Comparing Abilities of Different Lipid Measures in Diagnosis of Insulin Resistance: A Survey of Risk Factors of Non-Communicable Diseases (SuRFNCD-2007) Study

Abstract

Background:

Dyslipidemia is a distinguishing characteristic of subjects with insulin resistance. Our aim was to determine the predictive abilities and appropriate cutoffs of different lipid variables for insulin resistance.

Methods:

We used the data of the individuals without history of known diabetes mellitus, aged 25–64 years, from the third national Survey of Risk Factors of Non-Communicable Diseases (SuRFNCD-2007). Fasting blood was collected for lipids, glucose and insulin. Receiver operating characteristics (ROC) curves were used to evaluate the predictive abilities of lipid measures for insulin resistance reflected by a homeostasis model assessment value >1.775. The optimal cutoffs of the selected measures were determined using the maximum Youden index and the point closest to (0,1) on the ROC curve, and were compared with metabolic syndrome definitions to diagnose hyperinsulinemia (fasting insulin ≥10 in men and ≥11 U/L in women).

Results:

Among lipid measures, low-density lipoprotein cholesterol (LDL-C) was the most tenuously related parameter to insulin resistance. On the contrary, triglycerides and triglycerides to high-density lipoprotein cholesterol ratio (triglycerides/HDL-C) had the highest abilities to determine insulin resistance, regardless of sex and body mass index (BMI). The optimal triglycerides and triglycerides/HDL-C cutoffs were 145 mg/dL and 3.75 for men and 109 mg/dL and 3.00 for women, respectively, which had similar abilities to metabolic syndrome definitions in identifying hyperinsulinemia.

Conclusion:

Among lipid measures, triglycerides and the triglycerides/HDL-C ratio have the highest predictive abilities for insulin resistance, and the diagnostic values of their appropriate cutoffs are similar to those of metabolic syndrome definitions.

Introduction

M etabolic syndrome refers to a clustering of diabetes mellitus and coronary artery disease risk factors, including abdominal obesity, glucose intolerance, dyslipidemia, and elevated blood pressure.¹ Insulin resistance has been regarded as the possible underlying pathophysiologic pathway that weaves together components of metabolic syndrome.² However, because the assessment of insulin resistance in clinical settings is not feasible, it is not included as a criterion of metabolic syndrome.³

Dyslipidemia is a well-reported characteristic of individuals with type 2 diabetes.⁴ Triglycerides (TG) are associated with occurrence of type 2 diabetes in healthy men independent of age, body mass index (BMI), family history of diabetes and lifestyle parameters.⁵ The TG to high-density lipoprotein cholesterol ratio (TG/HDL-C) can be considered as a high-specificity index for diagnosis of coronary artery disease⁶ and is independently associated with all-cause mortality.⁷ Recently, we have shown that considering the TG/HDL-C ratio as a component of metabolic syndrome drives the metabolic syndrome definition to load on a single underlying factor, and this ratio deserves further considerations in metabolic derangements.⁸ TG and the TG/HDL-C ratio have been also reported to be closely associated with insulin resistance.^3,9 In this respect, it has been found that insulin resistance is associated with high TG levels independently of BMI and physical endurance capacity.¹⁰ Furthermore, the TG/HDL-C ratio compared with the total cholesterol to HDL-C ratio (TC/HDL-C) is a better indicator of insulin resistance and coronary artery disease.¹¹

Population-based studies comparing the predictive abilities of different lipid measures for insulin resistance are limited. In addition, it is not fully known which lipid profile index can better predict insulin resistance. Also, the appropriate cutoffs of lipid measures and their predictive values remain to be entirely clarified in a large national representative sample. Of note, population-based studies are the only studies that can be used for identifying cut points and determining positive and negative predictive values (PPV and NPV). This study benefits from a national population-based data set. On the other hand, ethnicity may influence the association of lipid profile measures with insulin resistance. For instance, TG and the TG/HDL-C ratio are not reliable markers of insulin resistance in African Americans.^3,12 Therefore, the aim of this study was to evaluate the association of different lipid profile measures with insulin resistance and also to determine the predictive abilities and appropriate cutoffs of lipid measures for insulin resistance in a large representative sample of Iranian men and women stratified by BMI.

Methods

Participants

This study was based on the data of the third national Survey of Risk Factors of Non-Communicable Diseases of Iran conducted in 2007 (SuRFNCD-2007). Details of the survey have been reported elsewhere.¹³ In brief, 4,233 individuals as a representative sample of 25- to 64-year-old Iranian adults were enrolled, using a cluster sampling method. The data were acquired and recorded in standardized questionnaires by trained health-care professionals in accordance with World Health Organization (WHO) recommendations. After excluding pregnant women, subjects with self-reported diabetes, subjects receiving lipid-lowering medications, participants who did not consent to blood sampling, and those with missing data, analyses were performed on the remaining 2,660 participants. The survey received ethics approval from the Center for Disease Control (CDC) of Iran and verbal informed consent was obtained from all participants before study commencement.

