Dietary insulin index, dietary insulin load and dietary patterns and the risk of metabolic syndrome in Hoveyzeh Cohort Study

Postprandial insulin secretion has been associated with metabolic disorders such as hyperlipidemia and type 2 diabetes. Therefore, we aimed to explore the relationship between dietary insulin indices and dietary pattern with the risk of Metabolic Syndrome (MetS). The participants of the present cross-sectional study were included among the individuals who participated in the Hoveyzeh Cohort Study (HCS). A total of 3905 Iranian adults, aged 35–70 years, are included in the current analysis. The Food Frequency Questionnaire (FFQ) is used to calculate the dietary Insulin Index (DII), Insulin Load (DIL), and dietary pattern. Dietary pattern was derived using Reduced-Rank Regression (RRR) based on intake of protein (g/day), fiber (g/day), fat (g/day), magnesium (mg/day), and dietary insulin index were considered as response variables. The Generalized Linear Model was used to obtain the odds ratio (OR) and 95% confidence interval (CI) for MetS based on gender, while considering quartiles of DIL, DII scores, and dietary pattern, adjusted for potential confounders. The mean ± SD of age and BMI of the participants in the top quartile of DIL were 45.72 ± 8.05 years and 28.25 ± 5.02 kg/m2, respectively. The mean ± SD of DII was 40.53 ± 4.06 and the mean ± SD of DIL was 117,986.1 ± 30,714.06. A significant positive association was observed between DIL and MetS in women after adjusting for confounding factors (OR: 1.51; 95% CI 1.16; 1.96). No significant association was seen between DIL, DII, and MetS among men. A derived dietary pattern characterized by high intakes of fruits, sugar, sweet deserts, Whole Grains, and dairy was associated with an increased risk of MetS in adjusted model2 among women (OR: 1.41; 95% CI 1.13; 1.75) and men in the same model (OR: 2.09; 95% CI 1.35; 3.21).However, the final model was significant just for men (OR: 2.08; 95% CI 1.35; 3.21) and not for women (OR: 1.24; 95% CI 0.96; 1.60). Our findings showed that adherence to a diet with a high insulin load can increase the risk of MetS in women. In addition, a derived dietary pattern by RRR indicated that a diet rich in fruits, sugar, sweet deserts, whole Grains, and dairy is related to increased risk of MetS in both men and women.


Study participants
The individuals who participated in the present cross-sectional study were selected from the Hoveyzeh cohort study (HCS).The HCS is a part of the PERSIAN cohort study that has been carried out in 18 different regions of Iran, including 180,000 Iranian adults 35 .Detailed information of HCS have been reported previously 36 .The HCS is a population-based cohort study conducted in Hoveyzeh County, focusing on common chronic diseases, disorders, and risk factors of non-communicable diseases (NCDs) in the Arab ethnicity from May 2016 to August 2018.At first, a total of 10,009 adults met the inclusion criteria and were living in the Hoveyzeh area, and they all agreed to take part in the cohort study.The requirements for participation included being of Iranian heritage, between the ages of 35 and 70, and residing in Hoveyzeh for at least 9 months out of the year.These individuals were invited to the cohort center located in the Hoveyzeh region to provide data.The trained interviewers gathered data on sociodemographic characteristics, physical activity, and dietary intake.Among 10,009 participants, individuals with diabetes (n = 2226), cardio vascular diseases (n = 1704), renal failure (n = 123), gallstone disease (n = 297), chronic lung disease (n = 585), any type of cancers (n = 37), pregnancy and/or lactating (n = 163), reported daily energy intakes outside the range of 800-4200 kcal/day (n = 856) 37 and incomplete FFQ (n = 113) were excluded.
Finally, of the initial 10,009 participants, 3905 persons were included in the final analysis (Fig. 1).The Ethics Committee of Jundishapur University of Medical Sciences in Ahvaz, Iran approved the Hoveyzeh Cohort Study (IR.AJUMS.REC.1399.738).All individual participants included in the study provided informed consent.

