Obesity is associated with insulin resistance and components of the metabolic syndrome in Lebanese adolescents

Background Prevalence of metabolic syndrome (MS) in obese adolescents has been reported to range between 18–42%, depending on country of origin, thus suggesting an ethnic-based association between obesity and MS. Aim This study aims to investigate the magnitude of the association between obesity, insulin resistance and components of MS among adolescents in Lebanon. Subjects and methods The sample included 263 adolescents at 4th and 5th Tanner stages of puberty (104 obese; 78 overweight; 81 normal weight). Anthropometric, biochemical and blood pressure measurements were performed. Body fat was assessed using dual-energy X-ray absorptiometry. Results According to International Diabetes Federation criteria, MS was identified in 21.2% of obese, 3.8% of overweight and 1.2% of normal weight subjects. The most common metabolic abnormalities among subjects having MS were elevated waist circumference (96.2%), low HDL (96.2%) and hypertriglyceridemia (73.1%). Insulin resistance was identified in all subjects having MS. Regression analyses showed that percentage body fat, waist circumference and BMI were similar in their ability to predict the MS in this age group. Conclusions MS was identified in a substantial proportion of Lebanese obese adolescents, thus highlighting the importance of early screening for obesity-associated metabolic abnormalities and of developing successful multi-component interventions addressing adolescent obesity.

Mediterranean Region( EMRO) than that reported from Europe and the US (Kelishadi 2007). The increase in the prevalenceo fo besityi ny ouths has been parallelled by an increase in the prevalenceofpaediatric metabolic syndrome (MS), ac onstellationo fm etabolic risk factors that include insulin resistance, hypertension,glucose intolerance and an abnormal lipid profile (Weiss et al.2004;Esmaillzadeh et al. 2006;Braga-Tavares and Fonseca2 010). Ar ecent study conducted in Turkey showed that, among 2-19 year old children and adolescents,aone-point increase in BMI Z -score resulted in a2-foldincrease in the prevalence of MS (Sen et al. 2008).
Based on the modifiedA TPIII criteria (Cook et al. 2003), prevalencee stimates of MS among obesea dolescents were found to varyb etween 18% in Spain and 42% in the US, suggesting an ethnic-based associationbetween obesityand MS (Ló pez-Capapé et al.2006;Dhuper et al. 2007). Limited data from the Middle-East suggest highr ates of MS and cardiometabolic risk factorsi nt he adolescent population. The prevalenceo fM Sa mong obesea dolescents in Iran is among the highest estimatesr eportedi nt he literature (41.9%) ( Esmaillzadeh et al. 2006). Similarly,i narecent study conducted on 203 overweight and obese subjects aged 6-17 years living in Norway,t he prevalence of MS was found to be higher among obese subjects with Middle Eastern origins than among their Norwegian counterparts (30.6% vs 20.8%) ( Kolsgaard et al. 2008). This association between adolescent obesityand MS indicators has not been investigated in manycountries of the Middle-East.
Althoughr ecent definitions of MS do not include direct measurement of insulin in children (Eyzaguirreand Mericq 2009), insulin resistance (IR) has been frequently reported to overlap with MS and to confer cardiometabolic disease risk distinct from that attributed to MS itself (Haffner et al. 1990;Dhuper et al. 2007;S teinbergere ta l. 2009). Ethnic disparities in insulinr esistanceh avea lso been reported in the literature,w ith insulin resistance suggested as being more frequent in Whitea nd African-Americanp eople and less common in Asiatic, Arabic andL atin American populations (Caceres et al. 2008).
Even though the clinical utilityofMSasadisease category faces considerable controversy,p articularly when applied to the paediatric population, the screening for MS and its components early in life allows for the identification of individualsa ts ignificantr isk for cardiovasculard isease and those in urgent need of lifestyle intervention. Recent evidence suggests the presenceo fe arly functional and morphologic changes to the heartand blood vessels among obese adolescents with MS (Chinali et al. 2008). Ye t, the plasticityo ft he cardiovascular system early in life allows for the reversal of cardiovascular damage and cardiac abnormalities in obese adolescents, but only if risksa re identified early and treated aggressively (Ippisch et al. 2008;Battistae ta l. 2009). Fort hese reasons, the identification of paediatricM Si ni ndividuals who haven ot yet developed cardiovasculardisease, is of great importance from apublic health perspective.
This study was conducted to investigate the magnitude of thea ssociation between obesity,i nsulin resistance and variousc omponents of the MS among adolescents (Tanners tages4 -5)i nL ebanon,asmallE astern Mediterranean country, where prevalencerates of paediatric obesitya re high, approaching those observed in developed countries such as the US (Sibai et al. 2003).Confirming MS prevalencei nL ebanese obese adolescents should lead to multi-component interventions targeting MS and its components in Lebanonand other parts of the Middle-East.

