Is Macronutrients Intake a Challenge for Cardiometabolic Risk in Obese Adolescents?

(1) Background: Pediatric obesity is an emerging public health issue, mainly related to western diet. A cross-sectional study was conducted to explore the association between macronutrients intake and cardiometabolic risk factors in obese adolescents. (2) Methods: Ninety-three Italian obese adolescents were recruited; anthropometric parameters, body composition, glucose and lipid metabolism profiles were measured. Macronutrients intake was estimated by a software-assisted analysis of a 120-item frequency questionnaire. The association between macronutrients and cardiometabolic risk factors was assessed by bivariate correlation, and multiple regression analysis was used to adjust for confounders such as age and sex. (3) Results: By multiple regression analysis, we found that higher energy and lower carbohydrate intakes predicted higher body mass index (BMI) z-score, p = 0.005, and higher saturated fats intake and higher age predicted higher HOmeostasis Model Assessment of insulin resistance (HOMA-IR) and lower QUantitative Insulin-sensitivity ChecK (QUICK) index, p = 0.001. In addition, a saturated fats intake <7% was associated with normal HOMA-IR, and a higher total fats intake predicted a higher HOMA of percent β-cell function (HOMA-β), p = 0.011. (4) Conclusions: Higher energy intake and lower carbohydrate dietary intake predicted higher BMI z-score after adjustment for age and sex. Higher total and saturated fats dietary intakes predicted insulin resistance, even after adjustment for confounding factors. A dietary pattern including appropriate high-quality carbohydrate and reduced saturated fat intakes could result in reduced cardiometabolic risk in obese adolescents.


Introduction
Obesity is a chronic condition characterized by an abnormal or excessive fat accumulation that presents a risk to health. From 1980 to nowadays, adult and childhood obesity has become a priority health issue in terms of prevalence and economic burden [1][2][3]. During the last three decades, in both developed and developing countries, obesity prevalence rates increased about 27% in adults and 47% in children, for a total of 2.1 billion individuals considered overweight or obese [4]. Such an alarming spread is mainly due to unhealthy westernized dietary habits within an "obesogenic environment" that promotes a sedentary lifestyle [5]. According to WHO data (2016), 41 million children under 5 years and over 340 million children and adolescents are overweight or obese. The WHO European Childhood Obesity Surveillance Initiative (COSI, fourth edition) reported an obesity prevalence among school-aged European children (6)(7)(8)(9) years) between 4% and 21% for boys and between 2% and 19% Tokyo, Japan) [20]. An oscillometric device was used to check blood pressure (BP), according to the National High Blood Pressure Education Program Working Group recommendations [21].

Biochemical Assessments
Blood samples were collected in standardized conditions: from 8:30 to 9, after 12 h of fasting, by cubital vein puncture. They were immediately sent to our clinical laboratory and analyzed to check for total cholesterol (TC), HDL-c, low-density lipoprotein cholesterol (LDL-c), TG, insulin, fasting glucose, glycated hemoglobin (HbA1c). US National Heart, Lung, and Blood Institute (NHLBI) lipid cutoff values, based on US normative data, were used to detect dyslipidemia [22]. Insulin, fasting glucose, and HbA1c levels were reported to our Clinical Laboratory range values.

Dietary Habits
Subjects' dietary habits were assessed through a food frequency questionnaire (FFQ) developed in 1990 at our Department, based on the original Block-FFQ [23] and revised in 2008 according to the full-length Block 2005 FFQ © (NutritionQuest, Berkeley, CA, USA). The FFQ is the most common method for dietary assessment used in large epidemiological studies [24]. The questionnaire consists of a list of 120 foods and beverages with response categories to indicate usual (daily, weekly, or monthly) frequency of consumption and portion (full, half, or double portion). Parents filled out the questionnaire about the patients' dietary habits during a 50 min interview held by a trained dietitian. Usual portion sizes were estimated using household measures and the weight (e.g., pasta) or unit (e.g., fruit juice) of the purchase. A 24 h recall was additionally recorded at the end of the interview to standardize the usual serving size.
Energy intake analysis and nutrient quantification were performed using an ad hoc PC software program (MetadietaVR, 2013; METEDAsrl, via S.Pellico 4, San Benedetto del Tronto, AP, Italy). The food consumption frequency for each item was converted to daily intake and compared to ageand sex-specific Italian national dietary reference values for energy and nutrients [25].

