Abstract
Abstract
Objectives
This study aimed to investigate the association between age-specific and sex-specific continuous metabolic syndrome severity score (cMetS-S) and the risk of developing type 2 diabetes mellitus (T2DM). Additionally, the study aimed to assess the added value of cMetS-S in predicting T2DM compared with traditional MetS criteria.
Design
The study used a longitudinal cohort design, following participants for 18 years.
Setting
The research was conducted within the Tehran Lipid and Glucose Study, a community-based study in Tehran, Iran.
Participants
A total of 6957 participants aged 20–60 years were included in the study.
Interventions/exposures
The cMetS-S of each participant was determined using age-specific and sex-specific equations and Cox proportional hazard regression models were used to analyse the association between cMetS-S and T2DM using continuous and quantile approaches.
Primary and secondary outcome measures
The outcome measure was the association between cMetS-S and the development of T2DM during the 18-year follow-up.
Results
A total of 1124 T2DM cases were recorded over 18 years of follow-up. In the fully adjusted model, a 1-SD increase in the cMetS-S was associated with future T2DM (HR 1.72; 95% CI 1.54 to 1.91). Men and women had HRs of 1.65 (95% CI 1.40 to 1.95) and 1.83 (95% CI 1.59 to 2.10) for T2DM per 1-SD increase in cMetS-S, respectively. Higher cMetS-S was associated with increased risk of diabetes in both prediabetic (HR 1.42;95% CI 1.23 to 1.64) and normoglycaemic individuals (HR 2.11;95% CI 1.76 to 2.54); this association was more significant in normoglycaemic individuals. Unlike the traditional-based MetS definitions, the cMetS-S improved diabetes prediction (p<0.001).
Conclusions
The cMetS-S is strongly associated with future diabetes in prediabetic and normoglycaemic individuals independent of MetS components during a long term. As the relationship between cMetS-S and T2DM is more pronounced in normoglycaemic individuals than in those with pre-diabetes, implementing the evaluation of cMetS-S can serve as an early identification tool for individuals at risk of T2DM prior to the onset of pre-diabetes.
Keywords: Diabetes & endocrinology, EPIDEMIOLOGIC STUDIES, EPIDEMIOLOGY
STRENGTHS AND LIMITATIONS OF THIS STUDY.
We evaluated our newly developed metabolic syndrome severity score (MetS-S) in a large population-based sample of Iranian adults who were followed up for almost two decades.
This is the first study to investigate the application of age-specific and sex-specific continuous MetS-S (cMetS-S) in type 2 diabetes mellitus (T2DM) prediction.
We displayed the association of standardised cMetS-S with T2DM on a continuous scale (for 1-SD increase), allowing for comparison and generalisation of the results from different studies.
We only included non-elderly participants, aged between 20 and 60 years.
This study was conducted solely among the urban residents of Tehran, therefore, its applicability to the rural population remains unknown and can be considered a limitation.
Introduction
Type 2 diabetes mellitus (T2DM) is a major global health issue affecting more than 400 million individuals worldwide. Despite being preventable, T2DM poses a significant burden on the health system and economy, with an annual global health expenditure of US$966 billion.1 2 Metabolic syndrome (MetS), a constellation of five cardiometabolic risk factors, is associated with an approximately fivefold increased risk for T2DM.3 According to the traditional joint interim statement (JIS) MetS criteria,4 an individual with abnormal values for three or more of the following components are categorised as having MetS: abdominal obesity, increased blood pressure, increased triglycerides (TG), low high-density lipoprotein cholesterol (HDL-C) and increased fasting plasma glucose (FPG). This risk factor cluster is seemingly driven by underlying physiological alterations involving insulin resistance, oxidative stress, systemic inflammation, adipocyte dysfunction and cellular function abnormalities.5
The traditional definition, which classifies the individuals as having MetS or not, has limited its clinical utility. This dichotomous definition does not exhibit the extent of abnormality of the MetS components, making it impossible to compare the metabolic abnormality within the populations with or without MetS. In addition, with minimal changes in one or two MetS components’ values, an individual may be identified as having MetS or not.6 The term pre-MetS (preMetS) was developed for these individuals who do not fit the criteria of MetS but have a significant risk for adverse health outcomes.7 8 Last but not least, tracking changes (aggravation or improvement) of MetS is not possible with the traditional criteria and without a definition for severity.
To address these gaps, a few researchers developed continuous MetS severity scores (cMetS-S) using confirmatory factor analysis (CFA) and have shown their association with diabetes.9 10 We developed an age-specific and sex-specific cMetS-S score using CFA to demonstrate MetS severity by considering the weighted contribution of MetS components to MetS in the West Asian population.11 In the current study, we aim to verify the association of our newly developed MetS-S with T2DM in individuals with pre-diabetes and normal glucose metabolism independent of the MetS components and to find whether cMets-S has added value in predicting T2DM over traditional MetS definition.
Methods
Study design and data source
This study was conducted within the framework of the Tehran Lipid and Glucose Study (TLGS), a prospective population-based study performed on a representative sample of residents of district 13 of Tehran with the aim of determining the prevalence of risk factors for non-communicable disease. A total of 15 005 individuals aged 3 years and over who were residence of district No.13 of Tehran and were under the coverage of three medical health centres, were selected using multistage cluster random sampling method (February 1999–August 2001). All members of each family, including those not having risk factors, were invited for baseline measurements and will be followed every 3 years for 18 years. This population was selected through a multistage stratified cluster random sampling technique from the population of district 13 in Tehran. At the time, Tehran was composed of 20 urban districts and made up a population of 6.7 million (Iran National Census 1996). District 13 was chosen mainly because city-wide data showed a high rate of stability in that district. Also, the age distribution in district 13 was representative of the overall population in Tehran. In the current study, out of 15 005 study population, 10 813 participants aged 20–60 years were included.
Participants with diabetes (n=626), cancer (n=39), use of corticosteroid drugs (n=112) and pregnant women (n=78) were excluded at baseline. Moreover, we excluded subjects with missing baseline covariates (n=1704) and missing follow-up (n=1297). Thus, a total of 6957 participants were included and followed by approximately 3-year intervals for incident T2DM for 18 years (online supplemental figure 1).