Biochemical and blood pressure assessment
Participants fasted for 10-12 h before collecting biological samples.Trained staff took blood, urine, hair, and nail samples immediately after enrollment.To separate serum from blood, samples were centrifuged at 1000 rpm for 15 min and stored at − 80 °C until analysis.Details of the procedure are included in the preliminary study 36 .Furthermore, we assessed Fasting Blood Sugar (FBS) utilizing the glucose oxidase technique.Enzymatic kits from Pars Azmoon in Iran were employed to determine Total Cholesterol (TC) and HDL levels.Additionally, we recorded the systolic and diastolic blood pressure of participants while seated, twice from each arm, employing standard mercury sphygmomanometers with a ten-minute interval between readings.The average of the two measurements was reported.

Anthropometric measurements and physical activity
Trained interviewers conducted face-to-face interviews using pretested questionnaires to gather demographic variables, medical histories, medication, and smoking habits.Anthropometric data were collected in the morning after biological sampling, as this reduced measurement errors or biases.The participants' height (cm) was determined by employing a stadiometer (Seca 206), whereas their weight (kg) was measured using a standing scale (Seca 755).Additionally, their waist, hip, and wrist circumferences in centimeters were gauged using Seca locked tape meters.Following the measurements, participants were treated to breakfast before they were interviewed using a questionnaire.A Modifiable Activity Questionnaire (MAQ), which previously modified and validated among Iranians were used for physical activity measurement 38 .The participants were asked to report their frequency and duration of engagement in light, moderate, hard, and very hard activities from the past year.

Dietary assessment
A 130-item Food Frequency Questionnaire (FFQ) was used to assess dietary intake over the previous year.It was divided into 29 food groups and classified as semi-quantitative.The cohort article stated that this FFQ was used for dietary assessment 36 .During the study, participants were asked to indicate their frequency of consumption for each food item, ranging from daily to yearly.The food intake was measured using household measures and then converted to grams per day 39 .The total energy and nutrient intake was calculated using Nutritionist IV.

Insulin indices
The food insulin indices were determined by measuring the insulin response in the bloodstream after consuming each food item.This involved comparing the area under the insulin response curve for 1000 kJ (239 kcal) of each food with that of a reference food.The resulting insulin index value for each food item was derived from these measurements 40 .
We used previously reported dietary Insulin Index values to calculate total dietary II and IL 18,40,41 .We matched the food items in our FFQ with these items.For the remaining foods, we used the II of similar items based on their nutritional content correlation.To calculate the total dietary IL, we multiplied the II value of each food by its energy intake and summed the values for all items 41 .The insulin load calculated as follow: Insulin load of a given food = Insulin index of that food × energy content of that food (kcal/day).The dietary II was then obtained by dividing the dietary IL by the total energy intake 41 .

Definition of MetS
Criteria for metabolic syndrome are abdominal obesity (waist circumference ≥ 102 in men and ≥ 88 in women), serum triglycerides (≥ 150 mg/dL), or use of hypertriglyceridemia drugs, and elevated serum levels.Reduced high density lipoprotein (HDL) cholesterol (< 40 mg/dl in men and < 50 in women) or medication for low HDL cholesterol, blood pressure ≥ 130/85 mmHg, or use of antihypertensive medications, fasting plasma glucose (FPG) ≥ 100 mg/dL or use of hyperglycemic agents 42 .

