Prevalence and Clinical Characteristics of Children and Adolescents with Metabolically Healthy Obesity: Role of Insulin Sensitivity

Obesity represents a major risk factor for metabolic disorders, but some individuals, “metabolically healthy” (MHO), show less clinical evidence of these complications, in contrast to “metabolically unhealthy” (MUO) individuals. The aim of this cross-sectional study is to assess the prevalence of the MHO phenotype in a cohort of 246 overweight/obese Italian children and adolescents, and to evaluate their characteristics and the role of insulin resistance. Homeostasis model assessment–insulin resistance (HOMA-IR), insulin sensitivity index (ISI), insulinogenic index (IGI) and disposition index (DI) were all calculated from the Oral Glucose Tolerance Test (OGTT). MHO was defined by either: (1) HOMA-IR < 2.5 (MHO-IRes), or (2) absence of the criteria for metabolic syndrome (MHO-MetS). The MHO prevalence, according to MHO-MetS or MHO-IRes criteria, was 37.4% and 15.8%, respectively. ISI was the strongest predictor of the MHO phenotype, independently associated with both MHO-IRes and MHO-MetS. The MHO-MetS group was further subdivided into insulin sensitive or insulin resistant on the basis of HOMA-IR (either < or ≥ 2.5). Insulin sensitive MHO-MetS patients had a better metabolic profile compared to both insulin resistant MHO-MetS and MUO-MetS individuals. These data underscore the relevance of insulin sensitivity to identifying, among young individuals with overweight/obesity, the ones who have a more favorable metabolic phenotype.


Introduction
Overweight and obesity prevalence has tremendously increased in recent years, not only among adults, but also among children and adolescents [1,2]. The global prevalence of childhood overweight and obesity, calculated on the basis of body mass index (BMI, Kg/m 2 ), has increased by over 50% in The physical examination was carried out by two trained investigators who measured weight and height for calculating BMI and BMI z-score [4,29], and the waist circumference (after gently exhaling, at the narrowest part between the lower rib and the iliac crest using a non-elastic flexible tape, recorded to the nearest 0.1 cm) and the waist to height ratio (WtHR) [30] were also calculated. Systolic and diastolic blood pressure (BP) was recorded before blood sampling with the subject in a sitting position and after a minimum of 5 min of acclimation. The mean of two BP measurements with 5 min interval was used in the analysis. Family and personal medical history, age of onset of obesity and the Tanner pubertal stage were recorded. According to the Tanner stage, patients were classified as pre-pubertal (stage I) and pubertal (stages II, III, IV, V) [31].
Blood samples were collected in the morning after 12 h fasting. Total and high-density lipoprotein (HDL) cholesterol (mmol/L), triglycerides, adiponectin and leptin levels were measured in the baseline sample. Plasma glucose and insulin levels were measured at baseline and 30, 60, 90 and 120 min after a 75 g oral glucose tolerance test (OGTT).
Plasma glucose was measured using the glucose oxidase method on a Beckman Glucose Analyzer 2 (Beckman Coulter, Inc., Fullerton, CA, USA), while plasma insulin was measured by the microparticle enzyme immunoassay (Abbott Laboratories, Abbott Park, IL, USA) method. Total cholesterol and triglycerides were evaluated via enzymatic methods (Instrumentation Laboratory, Milan, Italy). HDL cholesterol fraction was separated by the use of Mg 2+ Insulin sensitivity (insulin sensitivity index, ISI) was calculated according to the following Equation (3): 10,000/