Clinical variables and data extraction
Information on age, sex, education level, medication use, medical history and family history of T2DM was collected by interview-based questionnaires. Examinations and anthropometric and biochemical measurements were performed by healthcare professionals.12 Using standard protocols, weight, height and waist circumference (WC) measurements were performed to the nearest 0.1 kg, 0.1 cm and 0.1 cm in the clinic, respectively. Weight was measured, while subjects were minimally clothed without shoes, using digital scales (Seca 707, Seca, Hanover, Maryland, USA; range 0.1–150 kg) and recorded to the nearest 100 g. Height was measured in a standing position without shoes, using tape metre while shoulders were in a normal alignment. Body mass index (BMI) was calculated as weight (kg) divided by square of height (m2). WC was measured at umbilical level, using an unstretched tape metre, without any pressure to body surface and HC at the maximal level over light clothing. Systolic blood pressure and diastolic blood pressure (SBP/DBP) were measured twice using a standardised mercury sphygmomanometer on the right arm, with at least a 30 s interval after 15 min resting in the seated position; the mean of the two readings was reported as the individual’s blood pressure.
Venous blood samples for laboratory tests were drawn from individuals after 12–14 hour of overnight fasting between 07:00 and 09:00 hours FPG was measured using an enzymatic colourimetric method with glucose oxidase. TG and HDL-C were assayed using enzymatic colourimetric methods. A 2‐hour postchallenge plasma glucose (2‐h PCPG) test was performed with 75 g glucose for all participants who did not use any glucose-lowering medication. Total cholesterol (TC) was assayed using the enzymatic colourimetric method with cholesterol esterase and cholesterol oxidase. The analyses were carried out using Pars Azmon kits (Pars Azmon, Tehran, Iran) and a Vitalab Selectra E autoanalyzer (Vital Scientific, Spankeren, The Netherlands). The intra-assay and interassay coefficients of variation were both o2.2% for FPG, and 0.5 and 2% for TC, respectively. A detailed description of protocols and laboratory techniques was previously published
Definitions
Smoking status was classified into smokers (use of any tobacco product daily or occasionally) and non-smokers (never and ex-smokers). Education was categorised based on the highest level of education completed: <6 years, 6–12 years and ≥12 years. The family history of diabetes was considered positive when at least one first-degree relative had T2DM. Individuals with less than 600 min per week of activity in the metabolic equivalents of task scale (METS) were described as low physically active. Hypertension was defined as SBP≥140 mm Hg, DBP≥90 mm Hg or antihypertensive medication use. Pre-diabetes status was defined when 100≤FPG<126 mg/dL and 140≤2 h-PCPG<200 mg/dL. T2DM was defined by FPG≥126 mg/dL, 2 h-PCPG≥200 mg/dL or the use of glucose-lowering drugs.
The cMetS-S was calculated for each individual at baseline using age-based and sex-based formulas to indicate MetS severity (online supplemental table S1). These formulas were derived using the CFA approach for the five MetS components (WC, TG, HDL-C, SBP and FPG) by considering the weighted contribution of each component to the latent MetS factor on an age-specific and sex-specific basis.11 The resulting cMetS-S was then standardised at mean=0 with SD=1, reflecting z-scores for each individual, with higher scores showing higher MetS severity.
Patient and public involvement
Patients or the public were not involved in the research’s design, conduct, reporting or dissemination plans.
Statistical analysis
The characteristics of the individuals in the study are presented as mean values±SD for continuous and frequencies with corresponding percentages for categorical variables. An independent t-test was conducted for continuous, and a χ2 test was performed for categorical variables to compare the groups. The incident rate for T2DM was calculated by dividing the total number of cases by the overall follow-up duration in person-years. The association between cMetS-S and T2DM was analysed using continuous and tertile approaches with Cox proportional hazard regression models in total, prediabetic and normoglycaemic subjects. Model 1 was adjusted for age and sex. Model 2 was adjusted for age, sex, family history of diabetes, physical activity and obesity (BMI ≥30 kg/m2). Model 3 included adjustments in model 2 as well as adjustments for the antihypertension and lipid-lowering drugs and model 4 further adjusted for MetS components, including high WC, SBP, TG, FPG and low HDL-C. Restricted cubic spline regression analysis was employed to demonstrate the relationship between cMetS-S and the risk of T2DM and to check the linearity of this relationship. We also compared the model performance of JIS and international diabetes federation (IDF)-based MetS and cMetS-S for T2DM using the Akaike information criterion (AIC) and log-likelihood ratio. The statistical analyses were conducted by using STATA V.14 (Stata) and R V.3.0.3 (R Foundation for Statistical Computing, Vienna, Austria). A two-sided p<0.05 was considered statistically significant.
Development of cMetS-S: cMetS-S had been developed previously using CFA on the five identified MetS components: WC, FPG, SBP, TG and HDL-C to consider the weighted contribution of these components to the unobserved latent variable of MetS.11 Several CFAs were performed on the eligible TLGS participants aged 20–60 years, and the results were presented in total and on age-specific (20–39 years and 40–60 years) and sex-specific basis. All the five MetS components in the models were standardised at mean=0 and SD=1 over the entire sample. One-factor model CFA was performed, and it was assumed that the measurement errors of the five components were not correlated. The factor loadings were indicative of the magnitude of the association between each component and the unobserved latent variable of MetS. The factor loadings >0.3 were considered to show a moderate correlation. Models were developed in the total population and age and sex subgroups with and without the assumption of the equality of factor loadings across the age and sex subgroups, respectively. Factor scores were produced using proper linear combinations of the variables. We calculated factor scores and cMetS-S using linear regression analysis with unstandardised MetS components to allow for potentially higher standardised scores within sex and age groups. The standardised factor coefficients in the final models were applied to calculate the cMetS-S for each individual. For ease of use in clinical settings, the MetS components and their factor coefficients obtained from the CFA have been back transformed so that actual values of MetS components can be placed in the equations. The resulting cMetS-S score can be standardised in each population and interpreted as z-scores (mean=0; SD=1), with greater values representing higher MetS severity. The performance of the overall and age-specific and sex-specific models was compared with various fit indices.