Cardiometabolic Risk Assessment
HOmeostasis Model Assessment of insulin resistance (IR) index (HOMA-IR), HOMA of percent β-cell function (HOMA-β%), and QUantitative Insulin-sensitivity ChecK Index (QUICK index) calculated on fasting samples are useful tools in the clinical practice to identify subjects at risk for type 2 diabetes mellitus [26].
Although the glucose clamp method remains the reference standard for a direct measurement of insulin sensitivity, these quick and simple indexes are ideal for large and longitudinal studies and more acceptable to children and adolescents [27].
HOMA-IR: HOMA-IR is the most widely used measure of insulin resistance; it is calculated as the product of fasting plasma insulin (µU/mL) by fasting plasma glucose (mmol/L) divided by 22.5 [28].
HOMA-IR varies according to age and gender. Recently, HOMA-IR reference ranges have been published for a large population of normal-weight and obese young Caucasians. According to Shashaj et al., an HOMA-IR value above the 75th percentile in obese participants identifies adolescents with cardiometabolic risk factors [26].
Atherogenic index of plasma (AIP): AIP is calculated as Log [TG (mg/dL)/HDL cholesterol (mg/dL)]; it reflects the relationship between protective and atherogenic lipoproteins and may predict the risk of cardiovascular diseases in adults and adolescents [35].
Visceral adiposity index (VAI): VAI reflects fat distribution and metabolism and is calculated as:

Statistical Analysis
Descriptive data are reported as mean and standard deviation (SD) or median and min-max range, as appropriate. Normality tests were conducted to assess whether the variables were normally distributed. The association between macronutrients and cardiometabolic risk factors was assessed by Pearson's or Spearman's bivariate correlation coefficient, as appropriate. Forward stepwise multiple regression analysis was performed to predict each cardiometabolic risk factor from models including energy intake, amount of each macronutrient, age, and sex. Logarithmic transformation was performed on skewed data. Pearson's chi-squared test was used to assess the association between normal saturated fats intake and HOMA-IR. All values of p ≤ 0.050 were considered to indicate statistical significance (two-tailed test). The statistical package for social sciences (SPSS) package version 20.0 (SPSS Inc., Chicago, IL, USA) was used for statistical analysis.

Anthropometric Parameters in Obese Adolescents
Mean age at recruitment was 11 years. Fifty-six patients were females (60%). All patients were in pubertal stage. The median value of BMI z-score was 2.5. All patients had a waist circumference (median 91 cm) over 90 • percentile for age and sex [16]. The median value of WHtR, an index of central adiposity, resulted higher than normal [37]. In 45 patients (48%), WHtR values over the cutoff of 0.6 were found. Eighty (92%) adolescents had a tricipital skinfold measure over 95 • percentile for age and sex [38]. Data from bioimpedance segmental body composition analysis revealed that 85 patients (91.3%) had an FM% consistent with obesity, according to sex-and age-specific curves for body fat (Table 1) [39].

Macronutrients Intake in Obese Adolescents
In our cohort, the median energy intake, adjusted for sex and age, was normal compared to national dietary reference values (DRVs). The average daily protein intake was higher than the reference value (91 g/day vs. 39-50 g/day). The median lipid, saturated fat, and carbohydrate intakes (%) were within the reference range. Sugar intake (%) was higher than the reference value (18% vs. 15%). The intake of polyunsaturated fats was below the reference value (Table 2). Values are expressed as mean (SD) and median (25th-75th percentile).