√
[fasting plasma glucose (mg/dL) × fasting plasma insulin (mU/L)] × [mean OGTT glucose concentration (mg/dL) × mean OGTT insulin concentration (mU/L)] The disposition index (DI) was the product of IGI and ISI [32,33]. As already mentioned, numerous different criteria have been proposed for defining MHO. Table 1 summarizes the different criteria that have until now been used for the definition of the MHO status in children and adolescents.
In order to capture the most important metabolic characteristics of MHO individuals, we chose to use two different criteria to define the presence of MHO in our study [19,22]. Subjects were considered (A) "insulin sensitive" (MHO-IRes) or "insulin resistant" (MUO-IRes) if HOMA-IR was either < or ≥ 2.5, respectively [34][35][36], and (B) metabolically healthy (MHO-MetS) or unhealthy (MUO-MetS) in the absence or the presence of at least one of the criteria indicated by the National Cholesterol Education Program's Adult Treatment Panel (NCEP-ATP) for the MetS (adapted for children and adolescents) [34] (i.e., fasting plasma glucose ≥ 100 mg/dL, triglycerides ≥ 95th percentile, HDL cholesterol ≤ 5th percentile, systolic or diastolic blood pressure ≥ 95th percentile [20,37]). In addition, the MHO-MetS individuals were further subdivided into insulin sensitive MHO-MetS or insulin resistant MHO-MetS, according to the HOMA index being either < or ≥ 2.5, respectively (Table 2).  The study was conducted in accordance with the Declaration of Helsinki and its later amendments. The protocol was approved by the Ethics Committee of Garibaldi Hospital, Catania 2 (949/CE), and a written informed consent was obtained from the parents of all examined subjects before the commencement of the study.

Statistical Analysis
Data are expressed as mean ± standard deviations (SD) for continuous variables, and number of cases and percentage (%) for categorical variables. Continuous variables were compared between groups using either the unpaired t test, the one-way ANOVA or nonparametric analysis of covariance (ANCOVA), where appropriate. Differences in categorical variables were determined by Pearson's χhi-squared test. A multivariate logistic regression model accounting for possible confounders (BMI z-score, WC, WtHR, SBP, DBP, HDL cholesterol, triglycerides, fasting plasma glucose) was applied to examine the associations between each biochemical and clinical parameter and the MHO status. A Hosmer-Lemeshow post estimation test was used to assess the model's performance. The odds ratios' (OR) confidence interval (CI) was calculated at 95%. A p value < 0.05 was considered to be statistically significant.
Data were analyzed using StatView 5.01 (SAS Institute, Cary, NC, USA). Calculation of frequencies was carried out using the STATA 14.2 SE software (STATA Corp., College Station, TX, USA).

Results
The prevalence of MHO was clearly different according to the criteria used; lower for MHO-IRes (39 cases, 15.8%) and higher for MHO-MetS (92 cases, 37.4%).
The clinical, biochemical and anthropometric characteristics of the studied subjects are summarized in Table 3. Regardless of the definition used, both MHO-IRes and MHO-MetS patients showed a favorable metabolic profile compared to MUO; lower BMI z-score, WC and WtHR (although not statistically significant for MHO-MetS), lower mean systolic and diastolic blood pressure, lower fasting plasma insulin, glycaemia and triglycerides (although not statistically significant for MHO-IRes), as well as higher HDL cholesterol.
No difference was observed between the MHO and MUO groups regarding age, gender, being pre-pubertal/pubertal, pubertal stage, and mean age at the onset of obesity.

Multivariate Logistic Regression Analyses
Multivariate logistic regression analyses, accounting for the possible confounders related to the two definitions used for MHO (BMI z-score, WC, SBP, DBP, HDL cholesterol, triglycerides, fasting plasma glucose), showed that ISI was the strongest predictor of the MHO phenotype. In fact, it was the only parameter independently associated with MHO status regardless of the definition used; the OR was 7.49 (95% CI 3.80-14.84; p < 0.01) for MHO-IRes and 1.38 (95% CI 1.09-1.74; p < 0.01) for MHO-MetS (Table 4). On the contrary, as expected on the basis of the criteria used for the MetS definition, SBP, DBP and fasting plasma glucose were significantly negatively associated with MHO-MetS, whereas HDL cholesterol was positively associated. Finally, leptin levels were independently and negatively associated with MHO-IRes (OR 0.96, 95% CI 0.92-0.99, p = 0.04) ( Table 4).

