The Relationship between Snacking and Risk of Individual Components of Metabolic Syndrome in Normal-weight Adults: A Cross-sectional Study
Ahmad Zare Javid; PhD¹, Nasim Niknejad; MSc^2*, Hajieh Shahbazian; MD²,
Seyed Mahmoud Latifi; PhD², Bahar Niknejad; MD³ & Razie Hormoznejad; MSc^4
1 Nutrition and Metabolic Diseases Research Center & Hyperlipidemia Research Center, Ahvaz Jundishapur University of Medical Sciences, Ahvaz, Iran.
² Health Research Institute, Diabetes Research Center, Ahvaz Jundishapur University of Medical Sciences, Ahvaz, Iran.
³ Weill Cornell Medical College, New York, New York, USA.
⁴ Student Research Committee, School of Paramedicine, Ahvaz Jundishapur University of Medical Sciences, Ahvaz, Iran.

 
Introduction

etabolic syndrome (MetS) is associated with cardiovascular risk factors and type 2 diabetes mellitus which include abdominal obesity, elevated blood pressure (BP), fasting blood glucose (FBG), triglyceride (TG), and low level of high-density lipoprotein cholesterol (HDL-c).(Shahbazian et al., 2013) Approximately 50% of patients with type 2 diabetes and  severe coronary syndrome as well as 95% of patients with peripheral arterial disease suffer from MetS (Qadan et al., 2008, Zaliūnas et al., 2007).
Obesity is an important public health challenge worldwide and more than one billion people are predicted to be obese in 2030 (Esteghamati et al., 2009, Kelly et al., 2008). Obesity is associated with higher risk of several diseases such as non-alcoholic fatty liver disease, type 2 diabetes mellitus, MetS, and higher all-cause mortality (Kelishadi et al., 2008, Kwon et al., 2013, Phillips and Perry, 2015). Although the standard definition for obesity is to have high levels of body fat, epidemiologists believe that  the definition of obesity is having a body mass index (BMI) ≥ 30 (Romero-Corral et al., 2010).
Obesity is classified into three subtypes based on its metabolic risk which include: 1. Obese individuals with MetS, 2. Obese individuals without MetS, and 3. Normal-weight individuals with MetS (De Lorenzo et al., 2006). The condition of having obesity-related metabolic disturbances in spite of normal weight (BMI < 30 kg/m²) was first suggested by Ruderman in 1981 (Ruderman et al., 1981). Metabolically obese, normal weight (MONW) is a subgroup of obesity that cannot be revealed by routine measurements such as body weight, BMI, skin fold thickness, and body fat mass (Ruderman et al., 1981).
The prevalence of MetS in developed countries is reported as 25% (National Cholesterol Education Program Expert Panel on Detection and Treatment of High Blood Cholesterol in, 2002), while this rate has been estimated at 22.5-32.1% in different regions of Iran (Sarrafzadegan et al., 2011, Shahbazian et al., 2013, Sharifi et al., 2009, Zabetian et al., 2007). According to NHANES III, the prevalence rates of MONW condition among normal BMI men and women were 4.6% and 6.2%, respectively (Park et al., 2003). The high occurrence rates of MetS in normal-weight Iranian men and women in Tehran were respectively 9.9 and 11.0% that is a great cause of concern (Hadaegh et al., 2007b).
Evidences indicated that several risky characteristics reduce insulin sensitivity and increase visceral fat accumulation (Bednarek-Tupikowska et al., 2012, Katsuki et al., 2003, Ruderman et al., 1998). These risky characteristics include: high carbohydrate diet and high sucrose consumption (Ruderman et al., 1981), smoking and elevated hs-CRP (Kwon et al., 2013), reduced basal metabolic rate (De Lorenzo et al., 2006), low lean body mass (De Lorenzo et al., 2006, Marques-Vidal et al., 2010), low physical activity (Ruderman et al., 1998, Ruderman et al., 1981), and reduced VO₂max and aerobic fitness (Ruderman et al., 1998). The specific ethnicities, gender, and age groups (Bednarek-Tupikowska et al., 2012, Kelishadi et al., 2008)  also had a higher percentage in MONW individuals and ultimately lead to increased risk of CVD. In contrast, metabolically healthy obese (MHO) phenotype is a subgroup of obesity with normal metabolic profile despite having BMI ≥ 30 kg/m² (Primeau et al., 2011). Evidences further showed that arterial stiffness, carotid atherosclerosis, and risk of fatal CVD events are higher in MONW than MHO participants within 10 years (Yoo et al., 2014).
Diet, physical inactivity, and genetic susceptibility can be the causes of unhealthy metabolic profile among normal-weight adults (Hadaegh et al., 2007b, Roche et al., 2005). Sedentary lifestyle and  transition of nutrition from healthy food choices to westernized diet, such as higher consumption of industrial and processed foods, sweetened artificial beverages, and junk foods are the most significant factors contributing to the alarming trend of MetS (Naeem, 2012, Popkin et al., 2012).
Having specific dietary intake and eating behavior patterns are significant factors in initiation and development of insulin resistance and MONW (Choi et al., 2012). A few studies have examined the relationship between dietary patterns and MONW (Choi et al., 2012, O'Connor et al., 2015, Suliga et al., 2015). Suliga et al. reported that dietary patterns involving high consumption of fish and whole grains as well as reduced intake of sugar, sweets, refined grains, and cold cured meat are associated with lower risk of MONW (Suliga et al., 2015).
In recent years, increased frequency of both healthy and unhealthy snacking, portion sizes, trans and saturated fats, total fats, and added sugar content of snacks consumed by the public have been a major warning towards the ascending trend of obesity and metabolic disorders (Mirmiran et al., 2014, O'Connor et al., 2015). Snacking habit among USA adults increased from 71% to 97% and the contribution of snacking to the total energy intake increased from 18% to 24% from 1997 to 2006 (Piernas and Popkin, 2010). Diets with high-energy and low-nutrient density are also the major factor in rising trend of non-communicable diseases among Iranian population (Ghassemi et al., 2002). A recent study in Iranian population showed that high intake of energy-dense snacks, especially salty types is significantly associated with high rate of MetS after a 3-year follow-up (Mirmiran et al., 2014). Though there are various definitions for a ‘snack’, according to O’Conner et al., snacks can be defined as “energy-dense and nutrient-poor foods, commonly referred to as snack foods, and also low energy-density and high-fiber foods such as fruits” (O'Connor et al., 2015).
