Uric acid best predicts metabolically unhealthy obesity with increased cardiovascular risk in youth and adults§

Authors


  • Disclosures: All authors have read the journal's policy on disclosure of potential conflicts of interest. The authors have no potential conflicts of interest to declare.

  • Funding agencies: This work was funded by the “Zukunftsfond Steiermark” Project “STYJOBS-Extension”. Furthermore, the Austrian Nano-Initiative co-financed this work as part of the Nano-Health project (no. 0200), the sub-project NANO-PLAQUE being financed by the Austrian FWF (Fonds zur Förderung der Wissenschaftlichen Forschung, Project no. N212-NAN). Financial support of a technician and consumables associated with the set up of the STYJOBS EDECTA Cohort.

  • §

    Abbreviations: BMI, body mass index; CHE, cholinesterase; CI, Confidence interval; AST/GOT, Aspartat-Aminotransferase; ALT/GPT, Alanin-Aminotransferase; gamma-GT, Gammaglutaryltransferase; CCA, Common carotid arteries; CT, Computed tomography; CRP, C-reactive protein; HDL, High dense lipoprotein; HOMA-IR, Homeostatic model assessment – insulin resistance; IL-6, Interleukin-6; IMT, Intima media thickness; MetS, Metabolic syndrome; oxLDL, Oxidized low dense lipoprotein; SAT, Subcutaneous adipose tissue; US-CRP, Ultra sensitive c-reactive protein; UA, Uric acid; VAT, Visceral adipose tissue; WHR, Waist-to-Height Ratio.

Abstract

Objective:

The obesity prevalence is growing worldwide and largely responsible for cardiovascular disease, the most common cause of death in the western world. The rationale of this study was to distinguish metabolically healthy from unhealthy overweight/obese young and adult patients as compared to healthy normal weight age matched controls by an extensive anthropometric, laboratory, and sonographic vascular assessment.

Design and Methods:

Three hundred fifty five young [8 to < 18 years, 299 overweight/obese(ow/ob), 56 normal weight (nw)] and 354 adult [>18-60 years, 175 (ow/ob), 179 nw)] participants of the STYJOBS/EDECTA (STYrian Juvenile Obesity Study/Early DEteCTion of Atherosclerosis) cohort were analyzed. STYJOBS/EDECTA (NCT00482924) is a crossectional study to investigate metabolic/cardiovascular risk profiles in normal and ow/ob people free of disease except metabolic syndrome (MetS).

Results:

From 299 young ow/ob subjects (8-< 18 years), 108 (36%), and from 175 adult ow/ob subjects (>18-60 years), 79 (45%) had positive criteria for MetS. In both age groups, prevalence of MetS was greater among males. Overweight/obese subjects were divided into “healthy” (no MetS criterion except anthropometry fulfilled) and “unhealthy” (MetS positive). Although percentage body fat did not differ between “healthy” and “unhealthy” ow/ob, nuchal and visceral fat were significantly greater in the “unhealthy” group which had also significantly higher values of carotid intima media thickness (IMT). With MetS as the dependent variable, two logistic regressions including juveniles <18 years or adults >18 years were performed. The potential predictor variables selected with the exception of age and gender by t test comparisons included IMT, ultrasensitive c-reactive protein (US-CRP), IL-6, malondialdehyde (MDA), oxidized LDL, leptin, adiponectin, uric acid (UA), aldosterone, cortisol, transaminases, fibrinogen. In both groups, uric acid and in adults only, leptin and adiponectin, turned out as the best predictor.

Conclusion:

Serum levels of UA are a significant predictor of unhealthy obesity in juveniles and adults.