Results
A total of 6957 participants (42.17% male) with a mean age of 37.03±10.67 were included in the current study. Table 1 displays the baseline characteristics of the study population by cMetS-S quartiles. The prevalence of MetS was 31.02% and 26.41% according to the JIS and IDF criteria, respectively, with individuals in the upper quartiles having a higher prevalence of MetS. The mean values of age, BMI, WC, FPG, blood pressure and lipid indices (except HDL) increased on upper cMetS-S quartiles (p<0.001). The proportion of smokers and individuals who used antihypertension and lowering-lipid drugs was also greater towards the fourth cMetS-S quartile. In contrast, the proportion of individuals with low physical activity and a family history of diabetes did not differ across the cMetS-S quartiles. Baseline characteristics of the individuals are presented according to the presence of pre-diabetes (pre-diabetes vs normoglycaemia) in online supplemental table S2.
Table 1. Baseline characteristics of the study population according to continuous metabolic syndrome severity score (cMetS-S) quartiles.
| Characteristics | Overall | cMetS-S quartiles | P value | |||
| Q1 (−2.95, –0.72) | Q2 (−0.73, 0.00) | Q3 (−0.01, 0.71) | Q4 (0.72, 3.98) | |||
| Number of participants | 6957 | 1740 | 1739 | 1739 | 1739 | |
| Age (years) | 37.03±10.67 | 31.06±9.34 | 35.96±10.20 | 39.37±10.03 | 41.72±9.98 | <0.001 |
| Male | 2934 (42.17) | 463 (26.61) | 678 (38.99) | 827 (47.56) | 966 (55.55) | <0.001 |
| Body mass index (kg/m2) | 26.52±4.63 | 23.06±3.67 | 25.80±3.91 | 27.70±3.98 | 29.50±4.30 | <0.001 |
| Waist circumference (cm) | 86.65±11.83 | 75.94±8.75 | 84.34±9.24 | 90.14±9.02 | 96.17±9.69 | <0.001 |
| Education | 0.056 | |||||
| Illiterate/primary school | 4317 (62.05) | 1053 (60.52) | 1063 (61.13) | 1106 (63.60) | 1095 (62.97) | |
| High school | 1574 (22.62) | 383 (22.01) | 400 (23.00) | 388 (22.31) | 403 (23.17) | |
| Higher education | 1066 (15.32) | 304 (17.47) | 276 (15.87) | 245 (14.09) | 241 (13.86) | |
| Smokers | 967 (13.90) | 154 (8.85) | 228.0 (13.11) | 264 (15.18) | 321 (18.46) | <0.001 |
| Low physical activity | 4810 (69.14) | 1188 (68.28) | 1175.0 (67.5) | 1216 (69.93) | 1231 (70.79) | 0.149 |
| DM family history | 604 (10.44) | 141 (9.87) | 149.0 (10.37) | 174 (11.76) | 140 (9.72) | 0.62 |
| Hypertension | 1534 (22.05) | 99 (5.69) | 285.0 (16.39) | 444 (25.53) | 706 (40.60) | <0.001 |
| Dyslipidaemia | 2921 (41.99) | 6 (0.34) | 172.0 (9.89) | 1056 (60.72) | 1687 (97.01) | <0.001 |
| SBP (mm Hg) | 114.46±15.27 | 106.26±11.43 | 112.05±13.09 | 116.39±14.08 | 123.13±16.77 | <0.001 |
| DBP (mm Hg) | 76.30±10.27 | 70.97±8.68 | 74.88±9.32 | 77.72±9.54 | 81.64±10.38 | <0.001 |
| FPG (mg/dL) | 88.99±9.36 | 84.65±7.75 | 87.92±8.55 | 90.03±8.89 | 93.36±9.92 | <0.001 |
| Triglyceride (mg/dL) | 158.38±104.98 | 73.48±20.47 | 113.19±27.74 | 163.10±37.44 | 283.8±128.41 | <0.001 |
| HDL-C (mg/dL) | 42.01±10.79 | 50.37±10.87 | 43.78±9.11 | 39.40±8.52 | 34.47±7.47 | <0.001 |
| Anti-hypertensive drug use | 183 (2.63) | 9 (0.52) | 30.0 (1.73) | 50 (2.88) | 94 (5.41) | <0.001 |
| Lipid-lowering drug use | 78 (1.12) | 3 (0.17) | 5.0 (0.29) | 24 (1.38) | 46 (2.65) | <0.001 |
| MetS (JIS) | 2158 (31.02) | 7 (0.40) | 84.0 (4.83) | 633 (36.40) | 1434 (82.46) | <0.001 |
| MetS (IDF) | 1837 (26.41) | 7 (0.40) | 63.0 (3.62) | 500 (28.75) | 1267 (72.86) | <0.001 |
The categorical and continuous variables were reported as count (percentage) and mean±SD, respectively. Abbreviations: DM, diabetes; SBP, systolic blood pressure; DBP, diastolic blood pressure; FPG, , HDL-C, high-density lipoprotein cholesterol; MetS, metabolic syndrome; JIS, ; IDF, International Diabetes Federation.
DBPdiastolic blood pressureDMdiabetes mellitusFPGfasting plasma glucoseHDL-Chigh-density lipoprotein cholesterolIDFInternational Diabetes FederationJISJoint Interim StatementSBPsystolic blood pressure
A total of 1124 T2DM cases (16.15%) were recorded over 18 years of follow-up. T2DM incidence rates were about 12 per 1000 person-years in both men and women, with the highest cMetS-S quartile having an approximate sevenfold incidence rate of T2DM compared with the lowest quartile (table 2). After controlling for potential confounding variables, the risk of future T2DM increased with an HR of 1.87 (95% CI 1.74 to 2.00) per each SD increase in cMetS-S. Even after adjusting for MetS components (high TG, FPG, WC, SBP and low HDL-C), the cMetS-S was still associated with incident T2DM (HR 1.72; 95% CI 1.54 to 1.91 per 1-SD increase). This association was present in both men (HR 1.65; 95% CI 1.40 to 1.95) and women (HR 1.83; 95% CI 1.59 to 2.10). The results were consistent when assessing this association with the quartile approach, with the risk of T2DM increasing incrementally toward the upper cMetS-S quartile in the fully adjusted model (HR 6.45; 95% CI 4.48 to 9.27).