Study design and participants
This is ac ross-sectional study of adolescents attending public schoolsi nB eirut. The research wasa pproved by the Institutional Review Board, American University of Beirut.
As ample of 263 subjects was recruitedu sing am ultistage cluster samplingasfollows. Out of 74 publicschools in Beirut, eight were randomly selectedf rom the list of the Lebanese Ministry of Education. From grades 10, 11 and 12, aB MI-stratified samplew as conveniently recruited. Stratification was based on BMI status as defined by the newW HO growth standards (deO nise ta l. 2007). Accordingly,o besityw as defined as BMI . þ 2 Z -scores for sex and age, overweight as þ 1 , BMIZ-scores # þ 2 and normal weight as BMI # þ 1B MI Z -scores (de Onis et al. 2007). Written informedc onsent from parents and written informed assentf rom participants wass ought. Subjects were invited to visit the research centrea tt he Nutrition and Food Science Department at the American University of Beirutand were instructed to consume adaily diet providing 50 -60% of energyi ntake as carbohydrates for 3d ays prior to their visit and to avoid anyv igorous physical activity. Eligibilityo fs tudy subjects was verified. Subjects were eligible if they were healthy, at 4 th and 5 th Tanner stages of puberty, as determined by amedicaldoctor, and of Lebanese ancestry. Exclusioncriteriaincluded history of chronic illness or the use of medications that alter blood pressure, glucose or lipid metabolism.
Samples ize calculation wasb ased on the following: To detectam oderate correlation ( r )o farange of 0.2 -0.3 between obesitya nd componentso ft he metabolic syndrome, as ample of 68 -150 subjects was needed to provide an 80% power at p ¼ 0.05.

Data collection
Data collection was carried outi n2 008-2009o n participantsw ho matchedi nclusionc riteriaa nd had providedwritten assent as well as parental written informed consent. Anthropometric measurements were taken after voiding and included weight, height, WC and %BF which was measured by dual-energyX-ray absorptiometry(DXA) (DelphiW ,H ologic Inc, QDR software for windows version 11.2) based on region %BF.W aist circumference (WC) was measured using ac alibrated plasticm easuring tape, at the umbilicus level, with subjectss tanding and following normal expiration (Marfell-Jones et al. 2006). Fasting blood samples (10 ml/participant) were obtained by aphlebotomist for the analysis of serum glucose, insulin, triglycerides (TG) and HDL. Blood pressurew as measured by aregisterednurse using amercurysphygmomanometer, with the subjects seated andafter a5 -minute rest.