Introduction

Obesity is associated with reduced life expectancy largely because obese individuals are at increased risk of cardiovascular disease (CVD) (1, 2). However, an intriguing subgroup of obese individuals, referred to as having a metabolically healthy phenotype [Metabollically Healthy Obese (MHO)], remain resilient from developing CVD. Prevalence of MHO is surprisingly high. Primeau et al. reviewed 15 studies that report MHO prevalence ranging from 18 to 44%, depending on definitions used for obesity and metabolic health, as well as the different gender/ethnic groups studied (3). A recently published large Finnish cross sectional epidemiologic study (4) reported 9.2% of obese men and 16.4% obese women met criteria for MHO using Alberti et al. (5) criteria for Metabolic Syndrome (MetS). An Italian study which focused on insulin sensitivity identified 11% (men and women) as MHO phenotypes (6). Birth weight, ethnic background, age, and physical activity have been identified as factors that influence prevalence of MHO phenotype (3). In a 15-year follow-up study, subjects with MHO phenotype did not show increased all cause mortality, cancer and CVD risks (6). We completed an additional, comprehensive assessment to further characterize the metabolic profile of these two groups, and investigate if any obesity related anthropometric or laboratory parameters reliably differentiate MHO from non-MHO individuals.

Methods

Study participants are from the prospective, observational study STYJOBS/EDECTA (STYrian Juvenile Obesity Study/Early DEteCTion of Atherosclerosis; ClinicalTrials.gov Identifier NCT00482924), which investigates metabolic/cardiovascular parameters in obese individuals who were free of chronic health conditions except MetS. We included individuals aged between 8 and 60 years old (Tables 1 and 2). The inclusion criterion for the ow/ob youth (<18 years old) was BMI ≥ 85th-94th (obese > 95th) percentile, and for ow/ob adults BMI > 25 < 29.9 kg m−2 (obese > 30 kg m−2). Juvenile controls had a BMI between 5th and 84.9th percentile; adult controls a BMI between 18.5 and 24.9, respectively. The analysis of normal weighted controls was included in the design of this study to provide a reference for the metabolically healthy overweight/obese.

Table 1. Baseline characteristics, anthropometric and laboratory data not included in the metabolic syndrome algorithm of normal weighted, metabolically healthy (MHO) overweight/obese (i.e., no lab criterium of MS definition fulfilled) and “unhealthy” (nonMHO) overweight/obese (i.e. ≥3 criteria IDF criteria of MS fulfilled) subjects aged ≤ 18 years
 Normal weighted controls“Healthy” overweight/obese MHO phenotype“Unhealthy” overweight/obese nonMHO phenotype
n = 56n = 48n = 108
  • If normally distributed, results are expressed as mean ± SD, otherwise as median, 25th and 75th percentile. Differences between groups are analyzed by one-way ANOVA including Bonferroni post-hoc comparison.

  • The symbols (*) are indicators of significance between controls, MHO and nonMHO subjects; (+) indicate the significance between subjects with MHO and nonMHO phenotypes. Significance levels are given as

  • *

    P < 0.05,

  • **

    P < 0.01, and

  • ***

    P < 0.001, respectively.

  • IDF, international diabetes federation; BMI, body mass index; carotid IMT, SAT, subcutaneous adipose tissue; carotid intima-media thickness; US-CRP, ultra-sensitive C-reactive protein; IL-6, interleukin-6; ETP, endogenous thrombin potential.