To the best of our knowledge, limited facts have been found concerning the relationship between snacking and MONW condition or each MetS component in normal-weight Iranian population. Therefore, the present study targeted at investigating the relationship between snacking and each MetS component in normal-weight adults in southwest of Iran.
Materials and Methods
Study design and population: This study was performed with random cluster sampling in Ahvaz, southwest of Iran in 2015.  Six health centers were randomly chosen from 25 health centers, 55 households were then randomly selected from each health center and invited to take part in the examinations. A total of 780 individuals participated in the study. Male or female participants older than 20 years with normal BMI (18.5 ≤ BMI ≤ 24.9 kg/m²) were included for further investigation. Individuals with incomplete questionnaires or missing anthropometric, BP, or biochemical measurements were excluded. Finally, data collected from 328 normal-weight individuals were suitable for statistical analysis and written consent was obtained from all participants.
Other variables such as age, gender, ethnicity (Persian, Arabian, others), marital status, education, regular physical activity (defined as doing at least 30 minutes of brisk walking or high level activity three times or more per week) (Fan et al., 2016), and smoking status (never, former, and current) were obtained by a trained interviewer.
Measurements: Anthropometric indices including weight, height, and waist circumference (WC) were measured by trained technicians. Weight was measured by an analog scale (Seca, Hamburg, Germany) to the nearest 0.1 kg with light clothing. Height was then measured with barefoot by a wall-mounted stadiometer (Seca model 220, Hamburg, Germany) to the nearest 0.1 cm while head was in Frankfort horizontal plane and 4 sites touching the wall (head, shoulder blades, buttocks, and heels). BMI was calculated by dividing weight (kg) by the square of height (m²).  WC was measured just above the right iliac crest at the end of a normal expiration to the nearest 0.1 cm using a non-elastic tape measure. All anthropometric indices were measured on the basis of world health organization standards (WHO, 1987).
The BP of each participant was measured by a standard mercury column sphygmomanometer on the right arm at sitting position after a minimum of 5 minutes of rest. Systolic and diastolic BP (SBP and DBP respectively) were measured twice to the nearest 2 mmHg, and mean of BP was taken as participant’s arterial BP. The initial Korotkoff sound phase was accepted as SBP and disappearance of sound (the fifth phase) was recorded as DBP.
Blood samples were collected, centrifuged, stored in refrigerator, and then sent to Diabetes Research Center laboratory for biochemical analysis after 10 to 12 h of fasting. Serums
were analyzed for TG, FBG, and HDL-c levels
through enzymatic calorimetric method by an auto-analyzer with Pars Azmoon kits (Pars Azmon Inc., Tehran, Iran).
The information on dietary consumption was obtained using a validated 50-item non-quantitative food frequency questionnaire (FFQ). The FFQ used in this research was a short form of a 168-item semi-quantitative FFQ formerly used by Mirmiran et al. in Tehran Lipid and Glucose Study (Mirmiran et al., 2010). Reliability of the 50-item FFQ is optimal, with Cronbach’s alpha and Split-half reliability estimates of 0.825
and 0.732, respectively. Validity of FFQ for measuring intake of 50 food items was confirmed by 5 nutrition professionals. These results indicate that the 50-item FFQ is a valid and reliable tool for dietary assessment among people residing in southwest of Iran.
The FFQ considered dietary intakes as the frequency of consumption of a medium-size serving of each food item during the previous year on daily, weekly, monthly, or yearly bases. Afterwards, all frequencies were converted to daily intake frequency.
"Snack" was defined by O’Conner et al. as “energy-dense and nutrient-poor foods and also low energy-dense and high-fiber foods such as fruits” (O'Connor et al., 2015). Investigating each individual type of energy-dense snack compared
to total amount of snacks consumed will give
more distinct and accurate results concerning
its subsequent cardio-metabolic risks (Mirmiran et al., 2014).
MetS components: According to international diabetes federation (IDF) consensus worldwide, the definition of MetS components included: 1. Central obesity determined by WC ≥ 94 cm in males and ≥ 80 cm in females (Alberti et al., 2009), 2. Elevated TG level ≥ 150 mg/dL, 3. Reduced HDL-c level <4 0 mg/dL in males and < 50 mg/dL in females, 4. Elevated FBG level ≥ 100 mg/dL or drug treatment for elevated FBG, 5. Elevated BP, SBP ≥ 130 mmHg or DBP ≥ 85 mmHg or antihypertensive drug treatment.
As suggested by Hadaegh et al. WHO's WC cut-off points are not suitable for Iranian population (Hadaegh et al., 2007a). Hence, for abdominal obesity, the Middle East specific WC that is a compulsory component of MetS definition proposed by IDF was applied (Alberti et al., 2009).
Data Analysis: SPSS software version 22 was used to conduct the statistical analysis. Chi-square test and t-test were applied to compare the categorical and continuous baseline characteristics, respectively. Bonferroni post-hoc test was performed for further comparisons between different age groups and different education levels. The relationships between MetS components and the type of consumed snacks were evaluated using multiple binary logistic regression model adjusted for potential confounders including age, gender, ethnicity, marital status, education, smoking, physical activity, and BMI. In multiple regression model, different types of snacks were considered as continuous variables so they were entered in unique blocks.  Afterwards, the intake frequency of snack types with significant association with MetS components was divided into tertiles based on their 33^th and 66^th percentile values. Then, the relationships between MetS components and each tertile category of snacks were evaluated using multiple binary logistic regression model adjusted for potential confounders. The level of statistical significance was set at P-value < 0.05 and 95% confidence interval (CI) was used.