Table 2. Cox proportional HRs for incidence of type 2 diabetes in total and according to sex.
| Events | IR (95% CI) | Model 1 | Model 2 | Model 3 | Model 4 | |
| HR (95% CI) | HR (95% CI) | HR (95% CI) | HR (95% CI) | |||
| Men (n=2934) | ||||||
| CMetS-S* | 478 | 12.3 (11.3 to 13.5) | 1.89 (1.70 to 2.10) | 1.79 (1.61 to 1.99) | 1.79 (1.61 to 1.99) | 1.65 (1.40 to 1.95) |
| Quartile | ||||||
| Q1 (n=463) | 23 | 3.5 (2.3 to 5.3) | 1.0 (Reference) | 1.0 (Reference) | 1.0 (Reference) | 1.0 (Reference) |
| Q2 (n=678) | 62 | 6.7 (5.2 to 8.5) | 1.95 (1.14 to 3.33) | 1.75 (1.08 to 2.85) | 1.75 (1.08 to 2.85) | 1.62 (0.99 to 2.66) |
| Q3 (n=827) | 128 | 11.7 (9.8 to 13.9) | 3.16 (1.92 to 5.22) | 2.85 (1.81 to 4.48) | 2.84 (1.80 to 4.48) | 2.68 (1.61 to 4.47) |
| Q4 (n=966) | 265 | 21.8 (19.3 to 24.6) | 6.20 (3.83 to 10.04) | 4.86 (3.12 to 7.57) | 4.85 (3.11 to 7.56) | 4.24 (2.33 to 7.74) |
| Women (n=4023) | ||||||
| CMetS-S* | 646 | 12.0 (11.1 to 12.9) | 2.15 (1.96 to 2.35) | 2.00 (1.82 to 2.19) | 2.02 (1.83 to 2.22) | 1.83 (1.59 to 2.10) |
| Quartile | ||||||
| Q1 (n=1277) | 55 | 3.1 (2.4 to 4.0) | 1.0 (Reference) | 1.0 (Reference) | 1.0 (Reference) | 1.0 (Reference) |
| Q2 (n=1061) | 109 | 7.5 (6.2 to 9.1) | 2.27 (1.60 to 3.22) | 2.12 (1.52 to 2.94) | 2.12 (1.52 to 2.94) | 2.21 (1.57 to 3.10) |
| Q3 (n=912) | 202 | 16.7 (14.5 to 19.1) | 4.82 (3.47 to 6.70) | 4.18 (3.05 to 5.72) | 4.18 (3.05 to 5.72) | 4.88 (3.36 to 7.10) |
| Q4 (n=773) | 280 | 29.4 (26.1 to 33.0) | 8.39 (6.03 to 11.66) | 6.86 (4.99 to 9.45) | 6.88 (5.00 to 9.48) | 8.40 (5.31 to 13.28) |
| Total (n=6957) | ||||||
| CMetS-S* | 1124 | 12.1 (11.4 to 12.9) | 1.99 (1.86 to 2.13) | 1.86 (1.73 to 1.99) | 1.87 (1.74 to 2.00) | 1.72 (1.54 to 1.91) |
| Quartile | ||||||
| Q1 (n=1740) | 77 | 3.2 (2.6 to 4.0) | 1.0 (Reference) | 1.0 (Reference) | 1.0 (Reference) | 1.0 (Reference) |
| Q2 (n=1739) | 171 | 7.2 (6.2 to 8.3) | 2.13 (1.59 to 2.86) | 1.96 (1.49 to 2.57) | 1.96 (1.49 to 2.57) | 1.98 (1.50 to 2.61) |
| Q3 (n=1739) | 330 | 14.3 (12.9 to 15.9) | 4.06 (3.09 to 5.34) | 3.55 (2.74 to 4.59) | 3.55 (2.75 to 4.59) | 3.93 (2.91 to 5.31) |
| Q4 (n=1739) | 545 | 25.1 (23.1 to 27.3) | 7.29 (5.57 to 9.54) | 5.81 (4.50 to 7.51) | 5.82 (4.50 to 7.53) | 6.45 (4.48 to 9.27) |
Model 1: adjusted for age and sex.
Model 2: adjusted for age, sex, education, physical activity, family history of diabetes and obesity.
Model 3: adjusted for age, sex, education, physical activity, family history of diabetes, obesity, antihypertensive drug use, and lipid-lowering drug use.
Model 4: adjusted for age, sex, education, physical activity, family history of diabetes, anti-hypertensive drug use, lipid-lowering drug use and individual MetS components, including high waist circumference, high triglyceride, high fasting plasma glucose, high blood pressure and low high-density lipoprotein.
Continuous analysis (per 1--SD increase) of cMetS-S. Model 1: adjusted for age and sex. Model 2: adjusted for age, sex, education, physical activity, family history of diabetes, and obesity. Model 3: adjusted for age, sex, education, physical activity, family history of diabetes, obesity, anti-hypertensive drug use, and lipid-lowering drug use. Model 4: adjusted for age, sex, education, physical activity, family history of diabetes, anti-hypertensive drug use, lipid-lowering drug use, and individual metabolic syndrome components, including high waist circumference, high triglyceride, high , high blood pressure, and low high-density lipoprotein Abbreviations: CMetS-S, continuous metabolic syndrome severity score; HR, hazard ratio; , confidence interval; IR: Incidence rate per person-years.
CMetS-Scontinuous metabolic syndrome severity scoreIRincidence rate
As depicted in table 3, cMetS-S was associated with T2DM in both normoglycaemic and prediabetic participants. A 1-SD increase in cMetS-S increased the risk of T2DM in prediabetic (HR 1.42; 95% CI 1.23 to 1.64) and normoglycaemic (HR 2.11; 95% CI 1.76 to 2.54) participants in the fully adjusted model. This association was consistent when investigating cMetS-S based on quartile groups, with incrementally increased risk for T2DM in higher cMetS-S quartiles; the highest quartile of cMetS-S in normoglycaemic and prediabetic individuals had HRs of 6.12 (95% CI 3.75 to 9.99) and 3.84 (95% CI 2.17 to 6.82), respectively.