Statistical analysis
The mean ± SD and percentages were reported for total participants and MetS based on quartiles of DII and DIL for continuous and categorical variables, respectively.One-way analysis of variance was used to evaluate differences in quantitative variables and the χ 2 test was used to assess the distribution of categorical variables (Tables 1, 2).
Generalized linear model was used to obtain the odds ratio (OR) and 95% confidence interval (CI) for MetS across quartiles of DIL and DII scores.The participants were categorized by gender and their DIL and DII quartiles were evaluated.The initial model accounted for age and energy consumption, while the second model incorporated body mass index (BMI) to identify links between DIL, DII, and the susceptibility to MetS that were not influenced by obesity.The third model included further adjustments for marital status, wealth score, smoking habits, and physical activity levels (met-h/w) (Table 3).Participants in the first quartiles of DIL and DII were considered as the reference group.The categories of DII and DIL were considered as an ordinal variable to calculate the trend of odds ratios across quartiles of DIL and DII.
RRR analysis identifies dietary patterns as linear combinations of food variables that explain the variation in response variables.We used four nutrients intake, include of; fiber (g/day) 26 , magnesium (mg/day) 29 , protein, fat (g/day) 27 and total DII 30 as response variables, and 29 food groups as predictive variables (Supplementary Table S1).We considered the first dietary pattern for subsequent analysis, which explained the maximum total explained variance in all response variables.Food groups with factor loadings ≥ 0.11 43 were considered to describe the dietary pattern, but all foods contributed to calculating the total score of dietary patterns.Factor loadings indicate the correlation of each food group with the dietary pattern.The dietary pattern score was obtained by summing the product of standardized intakes of the 29 food groups and their corresponding weight.The first dietary pattern was characterized by higher intake of whole grains, sugar, dairy, fruits, sweet deserts and lower intake of solid oil, liquid oil, dry fruit, coffee, red meat (Fig. 2).
The number of dietary patterns extracted by RRR was determined by the number of response variables.However, the only first dietary pattern due to achieve higher explained variance was considered for further analysis (Supplementary Table S2).Correlations between extracted dietary pattern scores and nutrients serving as response variables were assessed using Spearman correlation coefficient (Supplementary Table S3).In this study, we analyzed four essential nutrients: protein (measured in grams per day), fat (g/day), fiber (g/day), and magnesium (mg/day), in addition to the dietary insulin index.To determine the odds ratio (OR) and 95% confidence interval (CI) for MetS based on gender, we utilized a generalized linear model with three adjusted models, examining quartiles of the first dietary pattern score.As a reference point, we used the lowest quartile (Table 4).Moreover, we provided an overview of the general characteristics of all participants by gender, examining quartiles of DII and DIL scores (Supplementary Table S4).We used SAS statistical software (SAS Studio 3.71; SAS Institute Inc.) to identify the dietary patterns and SPSS version 22 (SPSS Inc, Chicago, IL) was used for all other statistical analyses.Two-sided P values < 0.05 were considered statistically significant.

Ethical approval and consent to participate
This project was approved by the research council (research project number: NRC_9906) and the ethics committee (research ethics number: IR.AJUMS.REC.1399.738) of Ahvaz Jundishapur university of medical sciences.Participants were fully informed about the study objectives and methods, and were assured of the confidentiality of their information.An informed consent was obtained from all subjects.All methods were carried out in accordance with relevant guidelines and regulations.

Results
The average participants' age and BMI were 47.28 ± 8.88 years and 28.15 ± 5.20 kg/m 2 , respectively.MetS was prevalent 19.03% among men and 32.3% of women.The DII average for women was significantly higher than that of men (women: 40.7 ± 4.16, men: 40.23 ± 3.84, P = 0.001).However, the DIL average for men was significantly higher than that of women (men 128,320.28 ± 27,870.13 and women 112,157.98 ± 30,715.45P = 0.0001).General characteristics of the study population according to the quartiles of dietary DIL and DII are indicated in the Table 1.www.nature.com/scientificreports/Compared with the first quartile of DIL, Participants with MetS in the upper quartile of DIL were younger, had higher body weight and body mass index, were physically active, had higher triglyceride levels, and lower level of HDL cholesterol.Similarly, the whole population under study in the upper quartile of DIL were more likely to have higher body mass index and triglycerides.And in the same quartile, they were younger, married, had higher education level, higher income status, and more likely to smoke.
Dietary intakes of the study population based on the quartiles of DIL and DII are expressed in the Table 2.The dietary intake of whole population showed the same as Individuals with MetS in the top quartile of DIL, had higher intakes of macronutrients and micronutrients comprise of; Energy, Total Protein, fat, Carbohydrate, Total fiber, Calcium, Magnesium, Potassium and Sodium.Participant with MetS in the highest quartile of DII were more likely to have higher intake of calcium and lower intake of total fat.In addition, the whole population in the top quartile of DII showed a rise in intake of total protein, Carbohydrate and Calcium.
Table 3 shows the multivariate-adjusted ORs and 95% CIs for MetS across quartiles of DIL and DII in men and women.No significant association was seen between DIL, DII, and MetS among men.Additionally, there was no significant association between DII and MetS in women.Moreover, a significant positive association was observed between DIL and MetS after adjusting for confounding factors.These factors including age, energy intake present in Model1, BMI in addition with Model1 considered as Model2, marital status, wealth score, smoking status and physical activity in addition with Model 2 indicated as Model 3. Female participants in higher quartile of DIL showed higher weight, BMI or central obesity were mostly married.They appeared to consume more energy and other macronutrients with lower age.Although, they were most active and had lower serum levels of FBS and total cholesterol.Similarly, younger male participants in top quartile of DIL revealed higher weight and consumed more energy and other macronutrients, however were most active and married (Supplementary Table S4).
Regarding RRR analysis, five dietary patterns were identified, however dietary pattern 1 (DP1) explained the greatest proportion of variance for selected nutrients with 55.3%, and total explained variance of 86.2% (Supplementary Table S2).Therefore, we considered only DP1 for subsequent analysis.DP1 was characterized by high consumption of whole grain, sugar, dairy products, fruits and sweet deserts with highest positive loading values, and low intakes of solid oil, liquid oil, dry fruits, coffee and red meat with highest negative loading values (Fig. 2).
Odds ratio (95% CI) for MetS across quartiles of dietary pattern are provided in Table 4.After adjustments for potential confounder include of age, energy intake, body mass index, marital status, wealth score, smoking status, and physical activity, female in the top quartile of the dietary pattern had 41% increased odds of MetS compared with those in the bottom quartile (OR: 1.41; 95% CI 1.13; 1.75).However, male in contrast to female elevated OR value of MetS in the highest quartile of dietary pattern.

