Associations between surrogate measures of insulin resistance and waist circumference, cardiovascular risk and the metabolic syndrome across Hispanic and non-Hispanic white populations

Authors


María Teresa Martínez Larrad. E-mail: mmartinezl.hcsc@salud.madrid.org

Abstract

Aims  We evaluated the relations between surrogate indices of insulin resistance and waist circumference, metabolic syndrome and coronary heart disease risk across Hispanic and non-Hispanic white populations.

Methods  The study was a cross-sectional analysis of participants without diabetes in the San Antonio Heart Study, Mexico City Diabetes Study and Spanish Insulin Resistance Study. We evaluated commonly used indices of insulin resistance, including homeostasis model assessment, McAuley’s index, Gutt’s insulin sensitivity index, Avignon’s insulin sensitivity index and the Stumvoll index with and without demographics, the modified Matsuda index and the product of the triglycerides and glucose index. The metabolic syndrome was defined by American Heart Association/National Heart, Lung, and Blood Institute criteria and coronary heart disease risk by Framingham risk scores.

Results  The Stumvoll index with demographics and the Avignon’s insulin sensitivity index had the strongest correlations with waist circumference across populations. The triglycerides and glucose and McAuley’s indices had the most robust correlations with Framingham risk score. The triglycerides and glucose index had the greatest ability to detect individuals with the metabolic syndrome and ≥ 10% coronary heart disease risk. Some indices display significant variability in the strength of the relationship with adiposity and coronary heart disease risk across populations.

Conclusions  There are significant differences between insulin resistance indices regarding the ability to detect the metabolic syndrome and coronary heart disease risk across populations. Studies may need to consider the index of insulin resistance that best suits the objectives.

Introduction

Insulin resistance is a key physiopathological link between obesity, Type 2 diabetes mellitus and the metabolic syndrome [1–5]. We evaluated whether the relation of indices of insulin resistance with obesity, the metabolic syndrome and coronary heart disease risk is similar in the Spanish Insulin Resistance Study, the San Antonio Heart Study and the Mexico City Diabetes Study [6,7].

We examined the following indices of insulin resistance: (1) the homeostasis model assessment (HOMA-IR) [8] for its simplicity; (2) McAuley’s index [9] and the product of triglycerides and fasting glucose [10] for including a measure of dyslipidaemia; (3) Gutt’s index at 0 and 120 min [11] and Avignon’s insulin sensitivity index [12] for being the best predictors of future diabetes in the San Antonio Heart Study and the Mexico City Diabetes Study [13]; (4) Stumvoll indices with demographics and without demographics (0 and 120 min) [14] for being conceived with and without age and BMI, respectively; and (5) a modified Matsuda index [15] measurements at 0 and 120 min.

Research design and methods

The design and methods the Mexico City Diabetes Study, the San Antonio Heart Study and the Spanish Insulin Resistance Study have already been described in detail [7,16]. Briefly, the San Antonio Heart Study sampled both Mexican Americans and non-Hispanic white people from low-, middle- and high-income neighbourhoods of San Antonio, Texas [16]. The Mexico City Diabetes Study enrolled participants from low-income neighbourhoods of Mexico City [16]. The Spanish Insulin Resistance Study randomly sampled subjects from a census of small- and middle-size towns across Spain [7]. The present report includes information on subjects without diabetes aged 35–64 years: 1339 Mexican Americans and 774 non-Hispanic white people from the San Antonio Heart Study, 1897 participants in the Mexico City Diabetes Study and 808 participants in Spanish Insulin Resistance Study. All participants provided written informed consent and protocols were approved by local institutional review boards.

After a 12- to 14-h fast, blood specimens were collected prior to and 2 h after a 75-g oral glucose load. Immunoreactive insulin was measured in the San Antonio Heart Study and the Mexico City Diabetes Study [17] and immunospecific insulin in the Spanish Insulin Resistance Study. Diabetes mellitus and the metabolic syndrome were defined by the 2003 American Diabetes Association plasma glucose criteria [18] and the 2005 American Heart Association/National Heart, Lung, and Blood Institute criteria [19]. The 10-year coronary heart disease risk was computed by Framingham risk equations [20]. Surrogate indices of insulin sensitivity were calculated according to published formulas (see also Supporting Information, Table S1) [9–12,14,15].