Procedures and definitions

The weight and height of participants were determined in light clothing and without shoes. Waist circumference was measured using a constant tension tape at the end of a normal expiration, with arms relaxed at the sides, at mid-distance between iliac crest and rib cage on the mid-axillary line. The participants were instructed to rest for at least 5 min before having their blood pressure checked three times with at least 5-min intervals between each measurement. Venous blood samples were collected following 12 h of overnight fast. TG, total cholesterol (TC), HDL-C, and low-density lipoprotein cholesterol (LDL-C) were measured by enzymatic methods (Parsazmun, Karaj, Iran). Fasting plasma glucose was assessed by the glucose oxidase test (intra- and interassay coefficients of variation less than 2.1 and 2.6, respectively). Insulin was measured by radioimmunoassay, using an antibody with no cross-reaction against pro-insulin and C-peptide (Immunotech, Prague, Czech Republic). The intra- and interassay coefficients of variation were lower than 4.3 and 3.4, respectively.

The homeostasis model assessment of insulin resistance (HOMA-IR) was calculated as fasting insulin (U/L)−fasting plasma glucose (mg/dL)/405, as described by Matthews et al.¹⁴ According to our previous report, individuals with HOMA-IR values ≥1.775 were included in the category of insulin-resistant subjects. Insulin resistance was also defined in accordance with hyperinsulinemia using the 75^th percentile cutoff values of the fasting serum insulin concentration in our sample (i.e., ≥10 and ≥11 U/L for men and women, respectively). BMI was defined as weight (kg)/height(m)². Subjects were categorized into three BMI subgroups: 18.5≤BMI <25 (normal), 25≤BMI<30 (overweight) and BMI≥30 kg/m² (obese).

Metabolic syndrome was defined according to either the Adult Treatment Panel III (ATP III) or modified International Diabetes Federation (IDF) criteria. ATP III criteria allow the diagnosis of metabolic syndrome when three or more of the following conditions are satisfied: Presence of the abdominal obesity (waist circumference ≥102 cm and ≥88 cm in men and women, respectively), elevated blood pressure [systolic blood pressure (SBP) ≥130 mmHg and/or diastolic blood pressure (DBP) ≥85 mmHg), low HDL-C (<40 mg/dL and <50 mg/dL in men and women, respectively), TG ≥150 mg/dL, and fasting plasma glucose ≥100 mg/dL.^1,15 According to the modified IDF criteria, a person with metabolic syndrome must have abdominal obesity (waist circumference ≥90 cm for both men and women) plus any two or more of the following conditions: Elevated blood pressure (see above) or treatment of previously diagnosed hypertension, low HDL-C (see above) or on HDL-C therapy, TG ≥150 mg/dL or on TG therapy and fasting plasma glucose ≥100 mg/dL.^16,17

Statistical analysis

Data were weighted for sex, age, and residential area (urban/rural) strata according to the population of Iran (2006 national census). Complex sample survey analysis was performed based on the clusters of sampling protocol, strata (age groups, sex, and residential area) and the determined weights, using SPSS software (version 16.0; SPSS Inc., Chicago, IL). National estimates, made in the complex survey analysis mode, are expressed as mean±standard error of the mean (SEM). To improve the normality of skewed variables, natural log transformations were used in the subsequent analyses as appropriate. Design-based general linear models were used for the comparison of various lipid measures (i.e., TG, HDL-C, LDL-C, TC, non-HDL-C, TG/HDL-C, LDL-C/HDL-C, TC/HDL-C, and non-HDL-C/HDL-C) and clinical variables across BMI subgroups. Spearman correlation coefficients were estimated for the associations of lipid measures with HOMA-IR. Receiver operating characteristics (ROC) curves were used to estimate the predictive ability of lipid profile variables as well as waist circumference and BMI for insulin resistance defined by HOMA-IR >1.775. Areas under the ROC curves (AUCs) were compared by STATA software (ver 9.0, 2005). To determine the optimal thresholds of the lipid profile, the points on the ROC curves with maximum Youden index [sensitivity−(1−specificity)], and the points with shortest distance value from the point (0,1) [(1−sensitivity)²+(1−specificity)²] were calculated. Furthermore, the abilities of TG and the TG/HDL-C ratio to identify insulin resistance reflected by hyperinsulinemia were compared with conventional definitions of metabolic syndrome. The McNemar test was used to compare the sensitivities and specificities of appropriate cutoffs of TG and TG/HDL-C with metabolic syndrome definitions. PPV, NPV, and negative likelihood ratio (NLR) were also calculated. P<0.05 was considered statistically significant.