Table 3. Cox proportional HRs for incidence of type 2 diabetes in prediabetic and normoglycaemic individuals.
| Events | IR (95% CI) | Model 1 | Model 2 | Model 3 | Model 4 | |
| HR (95% CI) | HR (95% CI) | HR (95% CI) | HR (95% CI) | |||
| Normoglycaemic individuals (n=5638) | ||||||
| CMetS-S* | 533 | 6.9 (6.3 to 7.5) | 2.02 (1.81 to 2.25) | 1.90 (1.71 to 2.11) | 1.90 (1.71 to 2.12) | 2.11 (1.76 to 2.54) |
| Quartile | ||||||
| Q1 (n=1637) | 56 | 2.5 (1.9 to 3.2) | 1.0 (Reference) | 1.0 (Reference) | 1.0 (Reference) | 1.0 (Reference) |
| Q2 (n=1499) | 99 | 4.8 (3.9 to 5.8) | 2.00 (1.39 to 2.86) | 1.82 (1.30 to 2.53) | 1.81 (1.30 to 2.53) | 1.83 (1.30 to 2.58) |
| Q3 (n=1370) | 162 | 8.7 (7.4 to 10.1) | 3.64 (2.59 to 5.12) | 3.18 (2.32 to 4.38) | 3.20 (2.33 to 4.41) | 3.55 (2.38 to 5.29) |
| Q4 (n=1132) | 216 | 14.4 (12.6 to 16.4) | 6.47 (4.61 to 9.08) | 5.16 (3.73 to 7.13) | 5.12 (3.70 to 7.09) | 6.12 (3.75 to 9.99) |
| Prediabetic individuals (n=1319) | ||||||
| CMetS-S* | 591 | 38.1 (35.1 to 41.3) | 1.43 (1.30 to 1.56) | 1.36 (1.24 to 1.49) | 1.37 (1.25 to 1.50) | 1.42 (1.23 to 1.64) |
| Quartile | ||||||
| Q1 (n=103) | 21 | 14.8 (9.7 to 22.7) | 1.0 (Reference) | 1.0 (Reference) | 1.0 (Reference) | 1.0 (Reference) |
| Q2 (n=240) | 72 | 24.0 (19.0 to 30.2) | 1.59 (0.95 to 2.66) | 1.49 (0.92 to 2.43) | 1.50 (0.92 to 2.44) | 1.61 (0.98 to 2.63) |
| Q3 (n=369) | 168 | 38.3 (32.9 to 44.6) | 2.55 (1.58 to 4.12) | 2.30 (1.46 to 3.65) | 2.32 (1.47 to 3.67) | 2.75 (1.67 to 4.52) |
| Q4 (n=607) | 329 | 49.2 (44.1 to 54.8) | 3.36 (2.10 to 5.37) | 2.86 (1.81 to 4.50) | 2.89 (1.84 to 4.56) | 3.84 (2.17 to 6.82) |
Model 1: adjusted for age and sex.
Model 2: adjusted for age, sex, education, physical activity, family history of diabetes and obesity.
Model 3: adjusted for age, sex, education, physical activity, family history of diabetes, obesity, antihypertensive drug use, and lipid-lowering drug use.
Model 4: adjusted for age, sex, education, physical activity, family history of diabetes, antihypertensive drug use, lipid-lowering drug use and individual MetS components, including high waist circumference, high triglyceride, high fasting plasma glucose, high blood pressure and low high-density lipoprotein.
Continuous analysis (per 1-SD increase) of cMetS-S. Model 1: adjusted for age and sex. Model 2: adjusted for age, sex, education, physical activity, family history of diabetes, and obesity. Model 3: adjusted for age, sex, education, physical activity, family history of diabetes, obesity, anti-hypertensive drug use, and lipid-lowering drug use. Model 4: adjusted for age, sex, education, physical activity, family history of diabetes, anti-hypertensive drug use, lipid-lowering drug use, and individual metabolic syndrome components, including high waist circumference, high triglyceride, high , high blood pressure, and low high-density lipoprotein Abbreviations: CMetS-S, continuous metabolic syndrome severity score; HR, hazard ratio; , confidence interval; IR: Incidence rate per person-years.
CMetS-Scontinuous metabolic syndrome severity scoreIRincidence rate
The restricted cubic spline regression graphs were plotted according to the estimated HRs of participants for incident T2DM (figure 1). There was a non-linear positive association between the cMetS-S and future T2DM in all subgroups. The Kaplan-Meier curves showed poorer diabetes-free probability in higher cMetS quartiles over 18 years (online supplemental figure 2).
Figure 1. Restricted cubic spline regression analysis of continuous metabolic syndrome severity score (cMetS-S) with incident type 2 diabetes mellitus. Heavy central lines represent the estimated fully adjusted HRs in the logarithmic scale, with shaded ribbons denoting 95% CIs.
We conducted model fit analyses to determine whether cMetS-S could better predict T2DM than the traditional MetS definitions (JIS and IDF) and MetS components (table 4). Therefore, we created the initial model by including MetS components (high TG, WC, FPG, SBP and low HDL-C) and established T2DM risk factors. We then separately added cMetS-S, JIS and IDF-based MetS to the initial model in models 2, 3 and 4. Unlike the JIS- and IDF-based MetS definitions, adding cMetS-S to the initial model showed significant outperformance by reducing the AIC value by 94.5 on df=1 for diabetes (p<0.001).