Statistical analysis

Statistical analyses were performed using SAS statistical software (version 9.1; SAS Inc., Cary, NC, USA). Spearman’s partial correlation controlling for age and sex was used to measure the strength of the relationship between surrogate indices of insulin resistance and waist circumference (BMI or Framingham risk score). Correlation coefficients relating indices of insulin resistance to waist circumference, BMI and Framingham risk score were compared by the T2 method [21] on a spread sheet (Excel; Microsoft Corp., Redmond, WA, USA). The area under the receiver operating characteristic curve (AUC) was used to assess the ability of indices of insulin resistance to detect participants with ≥ 10% coronary heart disease risk or the metabolic syndrome. Statistical differences between areas under the curve were calculated as described by DeLong et al. [22].

Results

In the three studies, four populations were identified—two of them (Mexican Americans and non-Hispanic white people) in the San Antonio Heart Study. Age- and sex-adjusted mean (± se) BMI was 28.0 ± 0.1 kg/m2 in the Mexico City Diabetes Study, 28.9 ± 0.1 kg/m2 in Mexican Americans and 26.8 ± 0.2 kg/m2 in non-Hispanic white people in the San Antonio Heart Study, and 27.2 ± 0.2 kg/m2 in the Spanish Insulin Resistance Study. Framingham risk score was 9.0 ± 0.1% in the Mexico City Diabetes Study, 7.6 ± 0.1% in Mexican Americans and 6.7 ± 0.2% in non-Hispanic white people in the San Antonio Heart Study, and 7.8 ± 0.2% in the Spanish Insulin Resistance Study. The Supporting Information (Table S2) presents differences in demographic and metabolic characteristics between populations.

Avignon′s insulin sensitivity index and the Stumvoll index with demographics were more strongly related to waist circumference than was the HOMA-IR in all populations (Table 1). Other indices of insulin resistance had stronger, similar or weaker relationships with waist circumference as compared with HOMA-IR. In contrast, McAuley’s index and the triglycerides and glucose index had more robust correlations with Framingham risk score than did HOMA-IR in all four populations.

Table 1. Spearman’s partial correlation coefficients relating surrogate indices of insulin resistance to waist circumference and Framingham risk score
  HOMA-IR*Fasting insulinMcAuley’s indexTriglycerides and glucose indexGutt’s insulin sensitivity index (0, 120 min)Avignon’s insulin sensitivity indexStumvoll index without demographics (0, 120 min)Stumvoll index with demographicsMatsuda index
  1. Spearman’s partial coefficients by controlling for age and sex

  2. *P-value for the test of difference comparing correlation coefficients for homeostasis model assessment of insulin resistance (HOMA-IR) and each surrogate index of insulin resistance [21].

  3. P < 0.05; ‡P < 0.01; §P < 0.001.

Waist circumference
 Mexico (n = 1879)0.410.42†−0.430.27§−0.33§−0.52§−0.41−0.60§−0.41
 Mexican Americans (San Antonio) (n = 1339)0.510.49§−0.45§0.27§−0.41§−0.62§−0.47−0.67§−0.49
 Non-Hispanic white people (San Antonio) (n = 774)0.400.38‡−0.44†0.39−0.36−0.52§−0.38−0.67§−0.39
 Spanish Insulin Resistance Study (n = 808)0.280.27−0.36§0.32−0.31−0.48§−0.34†−0.74§−0.34†
Framingham risk
 Mexico0.210.20†−0.35§0.39§−0.22−0.24‡−0.23−0.22−0.23
 Mexican Americans (San Antonio)0.230.21§−0.38§0.45§−0.27−0.28§−0.26−0.30‡−0.27
 Non-Hispanic white people (San Antonio)0.250.24−0.43§0.48§−0.25−0.27−0.27−0.31−0.27
 Spanish Insulin Resistance Study0.110.11−0.40§0.49§−0.17−0.20§−0.18‡−0.33‡−0.19†

The triglycerides and glucose index had the greatest area under the curve for detecting individuals with ≥ 10% coronary heart disease risk (Fig. 1). The triglycerides and glucose index, McAuley’s index and the Stumvoll index with demographics differed little in the ability to identify individuals with the metabolic syndrome.

Figure 1.