MHO-MetS Subgroups According to Insulin Sensitivity
In order to deeply evaluate the impact of insulin sensitivity on MHO, we further subdivided patients with MHO-MetS into two groups: the insulin sensitive and the insulin resistant ones, on the basis of the HOMA-IR < or ≥ 2.5. Insulin sensitive MHO-MetS patients showed a more favorable clinical and metabolic phenotype, not only when compared to MUO-MetS, but also in comparison with insulin resistant MHO-MetS (lower BMI z-score, WC, WtHR and leptin levels) ( Table 5). This better metabolic condition is also noticeable when comparing insulin sensitive MHO-MetS OGTT-patterns and derived indices with those of insulin resistant MHO-MetS and MUO-MetS. Insulin sensitive MHO-MetS patients, in fact, exhibited lower fasting and post-load plasma insulin levels ( Figure 1B), higher ISI and DI, and lower IGI ( Figure 1C) compared to both insulin resistant MHO-MetS and MUO-MetS, and exhibited an insulin profile similar to those with MHO-IRes ( Figure 1A,B).  to height ratio; SBP, systolic blood pressure; DBP, diastolic blood pressure; HDL-C, HDL cholesterol; LDL-C, LDL cholesterol; FPG, fasting plasma glucose; OGTT, oral glucose tolerance test; IFG, impaired fasting glucose; IGT, impaired glucose tolerance; T2D, type 2 diabetes; HOMA-IR, Homeostasis model assessment-insulin resistance; ISI, insulin sensitivity index; IGI, insulinogenic index; DI, disposition Index. Bold text indicates a statistically significant difference (p value < 0.05). According to these three different definitions used, we then evaluated the prevalence of MHO in our population. Using HOMA-IR in addition to the MetS criteria allows us to identify a cluster of patients (10.6%) that really do not have any metabolic abnormality (Figure 2). According to these three different definitions used, we then evaluated the prevalence of MHO in our population. Using HOMA-IR in addition to the MetS criteria allows us to identify a cluster of patients (10.6%) that really do not have any metabolic abnormality (Figure 2).

Discussion
The first aim of the present study was to determine the prevalence of MHO in our cohort of children and adolescents using two currently applied criteria for the diagnosis of MHO [19,22]. The first is based on the HOMA-IR value (MHO-IRes) and the second is based on the MetS criteria