Table 4. The added value of continuous metabolic syndrome (MetS) severity score, MetS defined by JIS and IDF compared with when MetS components are in the model.
| Variables in model | −2 log | AIC | |
| Model (1) | Age+sex+education+physical activity+family history of DM+antihypertensive drug use+lipid-lowering drug use+high WC+high BP+high TG+high FPG+low HDL | 17 660.9 | 17 678.9 |
| Model (2) | Model 1+continuous MetS severity score | 17 566.7 | 17 586.7 |
| Model (3) | Model 1+MetS (JIS) | 17 660.9 | 17 680.9 |
| Model (4) | Model 1+MetS (IDF) | 17 659.5 | 17 679.5 |
Interpretation: Model (2) vs model (1): 17 661.2–17 566.7=94.5 on 1 df. This is significant (p<0.001); therefore, there is evidence that adding cMetS-S to the initial prediction model for diabetes improves the model fit. Model (3) vs model (1): 17 660.9–17 660.9<0.01 on 1 df. This is non-significant (p=0.52); therefore, there is evidence that MetS, defined by JIS, does not improve the initial model for diabetes that includes MetS components. Model (4) vs model (1): 17 660.9–17 659.5=1.4 on 1 df. This is non-significant (p=0.74); therefore, there is evidence that MetS, defined by IDF, does not improve the initial model for diabetes that includes MetS components.
AICAkaike information criterionBP, blood pressure; FPG, fasting plasma glucose; HDL, high-density lipoprotein cholesterol; IDF, international diabetes federationJIS, joint interim statement; TG, triglyceride; WC, waist circumference
Discussion
In this study, we, for the first time, evaluated the potential applicability of age-specific and sex-specific MetS-S to predict T2DM incidence. We found that an increase in cMetS-S significantly increases the risk of T2DM even after adjusting for potential confounders independent of the MetS components. The cMetS-S offers additional predictive value for T2DM beyond the MetS components. The association of the cMetS-S and future diabetes was established in men and women and both normoglycaemic and prediabetic groups. The strength of the association was notably higher in normoglycaemic individuals.
The traditional criteria of MetS have limited its clinical use due to the inability to define the extent of MetS severity and the risk of unfavourable health outcomes among individuals who do not have MetS. MetS is an assembly of distinct yet related pathophysiological conditions, and the association of MetS with diabetes can be extended to any individual or specific combinations of the MetS components.7 The term preMetS, defined as the presence of one or two MetS components, is associated with an increased risk of T2DM, which is not foreseen in the traditional criteria. Some researchers argue that the preMetS state is a better stage for initiating medical interventions.13 Multiple phenotypes and indices were created using MetS components, such as the triglyceride-glucose index,14 the hyper triglyceridaemic-waist phenotype15 and a classification based on metabolic health and obesity status.16 The multiplicity of these phenotypes and indices based on MetS components results in confusion since they fail to consider all the components involved in MetS. To overcome the limitations, cMetS-S was formulated as a continuous measure of severity based on CFA to consider the weighted contribution of all the MetS components. The cMetS-S enables clinicians to assess the severity of metabolic abnormality, track its change over time and assess the risk of unfavourable outcomes over time in the population regardless of MetS status.
Our findings demonstrate that the risk of T2DM was increased by 72% per 1-SD increase of the cMetS-S after adjusting for age, sex, education, physical activity, family history of diabetes, antihypertensive and lipid-lowering drug use and even the MetS components. We also conducted the analysis based on quantile classification in which the highest quartile of cMetS-S had a 3.84-fold increased risk for future T2DM. It also appeared that cMetS-S has a greater ability to predict T2DM than the MetS definitions based on JIS and IDF, as it offers additional value beyond the components of MetS. Similarly, Gurka et al found the highest quartile of MetS severity Z-score to be associated with a 2.24–5.30 fold increased risk of T2DM in white and black Americans, respectively, after adjusting for age, sex and MetS components.17 The congruent results from the two studies in the literature indicate that MetS severity defined as z-scores can not only help predict future T2DM independent of the MetS components but also simultaneously show the racial differences in attributed risk of T2DM.
Notably, the current and the American study were different in terms of methodology, adjustments and the studied population. The current study has the advantage of using age and sex-specific MetS severity equations to consider the age and sex variation of MetS component values and their attributed risk for T2DM, while in the study by Gurka et al, MetS severity was solely based on variations by sex. Pathophysiological and epidemiological studies have revealed age-dependent variations in the value and attributed risk of MetS components for diabetes.18,21
According to our results, the relationship between cMetS-S with T2DM was non-significantly higher in women. In the current study, the cMetS-S was linked to a 1.65-fold and 1.83-fold higher risk of T2DM in men and women, respectively. The underlying reason may be the sex variation in the baseline MetS components values, their contribution to MetS22 or their associated risk for diabetes.23 24 The difference in body fat composition, sex hormones and lipid metabolism are some potential explanations.25
Comparison within normoglycaemic and prediabetic subgroups showed the association of cMetS-S with T2DM with higher HRs in normoglycaemic group. Normoglycaemic individuals with a metabolic unhealthy profile may already harbour underlying pathophysiological processes that predispose them to a higher risk of developing diabetes, potentially explaining the stronger association observed in the current study. Studies have shown that insulin resistance can occur before the onset of pre-diabetes and diabetes, and it is often more pronounced in normoglycaemic individuals with metabolic abnormalities. This could lead to a stronger association between metabolic unhealthy profile and diabetes in this population. Research has indicated that individuals with a metabolic unhealthy profile, even in the absence of pre-diabetes, may have higher levels of ectopic fat deposition, such as visceral adipose tissue and hepatic fat content. These ectopic fat depots are known to contribute to insulin resistance and impaired glucose metabolism, potentially increasing the risk of developing diabetes in normoglycaemic individuals. Other contributing factors in normoglycaemic individuals are inflammatory markers, endothelial dysfunction due to dyslipidaemia and hypertension and genetic predispositions.26 27 The recent study and a meta-analysis of 61 studies included 71 196 participants with 11 771 T2D events, 412 metabolites from 37 metabolic pathways associated with T2D were reported among these amino acids, carbohydrates and lipid metabolite are most significant. These metabolites would perform significant mediation effects of BMI, dyslipidaemia, inflammatory markers on T2DM.28
To our knowledge, this is the first study to investigate the association of a cMetS-S with future T2DM in normoglycaemic and prediabetic subgroups. The cMetS-S was predictive of T2DM regardless of glycaemic status. The association between cMetS-S and future T2DM in normoglycaemic individuals was significantly higher (HR 2.11 vs HR 1.42, a 1.5-fold difference) than in individuals with pre-diabetes, suggesting the early-onset increased risk of MetS for T2DM independent of glucose metabolism impairment. Previous reports also suggested an association between MetS (excluding FPG as a component) with T2DM in both subgroups of individuals with normal fasting glucose (NFG) and impaired fasting glucose (IFG), which tended to increase incrementally on an increase in the number of MetS traits. The association was non-significantly higher in the NFG group.24 29 A clinical interpretation of the results from the previous studies is not feasible due to different possible combinations of MetS components. In the current study, we took a step further by investigating the MetS-S instead of the number of MetS components. We also used more specific definitions for glucose impairment by classifying the population into prediabetic and normoglycaemic individuals instead of individuals with NFG and IFG. Our findings suggest that the cMetS-S presents a unified quantitative measure of all MetS components that can identify the risk of future T2DM among prediabetic and normoglycaemic individuals in clinical and epidemiologic settings over long-term follow-up. This association is even significantly stronger in normoglycaemic individuals than in those with pre-diabetes. This score overcomes the limitations of the traditional MetS criteria as it was not defined on the outcome prediction basis (T2DM and cardiovascular disease (CVD)) and instead it was developed based on how MetS components cluster together as MetS by assigning weights to each MetS component using CFA and considering the variations of the MetS components among sex and age groups and the severity of MetS on a continuous scale. The study has some limitations. First, since there are variations in the severity of MetS across different racial groups, it is difficult to generalise the findings to other populations. Also to eliminate potential confounding variables that may be more prevalent among older adults, we confined our study to individuals aged between 20 and 60 years. Additionally, we did not consider the potential impact of lifestyle modifications on the severity of MetS over time, as this would have required additional data and resources.