Ethical consideration: The research was approved by the research ethics committee of Ahvaz Jundishapur University of Medical Sciences (ethical approval reference number 193043).
Results
A total population of 328 normal-weight adults including 157 (47.9%) males and 171 (52.1%) females were investigated in this study. The prevalence of MetS for total, male, and
female participants were 12.5%, 6.4%, and 18.1%, respectively. Mean ± SD of age was 39.2 ± 16.8 for all participants, 42.4 ± 18.0 in males, and 36.0 ± 15.0 in females (P = 0.001). Regarding the metabolic health, male participants had higher TG, SBP, DBP, WC, and lower HDL-c level when compared with female participants (P < 0.05). However, there was no significant difference in FBG and BMI between the two groups (P > 0.05). A greater percentage of smoking was observed in males than females (P = 0.0001), while no significant difference was observed between proportion of different education levels and regular physical activity between the two genders (P > 0.05). Furthermore, 89.6% of participants were educated, 50% of whom had university degree. Regarding the smoking status, 16.6% of males were current or past smokers while none of the female participants has ever been smokers. Regular physical activity was achieved only in 6.7% of participants (Table 1).
According to Table 2, 26.2% of participants were hyperglycemic. The proportion of hyperglycemia was significantly different based
on gender, age groups, and educational levels
(P < 0.05). The 20-34 years old participants with high FBG level were significantly lower in proportion than the other 3 age groups (P = 0.0001). The proportion of 35-49 years participants with high FBG level was also lower than ≥ 65 years participants (P = 0.008, chi-square test with Bonferroni correction). The proportion of hyperglycemic participants was significantly different among all 3 education levels (P = 0.0001). So, 20.1% of the study population had hypertriglyceridemia The proportion of participants with hypertriglyceridemia was significantly different between different genders and different age groups (P < 0.05); it was specifically lower in 20-34 years age group than the 50-64 and ≥ 65 years age groups (P = 0.002). Low HDL-c level was observed in 48.2% of individuals. The occurrence of low HDL-c was only different between males and females (P = 0.003). Hypertension was observed in 19.8% of participants. The proportion of hypertensive participants were significantly different between different age groups and different education levels (P = 0.0001). The proportion of hypertensive participants of 20-34 and 35-49 years age groups were significantly lower than that of both 50-64 and ≥ 65 years age groups (P = 0.0001). The proportion of hypertensive illiterate participants was significantly higher than that of 2 other education levels (P = 0.0001). Abdominal obesity was seen in 27.4% of participants. The proportion of abdominal obesity was significantly different between different genders, age groups, education levels, and smoking categories (P < 0.05). The proportion of abdominal obesity in 20-34 years age group was significantly lower than other 3 age groups (P < 0.05). The proportion of abdominal obesity was significantly different among all 3 education levels (P < 0.05). No other significant difference was observed for the proportion of MetS components among categories of demographic variables.
According to Table 3, there was a significant increasing trend for the occurrence of abdominal obesity and hypertension across increasing tertile of supermarket cakes and biscuits (OR = 1.23; 95% CI: 1.02–1.49) as well as chocolate (OR = 1.10; 95% CI: 1.03-1.18), respectively. Other snack types had no significant relationship with MetS components.
Discussion
In this study carried out on normal-weight adults, high intake of supermarket cakes and biscuits, as an important source of refined grains, is associated with higher risk of abdominal obesity. In addition, high intake of commercially available chocolate products is associated with high BP measurements.
It is noteworthy that although no significant relationship is observed between regular physical activity and MetS components, low percentage of physical activity (6.7%) in the studied population, compared to the global rates is a cause for concern.
Commercially available cakes and biscuits consumed by the general population are usually
a source of refined grains. A number of observational and interventional studies validated the dependencies between whole grains consumption and reduction in WC measurement when compared with refined grains (Jonnalagadda et al., 2011, Koh-Banerjee et al., 2004, Kristensen et al., 2012, Landberg et al., 2010, Shimizu et al., 2008). In a meta-analysis performed by Harland et al. consumers with higher intake of whole grains had lower WC and waist to hip ratio compared with consumers with lower amounts of intake (Harland and Garton, 2008). On the other hand, Halkjaer et al. found no significant relationship between the consumption of whole grains and forthcoming WC change, though the consumption of refined grains revealed a positive relationship with WC change in women (Halkjær et al., 2006). A 12-week whole grains dietary intervention conducted on 40-65 years old adults with MetS resulted in no change in WC measurements when compared to the refined grains group (Giacco et al., 2013).
The results of present study revealed a positive relationship between consumption of cakes and biscuits and WC in normal-weight adults. O’Neil et al. reported that the relationship between the consumption of whole grains and decreased WC reduced to nil after adjustment for cereal fiber (O'Neil et al., 2010). The probable cause for the increase in WC is low fiber and high content of refined grains in common cakes and biscuits obtainable in stores.  Furthermore, supermarket cakes and biscuits may have high amounts of added sugar, fat, and sodium (Williams, 2012). Dietary guidelines usually differentiate cereal products with much added fat and sugar content (e.g., cakes, biscuits, pizza and pastries) from
plain cereal products (Williams, 2012). One of
the fundamental mechanisms for reducing the percentage of abdominal fat by the consumption of whole grains is a reduction in insulin and glucose responses and the subsequent enhancement in lipid oxidation and lipolysis (Pauline and Rimm, 2003, Pol et al., 2013).