Results

Table 1 shows that all lipid profile measures were worsened in proportion to BMI (P<0.001). Additionally, all lipid parameters were also associated with HOMA-IR (Table 2). Although LDL-C was correlated with HOMA-IR, it had the lowest correlation coefficients among lipid parameters. TG, TG/HDL-C, and TC/HDL-C were the most highly related to HOMA-IR. After stratifying by BMI, the correlation coefficients of TG and TG/HDL-C ratio remained in significant levels in different subgroups. HDL-C was inversely correlated with HOMA-IR except for those with BMI<25 kg/m². Generally, in men the associations of lipid profiles with HOMA-IR were weaker in obese subjects, but in women they were loosely associated in those with BMI<25 kg/m².

Table 1.

Principal Characteristics of the Study Population Stratified by Sex and BMI

		All	18.5 ≤BMI <25	25 ≤BMI <30	BMI ≥30
Men		n=1,245	n=530	n=507	n=208
	TG (mg/dL)	158.9±2.9	128.2±3.1	172.8±4.6	194.8±6.8^*
	HDL-C (mg/dL)	34.7±0.3	36.2±0.3	33.6±0.5	34.2±1.4^*
	LDL-C (mg/dL)	129.4±0.9	124.8±1.3	131.5±1.4	137.9±2.5^*
	TC (mg/dL)	194.6±1.0	186.4±1.5	198.0±1.6	209.8±2.4^*
	Non-HDL-C (mg/dL)	159.7±1.0	150.0±1.5	164.6±1.7	175.1±2.6^*
	TG/HDL-C	5.24±0.12	4.07±0.14	5.93±0.21	6.89±0.38^*
	LDL-C/HDL-C	3.92±0.03	3.63±0.05	4.11±0.05	4.32±0.09^*
	TC/HDL-C	6.00±0.05	5.44±0.06	6.34±0.09	6.75±0.14^*
	Non-HDL-C/HDL-C	5.11±0.11	4.44±0.07	5.61±0.27	5.76±0.14^*
	Fasting plasma glucose (mg/dL)	91.8±0.9	87.0±0.8	94.8±1.8	98.2±2.4^*
	Fasting insulin (U/L)	9.08±0.16	7.94±0.19	9.63±0.28	10.7±0.51^*
	HOMA-IR	2.19±0.07	1.77±0.06	2.38±0.10	2.98±0.30^*
	Systolic blood pressure (mmHg)	127.1±0.4	121.7±0.4	128.5±0.6	133.4±1.0^*
	Diastolic blood pressure (mmHg)	81.0±0.3	76.7±0.3	81.7±0.4	85.3±0.7^*
	Waist circumference (cm)	90.1±0.3	80.4±0.3	93.2±0.3	104.9±0.6^*
	BMI (kg/m²)	25.8±0.1	22.0±0.05	27.1±0.04	33.1±0.23^*
Women		n=1,415	n=470	n=515	n=430
	TG (mg/dL)	140.4±2.1	112.7±2.8	143.4±3.6	163.6±3.6^*
	HDL-C (mg/dL)	39.1±0.3	40.6±0.5	39.1±0.4	37.8±0.4^*
	LDL-C (mg/dL)	137.0±1.0	125.9±1.5	138.4±1.6	146.8±1.7^*
	TC (mg/dL)	203.4±1.1	188.6±1.7	205.2±1.8	216.5±1.9^*
	Non-HDL-C (mg/dL)	164.6±1.1	148.7±1.8	166.4±1.9	178.9±1.9^*
	TG/HDL-C	4.14±0.09	3.21±0.13	4.26±0.18	4.92±0.16^*
	LDL-C/HDL-C	3.69±0.03	3.31±0.06	3.70±0.06	4.06±0.06^*
	TC/HDL-C	5.52±0.05	4.95±0.08	5.54±0.08	6.06±0.08^*
	Non-HDL-C/HDL-C	4.54±0.05	3.98±0.08	4.59±0.08	5.06±0.08^*
	Fasting plasma glucose (mg/dL)	90.0±0.6	87.4±0.9	89.9±1.0	92.8±0.9^*
	Fasting insulin (U/L)	9.74±0.14	8.63±0.20	9.90±0.24	10.69±0.29^*
	HOMA-IR	2.21±0.04	1.90±0.07	2.27±0.08	2.46±0.08^*
	Systolic blood pressure (mmHg)	124.2±0.5	115.1±0.6	122±0.7	129.1±0.9^*
	Diastolic blood pressure (mmHg)	82.4±0.3	77.1±0.4	80.9±0.4	85.3±0.4^*
	Waist circumference (cm)	90.3±0.3	77.3±0.3	88.5±0.3	101.2±0.4^*
	BMI (kg/m²)	28.1±0.1	22.1±0.06	27.4±0.05	34.1±0.19^*

Data are mean±standard error of the mean (SEM).