Baseline characteristics, anthropometry
Age (years)14.0 ± 3.011.7 ± 3.312.8 ± 2.7
Females/males29/2730/1847/61
Birthweight (g)3252 ± 4373426 ± 443479.2 ± 582
BMI (kg m2)17.9 ± 2.628.0 ± 5.0***31.4 (28.8–36.5)***+++
BMI mother (kg m2)23.5 ± 3.625.7 (22.3–33.9)***28.8 ± 7.1***
BMI father (kg m2)26.4 ± 2.826.9 (24.6–32.1)*29.1 ± 4.7***
Lipometry body fat (%)18.8 ± 9.337.3 ± 9.7***38.8 ± 10.4***
Lipometry SAT thickness neck (mm)2.0 (1.3–4.5)11.5 (7.8–16.6)***16 (12.4–19.6)***+++
Lipometry visceral adipose tissue (kg)1.5 (1.0–1.9)3.6 (2.9–5.3)***5.9 (4.8–7.6)***+++
Carotid IMT (cm)0.05 (0.04–0.06)0.07 (0.06–0.07)***0.08 (0.07–0.08)***+
Organised physical activity50%28.2%*26.6%
Nonorganized physical activity65.4%53.2%46.6%
Laboratory parameter
Homa Index1.2 (0.7–2.3)2.1 (1.2–3.4)4.5 (2.5–8.2)***++
US-CRP (mg l−1)0.4 (0.2–0.5)1.7 (0.6–3.4)***2.7 (1.1–5.2)***
IL-6 (pg ml−1)1.9 (1.5–2.5)3.0 (1.8–5.4)3.8 (2.4–5.3)***
LDL-cholesterol (mg dl−1)99.0 (84–119)103 (89–123)104 (89–116)
Oxidized LDL (mmol l−1)34.4 ± 9.441.3 ± 15.0*46.3 ± 16.1***
Malondialdehyde (μmol l−1)0.4 (0.3–0.6)0.4 (0.3–0.5)0.6 (0.4–0.8)*++
Myeloperoxidase (ng ml−1)13.7 (10–19)16.3 (12–29)17 (13–33)
Leptin (ng ml−1)1.8 (0.7–8.2)33.1 ± 16.5***37.7 ± 16.0***
Adiponectin (μg ml−1)14.0 ± 6.811.7 ± 5.99.5 ± 5.0***
Uric acid (mg dl−1)4.5 ± 1.05.1 ± 1.1*6.1 ± 1.5***+++
Creatinin (mg dl−1)0.7 (0.6–0.8)0.6 (0.55–0.72)0.7 (0.6–0.8)
Aldosteron (ng l−1)76.2 (65–114)80.6 (55–120)78.9 (48.2–129.4)
Renin (ng l−1)35.2 ± 30.132.7 ± 13.225.3 ± 8.1
Cortisol (ng ml−1)181.1 ± 173.373.4 (46–103)***70.0 (49.0–124.6)***
AST/GOT (U l−1)26 (22–35)29 (25–34)29 (24–35)
ALT/GPT (U l−1)17 (12–24)21 (17–32)25 (20–36)**
gammaGT (U l−1)11 (9–15)14 (11–19)19 (16–25)***++
Cholinesterase (U l−1)8102 ± 17289996 (8985–11191)***10172 (8948–11482)***
Alcaline phosphatase (U l−1)151.2 ± 91.7247 (112–305)**210 (140–299)***
Fibrinogen (mg dl−1)301.4 ± 101.6334 ± 88.6370.6 ± 66.5
ETP (extinction)327.2 ± 7.4369 ± 39386.4 ± 54.6
Table 2. Baseline characteristics, anthropometric and laboratory data not included in the metabolic syndrome algorithm of normal weighted, metabolically healthy (MHO) overweight/obese (i.e., no lab criterium of MS definition fulfilled) and “unhealthy” (nonMHO) overweight/obese (i.e., ≥3 criteria IDF criteria of MS fulfilled) subjects aged > 18 years
 Normal weighted controls“Healthy” overweight/obese MHO phenotype“Unhealthy” overweight/obese nonMHO phenotype
n = 176n = 31n = 79
  • If normally distributed, results are expressed as mean ± SD, otherwise as median, 25th and 75th percentile. Differences between groups are analysed by one-way ANOVA including Bonferroni post-hoc comparison.

  • The symbols (*) are indicators of significance between controls, MHO and nonMHO subjects; (+) indicate the significance between subjects with MHO and nonMHO phenotypes. Significance levels are given as

  • *

    P < 0.05,

  • **

    P < 0.01, and

  • ***

    P < 0.001, respectively.

  • IDF, international diabetes federation; BMI, body mass index; carotid IMT, SAT, subcutaneous adipose tissue; carotid intima-media thickness; US-CRP, ultra-sensitive C-reactive protein; IL-6, interleukin-6; ETP, endogenous thrombin potential.