Glucose and Lipid Metabolism in Obese Adolescents
The median values of fasting glucose, insulin, and HbA1c serum levels were within the normal values. IFG was found in 3.2% of the patients, and Impaired Glucose Tolerance (IGT) in 2.1% of them. Thirteen patients (about 14%) had HbA1c levels consistent with prediabetes. No one was diagnosed with diabetes type 1 in our population. In 60.2% of the patients, the HOMA-IR value was above the 75th percentile of the reference population and therefore associated with an increase of cardiometabolic risk (Table 3) [26]. Values are expressed as mean (SD) and median (25th-75th percentile).
The median values of TC, LDL-c, HDL-c, TG resulted normal compared to the reference values ( Table 4). The TG/HDL ratio and AIP (SD) were calculated (  Values are expressed as mean (SD) and median (25th-75th percentile).

Association between Macronutrients and Cardiometabolic Risk Factors
By bivariate correlation, energy intake and total fats intake were both significantly and positively associated with BMI z-score and HOMA-β. Protein intake was positively associated with BMI z-score and HOMA-IR and negatively associated with the QUICK index. Saturated fats intake was positively associated with HOMA-IR and HOMA-β and negatively associated with the QUICK index. Carbohydrates percentage was inversely related to BMI z-score and HOMA-β. No significant association was found between macronutrients intake and TyG index, TG/HDL, AIP, and VAI (Table 5).  Multiple regression was performed to predict the BMI z-score from energy intake and macronutrients intake after adjusting for age and sex. A model comprising energy intake and carbohydrate intake predicted the BMI z-score. Energy intake was positively associated with BMI z-score, while carbohydrate intake (g) was negatively associated with it ( Table 6). A model comprising saturated fats intake (g) and age predicted HOMA-IR, p = 0.001, adj. R 2 = 0.123. Both variables added statistically significantly to the prediction, p < 0.050 (Table 7). Patients with normal saturated fats intake according to age-and sex-specific Italian national dietary reference values for energy and nutrients (<10% of total energy, E, [25]) did not have a significantly higher probability of having a normal HOMA-IR. On the contrary, patients with saturated fats intake <7% of E (as recommended by the National Lipid Association Expert Panel on Familial Hypercholesterolemia [40]) had a significantly higher probability of having a normal HOMA-IR, p = 0.050 (Table 8). Only total fats intake (g) predicted HOMA-β after adjusting for energy intake, other macronutrients intake, age, and sex; p = 0.011, adj. R 2 = 0.060 (Table 9). A model comprising age and saturated fats intake predicted the QUICK index using multiple regression analysis, F (2, 90) = 7.271, p = 0.001, adj. R 2 = 0.120. Both variables added statistically significantly to the prediction, p < 0.050 (Table 10). By multiple regression analysis, a model containing energy intake, macronutrients intake (g), age, and sex could not predict TyG index, TG/HDL ratio, AIP, or VAI.