 

^a :  χ² test for categorical and Student t-test for continuous variables, ^b: number (%),^c: mean±SD
 

 

^a: fasting blood glucose ≥100 mg/dL or drug treatment for elevated it; triglyceride ≥ 150 mg/dL; High density lipoprotein < 40 mg/dL in male and < 50 mg/dL in female; Bood pressure: Systolic ≥ 130 mmHg or Diastolic ≥ 85 mmHg or antihypertensive drug treatment; Waist circumference ≥ 94 cm in male and ≥ 80 cm in female ^b : All values are frequency (%), ^c:  χ² test., Significant differences in multiple comparisons between different age groups (P-value < ), and education levels (P-value < ) obtained by post-hoc Bonferroni test, ^d: Smoking: Yes, current or past smoker; No, never smoker.
 

^a: fasting blood glucose ≥100 mg/dL or drug treatment for elevated it; triglyceride ≥ 150 mg/dL; High density lipoprotein < 40 mg/dL in male and < 50 mg/dL in female; Bood pressure: Systolic ≥ 130 mmHg or Diastolic ≥ 85 mmHg or antihypertensive drug treatment; Waist circumference ≥ 94 cm in male and ≥ 80 cm in female, ^b: OR (95% CI) are obtained by multiple binary logistic regression adjusted for age, gender, ethnicity, marital status, education, smoking, physical activity, and BMI, ^c : ORs in T2 and T3 rows indicate the risk of incident metabolic syndrome component among those in the 2^nd and 3^rd tertile of the independent variable, respectively, compared with those in the 1^st tertile. T1: 1^st tertile; T2: 2^nd tertile; T3: 3^rd tertile. ^d :  Significant correlation with the MetS component, NS: Non-significant