Discussion
Results of the present study revealed that adherence to a diet with a high insulin load (DIL) was significantly associated with an increased odds of MetS in women by 51%.No significant association was seen between dietary insulin indices and odds of MetS in men.Additionally, men and women in top quartile of DIL were younger and married, had greater consumption of energy and other macronutrients with obese BMI level.The whole population under study were younger in higher quartile of DIL with higher weight and BMI level.However, they had lower FBS, total cholesterol serum level and lower HDL level as well.
In one study, they found that higher adherence to dietary diversity was associated with higher energy intake and macronutrients.The increasing risk of having MetS in the higher dietary diversity for old female could be Table 4. Odds ratio (95% confidence interval) for metabolic syndrome by gender across quartiles of dietary pattern score in whole population, Howeyzeh cohort study, Khuzestan, Iran. a Adjusted for age and energy intake.b Adjusted for model 1 + body mass index.c Adjusted for model 2 + marital status, wealth score, smoking status, and physical activity (met-h/w).www.nature.com/scientificreports/attributed to a higher likelihood of abdominal obesity, which could be attributed to different food consumption in higher categories of dietary diversity 44 .Eating a modern diet with pork, poultry, vegetables, seafood, pastries, and snacks can lead to higher energy intake, BMI, waist circumference, TG, and FPG for both men and women 45 .
Although the precise causes of Metabolic Syndrome (MetS) are unclear, insulin resistance and hyperinsulinemia have been shown to be involved in the etiology of Metabolic Syndrome (MetS) and its components 46 .It is well known that diet can play a pivotal role in insulin secretion.Recent observational studies have shown that diets that increase the insulin response may lead to several metabolic disorders 13,30,47 .Various methods exist to evaluate a diet's insulinemic potential, but the DIL and DII indices have gained significant attention in recent times.These indices are based on the insulin response and can accurately measure the secretion of insulin in response to foods that are rich in carbohydrates and proteins 48 .A prospective cohort study indicated that a positive association between DII and DIL with insulin resistance in men and women 49 .
The results of the Shahedieh cohort study indicate that following a diet with a high DIL and DII can increase the risk of MetS in women.Similarly, the Tehran Lipid and Glucose Study found a positive association between higher DII, DIL and MetS risk in women.Additionally, among Iranian adolescents, higher DIL and DII were strongly linked to increased risk of metabolically unhealthy obesity.Moreover, among women hs-CRP play a potential mediatory role in the association between DIL and MetS 21,23,24 .
Despite the few longitudinal study on the association of the dietary DII and DIL with MetS risk, there are several studies reported the controversial findings on the association of dietary DII or DIL with MetS risk factors, including hyperglycemia, insulin resistance, obesity, and dyslipidemia 30,47,49,50 .Results of two cohort studies showed that a diet with high DII and DIL was positively related with increased serum levels of TG and inversely associated with serum HDL-C concentrations 13 .However, this study could not find any significant relationship between the DII and DIL and the biomarkers of glycemic control (c-peptide and hemoglobin A1c).In addition, a cross-sectional study by Hajhashemy et al. 25 revealed that the DII and DIL were strongly associated with odds of metabolically unhealthy obese.In contrast, Ghorbaninejad et al. could not find a significant relationship between DII and DIL with the risk of metabolic syndrome and obesity 22 .Likewise, in a cohort study conducted within the framework of SEPAHAN data set, Anjom-Shoae et al. showed that higher DII and DIL were not associated with the risk of abdominal obesity 30 .The reasons for these discrepancies between studies may be due to different methods for calculating DII and DIL, different food processing and cooking methods in different cultures and considering different confounders.In the present study, the significant positive relationship between DIL and MetS was only seen in women.The exact mechanism of this sex difference is unknown.However, it may be due to the influence of sex hormone levels and androgen/estrogen balance and their effect of appetite and body composition [51][52][53] .One potential explanation could be the lower prevalence of MetS in men compared to women, resulting in larger confidence intervals.The specific mechanisms and pathways by which DII and DIL impact MetS are currently unknown.Whilst, female subjects had 44.6% obesity this may be increased the susceptibility of MetS.Gender-associated factors corelated with genetic and biological signaling pathways are primarily determined by hyperandrogenism, insulin resistance, and associated increased abdominal obesity and decreased postmenopausal HDL cholesterol.Gender-related factors are sensitive to social and cultural behaviors, dietary habits, and psychosocial factors.Women are more likely than men to develop MstS in response to work stress and low socioeconomic status 54 .Most married women in the present study were in the highest quartile of DII and DIL.Additionally, a greater percentage of obese women were placed in the highest quartile of DIL.
Likewise another study found out women who have ever been married are more likely to be overweight or obese due to factors such as older age, wealthy households, higher education, food security, television watching, and urban unemployment 55 .Retrospective cohort showed that individuals who get married are more likely to gain weight due to changes in their eating habits after marriage, which can lead to inconsistent weight control 56 .A community-based cohort study found that women who were married had a higher likelihood of being obese and having abdominal obesity.This could be attributed to getting married leading to more frequent meals and late-night snacking, as well as the chance to eat together, which results in an overall increase in energy intake 57 .