P<0.001 for the comparison of variables across BMI subgroups.

BMI, body mass index; TG, triglycerides; HDL-C, high-density lipoprotein cholesterol; LDL-C, low-density lipoprotein cholesterol; TC, total cholesterol; HOMA-IR, homeostasis model assessment of insulin resistance.

Table 2.

Spearman Correlation Coefficients of Lipid and Anthropometric Measures with HOMA-IR

		All	18.5 ≤BMI <25	25 ≤BMI <30	BMI ≥30
Men
	TG	0.244^***	0.172^***	0.181^***	0.158^*
	HDL-C	−0.122^***	−0.049	−0.112^**	−0.134^*
	LDL-C	0.083^*	0.061	0.040	0.049
	TC	0.128^***	0.092^*	0.006	0.057
	Non-HDL-C	0.148^***	0.100^*	0.029	0.072
	TG/HDL-C	0.236^***	0.176^***	0.180^***	0.152^*
	LDL-C/HDL-C	0.128^***	0.080	0.030	0.044
	TC/HDL-C	0.190^***	0.128^**	0.099^*	0.120
	Non-HDL-C/HDL-C	0.176^***	0.096^*	0.095^*	0.115
	Waist circumference (cm)	0.248^***	0.082^*	0.075	0.176^**
	BMI (kg/m²)	0.278^***	—	—	—
Women
	TG	0.227^***	0.100^*	0.233^***	0.174^***
	HDL-C	−0.154^***	−0.039	−0.184^***	−0.155^***
	LDL-C	0.067^*	−0.012	0.093^*	−0.036
	TC	0.083^**	−0.002	0.090^*	−0.010
	Non-HDL-C	0.119^***	0.006	0.141^**	0.022
	TG/HDL-C	0.230^***	0.107^*	0.255^***	0.188^***
	LDL-C/HDL-C	0.143^***	0.014	0.185^***	0.036
	TC/HDL-C	0.190^***	0.044	0.224^***	0.119^*
	Non-HDL-C/HDL-C	0.186^***	0.038	0.222^***	0.119^*
	Waist circumference (cm)	0.215^***	0.123^**	0.047	0.138^**
	BMI (kg/m²)	0.232^***	—	—	—

P<0.05.

P<0.01.

***

P<0.001.

HOMA-IR, homeostasis model assessment of insulin resistance; TG, triglycerides; HDL-C, high-density lipoprotein cholesterol; LDL-C, low-density lipoprotein cholesterol; TC, total cholesterol; BMI, body mass index.

Among lipid parameters LDL-C was diagnosed as the weakest determinant of insulin resistance (Table 3). The AUCs of TG and the TG/HDL-C ratio were significantly greater than those of other lipid measures (P<0.01). Discarding TG and the TG/HDL-C ratio, TC/HDL-C had the highest AUC among lipid profile measures in men (P<0.05). In women, the AUCs of TC/HDL-C and non-HDL-C/HDL-C were higher than those values of HDL-C, LDL-C, TC, non-HDL-C, and LDL-C/HDL-C ratio (P<0.05). The AUCs of TG and TG/HDL-C were not statistically different from those seen with waist circumference and BMI. After stratifying by BMI, the AUCs of TG and TG/HDL-C ratio remained at significant levels in all BMI subgroups of men and women except for TG in obese men.

Table 3.

Areas Under the ROC Curves (95% Confidence Interval) of Lipid and Anthropometric Measures for Insulin Resistance Reflected by HOMA-IR >1.775