Baseline characteristics, anthropometry
Age, age range (years)32.5 ± 10.5 (18.1–60)35.7 ± 12.9 (18.1–60)37.7 ± 10.5 (18.1–60)
Females/males125/5122/934/45
Birthweight (g)3352 ± 5203368 ± 3843418 ± 428
BMI (kg m2)21.6 ± 1.927.8 (26.7–33.4)***33.6 (29.9–39.4)***+++
BMI mother (kg m2)24.9 ± 3.926.0 ± 5.626.3 (24–30)***
BMI father (kg m2)26.3 ± 3.427.0 ± 3.529.1 ± 5.4***
Lipometry body fat (%)19.7 ± 7.227.5 ± 8.1***24.5 ± 5.3***
Lipometry SAT thickness neck (mm)3.3 (2.2–5.1)7.3 (5.2–11.7)***9.5 (7.3–12.2)***++
Lipometry visceral adipose tissue (kg)2.2 (1.7–2.7)4.3 (3.3–5.5)***6.2 (5.2–7.3)***+++
Carotid IMT (cm)0.06 (0.05–0.07)0.06 (0.06–0.07)0.07 (0.07–0.08)***++
Organised physical activity57.3%48.3%26.6%*
Nonorganized physical activity48.3%51.6%54.4%
Laboratory parameter
Homa index1.3 (0.7–1.9)1.8 (0.9–2.9)3.4 (2.0–5.0)***+++
US-CRP (mg l−1)0.8 (0.5–1.9)2.6 (0.9–4.1)***3.5 (1.6–7.4)***
IL-6 (pg ml−1)1.5 (1.5–2.2)2.2 (1.5–3.3)2.7 (1.6–4.5)***
LDL-cholesterol (mg dl−1)96.0 (80–125)121 (95–147)§§125 (96–142)
Oxidized LDL (mmol l−1)44.1 ± 16.953.0 ± 17.769.0 ± 25.1***+++
Malondialdehyde (μmol l−1)0.5 (0.3–0.5)0.4 (0.4–0.5)0.4 (0.4–0.6)
Myeloperoxidase (ng ml−1)11.5 (9–16)13.2 (9.7–20)14 (10–26)**
Leptin (ng ml−1)8.4 ± 7.926.2 ± 16.7***29.2 ± 17.4***
Adiponectin (μg ml−1)12.0 ± 5.611.1 ± 3.27.3 ± 3.2***++
Uric acid (mg dl−1)4.5 ± 1.15.2 ± 1.2***6.2 ± 1.3***+++
Creatinin (mg dl−1)0.84 (0.76–0.94)0.82 (0.70–0.92)0.9 (0.77–1.02)*+
Aldosteron (ng l−1)88.3 (53–149)80.4 (65–116)72.7 (51.8–95.3)*
Renin (ng l−1)21.4 ± 30.915.6 ± 11.011.1 (6.7–21.5)
Cortisol (ng ml−1)181.0 (125–293)165.4 (126–270)158.6 (96.4–220.5)**
AST/GOT (U l−1)25 (21–29)26 (22–31)30 (25–35)*
ALT/GPT (U l−1)19 (15–24)25 (19–34)**35 (24–53)***++
gammaGT (U l−1)17 (13–22)18 (13–25)33 (25–57)***+
Cholinesterase (U l−1)7248 ± 14127737 (6926–8769)9651 (8422–10747)***+++
Alcaline phosphatase (U l−1)57 ± 1761 ± 1669.6 ± 18
Fibrinogen (mg dl−1)282 ± 60338.8 ± 63.8**353.7 ± 81.6***
ETP (extinction)383 ± 54418.5 ± 53.9*430.1 ± 53.7***

Participants were excluded if they received medications or hormone replacement therapy. Written consent was obtained for all adult participants, and parental consent was obtained for youth.

Procedures

Standard anthropometric data (height, weight, waist-, hip-circumference, waist to hip-, waist to height-ratio) were obtained from each subject as described elsewhere (7). Waist-circumference was measured midway between the lower costal margin and the iliac crest; hip-circumference at the maximum circumference over the buttocks (7), and resting blood pressure in a sitting position in the right arm at the end of the physical examination.

Determination of metabolic syndrome

For youth MetS was determined according to the updated criteria of Alberti et al. (8) and for adults according to Zimmet et al. (9); i.e., waist circumference ≥ 90th percentile (for youth); male ≥ 94 cm, female ≥ 80 cm (for adults); and two of the four criteria fulfilled: fasting glucose ≥ 100 mg dl−1 (5.6 mmol l−1); triglycerides ≥ 150 mg dl−1 (1.7 mmol l−1); HDL-Cholesterol < 40 mg dl−1 (1.03 mmol l−1), males; < 50 mg dl−1 (1.29 mmol l−1), females; blood pressure ≥ 130 systolic or ≥ 85 mmHg diastolic.

