Nutrient Patterns and Their Relation With Obesity in Iranian Adults: A Population-based Study

Objective: In spite of growing evidence on the associations between nutrient patterns and obesity. A few study examined the association between patterns of nutrient intake and obesity we aimed to explore the association between nutrient pattern and obesity in Iranian adults. Results: In this cross-sectional study, a total of 850 subjects aged years 20 to 59 were randomly selected. Our statistical analysis revealed three major nutrient patterns that show the principle factor loading of nutrient intake. We observed a signicant association between quintiles in men for Fasting Blood Sugar (FBS) (P<0.006) in the rst nutrient pattern. Moreover, we identied a signicant difference between quintiles in the rst nutrient pattern in women for obesity (P=0.007), Systolic Blood Pressure (SBP) and Diastolic Blood Pressure (DBP) (P<0.02). In the third nutrient pattern, we identied a signicant association for SBP in women (P<0.02). of dietary consumption and components of obesity were done in the nutrient pattern categories using Analysis of Variance (ANOVA) test and reported using mean ± standard deviation (SD) values. To determine any association between nutrient patterns and general or abdominal obesity, we used binary logistic regression, with three adjustment steps: First, unadjusted and then in the second model, we further controlled for age, total energy intake. Finally in the third model, additionally adjusted for current smoking, job status, education level, and physical activity. These analyses were done for both genders. In these analyses, the rst quintile of the nutrient pattern scores was considered as the reference category. Clinical and demographical characteristics of participants based on obesity status were compared according to the classication of people into two groups of healthy and obese people separately. For this purpose, we used the independent sample t-test and Chi-square test. All statistical analyses were performed with Statistical Package for Social Science (SPSS version 24.0). Statistical signicance was dened as P ≤ 0.05.


Introduction
Obesity and overweight are identi ed as two major public health concerns which its prevalence is increasing at an alarming rate in both developed and developing countries [1]. According to the World Health Organization (WHO) report, in 2016, 39% and 13% of adults in the world were overweight and obese, respectively [1].
Some studies demonstrated a direct relationship between certain nutrients, particularly dietary fat [2] and carbohydrates [3], and the risk of obesity, while other studies indicated an inverse association between intake of dietary proteins [4] and ber [5] or individual micronutrients, including vitamins A, B, C [6,7] and D [8], and minerals, such as calcium [9], with obesity.
The preceding study indicated that a nutrient pattern with high thiamine, betaine, starch, folate, iron, selenium, niacin, calcium, and manganese was associated with a lower risk of general obesity, while a pattern characterized by high amounts of glucose, fructose, sucrose, vitamin C, potassium, total dietary ber, copper and vitamin K was related with greater odds of general obesity, in men, but not in women [10].
Another study showed that nutrient pattern which was high in micronutrients and vitamins reduced odd of obesity and nutrient pattern characterized by high saturated and mono Saturated Fatty Acids (SFAs) was associated with a higher risk of obesity [11]. Studying patterns of nutrient intake to provide insights into combinations of nutrients might affect obesity risk. Therefore, we aimed to investigate the association of nutrient patterns with obesity as a prevalent condition.

Subjects
This cross-sectional study was conducted among 850 apparently healthy individuals of both sex, aged 20-59 years old who referred to Health Centers in Tehran. Participants with a history of diabetes, cancer, and CVD were excluded because of possible changes in their diet. Also, people in the age range of 20 to 59 years old,

Assessment of dietary intake
The usual diet of participants was assessed with three 24-hours recall. The rst 24-hours recall was collected by interview and the other two recalls by a phone call to the participants on random days of the week. Finally, we extracted the meals and food groups from these questionnaires.

Assessment of anthropometric measures
Weight was measured by a digital scale with a sensitivity of 0.1 kg (seca 808, Germany), while the subjects were minimally clothed and not wearing shoes. Height measurement by wall stadiometer with a sensitivity of 0.1 cm (Seca, Germany). BMI was calculated and expressed in kg/ . Waist (WC) and hip circumference (HC) were measured in the smallest girth and the largest girth, respectively, with accuracy nearest to 0.1 cm.

Assessment of biomarkers
A blood sample was drawn about 10 cc between 7 am to 10 am from all study participants after they fasted overnight for 12 h. After testing blood sample, people with a blood sugar above 126 mg, individuals with a history of diabetes and those taking blood sugar lowering medications, they are considered as diabetic patients. Serum glucose and lipids are measured using enzymatic methods, commercial kits (Pars test, Iran), and automatic devices (Selecta E, vitalab, Netherland). These tests are performed on the same day.

