of CHG value for diagnosis of type 2 diabetes mellitus was 5.57 which indicated lower sensitivity (70.38%) and higher specificity (89.82%) values compared to TyG. Area under curve (AUC) of CHG cut‐off was 0.864 (0.857, 0.871) compared with 0.825 (0.818, 0.833) for TyG value (Figure 1). The sensitivity and specificity CHG‐BMI cut‐off in diagnosis of type 2 diabetes mellitus were 71.91% and 63.87% which was lower compared to CHG. Area under curve (AUC) of CHG‐BMI cut‐off was 0.735 (0.726, 0.744) (Figure 2). The sensitivity and specificity CHG‐WC cut‐off in diagnosis of type 2 diabetes mellitus were 69.87% and 74.39% which were lower compared to CGH. AUC of CHG‐BMI cut‐off was 0.790 (0.781, 0.798) (Figure 3). The positive likelihood ratio (LR+) was 6.91 for CHG compared to 3.47 for TyG value.

Table 2.

Sensitivity and specificity of indices cut‐offs

Variable	Cut‐off value	Sensitivity (%)	Specificity (%)	+LR	−LR	AUC (95% CI)	P‐value*
TyG	8.9 †	72.29	79.15	3.47	0.35	0.825 (0.818, 0.833)	<0.001
TyG‐BMI	244.02 †	69.62	59.18	1.71	0.51	0.698 (0.688, 0.707)	<0.001
TyG‐WC	847.82 †	73.12	64.83	2.08	0.41	0.750 (0.741, 0.759)	<0.001
CHG	5.57 †	70.38	89.82	6.91	0.33	0.864 (0.857, 0.871)	<0.001
CHG‐BMI	152.6 †	71.91	63.87	1.99	0.44	0.735 (0.726, 0.744)	<0.001
CHG‐WC	538.06 †	69.87	74.39	2.73	0.41	0.790 (0.781, 0.798)	<0.001

±LR, positive/negative likelihood ratios; BMI, body mass index; CHG, TC‐HDL‐glucose index; TyG, triglyceride‐glucose index; WC, waist circumference.

^* P‐value is computed based on t‐tests.

^† All indexes were significantly different (P < 0.005).

Figure 1.

Receiver operating characteristic (ROC) curves of cholesterol, high density lipoprotein, and glucose (CHG) and (TyG).

Figure 2.

Receiver operating characteristic (ROC) curves of cholesterol, high density lipoprotein, and glucose‐body mass index (CHG‐BMI) and triglyceride‐glucose index‐BMI (TyG‐BMI).

Figure 3.

Receiver operating characteristic (ROC) curves of cholesterol, high density lipoprotein, and glucose‐waist circumference (CHG‐WC) and triglyceride‐glucose index‐WC (TyG‐WC).

DISCUSSION

In this study we are investigated the accuracy and sensitivity of CHG as a new index in diagnosis of type 2 diabetes mellitus. Our findings showed that CHG had a lower sensitivity and higher specificity values compared to TyG.

According to the prevalence of diabetes in worldwide nations especially developing countries ¹⁴ , ¹⁵ and moreover, expensive and limit‐accessibility of diagnostic tests such as HOMA‐IR to diagnosis of IR in most undeveloped countries, it is necessary to use an alternative, globally available and cheap diagnostic test that derived from fasting status with high sensitivity and specificity for early detection of IR and type 2 diabetes mellitus in undeveloped and low‐income population that leads to decreasing burden of disease via modulation of lifestyle or medications consequently ⁶ .

Obesity is related to changing adipokines secretion and dysfunction of adipose tissue that it leads to different metabolic disorders ¹⁶ . Furthermore, it has been reported that visceral adipose tissue is a better predictor for IR and type 2 diabetes mellitus ¹⁷ . IR is a common metabolic disorder in obese patients which is responsible for progression of dyslipidemia ¹⁶ . Dyslipidemia is a significant relationship with obesity and other complications such as type 2 diabetes mellitus, cardiovascular disease, and cancers. Lipid abnormality is an increasing in type 2 diabetes mellitus patients. Following IR, increasing level of TG, LDL, VLDL, and decreasing level of HDL has been observed in type 2 diabetes mellitus patients. It suggested that IR rather than hyperglycemia involved in dyslipidemia ¹⁸ . Previous studies have shown that higher ratio of TC to HDL is an important predictor of type 2 diabetes mellitus ¹⁹ , ²⁰ , ²¹ . According to the previous findings can be used the ratio TC to HDL as a surrogate of type 2 diabetes mellitus detection. Until now, the exact underlying mechanism of dyslipidemia in diabetes is not well known ¹⁸ . The possible mechanism of diabetic dyslipidemia is reduction of HDL, increased IDL, and VLDL followed by obesity. Relative insulin deficiency consequence of insulin resistance, result in dysregulation in expression and activity of hormone‐sensitive lipase (HSL), apo‐lipoprotein B100 (apo‐B100), microsomal triglyceride transfer protein (MTP), lipoprotein lipase (LPL), and hepatic triglyceride lipase (HTGL) and maybe diabetic dyslipidemia consequently ²² . According to the LR and AUC of CHG‐BMI and CHG‐WC indexes, these have shown the lower correlation with IR compared to CHG index. So, CHG index is an appropriate index to identification of IR compared with other indexes. It is necessary to using available and cheap tools to detection of IR in low‐income countries and poor socio‐economic condition. In this regard, CHG index is a simple and non‐expensive tool that derived from fasting state and it does not require measurement of HOMA‐IR or insulin. It suggested that CHG could be used as a surrogate of IR.

Limitation

The strength point of our study was the assessment of an index with high accuracy. The main limitation of our study was the sensitivity and specificity of CHG index were not determined using HOMA‐IR, HbA1c (hemoglobin A1C), and serum concentration of insulin as a gold standard comparator. However, further studies need to evaluation of CHG reliability in detection of IR compared to HOMA‐IR and gold standard.

DISCLOSURE

The authors declare no conflict of interest.

Approval of the research protocol: All the participants consented to take part in the study by signing written informed consent. The study protocol was reviewed and all methods are approved by the Ethics Committee of Mashhad University of Medical Sciences with approval number IR.MUMS.REC.1386.250 (approved on July 2007). All methods were carried out in accordance with relevant guidelines and regulations.

Informed consent: All the participants consented to take part in the study by signing written informed consent. Also, it is not applicable to the Consent of Image Publication for this manuscript. The figures were designed only in this manuscript for presenting the results of the current paper.

Registry and the registration no. of the study/trial: The study was approved on July 2007 with approval number IR.MUMS.REC.1386.250.

Animal studies: N/A.