The newly developed age-specific and sex-specific cMetS-S strongly correlates with future T2DM independent of MetS components over long-term period especially in normoglycaemic individuals, indicating the potential clinical applicability of the cMetS-S in detecting those at risk of diabetes before the onset of pre-diabetes. The cMetS-S also provides additional value over MetS components for diabetes risk prediction contrary to the tradition-based MetS definition. The cMetS-S has significant potentials for widespread use in clinical settings as a tool with significant ability to predict future health outcomes, track the MetS severity changes over time and monitor response to treatment in all individuals regardless of MetS status.
supplementary material
Acknowledgements
The authors wish to thank the study participants and the TLGS research team for their passionate support.
Footnotes
Funding: This study was supported by Grant No 43009357-4 in Shahid Beheshti University of Medical Sciences.
Prepublication history and additional supplemental material for this paper are available online. To view these files, please visit the journal online (https://doi.org/10.1136/bmjopen-2023-078701).
Provenance and peer review: Not commissioned; externally peer reviewed.
Patient consent for publication: Consent obtained directly from patient(s).
Ethics approval: This study conformed to the ethical guidelines of the Declaration of Helsinki. It was approved by the National Research Council of the Islamic Republic of Iran, the Human Research Review Committee of the Endocrine Research Center of Shahid Beheshti University of Medical Sciences, Tehran, Iran (IR.SBMU.ENDOCRINE.REC.1401.081). All participants provided informed written subjects. It was not appropriate or possible to involve patients or the public in the design, or conduct, or reporting, or dissemination plans of our research.
Data availability free text: Datasets generated during and/or analysed during the current study are not publicly available due to institutional policies but are available from the corresponding author on reasonable request.
Patient and public involvement: Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.
Correction notice: This article has been corrected since it was published. Author name ‘Mohammadjavad’ was misspelled.
Contributor Information
Atieh Amouzegar, Email: Amouzegar@endocrine.ac.ir.
Mohammadjavad Honarvar, Email: mj.honarvar19@gmail.com.
Safdar Masoumi, Email: safdar.masoomi@yahoo.com.
Sadaf Agahi, Email: s-agahi@student.tums.ac.ir.
Fereidoun Azizi, Email: azizi@endocrine.ac.ir.
Ladan Mehran, Email: lmehran@endocrine.ac.ir.
Data availability statement
Data are available on reasonable request.
References
- 1.International Diabetes Federation IDF diabetes atlas. Secondary IDF diabetes atlas. 2021. https://www.diabetesatlas.org Available.
- 2.Sun H, Saeedi P, Karuranga S, et al. IDF Diabetes Atlas: Global, regional and country-level diabetes prevalence estimates for 2021 and projections for 2045. Diabetes Res Clin Pract. 2022;183:109119. doi: 10.1016/j.diabres.2021.109119. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Ford ES, Li C, Sattar N. Metabolic syndrome and incident diabetes: current state of the evidence. Diabetes Care. 2008;31:1898–904. doi: 10.2337/dc08-0423. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Alberti K, Eckel RH, Grundy SM, et al. Harmonizing the metabolic syndrome: a joint interim statement of the International Diabetes Federation Task Force on Epidemiology and Prevention; National Heart, Lung, and Blood Institute; American Heart Association; World Heart Federation; International Atherosclerosis Society; and International Association for the Study of Obesity. Circulation. 2009;120:1640–5. doi: 10.1161/CIRCULATIONAHA.109.192644. [DOI] [PubMed] [Google Scholar]
- 5.Fahed G, Aoun L, Bou Zerdan M, et al. Metabolic Syndrome: Updates on Pathophysiology and Management in 2021. Int J Mol Sci. 2022;23:786. doi: 10.3390/ijms23020786. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Gurka MJ, Lilly CL, Oliver MN, et al. An examination of sex and racial/ethnic differences in the metabolic syndrome among adults: a confirmatory factor analysis and a resulting continuous severity score. Metab Clin Exp. 2014;63:218–25. doi: 10.1016/j.metabol.2013.10.006. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Cho A-R, Kwon Y-J, Kim J-K. Pre-Metabolic Syndrome and Incidence of Type 2 Diabetes and Hypertension: From the Korean Genome and Epidemiology Study. J Pers Med. 2021;11:700. doi: 10.3390/jpm11080700. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Yin Q, Chen X, Li L, et al. Apolipoprotein B/apolipoprotein A1 ratio is a good predictive marker of metabolic syndrome and pre-metabolic syndrome in Chinese adolescent women with polycystic ovary syndrome. J Obstet Gynaecol Res. 2013;39:203–9. doi: 10.1111/j.1447-0756.2012.01907.x. [DOI] [PubMed] [Google Scholar]