Discussion
The first aim of the present study was to determine the prevalence of MHO in our cohort of children and adolescents using two currently applied criteria for the diagnosis of MHO [19,22]. The first is based on the HOMA-IR value (MHO-IRes) and the second is based on the MetS criteria (MHO-MetS). Our data showed that the prevalence of MHO in children with overweight or obesity was substantially different depending on the criteria used (15.8% vs. 37.4%, respectively). The frequency of MHO-IRes in our cohort of patients is similar to that previously reported in adult individuals, using the same HOMA-IR cut-off value (<2.5) [34][35][36]. Considering the possible development of transient insulin resistance during puberty, other authors used a higher HOMA-IR cut-off (<3.16) in adolescents and observed a higher prevalence of MHO [19]. Indeed, using the same HOMA-IR cut-off value, we would have observed a similarly high prevalence of MHO (25.2%). This percentage, however, remains noticeably lower compared to that of MHO-MetS, confirming the need to assess the level of insulin sensitivity in order to adequately characterize the MHO status. To point out the role of puberty in the MHO/MUO status, we calculated the percentages of each Tanner stage in our cohort, without finding any significant difference.
Using the NCEP-ATP criteria for MetS, modified for percentiles cut-offs in children and adolescents [37], we found a similar prevalence of MHO-MetS compared to recent studies in adolescents [22].
Regardless of the criteria used for identifying the MHO condition, MUO individuals always had an unfavorable metabolic profile compared to MHO (higher BMI and visceral adiposity, and worse glycemic, lipid and blood pressure profile). Furthermore, serum leptin levels were also significantly higher in MUO compared to MHO, confirming the role of this adipokine in determining insulin resistance and metabolic impairment independently of the age and BMI-z score of the population studied [38]. In our series, adiponectin levels did not differ significantly between MUO and MHO individuals, at variance with the positive correlation reported by others between serum adiponectin concentrations and the MHO phenotype in children and adolescents with overweight or obesity [39,40]. This difference might be due to different ethnicities or to the assay utilized for the determination of adiponectin (radioimmunoassay-RIA vs. enzyme-linked immunosorbent assay-ELISA).
The second aim of the present study was to identify the role of insulin resistance in determining clinical and metabolic features of the MHO phenotype in young individuals. We found that ISI was the strongest predictor of MHO, being independently associated with both MHO-IRes and MHO-MetS. This observation supports the crucial role of insulin sensitivity in the pathophysiology of MHO. As already demonstrated by "clamp-based" studies, ISI calculated on the basis of post-load plasma glucose and insulin levels provides a good approximation of both hepatic and skeletal muscle disposition of glucose, and correlates well with insulin sensitivity measured directly using the euglycemic-hyperinsulinemic clamp [41,42]. ISI, however, cannot be used for routine clinical practice, because it requires an OGTT, and therefore cannot be proposed on a large-scale basis. Nevertheless, insulin resistance can be more easily assessed by calculating HOMA-IR, which in our study was negatively associated with MHO regardless of the criteria used.
Given this relevant role of insulin sensitivity in MHO, we further subdivided our young MetS patients into either insulin sensitive MHO-MetS or insulin resistant MHO-MetS. Considering the degree of insulin resistance, MHO-MetS patients that were insulin sensitive had a significantly better metabolic profile, not only compared to MUO, but also compared to insulin resistant MHO-MetS (e.g., patients classified as "healthy" according to MetS criteria). These findings suggest that using the MetS classification in identifying MHO patients can overestimate the prevalence of MHO phenotypes in those young subjects with overweight/obesity whose condition is predominantly characterized by impaired insulin sensitivity. Insulin resistance represents one of the main etiological features leading to metabolic impairment, irrespective of the excess of adipose tissue [43,44]. This condition, in fact, although strongly related to obesity, could affect also normal weight young individuals [45][46][47][48]. Moreover, insulin resistance per se is an independent predictor of cardiovascular risk both in childhood [48] and in adulthood [49], strongly related to atherogenic dyslipidemia (low HDL cholesterol and hypertriglyceridemia) and hypertension, and deeply involved in the pathogenesis of T2D, representing the "trigger" of MetS. For these reasons, the presence of insulin resistance, along with the absence of other components of MetS in childhood, should be considered a risk factor for metabolic disorders [50] and cardiovascular diseases [51]. It has been demonstrated that insulin resistance affects myocardial function [52][53][54]. Recently, Corica et al. showed the negative effects of insulin resistance on cardiovascular remodeling and subclinical myocardial dysfunction in a cohort of children with obesity, and suggested that insulin resistance represents a strong predictor of subclinical myocardial dysfunction in obese, non-diabetic children [55].
Therefore, in light of this evidence, the definition of MHO must also take into account the presence of insulin resistance.
Another significant finding is that in insulin sensitive MHO-MetS patients, DI is higher compared to both MUO and insulin resistant MHO-MetS individuals. This index reflects the ability of the pancreatic β-cells to compensate for insulin resistance [56]. A low DI, in fact, is considered an early marker of insufficient β-cell compensation, correlated, therefore, with the risk of future diabetes [57]. The decreased DI, observed in both MUO-MetS and insulin resistant MHO-MetS patients, suggests therefore that these young individuals do not only presently have an unfavorable metabolic profile, but also a higher risk of developing T2D in the future.

Conclusions
Despite several limitations, such as the lack of information regarding long-term outcomes (e.g., obesity complications and related morbidities) due to the cross-sectional design of our study, our study offers important insights into the MHO phenotype during childhood and adolescence. Clinicians should consider the heterogeneity of pediatric obesity and the crucial role of insulin sensitivity in determining MHO. In particular, we believe that, in addition to the absence of the metabolic syndrome criteria, the degree of insulin sensitivity should be assessed in overweight/obese children and adolescents in order to identify those at low risk. We demonstrated, in fact, that if we exclusively used the metabolic syndrome criteria for the MHO definition, there would have been an overestimation of the number of patients without metabolic alterations. We propose, therefore, to use both the described criteria in order to diagnose as MHO only those individuals at really low metabolic risk.
The possibility to accurately distinguish young individuals with overweight/obesity by assessing insulin sensitivity could provide better and more appropriate therapeutic options, with a different intensity of intervention according to the risk level, and more structured interventions (e.g., multidisciplinary obesity management and/or bariatric surgery) in individuals at higher risk.
Further longitudinal studies are needed so as to better characterize the evolution and outcomes of MHO young patients.