	All	18.5 ≤BMI <25	25 ≤BMI <30	BMI ≥30
Men
TG	0.614 (0.583–0.645)^***	0.624 (0.574–0.675)^***	0.551 (0.500–0.601)^*	0.545 (0.466–0.625)
HDL-C	0.442 (0.410–0.473)^***	0.466 (0.415–0.517)	0.466 (0.415–0.516)	0.433 (0.353–0.513)
LDL-C	0.510 (0.478–0.542)	0.529 (0.477–0.581)	0.452 (0.402–0.502)	0.468 (0.388–0.549)
TC	0.542 (0.510–0.574)^**	0.560 (0.508–0.612)^*	0.474 (0.423–0.524)	0.488 (0.408–0.569)
Non-HDL-C	0.556 (0.525–0.587)^***	0.559 (0.507–0.611)^*	0.487 (0.437–0.536)	0.519 (0.440–0.597)
TG/HDL-C	0.614 (0.583–0.645)^***	0.612 (0.560–0.663)^***	0.566 (0.516–0.617)^**	0.590 (0.507–0.674)^*
LDL-C/HDL-C	0.549 (0.518–0.581)^**	0.542 (0.490–0.594)	0.490 (0.439–0.541)	0.526 (0.445–0.607)
TC/HDL-C	0.596 (0.570–0.622)^***	0.571 (0.519–0.624)^*	0.521 (0.471–0.572)	0.535 (0.455–0.616)
Non-HDL-C/HDL-C	0.578 (0.547–0.609)^***	0.564 (0.512–0.615)^*	0.515 (0.466–0.565)	0.556 (0.479–0.634)
Waist circumference	0.616 (0.585–0.647)^***	0.525 (0.472–0.578)	0.517 (0.466–0.568)	0.600 (0.523–0.678)^*
BMI	0.637 (0.606–0.667)^***	—	—	—
Women
TG	0.622 (0.592–0.651)^***	0.568 (0.512–0.623)^*	0.598 (0.548–0.649)^***	0.596 (0.541–0.651)^***
HDL-C	0.437 (0.407–0.468)^***	0.509 (0.450–0.568)	0.430 (0.379–0.481)^**	0.418 (0.363–0.472)^**
LDL-C	0.539 (0.509–0.570)^*	0.522 (0.466–0.578)	0.556 (0.505–0.608)^*	0.455 (0.399–0.511)
TC	0.560 (0.530–0.591)^***	0.538 (0.483–0.593)	0.568 (0.517–0.619)^**	0.482 (0.426–0.538)
Non-HDL-C	0.572 (0.542–0.602)^***	0.536 (0.481–0.591)	0.580 (0.529–0.630)^**	0.504 (0.448–0.560)
TG/HDL-C	0.620 (0.591–0.650)^***	0.574 (0.502–0.645)^*	0.611 (0.561–0.661)^***	0.605 (0.550–0.659)^***
LDL-C/HDL-C	0.573 (0.543–0.604)^***	0.518 (0.460–0.576)	0.589 (0.538–0.640)^**	0.512 (0.455–0.568)
TC/HDL-C	0.602 (0.572–0.631)^***	0.536 (0.479–0.594)	0.606 (0.556–0.656)^***	0.550 (0.494–0.605)
Non-HDL-C/HDL-C	0.597 (0.568–0.627)^***	0.526 (0.469–0.583)	0.609 (0.559–0.658)^***	0.558 (0.503–0.612)^*
Waist circumference	0.618 (0.588–0.647)^***	0.557 (0.502–0.613)^*	0.532 (0.481–0.584)	0.572 (0.517–0.627)^*
BMI	0.626 (0.596–0.655)^***	—	—	—

^*P<0.05.

P<0.01.

***

P<0.001.

ROC, receiver operating characteristics; HOMA-IR, homeostasis model assessment of insulin resistance; TG, triglycerides; HDL-C, high-density lipoprotein cholesterol; LDL-C, low-density lipoprotein cholesterol; TC, total cholesterol; BMI, body mass index.

Youden index values and the distance from the top left corner of the ROC curves of TG and TG/HDL-C for diagnosis of insulin-resistant subjects (i.e. subjects with HOMA-IR <1.775) are depicted in Fig. 1. The highest Youden index and shortest distance on the ROC curves were yielded at the TG cutoff points of 145 and 109 mg/dL and TG/HDL-C ratios of 3.75 and 3 in men and women, respectively. In men, 53.5% and 51.3% of subjects with TG >145 mg/dL and TG/HDL-C >3.75 were insulin resistant, respectively. These values in women were equal to 61.4% and 61.8% for TG >109 mg/dL and TG/HDL-C >3, respectively.

FIG. 1.

The optimal cutoff points of triglycerides (TG) and tryglycerides to high-density lipoprotein cholesterol (TG/HDL-C) ratio for diagnosis of insulin resistance. (A) Results in men; (B) results in women.

Table 4 compares the abilities of the suggested TG and TG/HDL-C cutoff points with TG >150 mg/dL and metabolic syndrome definitions to predict insulin resistance defined by hyperinsulinemia. McNemar analysis revealed that the sensitivities of identified cutoffs of TG and TG/HDL-C are statistically higher than those seen with metabolic syndrome definitions (P<0.001) except for TG in men, which was not significantly different from IDF criteria (P>0.05). TG had higher and lower specificity compared to the TG/HDL-C ratio in men and women, respectively (P<0.001). In addition, the PPV and NPV of TG and TG/HDL-C were comparable to ATP III and IDF definitions in diagnosis of insulin resistance. In women, the NLR of TG and TG/HDL-C ratio were slightly superior to the metabolic syndrome definitions especially IDF criteria.