Results
Clinical and demographical characteristics of participants based on obesity status are shown in Table 1. The mean age, weight, BMI, WC, Triglycerides (TG), DBP, SBP (P < 0.001 for all comparisons), WHR (P = 0.002), FBS (P = 0.01) and Total Cholesterol (TC) (P = 0.03) among obese subjects were higher than healthy subjects. Additional le 1: Table S1 shows the principal factor loading of nutrients intake. Positive loading demonstrated strong associations between the nutrient groups and nutrient patterns, while negative loading demonstrated negative associations. The rst nutrient pattern was characterized by high positive loading for the consumption of vitamins B1, B2, B6, B5, B3, B12, zinc, iron, SFAs, and protein. The second nutrient pattern showed positive loading for the consumption of zinc, SFAs, vitamin E, α-tocopherol, oleic acids, Polyunsaturated Fatty Acids (PUFAs), β-carotene, linolenic acids, Monounsaturated Fatty Acids (MUFAs), total fat. The third nutrient pattern was represented mainly by positive loading for potassium, magnesium, phosphorus, calcium, protein, carbohydrate, vitamin C, and folate.
Additional le 2: Table S2 shows the nutrient intake based on quintiles of nutrient patterns. In the rst nutrient pattern, the individuals in the fth quintile had a higher intake of vitamins B1, B2, B3, B5, B6, B12, zinc, iron, SFAs, molybdenum, vitamin E, PUFA, total fat, potassium, magnesium, phosphorus, calcium, carbohydrate, sodium, cholesterol, protein and had a lower vitamin K, caffeine and β-carotene. In the second nutrient pattern individuals in the rst quintile had less consumption of zinc, SFAs, vitamin E, α-tocopherol, oleic acid, PUFA, β-carotene, linolenic acid, MUFA compared to the fth quintile. Furthermore, in the third nutrient pattern, the individuals in the fth quintile had a higher intake of potassium, magnesium, phosphorous, calcium, protein, carbohydrate, vitamin C, folate compared to other quintiles. Components of obesity across based on gender and across quintiles of nutrient patterns are shown in Table 2. We observed a signi cant association between quintiles in men for Fasting Blood Sugar (FBS) (P < 0.006) in the rst nutrient pattern. Moreover, we identi ed a signi cant difference between quintiles in the rst nutrient pattern in women for SBP and DBP (P < 0.02). In the third nutrient pattern, we identi ed a signi cant association for SBP in women (P < 0.02).
There were no signi cant differences between other components of obesity based on quintiles of nutrient patterns. We examined the association between obesity, according to quintiles of nutrient patterns for both genders, by using an unadjusted, partially, and fully adjusted model. For every three nutrient patterns, we identi ed no effect of the nutrient pattern for men and women, even after adjustment for confounders in the second and third nutrient patterns. We identi ed a signi cant difference between quintiles in the rst nutrient pattern in women (Model 1, P = 0.01). The result was similar after adjustment for confounding factors (Model 2, P = 0.01 and Model 3, P = 0.007) ( Table 3). Characteristics of study participants across quintiles of major nutrient pattern scores are shown in Additional le 3: Table S3. There were signi cant associations for gender between quintiles in the rst nutrient pattern (P < 0.006) and second nutrient pattern (P < 0.01), but not in the third (P < 0.09). Moreover, there was a signi cant association for job status between quintiles in the rst nutrient pattern (P < 0.05), but not in the second and third. There was no signi cant association for smoking, education, marriage, and physical activity between quintiles in the three nutrient patterns.

Statistical method
To identify the nutrient pattern in our study population, the principal component analysis was used. Also, we applied factor analysis with the varimax procedure to derive nutrient patterns based on the nutrients' loading factors. Comparison of dietary consumption and components of obesity were done in the nutrient pattern categories using Analysis of Variance (ANOVA) test and reported using mean ± standard deviation (SD) values. To determine any association between nutrient patterns and general or abdominal obesity, we used binary logistic regression, with three adjustment steps: First, unadjusted and then in the second model, we further controlled for age, total energy intake. Finally in the third model, additionally adjusted for current smoking, job status, education level, and physical activity. These analyses were done for both genders. In these analyses, the rst quintile of the nutrient pattern scores was considered as the reference category. Clinical and demographical characteristics of participants based on obesity status were compared according to the classi cation of people into two groups of healthy and obese people separately. For this purpose, we used the independent sample t-test and Chi-square test. All statistical analyses were performed with Statistical Package for Social Science (SPSS version 24.0). Statistical signi cance was de ned as P ≤ 0.05.