- 9.DeBoer MD, Gurka MJ, Golden SH, et al. Independent Associations Between Metabolic Syndrome Severity and Future Coronary Heart Disease by Sex and Race. J Am Coll Cardiol. 2017;69:1204–5. doi: 10.1016/j.jacc.2016.10.088. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Low S, Khoo KCJ, Wang J, et al. Development of a metabolic syndrome severity score and its association with incident diabetes in an Asian population-results from a longitudinal cohort in Singapore. Endocrine. 2019;65:73–80. doi: 10.1007/s12020-019-01970-5. [DOI] [PubMed] [Google Scholar]
- 11.Honarvar M, Masoumi S, Mehran L, et al. Development and validation of a continuous metabolic syndrome severity score in the Tehran Lipid and Glucose Study. Sci Rep. 2023;13:7529. doi: 10.1038/s41598-023-33294-w. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Azizi F, Ghanbarian A, Momenan AA, et al. Prevention of non-communicable disease in a population in nutrition transition: Tehran Lipid and Glucose Study phase II. Trials. 2009;10:5. doi: 10.1186/1745-6215-10-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Nichols GA, Moler EJ. Diabetes incidence for all possible combinations of metabolic syndrome components. Diabetes Res Clin Pract. 2010;90:115–21. doi: 10.1016/j.diabres.2010.06.011. [DOI] [PubMed] [Google Scholar]
- 14.Guerrero-Romero F, Simental-Mendía LE, González-Ortiz M, et al. The product of triglycerides and glucose, a simple measure of insulin sensitivity. Comparison with the euglycemic-hyperinsulinemic clamp. J Clin Endocrinol Metab. 2010;95:3347–51. doi: 10.1210/jc.2010-0288. [DOI] [PubMed] [Google Scholar]
- 15.Lemieux I, Pascot A, Couillard C, et al. Hypertriglyceridemic waist: A marker of the atherogenic metabolic triad (hyperinsulinemia; hyperapolipoprotein B; small, dense LDL) in men? Circulation. 2000;102:179–84. doi: 10.1161/01.cir.102.2.179. [DOI] [PubMed] [Google Scholar]
- 16.Blüher M. Metabolically Healthy Obesity. Endocr Rev. 2020;41:bnaa004. doi: 10.1210/endrev/bnaa004. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Gurka MJ, Golden SH, Musani SK, et al. Independent associations between a metabolic syndrome severity score and future diabetes by sex and race: the Atherosclerosis Risk In Communities Study and Jackson Heart Study. Diabetologia. 2017;60:1261–70. doi: 10.1007/s00125-017-4267-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Lee M-K, Han K, Kwon H-S. Age-specific diabetes risk by the number of metabolic syndrome components: a Korean nationwide cohort study. Diabetol Metab Syndr. 2019;11:112. doi: 10.1186/s13098-019-0509-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Wang T, Zhao Z, Wang G, et al. Age-related disparities in diabetes risk attributable to modifiable risk factor profiles in Chinese adults: a nationwide, population-based, cohort study. Lancet Healthy Longev. 2021;2:e618–28. doi: 10.1016/S2666-7568(21)00177-X. [DOI] [PubMed] [Google Scholar]
- 20.Kuk JL, Ardern CI. Age and sex differences in the clustering of metabolic syndrome factors: association with mortality risk. Diabetes Care. 2010;33:2457–61. doi: 10.2337/dc10-0942. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Carr MC. The emergence of the metabolic syndrome with menopause. J Clin Endocrinol Metab. 2003;88:2404–11. doi: 10.1210/jc.2003-030242. [DOI] [PubMed] [Google Scholar]
- 22.Rajca A, Wojciechowska A, Śmigielski W, et al. Increase in the prevalence of metabolic syndrome in Poland: comparison of the results of the WOBASZ (2003-2005) and WOBASZ II (2013-2014) studies. Pol Arch Intern Med. 2021;131:520–6. doi: 10.20452/pamw.15975. [DOI] [PubMed] [Google Scholar]
- 23.Onat A, Hergenç G, Keleş I, et al. Sex difference in development of diabetes and cardiovascular disease on the way from obesity and metabolic syndrome. Metab Clin Exp. 2005;54:800–8. doi: 10.1016/j.metabol.2005.01.025. [DOI] [PubMed] [Google Scholar]
- 24.Mukai N, Doi Y, Ninomiya T, et al. Impact of metabolic syndrome compared with impaired fasting glucose on the development of type 2 diabetes in a general Japanese population: the Hisayama study. Diabetes Care. 2009;32:2288–93. doi: 10.2337/dc09-0896. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Tramunt B, Smati S, Grandgeorge N, et al. Sex differences in metabolic regulation and diabetes susceptibility. Diabetologia. 2020;63:453–61. doi: 10.1007/s00125-019-05040-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Koay YC, Coster ACF, Chen DL, et al. Metabolomics and Lipidomics Signatures of Insulin Resistance and Abdominal Fat Depots in People Living with Obesity. Metabolites. 2022;12:1272. doi: 10.3390/metabo12121272. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Dong Q, Sidra S, Gieger C, et al. Metabolic Signatures Elucidate the Effect of Body Mass Index on Type 2 Diabetes. Metabolites. 2023;13:227. doi: 10.3390/metabo13020227. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Morze J, Wittenbecher C, Schwingshackl L, et al. Metabolomics and Type 2 Diabetes Risk: An Updated Systematic Review and Meta-analysis of Prospective Cohort Studies. Diabetes Care. 2022;45:1013–24. doi: 10.2337/dc21-1705. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Kurotani K, Miyamoto T, Kochi T, et al. Metabolic syndrome components and diabetes incidence according to the presence or absence of impaired fasting glucose: The Japan Epidemiology Collaboration on Occupational Health Study. J Epidemiol. 2017;27:408–12. doi: 10.1016/j.je.2016.08.015. [DOI] [PMC free article] [PubMed] [Google Scholar]