Table 4.

Comparison of Optimal TG and TG/HDL-C Ratio Cutoff Points with TG >150 mg/dL, ATP III, and IDF Declarations for Predicting Insulin Resistance Revealed by Hyperinsulinemia

		Sensitivity	Specificity	PPV	NPV	NLR
Men
	TG >145	0.543	0.616	0.320	0.802	0.741
	TG >150	0.520	0.637	0.323	0.799	0.754
	TG/HDL-C >3.75	0.630	0.524	0.306	0.809	0.706
	ATP III	0.418	0.757	0.365	0.796	0.769
	IDF	0.521	0.684	0.354	0.811	0.701
Women
	TG >109	0.668	0.482	0.300	0.813	0.688
	TG >150	0.422	0.704	0.322	0.785	0.821
	TG/HDL-C >3.00	0.637	0.526	0.309	0.813	0.691
	ATP III	0.552	0.611	0.321	0.803	0.734
	IDF	0.464	0.677	0.324	0.791	0.791

TG, triglycerides; HDL-C, high-density liporprotein cholesterol; ATP III, Adult Treatment Panel III; IDF, International Diabetes Federation; PPV, positive predictive value; NPV, negative predictive value; NLR, negative likelihood ratio.

Discussion

This study inspected the ability of lipid profile measures to predict insulin resistance in an Iranian population. All lipid measures were worsened in proportion to BMI. LDL-C was the most tenuously related parameter to insulin resistance. Among lipid measures, only the correlation of TG and the TG/HDL-C ratio with HOMA-IR extended to all BMI subgroups. Generally, the abilities of these two parameters to diagnose insulin-resistant subjects were higher than other lipid measures, regardless of sex and BMI. We also found that women at remarkably lower values of TG and TG/HDL-C are prone to insulin resistance. Furthermore, the diagnostic accuracies of identified TG and TG/HDL-C cutoffs were similar to that of ATP III and modified IDF definitions.

Population studies comparing the association of lipid measures with insulin resistance are rare, and no population-based study has yet reported the appropriate cutoff points of lipid measures for diagnosis of insulin resistance. McLaughin et al. have proposed that overweight subjects with TG ≥130 mg/dL or TG/HDL-C ≥3 are sufficiently insulin resistant to have higher tendency for various unfavorable complications.¹⁸ Another study showed that fasting insulin and TG/HDL-C are similarly associated with the specific measure of insulin-mediated glucose uptake.¹⁹ However, it has been shown that there are interethnic differences in the association of lipid profile measures with insulin resistance.^3,12 For instance, although African Americans compared to Caucasians have lower levels of TG and TG/HDL-C ratio, they have higher tendency toward insulin resistance.^20

–23 Similarly, in an East African population, the associations between these two lipid measures and insulin resistance were weaker in subjects of African origin.²⁴ In contrast, insulin resistance is accompanied by higher TG and lower HDL-C levels independently of waist circumference and BMI in South Asian women.²⁵ Our study provides more insight about the association of lipid measures with insulin resistance in a Middle Eastern population.

Although the TC/HDL-C ratio was generally highly correlated with insulin resistance, this correlation did not reach significant levels in all BMI strata. In this respect, some previous observations have revealed that the TC/HDL-C ratio compared to TG and the TG/HDL-C ratio is only moderately correlated with insulin resistance.^11,19,26 Furthermore, in a recent study, the LDL-C/HDL-C ratio has been suggested as a simple measure to evaluate insulin resistance in nonobese individuals, which is in contrast to our findings.²⁷

Metabolic syndrome definitions are regarded as potent predictors of insulin resistance and new-onset type 2 diabetes.^28

–31 This study for the first time reports the PPV and NPV of appropriate TG and TG/HDL-C cutoff points in a large national representative sample and further shows that these values are similar to those seen with metabolic syndrome definitions. In addition, except for the sensitivity of TG compared to IDF in men, the sensitivities of TG and TG/HDL-C were stronger than metabolic syndrome definitions. On the other hand, these two lipid profile measures are closely related to coronary artery disease and all-cause mortality independent of diabetes status.^6,7 Thus, given the relative simplicity and applicability of measuring TG and the TG/HDL-C ratio compared to the criteria of metabolic syndrome, they can be readily used in clinical settings.