A diet with high DII may facilitate obesity by enhancing more insulin secretion, which may reduce fat oxidation and increase carbohydrate oxidation, therefore, such a diet could increase fat storage, abdominal obesity and enhanced risk of MetS 25,58 .
Furthermore, foods that cause high insulin levels are quickly digested and absorbed, resulting in a rapid increase in blood glucose and insulin and a subsequent decrease in glucose levels 59 .This rapid decrease in blood glucose can reduce satiety, restoring hunger sensation, thus contribute to excessive calorie intake, which could lead to an increased risk of abdominal obesity and MetS 59,60 .However, all insulinemic foods with high-DII are not quickly transformed into glucose and directly enhance insulin secretion.Iranian people receive more than 55% of energy from the carbohydrates-rich foods which is considerably higher than the amounts consumed in other populations 61 .Also, the main sources of carbohydrate intake among the Iranians are refined grains such as rice, white bread and potato 62 which are considered as the insulinogenic foods that can increased risk of insulin resistant and MetS 63 .In the current study, individuals in the fourth quartile of DIL significantly consumed more carbohydrate than those in the first quartile.Therefore, higher consumption of the insulinogenic foods or high insulin index foods including rice, white bread, and potato may play a crucial role in decreasing the MetS risk.
In the current study, five dietary patterns were identified based on RRR, but the first pattern, which explained 55.3 of the response variation, was considered as the main pattern.This pattern was characterized by higher intakes of fruits, sugar, sweet deserts, whole Grains and dairy and less intakes of solid oils, liquid oils, dry fruits, coffee and red meat.A significant positive association was found between this dietary pattern and risk of MetS.Our research aligns with a cross-sectional study of young and middle-aged Taiwanese adults with dyslipidemia and abnormal FPG.It found that those who consumed a western-style diet, high in desserts and sugary foods, were at increased risk for general obesity, central obesity, and high body fat 64 .The relation of Gender differences in the prevalence of metabolic syndrome may be due to physiological differences such as hormones, social and psychological stressors, and lifestyle differences 65 .
Regarding dietary pattern, previous studies using RRR have used different indicators as response variables to generate dietary patterns predictive of the risk of metabolic diseases.Some studies used plasma and blood biomarkers as response variables 66,67 and others used nutrients from food as response variables 68 .The use of dietary nutrients as response variables could result in a greater variation of responses as they are more proximal in the causal chains.However, the assumption of independence for response variables may be violated when using predictors and responses from dietary assessment tools 69 .Our results in regarding the association of dietary pattern and MetS risk are comparable with those identified in previous studies.A study in Sweden found that a diet high in sugar-sweetened beverages, reduced-fat milk, artificially sweetened beverages, and sweets is linked to a higher risk of MetS 70 .A study by Asadi et al. that conducted in Iran reported a positive association between a dietary pattern rich in sugar and sweetened beverage and risk of MetS 71 .Results from a cross-sectional study in Germany revealed that a dietary pattern with greater loading on high-sugar beverages, eggs, potatoes, beer and sweets was associated with risk of MetS 72 .The effect of food groups on body weight and obesity could define their relationship with MetS, a significant risk factor 73,74 .Surprisingly, fruits and whole grains had a high load in the identified pattern.In contrast with our results, epidemiological studies indicated that diet rich in fruits and whole grains is associated with reduced risk of MetS [75][76][77] .This discrepancy could be explained by differences in socio-demographic characteristics, behavioral and lifestyle factors.It should be noted that in our study we used fiber and magnesium as response variables that are found in abundance in fruits and whole grains.Epidemiological evidences reported controversial findings in the relationship between dairy with MetS and its components.Some studies showed an inverse association between dairy and risk of MetS 78,79 , while others failed to show such association 80,81 .A review on the relationship between dairy fat content and risk of metabolic disease does not support the hypothesis that dairy fat or high-fat dairy foods contribute to obesity or cardiometabolic risk 82 .
The main strength of present study was the large sample size and ethnicity origin of Arab.Other strengths were the use of food-frequency and reliable and validate physical activity questionnaires and the collection of a broad range of confounding factors.However, the present study had also some limitations.First, we used the US Department of Agriculture (USDA) databank because the Iranian food composition table was not complete in some food items and micronutrients, which possibly can cause measurement error.Second, the DII of some food items in FFQ were not available in the reference list and we used the DII of similar foods.Third, our study was not prospective design and we used data of the first phase of cohort (cross-sectional data).Therefore, the results of current study should be interpreted with caution and generalizability of findings will reduce due to Arab population-based.Hence, despite considering for a wide variety of confounders in our analysis, it is not possible to confidently rule out unmeasured or unknown confounding effects.