Extended anthropometry (Lipometer®)

The subcutaneous adipose tissue (SAT) distribution was analyzed by a patented optical device (EU Pat.Nr. 0516251) on 15 anatomically well-defined body sites (10) distributed from the neck to calf bilaterally, and then averaged for both body sides. Calibration and evaluation were done using computed tomography (CT) as the reference method (10).

Carotid ultrasound

The ultrasound protocol involved scanning of the bulbous near common carotid arteries on both sides with a 12-to-5-MHz broad-band linear transducer on a HDI 5000 (ATL, Bothell, WA, USA) as described elsewhere (11-14).

Laboratory analysis

Fasting blood samples were collected from 08:00 to 10:30 h. Interleukin-6 was analyzed by an electrochemiluminescence immunoassay (Roche Diagnostics). Ultrasensitive-CRP (US-CRP) was analyzed with a Tina-quant® C-reactive protein latex ultrasensitive assay (Roche diagnostics). Leptin and total adiponectin were determined in plasma by ELISA from Biovendor Laboratory Medicine (Brno, Czech Republic). Cholesterol, HDL-cholesterol, and triglycerides were measured by enzymatic photometric methods, LDL cholesterol calculated by the Friedewald formula. Oxidized low dense lipoprotein (oxLDL) was measured by ELISA (Mercodia oxidized LDL Competitive ELISA, SE-754 50 Uppsala, Sweden). Intra- and inter-assay variation coefficients for all ELISAs in our study were below 10%. Plasma glucose was measured by the glucose hexokinase method. Malondialdehyde (MDA) was determined by high-performance liquid chromatography with spectrofluorimetric detection as described elsewhere (15); myeloperoxidase (MPO) by an automated chemiluminescent microparticle immunoassay (Architect MPO assay, Abbott Diagnostics, Abbott Park, IL) (16).

Homeostatic model assessment—insulin resistance (HOMA-IR) was calculated using the formula by Mathews et al. (17). A cut off >2.5 was used for identification of insulin resistance. Alkaline phosphatase (AP), aspartate transaminase (AST/GOT), alanine transaminase (ALT/GPT), gamma-GT (gGT), cholinesterase (CHE), creatinine, and UA were measured on a Cobas 8000, fibrinogen with LIATEST reagents using a STAGO STA-R Evolution coagulation analyzer (Roche Diagnostics, Germany).

Statistics

All statistical analyses were carried out using PASW Statistics 18.0 for Windows. Normal distribution of all clinical and biochemical measures was controlled by the Kolmogorov–Smirnov test. In cases of skewed distribution of variables logarithmic transformation was applied for significance calculation. Differences between groups were analyzed by unpaired t test. When more than two groups were compared, analyses of variance (ANOVA), with a three-stage factor [normal weighted controls, healthy-ow/ob (MHO), unhealthy-ow/ob(nonMHO)] was performed. Differences were calculated by Bonferroni post-hoc test. A stepwise logistic regression analysis (with forward and backward selection, respectively) was applied to investigate association of MetS as dependent variable with laboratory measures as independent variables. Additionally, Receiver-Operator-Curve (ROC) analysis was performed to confirm the results of the binary logistic regression analysis.

Ethics

STYJOBS/EDECTA was approved by the ethical committee of the Medical University of Graz, and conducted in compliance of human studies with the Helsinki Declaration of 1975 as revised in 1996.

Results

We analyzed 474 overweight/obese (ow/ob) and 235 normal-weight (nw) controls. Participants were categorized as juvenile (<18 years) or adult (>18). From 299 juvenile ow/ob subjects, 108 (36%), and from 175 adult ow/ob subjects, 79 (45%) had positive criteria for MetS. In both groups, more males had MetS (Tables 1 and 2).