Biochemical analysis
Plasma glucose was analysed using the COBAS 6000 analyzer (Roche Diagnostics, Switzerland). Serum insulin was determined by the use of an antibody radioimmunoassay kit (Linco Research, Missouri, USA). Serum TG was determined using the COBAS 6000 analyzer (Roche Diagnostics, Switzerland)w hile serum HDLw as colourimetrically determined by the use of the HDLs eparationt ab (Union Carbide Corp.P leasantville, NY,U SA).
Diagnostic criteria for the metabolic syndrome MS was defined based on the recently published harmonized metabolics yndrome definition( Albertie ta l. 2009), according to which three abnormal findingso ut of five would qualify the subject for the MS. The cut-off values for the individual metabolic abnormalities are based on the IDF criteria (Zimmet et al. 2007), which recommend the use of the adult IDF cut-off valuesf or subjects aged 16 yearsa nd older, as follows:a bdominal obesity (WC $ 94 cm for males and $ 80 cm for females); elevated TG ( $ 150 mg/dL); low levels of HDL ( , 40 mg/dL for males and , 50 mg/dL for females); highs ystolic blood pressure( SBP) ( $ 130 mmHg) or highd iastolic blood pressure( DBP) ( $ 85 mmHg);h ighf asting glucose ( $ 100 mg/dL). Fors ubjects aged 10 -16y ears,t he IDF recommends the use of the same diagnosticc riteriae xcept for WC and HDL,w here WC $ 90 th percentile for age and sex, and HDL , 40 mg/dL were useda sc ut-off values. For M ETABOLIC SYNDROME IN L EBANESEADOLESCENTS 123 comparativep urposes, the data were also analysed according to the modified ATPI II definition (Cook et al. 2003).
As reference values specific for adolescents in Lebanon are not available, hyperinsulinemia was determined based on cut-off points specific for the Tanner stage of maturation (fastingi nsulin $ 30 mIU/L for Tanner stage 4; fasting insulin $ 20 mIU/L forT anners tage 5) proposedb y the literature (Goran and Gower 2001;Sen et al. 2008). Insulin resistance( IR) wasd etermined using the homeostatic model assessment of insulinr esistance (HOMA-IR) defined as: Values exceeding 3.16 indicated the presenceofIR (Keskin et al. 2005).

Statistical analysis
Statistical analysis was performed using the Statistical Analysis Package forS ocial Sciences (SPSS, version 16.0) and the level of significance was set at p , 0.05. Frequencies and descriptive statistics were performed fort he different variables understudy including MS prevalence according to different definitionsa sw ell as the prevalence of individual components of MS. Means and standard deviations were computed for anthropometric and biochemical variables. Mean differences were compared for significanceb etween groups, using one-way ANOVA .P earsonc orrelations were performedbetween obesityi ndicators and the different MS components. The associationb etween MS abnormalities (dependant variables) and adiposityi ndicators including %BF,W Cand BMI (independent variables) was assessed by age and sex-a djusted linear regression analyses. Natural logarithm transformation was performed when the dependent variable was not normally distributed. Determinants of MS were evaluatedu sing age-and sex-adjusted logistic regression analysis.

RESULTS
The study sample consisted of 263 adolescents (112 males and 151 females) and included104 obese,78overweight and 81 normal weight subjects.A nthropometric, biochemical and blood pressurem easurements of the studys ubjects are shown in Table I. Genders combined, mean serum TC, LDL-C, TG,i nsulin levels, systolic and diastolic blood pressurew eres ignificantly higher and HDL-C significantly lower in obesea sc ompared to normal weight adolescents. The most common abnormalitya mong obesea dolescents was elevated WC (90.4%), followed by IR (88.5%) and hyperinsulinemia (73.1%), while hyperglycemia wast he least common abnormality( 3.8%) ( Table II). The number of metabolic abnormalities increased with obesity ( Table III), whereby1 9.2% of obese subjects had three metabolic abnormalities as compared to 3.8% of overweight subjects and 1.2% of the normal weight subjects. The presenceo ff our metabolic abnormalities was only noted among obese subjects (1.9%). Based on IDF criteria, the prevalence of MS was 21.2%i no beses ubjects, 3.8% in overweight subjects and 1.2% in the normal weight subjects. Analysing the data according to the modified ATPI II definition (Cook et al. 2003)y ielded as lightly higher prevalenceo fMSinthe obese study sample (24.0%), while MS prevalence among overweighta nd normal weight subjects remainedessentially the same(3.8% in overweight and 1.2% in normal weight subjects). Gender disparities were noted with both the IDF (27.1% in obeseb oys and 16.1% in obese girls) and ATPIII (31.3%inobeseboys and 17.9% in obese girls) definitions, with obese boys showing higher prevalence rates of MS. Table I. Baseline characteristics of the adolescent subjects.
IR wasidentified in 88.5% of obese,84.6% of overweight and 24.7% of normal weight adolescents (Table II). Similarly, IR and hyperinsulinemia were present in 100% and 73.1% of subjects diagnosed with MS, respectively (Table IV).