Conclusion
In conclusion, findings from this population-based cohort study showed that adherence to a high insulin loading diet in women was associated with greater risk of MetS.Additionally, dietary pattern derived from RRR suggested that diet rich in fruits, sugar, sweet deserts, whole Grains and dairy products were associated with increased risk of MetS.Further observational studies with long-term follow up are required to substantiate these findings. https://doi.org/10.1038/s41598-024-52263-5www.nature.com/scientificreports/ 14:1968 | https://doi.org/10.1038/s41598-024-52263-5www.nature.com/scientificreports/

Table 1 .
General characteristics of total participants across quartiles of DII and DIL scores and participants with MetS, Hoveyzeh cohort study, Khuzestan, Iran 1 .DII Dietary insulin index, DIL Dietary insulin load, MetS Metabolic syndrome, Met Metabolic equivalent, FBG Fasting blood glucose, HDL-C High density lipoprotein-cholesterol, TG Triglyceride, WHR Waist hip ratio.Bold p-values are statistically significant.1 Continuous values are shown as mean ± standard deviation and categorical values are shown as percent. 2

Table 2 .
Dietary intakes of total participants across quartiles of DII and DIL scores and participants with MetS, Hoveyzeh cohort study, Khuzestan, Iran 1 .

Table 3 .
Odds ratio (95% confidence interval) for metabolic syndrome by gender across quartiles of DII and DIL scores, Howeyzeh cohort study, Khuzestan, Iran.DII Dietary insulin index, DIL Dietary insulin load.Bold p-values are statistically significant.a Adjusted for age and energy intake.b Further Adjusted for body mass index.c Further Adjusted for marital status, wealth status, smoking status, and physical activity (met-h/w).