Overweight/obese subjects were divided into a healthy (MHO) and an unhealthy ow/ob (nonMHO) subgroup. MHOs had a fasting glucose <100 mg dl−1, triglycerides <150 mg dl−1, HDL-cholesterol >45 mg dl−1, systolic blood pressure (SBP) <130, diastolic blood pressure (DBP) <85 mmHg; nonMHOs had > 2 of these criteria pathologically altered (Tables 1 and 2). Both, MHOs and nonMHOs had an increased waist circumference (<18 years: ≥ 90th percentile, >18 years: males ≥ 94 cm, females ≥ 80 cm). The data are outlined separately for juveniles (Table 1) and adults (Table 2), respectively.

In both juveniles and adults, percent body fat did not differ between MHOs and nonMHOs, although lipometry defined nuchal SAT thickness and visceral adipose tissue (VAT) mass were significantly increased in nonMHOs. Further, compared to MHOs, nonMHOs had significantly higher values of IMT, HOMA-index, UA, and gGT. Adiponectin was significantly decreased relative to normal controls. Ultrasensitive CRP, IL-6 and Leptin were increased in both juveniles and adults but MHOs did not differ significantly from nonMHOs (Tables 1 and 2).

Malondialdehyde (MDA) was found only in nonMHO juveniles significantly increased whereas oxidized LDL was strongest increased in adult nonMHOs as was seen for ALT/GPT, CHE, Creatinin, Fibrinogen and ETP (Tables 1 and 2).

To identify potential confounders prior to their inclusion in the logistic regression model, all relevant study variables were tested by an unpaired t test for significant differences with MetS no/yes as categorization in juveniles (Table 3), and adults (Table 4), respectively. Only variables demonstrating a significant difference (P < 0.05) by t tests were included in the logistic regression model. Controlling for an influence of gender by a separate analysis of females and males did not change the set of variables (not shown). If highly inter-related variables like US-CRP and IL-6 showed significant differences, only one representative variable with the lowest P value was included in the logistic regression analysis, respectively. Thus, two logistic regression models with MetS as dependent variable, one in juveniles <18 years or one in adults >18 years were performed (Table 5).

Table 3. Unpaired t test comparison for clinical/anthropometric/laboratory variables for the categories metabolic syndrome “no” and metabolic syndrome “yes” in probands aged ≤ 18 years. If variables were not normally distributed, analysis of log-transformed data was performed
N = 355P value
  • Significance levels are given as

  • *

    P < 0.05,

  • **

    P < 0.01,

  • ***

    P < 0.001.

Age (years)0.89
BMI (kg m2)0.001***
Lipometry body fat (%)0.026*
Lipometry SAT thickness neck (mm)0.001***
Lipometry visceral adipose tissue (kg)0.001***
Organized physical activity0.241
Nonorganized physical activity0.124
Carotid IMT (cm)0.001***
US-CRP (mg l−1)0.022*
IL-6 (pg ml−1)0.001***
LDL-cholesterol (mg dl−1)0.694
Oxidized LDL (mmol l−1)0.008**
Malondialdehyde (μmol l−1)0.001***
Myeloperoxidase (ng ml−1)0.629
Leptin (ng ml−1)0.001***
Adiponectin (μg ml−1)0.001***
Aldosteron (ng l−1)0.593
Renin (ng l−1)0.166
Cortisol (ng ml−1)0.397
Uric acid (mg dl−1)0.001***
Creatinine (mg dl−1)0.116
AST/GOT (U l−1)0.07*
ALT/GPT (U l−1)0.002**
gammaGT (U l−1)0.001***
Cholinesterase (U l−1)0.002**
Fibrinogen (mg dl−1)0.06
ETP (extinction)0.112
Antithrombin (%)0.463
Table 4. Unpaird t test comparison for clinical/anthropometric/laboratory variables for the categories metabolic syndrome “no” and metabolic syndrome “yes” in probands aged ≤ 18 years. If variables were not normally distributed, Mann–Withney U test was performed
N = 354P value
  • Significance levels are given as

  • *

    P < 0.05,

  • **

    P < 0.01,

  • ***

    P < 0.001.