Discussion
The present study resulted in a signi cant positive association between quintiles in men for FBS in the rst nutrient pattern and a signi cant difference between quintiles in the rst nutrient pattern in women for blood pressure. Moreover, the women in the top quintile of the rst nutrient pattern compared to the rst quintile have more chance to be obese even after adjustment for covariates.
Some community-based studies are done to investigate the association between nutrient patterns and obesity. For instance, in one study vegetarians consume more legumes, fruit, and whole grains and less re ned cereals, desserts, fried foods, fruit juice less had a low chance to be overweight/obese compared to those consuming other nutrient patterns groups [12]. A Chinese study revealed that the traditional south pattern including rice as the key staple food with chicken and vegetables was associated with a lower risk of general and abdominal obesity [13].
In our study, the rst nutrient pattern loaded mainly on vit (B1, B2, B6, B5, B3, B12) and zinc. The association of low levels of B vitamins such as B1, B2, and B6 and obesity was reported in some publications [14][15][16]. Low thiamine, niacin, and pyridoxine concentrations are linked to metabolic dysfunction like obesity [14][15][16]. Obese patients may have a greater amount of body thiamine reservation in cells that trigger lower plasma thiamine concentration [15]. Elevated total body water in obese subjects may also result in dilution of the extracellular compartment, further decreasing the plasma vitamin levels [14]. However, in a randomized placebo-controlled clinical trial conducted by Dollerup et al [17], supplementation with 1000 mg nicotinamide riboside two times per day for 12 week period had no signi cant effect on the body composition measures [18]. Moreover, the existence of abdominal obesity and elevated body fat were positively related to serum zinc concentrations in men. But, this association was not found in women [19]. In contrast, a cross-sectional study observed an inverse association between the BMI and plasma zinc levels [20].
Our second nutrient pattern (mainly representative of vitamin E, α-tocopherol, and unsaturated fatty acids such as oleic acid and PUFA s ) is positively associated with body weight and height. In contrast to our result, Amini MB et al concluded that serum vitamin E is negatively associated with WC, and Hip Circumference (HC). The reduction observed in body weight was not signi cant [21]. The ratio of omega-6/ omega-3 PUFA s is a matter of importance. A diet which contains usually meat rather than sh can lead to an imbalance in that ratio. The fat reservation resulted from dietary omega-6 PUFA s may have more progression in comparison with long-chain omega-3 or SFA s when consumed besides a relatively high carbohydrate diet [22]. In a randomized cross-over study, overweight or obese men substituting a diet high in SFA with MUFA in four weeks duration, signi cantly decreased body weight and fat mass whereas the former amount of calories and fat being used [23].
The third nutrient pattern mainly containing potassium, magnesium, phosphorus, calcium, and protein. It could be expected that this nutrient pattern is likely the Mediterranean diet. This diet mostly includes a high intake of fruits and vegetables which are a rich source of minerals mentioned in our third food pattern. In a cohort study, men who had high adherence to the Mediterranean diet had a low risk of becoming obese up to 29% [24].

Conclusion
In conclusion, we observed a signi cant association between quintiles in men for FBS in the rst nutrient pattern. Moreover, we identi ed a signi cant difference between quintiles in the rst nutrient pattern in women for obesity, SBP and DBP. In the third nutrient pattern, we identi ed a signi cant association for SBP in women. More researches for nutrient patterns in the isocaloric clinical trials seem to be helpful to extend the presented data on the Iranian habitual diet in order to found nutritional recommendations to prevent obesity in the community.

Limitation
To state the strengths and limitations of our study, although the sample size was small, it was enough according to statistical calculations. We aware some limitations are inevitable. For instance, the effects of cooking on the bioavailability of the nutrients were not possible to measure, however, we try to control for possible potential confounding. The cross-sectional design of the study does not permit inference about causality. It is critical to note that because the nutrient intake was evaluated by an FFQ, recall bias is unavoidable.