Our results show that among studied variables, waist circumference and BMI are also strongly correlated with insulin resistance. It has been demonstrated that among anthropometric measurements (e.g., BMI, waist-to-hip ratio, and other measures of total body fat) waist circumference is the strongest independent determinant of insulin resistance.³² In type 2 diabetes, waist circumference compared to the metabolic syndrome definition is suggested to be a better predictor of insulin resistance.³³ In this respect, we found that the abilities of TG and TG/HDL-C to determine insulin resistance are similar to the ability of waist circumference.

The major limitation, relevant to our results interpretation, is use of cross-sectional data, which limits our ability to infer a causal relationship between lipid profile measures and insulin resistance. On the contrary, this study also has several strengths. We stratified our analyses by sex and BMI and used a large representative sample of an Iranian population. For the first time in a population-based study, we reported the optimal cutoff points of lipid measures and compared their diagnostic abilities with metabolic syndrome definitions. Moreover, because the Youden index and shortest distance compared to other methods are least dependent on the population case prevalence, they are more suitable for diagnosis of a generalizable cut point.³⁴ However, of note, depending on the clinical situation, different cutoffs along with corresponding sensitivities and specificities would be considered acceptable.

In conclusion, among lipid profile measures, TG and the TG/HDL-C ratio have the highest abilities for diagnosis of insulin-resistant subjects regardless of sex and BMI. LDL-C is the weakest determinant of insulin resistance. In addition, we showed that the diagnostic values of appropriate TG and TG/HDL-C cutoff points for prediction of insulin resistance in an Iranian population are comparable to those of ATP III and IDF definitions. More studies are needed to determine how measurements of lipids can help the clinicians to detect subjects who are prone to insulin resistance.

Footnotes

Author Disclosure Statement

The authors declare no conflicts of interests.

References

Grundy

, Brewer

Jr , Cleeman

et al. Definition of metabolic syndrome: Report of the National Heart, Lung, and Blood Institute/American Heart Association conference on scientific issues related to definition. Circulation, 2004; 109:433–438.

Meigs

. Invited commentary: Insulin resistance syndrome? Syndrome X? Multiple metabolic syndrome? A syndrome at all? Factor analysis reveals patterns in the fabric of correlated metabolic risk factors. Am J Epidemiol, 2000; 152:908–911.

Kim-Dorner

, Deuster

, Zeno

et al.

Should triglycerides and the triglycerides to high-density lipoprotein cholesterol ratio be used as surrogates for insulin resistance?

Metabolism, 2010; 59:299–304.

Mooradian

. Dyslipidemia in type 2 diabetes mellitus. Nat Clin Pract Endocrinol Metab, 2009; 5:150–159.

Tirosh

, Shai

, Bitzur

et al. Changes in triglyceride levels over time and risk of type 2 diabetes in young men. Diabetes Care, 2008; 31:2032–2037.

Bertoluci

, Quadros

, Sarmento-Leite

et al. Insulin resistance and triglyceride/HDLc index are associated with coronary artery disease. Diabetol Metab Syndr, 2010; 2:11.

Bittner

, Johnson

, Zineh

et al.

The triglyceride/high-density lipoprotein cholesterol ratio predicts all-cause mortality in women with suspected myocardial ischemia: A report from the Women's Ischemia Syndrome Evaluation (WISE)

Am Heart J, 2009; 157:548–555.

Esteghamati

, Zandieh

, Khalilzadeh

et al. Clustering of metabolic syndrome components in a Middle Eastern diabetic and non-diabetic population. Diabetol Metab Syndr, 2010; 2:36.

, Ford

, Meng

et al.

Does the association of the triglyceride to high-density lipoprotein cholesterol ratio with fasting serum insulin differ by race/ethnicity?

Cardiovasc Diabetol, 2008; 7:4.

10.

Laws

, Reaven

. Evidence for an independent relationship between insulin resistance and fasting plasma HDL-cholesterol, triglyceride and insulin concentrations. J Intern Med, 1992; 231:25–30.

11.

Kannel

, Vasan

, Keyes

et al.

Usefulness of the triglyceride-high-density lipoprotein versus the cholesterol-high-density lipoprotein ratio for predicting insulin resistance and cardiometabolic risk (from the Framingham Offspring Cohort)

Am J Cardiol, 2008; 101:497–501.

12.

Sumner

, Finley

, Genovese

et al. Fasting triglyceride and the triglyceride-HDL cholesterol ratio are not markers of insulin resistance in African Americans. Arch Intern Med, 2005; 165:1395–1400.

13.

Esteghamati

, Meysamie

, Khalilzadeh

et al. Third national Surveillance of Risk Factors of Non-Communicable Diseases (SuRFNCD-2007) in Iran: Methods and results on prevalence of diabetes, hypertension, obesity, central obesity, and dyslipidemia. BMC Public Health, 2009; 9:167.