Age (years)0.01*
BMI (kg m2)0.001***
Lipometry body fat (%)0.002***
Lipometry SAT thickness neck (mm)0.001***
Lipometry visceral adipose tissue (kg)0.001***
Organized physical activity0.001***
Nonorganized physical activity0.582
Carotid IMT (cm)0.001***
US-CRP (mg l−1)0.001***
IL-6 (pg ml−1)0.002***
LDL-cholesterol (mg dl−1)0.002**
Oxidized LDL (mmol l−1)0.001***
Malondialdehyde (μmol l−1)0.484
Myeloperoxidase (ng ml−1)0.01**
Leptin (ng ml−1)0.001***
Adiponectin (μg ml−1)0.001***
Aldosteron (ng l−1)0.03*
Renin (ng l−1)0.144
Cortisol (ng ml−1)0.02*
Uric acid (mg dl−1)0.001***
Creatinine (mg dl−1)0.023*
AST/GOT (U l−1)0.016**
ALT/GPT (U l−1)0.001***
gammaGT (U l−1)0.001***
Cholinesterase (U l−1)0.001***
Fibrinogen (mg dl−1)0.001***
ETP (extinction)0.002***
Antithrombin (%)0.575
Table 5. Results of the binary logistic regression model (dependent variable metabolic syndrome no/yes) including clinical non-anthropometric/ laboratory variables
PredictorsORP value95% CI Exp(B)R2 (Nagelkerkes)
  • a

    Independent variables included in this model: age, gender, carotis-IMT, IL-6, MDA, leptin, adiponectin, uric acid, gammaGT. OR = Odds Ratio per unit increase, CI = Confidence Interval.

  • b

    Independent variables included in this model: age, gender, carotis-IMT, US-CRP, oxidized LDL, leptin, adiponectin, aldosteron, cortisol, uric acid, cholinesterase, fibrinogen. OR = Odds Ratio per unit increase, CI = Confidence Interval.

Probands aged ≤ 18 years; Criterion: Metabolic syndrom (= nonMHO phenotype) yes/noa
Total (N = 355); uric acid (mg dl−1)2.17<0.0011.5–3.20.42
Probands aged > 18 years; Criterion: Metabolic syndrom yes/nob
Total (N = 355)    
Uric acid (mg dl−1)3.4<0.0011.4–8.00.68
Leptin (ng ml−1)1.13<0.0011.0–1.2 
Adiponectin (μg ml−1)0.670.0030.5–0.9 

Based on the P values shown in Table 3 for juveniles, the set of independent variables included age, gender, carotid-IMT, IL-6, MDA, leptin, adiponectin, UA, and gGT. For adults, the independent variables were age, gender, carotid-IMT, US-CRP, oxLDL, leptin, adiponectin, aldosterone, cortisol, UA, ALT/GPT, and fibrinogen (Table 3). Variables used for the MetS definition were not included in the analysis (i.e., anthropometry, glucose, HDL-cholesterol, triglycerides, blood pressure). The only variable that predicted nonMHO phenotype in both juveniles and adults was the UA (Table 5). In adults, leptin and adiponectin were also selected as significant predictors in the logistic regression model (Table 5). ROC analysis confirmed these results as the area under the curve (AUC) value was highest for the selected variables in both subgroups (≤18 years, AUC: 0.79, >18 years, AUC: 0.91).

Discussion

To the best of our knowledge, our study is the first that comprehensively distinguishes a MHO from a nonMHO phenotype in both juveniles and adults. Comparatively higher to other studies (4, 6), 28.2% of the adult, and 30.7% of the pediatric ow/ob subjects were found to be metabolically healthy (MHO-phenotype).

Comprehensive metabolic assessment identified differences between the MHO and non-MHO groups beyond metabolic syndrome criteria. Anthropometrically, irrespective of age, the MHO phenotype was characterized by lower visceral fat content and significantly decreased nuchal SAT at a given proportion of body fat. This result is supported by previous findings that high visceral fat stores and their clinical surrogate, waist circumference, are determinants of an adverse metabolic phenotype (3, 4, 18). The increased nuchal subcutaneous fat observed in the non-MHO phenotype is consistent with observations in the Framingham Heart Study that neck circumference is the best predictor for cardiometabolic risk (19). In a previous study, we have demonstrated a robust association between nuchal SAT thickness and low molecular weight/total adiponectin ratio (20). Accordingly, adiponectin levels were lower in the non-MHO phenotype compared to MHO.