DISCUSSION
This study is the first, to ourk nowledge, to examinet he association between obesitya nd MS indicators among adolescents from the Levant area in the Middle-East. It showed that obesityi ncreases the risko fd eveloping MS among Lebanese adolescents (13 -19years) of Tanner stages 4-5. The prevalence of MS was estimated at 21.2%inobese adolescents using IDF criteria, with obese boys presenting a higher rate as compared to obese girls. Abdominal obesity, reducedH DL-levels and hypertriglyceridemiaa ppeared as the main abnormalities contributing to the syndrome.
Various definitions of paediatricM Sh aveb een used in the literature with varied sets of criteria and different cut-off values for each risk factor,making disparitiesacross studies difficult to appraise (Cook et al. 2003;de Ferranti et al. 2004;Weisse ta l. 2004;K arachaliou et al. 2008). For comparison purposes, the prevalence of MS in this study was re-calculated using the ATPI II definition (Cook et al. 2003) and results were contrasted with those obtained from ther egiona nd elsewhere. Findings indicate thatt he prevalenceo fM Sa mong obese adolescents in Lebanon (24%) is notably high, exceeding that reported from developed countries such as Italy (16.5%) (Caranti et al. 2008) and Spain (18.0%) (Ló pez-Capapé et al. 2006), while being lower than that reported from the US (42%) (Dhuper et al. 2007), the UAE( 44%) ( Eapen et al. 2010) and Iran (41.9%) (Esmaillzadeh et al. 2006). This discrepancy in MS prevalencea mong obesea dolescents is in line with the assumptions of an ethnic-basedassociationbetween obesity and MS indicators.S everal studies haves uggested that the impacto fo besity, particularly abdominal obesityo nM S   (Kolsgaard et al. 2008). It is in this context that the use of criteria specific to race or ethnic group for MS in youths is becoming highly warranted (Sumner 2008).
The results of the present study clearly show that the presenceofindividual cardiometabolic risk factors was high in obeses ubjects,w ith9 7.1%o fo bese adolescents presenting with at least one MS abnormality( data not shown). These findingsm ay place obese adolescents at ah igher riskf or early manifestations of atherosclerosis, as suggested by Whincup et al. (2005), whos howedt hat brachial arterydispensabilitydecreased in obese adolescents as adiposity, IR, diastolic blood pressurea nd the other criteriaofM Si ncreased.
The findings of thisstudysuggestgender disparities in the prevalenceo fM S, with boys presenting higher prevalence rates as compared to girls. This higher prevalence of MS in boys is in agreement with previous studies conducted in the US (Cook et al. 2003;Weiss et al. 2004), but contradictory to those reported from Iran, another Middle-Eastern country (Esmaillzadeh et al. 2006). One reasonb ehind this gender disparityinour studymay be the higher prevalenceofsevere obesity( BMI . þ 3 Z -scores for sex and age) among boys (8.0%)ascompared to girls (5.3%).
In addition to abdominal obesity, which was present in the majorityo fs ubjects having the MS, thiss tudy showed that lowH DL and elevatedT Gc ontributed most to MS prevalence(96.2% and73.1%,respectively). Further analysis of the data by logistic regression showedt hat the TG/HDL ratio was as trongd eterminant of MS in the study sample (OR ¼ 4.3, 95% CI ¼ 2.76 -6.72). The highp revalence of highTGcoupled with low HDL (i.e. ahighTG/HDL ratio) among obese subjects has been noted previously as astrong predictor of MS in Ve nezuelan children (6 -12y ears) (Quijadaetal. 2008), prompting investigators to propose the use of the TG/HDL ratio as apredictive indicator of MS in children and adolescents (Hannon et al. 2006;Dhuper et al. 2007;Li et al. 2008). It is worth noting that, in an ational study on the prevalenceofMSamong Lebanese adults, low HDL appeared as the most prevalent metabolic abnormality in the studied population (49.3%) (Sibai et al. 2008). This is supported by previous findingsw here reduced HDL levels were reporteda st he most common metabolic abnormalityamong adult Arab Americans (Jaber et al. 2004).