14.

Matthews

, Hosker

, Rudenski

et al. Homeostasis model assessment: insulin resistance and beta-cell function from fasting plasma glucose and insulin concentrations in man. Diabetologia, 1985; 28:412–419.

15.

Genuth

, Alberti

, Bennett

et al. Follow-up report on the diagnosis of diabetes mellitus. Diabetes Care, 2003; 26:3160–3167.

16.

International Diabetes Federation. The IDF consensus worldwide definition of the metabolic syndrome [article online] 2006. http://www.idf.org/webdata/docs/IDF_Meta_def_final.pdf/. 2010 August 10.

17.

Esteghamati

, Abbasi

, Rashidi

et al.

Optimal waist circumference cut-offs for the diagnosis of metabolic syndrome in Iranian adults: Results of the third national survey of risk factors of non-communicable diseases (SuRFNCD-2007)

Diabet Med, 2009; 26:745–746.

18.

McLaughlin

, Abbasi

, Cheal

et al. Use of metabolic markers to identify overweight individuals who are insulin resistant. Ann Intern Med, 2003; 139:802–809.

19.

McLaughlin

, Reaven

, Abbasi

et al.

Is there a simple way to identify insulin-resistant individuals at increased risk of cardiovascular disease?

Am J Cardiol, 2005; 96:399–404.

20.

Hyatt

, Phadke

, Hunter

et al. Insulin sensitivity in African-American and white women: association with inflammation. Obesity (Silver Spring), 2009; 17:276–282.

21.

Kim-Dorner

, Simpson-McKenzie

, Poth

et al. Psychological and physiological correlates of insulin resistance at fasting and in response to a meal in African Americans and Whites. Ethn Dis, 2009; 19:104–110.

22.

Despres

, Couillard

, Gagnon

et al. Race, visceral adipose tissue, plasma lipids, and lipoprotein lipase activity in men and women: The Health, Risk Factors, Exercise Training, and Genetics (HERITAGE) family study. Arterioscler Thromb Vasc Biol, 2000; 20:1932–1938.

23.

Sumner

, Vega

, Genovese

et al. Normal triglyceride levels despite insulin resistance in African Americans: role of lipoprotein lipase. Metabolism, 2005; 54:902–909.

24.

Bovet

, Faeh

, Gabriel

, Tappy

. The prediction of insulin resistance with serum triglyceride and high-density lipoprotein cholesterol levels in an East African population. Arch Intern Med, 2006; 166:1236–1237.

25.

Palaniappan

, Kwan

, Abbasi

et al. Lipoprotein abnormalities are associated with insulin resistance in South Asian Indian women. Metabolism, 2007; 56:899–904.

26.

Brehm

, Pfeiler

, Pacini

et al. Relationship between serum lipoprotein ratios and insulin resistance in obesity. Clin Chem, 2004; 50:2316–2322.

27.

Kawamoto

, Tabara

, Kohara

et al. Low-density lipoprotein cholesterol to high-density lipoprotein cholesterol ratio is the best surrogate marker for insulin resistance in non-obese Japanese adults. Lipids Health Dis, 2010; 9:138.

28.

Lorenzo

, Okoloise

, Williams

et al. The metabolic syndrome as predictor of type 2 diabetes: The San Antonio heart study. Diabetes Care, 2003; 26:3153–3159.

29.

Lee

. Usefulness of the metabolic syndrome criteria as predictors of insulin resistance among obese Korean women. Public Health Nutr, 2010; 13:181–186.

30.

Cheal

, Abbasi

, Lamendola

et al. Relationship to insulin resistance of the adult treatment panel III diagnostic criteria for identification of the metabolic syndrome. Diabetes, 2004; 53:1195–1200.

31.

Hanley

, Karter

, Williams

et al. Prediction of type 2 diabetes mellitus with alternative definitions of the metabolic syndrome: The Insulin Resistance Atherosclerosis Study. Circulation, 2005; 112:3713–3721.

32.

Wahrenberg

, Hertel

, Leijonhufvud

et al. Use of waist circumference to predict insulin resistance: Retrospective study. BMJ, 2005; 330:1363–1364.

33.

Blaha

, Gebretsadik

, Shintani

et al. Waist circumference, not the metabolic syndrome, predicts glucose deterioration in type 2 diabetes. Obesity (Silver Spring), 2008; 16:869–874.

34.

Kelly

, Dunstan

, Lloyd

et al. Evaluating cutpoints for the MHI-5 and MCS using the GHQ-12: A comparison of five different methods. BMC Psychiatry, 2008; 8:10.