A comprehensive logistic regression model identified uric acid (UA) as the best predictor of MetS (i.e., nonMHO phenotype) with odds ratio of 2.2 (CI: 1.5-3.2) in juveniles and 3.4 (CI: 1.4-8.0) in adults per unit increase (i.e., 1 mg dl−1). Thus, UA emerges as a considerable discriminator between the two obesity phenotypes. Hyperuricemia has already been shown to be associated with obesity and metabolic syndrome in adults (21) and children (22), and has been identified as a possible predictor of cardiovascular disease (23). Increased UA levels are also associated with increased blood pressure through effects on the renin-angiotensin system and increased insulin resistance. UA has atherogenic effects through upregulated expression of platelet-derived growth factor (24), increased vascular smooth muscle proliferation (24, 25), and monocyte chemoattractant protein 1 synthesis (26).

The emergence of UA as a possible biomarker to distinguish these two phenotypes has implications for both obesity treatment and research. UA can be investigated in longitudinal research with younger patients to determine whether it is a biomarker of future non-MHO status prior to development of adolescent metabolic syndrome criteria. If non-MHO status can be identified before clear metabolic syndrome abnormalities, it will be possible to target appropriate high risk groups for cardiovascular disease prevention. Individuals with non-MHO phenotype may benefit from specific targeted interventions that address possible metabolic effects of UA. For example, UA levels are affected by alcohol levels. Dietary interventions may also be useful in weight loss to address increased salt sensitivity potentiated by increased UA levels. However, sustained low UA levels are associated with neurologic diseases like Multiple Sclerosis (27) and Alzheimer's disease (28-30). Although, it is still controversial whether UA is beneficial per se in these diseases or just an innocent bystander, interventions to lower UA below normal levels to reduce cardiovascular disease risk may increase certain neurological risks.

In adults only, adiponectin (decreased) and leptin (increased) were also identified as predictors for MetS in the same model with UA. This fact underlines the significance of these two adipokines for the generation of metabolically unfavorable obesity. However, compared to UA [3.4 (CI: 1.4-8.0)], the ORs per unit change were lower, for leptin [1.1 (CI:1.0-1.2)], and for adiponectin [1.5 (CI:2-1.1)], respectively.

Our study has limitations that need to be acknowledged: First, BMI was different between the MHO and nonMHO group. However, there was no significant difference between per cent body fat as reflected by similar leptin levels in these two phenotypes. This is in accordance with data presented by Klöting et al. (31), and is most probably caused by an increased intrabdominal fat content in nonMHOs, which cannot be detected reliablly by lipometry. However, the indirect lipometry calculated abdominal fat content was found increased in the nonMHOs suggesting this fact. Second, the cross-sectional nature of our study allows no conclusions in regard to the CVD prognosis of the healthy obesity phenotypes. Thus, longitudinal studies are warranted in this regard.

Taken together, obesity is not a unanimous description of a clinical entity. Our study adds to the growing literature that increases the understanding of healthy obesity and may lead to future individualized therapeutic approaches. It seems important to distinguish metabolically healthy from unhealthy obese subjects to allow for cost-effective individualized approaches of treatment and prevent recommending excessive dieting, pharmacological and surgical treatment in certain healthy obese individuals. Future longitudinal and then interventional studies would be necessary to consolidate the position of UA as a clinically relevant indicator of the obesity phenotype, and, perhaps, even a risk modifying therapeutic target.

Acknowledgements

We gratefully acknowledge the laboratory work of Karin Glänzer, Anja Stoisser, Melanie Korbelius, Nina Fuchs, Ingrid Reiterer, Petra Legen, Ulrike Knebel, and Elke Bernhard. Further, the statistical advice of Dr. Franz Quehenberger (Institute for Medical Informatics, Statistics and Documentation, Medical University of Graz) is gratefully acknowledged.

Ancillary