In this study, 73.1%o fo bese adolescentsw ere hyperinsulinemic and8 8.5% insulin-resistant, based on   HOMA-IR. Thesefi ndingsa re similar to those reported from France ( Druet et al. 2006) where 71.8%o fo bese children and adolescents were insulin-resistant, but higher than estimatesreported from Bolivia (39.4%) (Caceres et al. 2008 ( Juarez-Lopez et al. 2010). Hyperinsulinemia serves to compensate for IR to maintaing lucose homeostasis (Eyzaguirrea nd Mericq 2009), as evidenced by the fact that the least common abnormalityi nt he study subjects wase levated fasting bloodg lucose. Adolescents with hyperinsulinemia and IR are prone to several health problems, including systemic inflammation, endothelial dysfunction, hypertension, polycystic ovarysyndrome and early atherosclerosis, all of which can appear during childhood anda dolescencei n obese individuals (Eyzaguirreand Mericq 2009;Steinberger et al. 2009;Juarez-Lopez et al. 2010). IR has also been suggested to play apivotal role in the pathogenesis of MS as evidenced by the fact that, in this study, IR was found in all subjects having MS.T hisc entral role of IR in the development of MS is corroboratedb yr esults from other studies in American, Mexican and Bolivian youth suggesting that MS is more likely to develop in childrenand adolescents with hyperinsulinemia (Cruz and Goran 2004;Caceres et al. 2008;Shaibi and Goran 2008;J uarez-Lopez et al. 2010). IR has thus been suggested as being as ignificantp redictor of the paediatricMSand was proposed as adiagnostictoolin children and adolescents (Tresaco et al. 2005;Lee et al. 2007).
In this study, regressiona nalyses showed that %BF, BMI and WC were relatively similar in their ability to predict MS and that body fat was not superior to either BMI or WC in predicting metabolic risk in obesea dolescents. Thesefi ndingsa re in agreement with those reported by Neovius et al. (2009) in Swedish adolescents and Lindsay et al. (2001)inPima Indian adolescents, thus underscoring the utilityofBMI as an adiposityi ndicator and apredictor of cardiovascularr isk factors in this age group.B MI was suggested by Sen et al. (2008)a st he most important determinant of MS in as tudy on obesec hildrena nd adolescents in Turkey,wherebyaone-point increase in BMI Z -score resulted in a2 -foldi ncrease in the prevalenceo f MS (Sen et al. 2008).

CONCLUSIONS
In conclusion, this study is, to our knowledge, the first in the Levant area and in Lebanon to investigate the magnitude of the associationbetween obesityand components of the MS in an adolescent population group.T he findingss howed that the prevalenceo fm etabolic abnormalities and their clustering into MS increasedwith obesityand that BMI, WC and %BF were significant predictors of MS in thisa ge group.T hesefi ndings highlightt he importance of early screening for metabolic abnormalities in adolescent obesity and stress the importance of early weight management intervention strategies as the burden of comorbidities may be reversed if health risks are identified and treated early in life (Battista et al. 2009).
The results of this study should be considered in light of the following limitations. Elevated WC was determined, as recommended, according to cut-off points specific for Europids duet ot he unavailability of WC cut-off values specific for ourp opulation. Hyperinsulinemia was also identified based on international cut-off points due to the lack of an upper limito fi nsulinvaluesf or Lebanese population. Finally,wehaveused fasting glucose and insulin values in the HOMA equation to estimate IR, rather than other more accurate techniquess uch as the euglycemic clamp.