Discussion
Thirteen adolescents (14%) were diagnosed with MetS in our population. This finding is comparable to the overall prevalence of MetS in other cross-sectional studies conducted in obese adolescents, with rates ranging from 10% to 38% [41][42][43]. The real prevalence of this condition in children and adolescents is hard to estimate due to the lack of a consensus on its definition [44].
In our study, energy intake and carbohydrate dietary amount predicted BMI z-score after adjustment for age and sex, while protein and total fats intakes were predictors only in unadjusted analysis. Therefore, our data suggest that diets with higher energy intake and lower carbohydrate content are a substantial risk factor for developing obesity in adolescents. Adolescent obesity management requires complex efforts, both with lifestyle and behavioral interventions and with nutritional support. The results of our study suggest that a nutritional intervention in obese adolescents should be focused on reducing the energy intake, in agreement with current guidelines that suggest consuming fewer calories, fewer sugars, less saturated fats, more unsaturated fats (including vegetable oils) and eating vegetables and fruits daily [45]. According to a recent metanalysis of nutritional intervention studies among obese children and adolescents, weight reduction can be achieved with a low-energy intake diet, independently of macronutrients distribution [46]. However, our data also suggest that a higher carbohydrate intake is protective against adolescent obesity. Data on optimal carbohydrates intake in adolescents are lacking. A recent systematic review and meta-analysis of eleven observational studies enrolling a total of 153891 obese adults investigated the relationship between carbohydrate intake and obesity. Six studies linked a higher carbohydrate intake to a reduced risk of obesity, while five studies showed an increased risk, with the pooled odds ratio being non-significantly different from 1. The authors suggested that these inconclusive results could be due to differences in the quality of the carbohydrates (refined vs. unrefined) mainly represented in the different studies, and this view seems supported by the literature showing an increased risk of obesity conferred by diets rich in refined sugars, as opposed to diets rich in unrefined sugars [47]. Moreover, diets rich in refined, high-glycemic-index carbohydrates have been linked to an increased risk of myocardial infarction in adults [48].
Moreover, in our cohort of obese adolescents, higher total and saturated fats dietary intakes predicted insulin resistance, even after adjustment for age and sex. An increased intake of total fats was associated with a raised HOMA-β, while an elevated intake of saturated fats was associated with a raised HOMA-IR and a reduced QUICK index. A causal link between total and saturated fats dietary intakes and altered glucose metabolism could be traced in the pro-inflammatory effect of dietary fats. In a cohort of 219 girls aged 12 to 17 years, a western dietary pattern as opposed to a Mediterranean dietary pattern was associated with reduced levels of adiponectin [49]. A cross-sectional study including 532 European adolescents also found that high sugar and fat dietary intakes were associated to higher levels of inflammation [50]. Studies addressing the association between macronutrients intake and glucose metabolism among obese adolescents are scarce. In a large cohort of adolescents (with subjects included independently of their BMI), an energy-dense, high-fat, and low-fiber dietary pattern was positively associated with higher levels of insulin and HOMA-IR [51]. Moreover, a recent Brazilian cross-sectional study found that in a cohort of adolescents (again, including subjects independently of their BMI), an elevated saturated fat intake was associated with insulin resistance [52]. In adults, different studies on large cohorts and with long follow-up have demonstrated that a low-fat (but high-carbohydrate) diet is effective in the prevention of type 2 diabetes mellitus, reporting reductions in its incidence of up to 58% [53][54][55][56]. In particular, a recent systematic review and meta-analysis of observational studies in adults showed that the effect on the cardiovascular risk of replacing saturated fats in the diet depends on the type of quality carbohydrate used for replacement. Only the replacement with low-glycemic-index carbohydrates is able to decrease the risk [57]).
According to our knowledge, few studies evaluated the association between macronutrients intake and cardiometabolic risk factors in obese adolescents, and in this view our results could represent a novelty in the field.

Conclusions
Our results are hypothesis-generating and suggest that in obese adolescents, a dietary intervention targeted at reducing calories intake while increasing high-quality carbohydrates intake (but with limited sugars) could reduce the ponderal excess. Moreover, lowering fat intake could result in improved glucose tolerance, reduced risk of developing diabetes mellitus, and thus, reduced cardiometabolic risk in adulthood. In particular, dietary patterns with saturated fats intake <7% E (according to the heart-healthy fat-modified diet [40]) could be considered for obese adolescents to reduce their cardiovascular risk. Further studies are therefore needed to assess the impact of nutritional interventions on cardiometabolic risk factors in adolescents.

Limitations
Our study has some limitations, including its observational nature and the lack of a control group for ethical reasons, considering blood sampling as an invasive procedure for healthy adolescents. Besides, the number of study participants was limited; therefore, the results of our analyses need to be replicated in a larger cohort. The cross-sectional design could not determine causal relationships between the analyzed variables. Physical activity was not assessed, and therefore the analyses did not include it. In addition, there are no unique definitions for metabolic syndrome and cardiometabolic risk assessment in adolescents. Another limitation is that the gold standard to assess food intake is the 3-day food record; however, the Frequency Food Questionnaire is largely used and in the present study was associated with a 24 h recall to standardize the usual serving size. Moreover, data regarding the socioeconomic and educational levels of the families, their knowledge of healthy diet, and their access to quality food are lacking.

Conflicts of Interest:
The authors declare no conflict of interest.