 (a) Mexico City Diabetes Study; (b) San Antonio Mexican Americans; (c) San Antonio non-Hispanic white people; (d) Spanish Insulin Resistance Study. The dependent variable was ≥ 10 coronary heart disease risk in (a) to (d) and the metabolic syndrome in (e) to (h). In each of the panels, the triglycerides and glucose index (TyG) is presented by the wide solid line, the Stumvoll index with demographics (Stv-Dem) by the narrow solid line, McAuley’s index by the dotted line and homeostasis model assessment of insulin resistance (HOMA-IR) by the dashed line. P-values for test of difference between the areas under curve (AUC) of the triglycerides and glucose index and each one of the other surrogate insulin resistance indices using the method developed by DeLong et al. [22]. Ref indicates referent area under the curve; *P < 0.05; †P < 0.01; ‡P < 0.001.

Discussion

Our study has several novel findings: (1) surrogate indices of insulin resistance differ in the strength of the relationship with adiposity and coronary heart disease risk across Hispanic and non-Hispanic white populations; (2) Avignon’s insulin sensitivity index and the Stumvoll index with demographics are the strongest correlates of adiposity, whereas the triglycerides and glucose index is the best correlate of coronary heart disease risk; (3) the triglycerides and glucose index, McAuley’s index and the Stumvoll index with demographics are quite comparable in detecting individuals with the metabolic syndrome.

Despite differences, surrogate indices of insulin resistance are adequate measures for clinical and epidemiological studies [23–25]. Gutt’s index at 0 and 120 min has been shown to have the strongest association with future development of diabetes in the San Antonio Heart Study, the Mexico City Diabetes Study and the Insulin Resistance Atherosclerosis Study [23]. However, indices of insulin resistance may reflect other important domains for diabetes and cardiovascular disease, including adiposity, dyslipidaemia, hepatic glucose production and β-cell dysfunction [26]. We have previously reported that Avignon′s insulin sensitivity index and the Stumvoll index with demographics have strong correlations with adiposity as compared with directly measured insulin sensitivity by clamp studies [27]. In the present study, Avignon′s insulin sensitivity index and the Stumvoll index with demographics also have the most robust relationships with adiposity in all four populations, probably because these indices include a measure of adiposity in their respective formulas. Moreover, the Stumvoll index with demographics has a large area under the curve for detecting individuals with the metabolic syndrome, as previously described [27]. The strength of the relationship between indices of insulin resistance and adiposity may not be the only important factor for identifying subjects with the metabolic syndrome. In this regard, two indices with a more modest relationship with adiposity, McAuley’s index and the triglycerides and glucose index, are comparable with the Stumvoll index with demographics. McAuley’s index and the triglycerides and glucose index may perform well because triglyceride concentration, a measure of dyslipidaemia, is used in their respective formulas.

The triglycerides and glucose index is the surrogate index with the strongest correlation with coronary heart disease risk. As previously reported, McAuley’s index also has a robust relationship with Framingham risk score [27]. Having a measure of dyslipidaemia in the formula may confer advantage to these two indices for detecting individuals with moderately high coronary heart disease risk. A modified Matsuda index derived from measurements at 0 and 120 min is not better than HOMA-IR. However, we cannot determine the performance of the original Matsuda’s formula (only measurements at 0 and 120 min were available in the Mexico City Diabetes Study and the Spanish Insulin Resistance Study). Longitudinal studies are needed to validate these findings.

This study has some limitations. First, the design is cross-sectional therefore we cannot establish cause–effect or temporal relationships. Second, we do not have a direct measure of insulin resistance. However, our results are similar across populations that differ in terms of lifestyle, anthropometric and metabolic characteristics and coronary heart disease risk. Third, insulin assay in the Spanish Insulin Resistance Study was different from that in the Mexico City Diabetes Study and the San Antonio Heart Study. Differences in insulin assay may be responsible for the performance of the insulin resistance indices.

In summary, studies may need to consider the index of insulin resistance that best suits the objective of the study. Avignon’s insulin sensitivity index and the Stumvoll index with demographics may be advantageous in studying adiposity; the triglycerides and glucose index, McAuley’s index and the Stumvoll index with demographics in examining the metabolic syndrome; and the triglycerides and glucose index and McAuley’s index in assessing coronary heart disease risk. Some indices display significant variability in the strength of the relationship with adiposity and coronary heart disease risk across populations. These differences may be in part related to the type of insulin resistance index, population characteristics and insulin assay.

Funding sources

None.

Competing interests

Nothing to declare.

Acknowledgements

We thank Milagros Pérez Barba for technical assistance and CIBER in Diabetes and Associated Metabolic Disorders (ISCIII, Ministerio de Ciencia e Innovación) for financial support.

Ancillary