Introduction

Endothelial dysfunction is characterized by a reduction in nitric oxide (NO) and a loss in endothelial cell properties1. This disorder is one of the main mechanisms of developing cardiovascular diseases2, certain cancers3, and metabolic syndrome (MetS)4. Moreover, the role of endothelial dysfunction in pathology of type 2 diabetes and insulin resistance (IR)5 has been investigated in several studies6,7,8. IR is described as an impaired response of skeletal muscles and liver to circulating insulin9. IR can be responsible in etiology of a variety of diseases from hepatic steatosis10 to thyroid disorders11 and Alzheimer’s diseases12.

An unhealthy lifestyle, consisting of smoking, insufficient physical activity13, and unhealthy dietary patterns are claimed to be the most prevalent risk factors for IR and endothelial dysfunction14,15. The universal characteristics of multiple nutrients have brought a new insight. For instance, it has been suggested that IR could be affected by several nutrients such as vitamin D, chromium16, magnesium17, fiber18, dietary fats such as polyunsaturated and omega 3 fatty acids19,20 and specific polyphenols such as anthocyanins21, resveratrol22, and quercetins23,24. Additionally, endothelial dysfunction could be influenced by nutrients such as magnesium25, flavanol26, vitamins C and E27, and lycopene28.

Prior studies have mainly focused on the association of one specific nutrient with outcomes such as IR, and endothelial dysfunction. Since nutrients usually are not consumed distinctly, evaluating the association between special combinations of different nutrients and outcomes of interest might provide a better insight. However, few studies have been carried out to evaluate the association between nutrient patterns and IR and endothelial function. For instance, a prospective cohort study has assessed the association between 5 nutrient patterns and risk of insulin-related disorders24. They illustrated that higher adherence to a nutrient pattern, rich in vitamins A, C, B6, potassium, and fructose had favorable effects on insulin, homeostasis model assessment-Insulin resistance (HOMA-IR), and Homeostatic Model Assessment of insulin Sensitivity (HOMA-S), during 3 years of follow-up. Thus, the present study aimed to estimate the association of nutrient patterns with endothelial function and IR in Iranian women.

Methods

Study design and participants

This cross-sectional study was conducted on a population of 368 female nurses working in 7 hospitals in Isfahan city. A multi-stage cluster random sampling method was used for selecting these participants. Serum insulin levels (with an SD of 6.54 among Iranians) were considered as the main dependent variable for estimating the total sample size29. Then, by considering type 1 error of 5%, and design effect of 1.25, a total number of 407.5 subjects were estimated to be required for this study. First, 510 females older than 30 years were randomly invited to participate in the study; 30 nurses rejected to take part in the study. So, 480 women agreed to participate in our study. We excluded 2 participants that did not complete over 70 items of dietary questionnaire. Moreover, 9 women with a total energy intake of less than 800 or over 4200 kcal/day, 26 women with a previous history of diabetes, cancer, stroke, and CVD, 16 women consuming medications that could change serum glucose values and 59 subjects with incomplete data were excluded from the study. Finally, this analysis was carried out on data from 368 female nurses. Each participant signed a written consent form. All methods of the current study were carried out according to the relevant guidelines and regulations. The present study’s approach has been approved by the ethics committee of the Tehran University of Medical Science (IR.TUMS.MEDICINE.REC.1400.178).

Dietary assessment

A validated semi-quantitative dish-based FFQ was applied for assessment of common food intakes30. This FFQ included 106 food items and dishes and the participants were asked to report how often they have used these food items during the last year. Nine options ranging from “no or less than once in a month” to “more than 12 times in a day” were considered for each food item. A trained nutritionist instructed people on how to complete the FFQ. The validity and reliability of the FFQ were previously reported30. Additionally, the validity and reproducibility of the applied FFQ in the measurement of the average consumption of foods31, food groups32, and nutrients33 have been proven in the previous investigations. The US Department of Agriculture database (USDA) was used to calculate the total daily energy and nutrient intakes of each participant. Nutrient contents of some special foods were added to this software. The total energy and nutrient intake of each individual was computed by adding up energy intake and nutrients of all food items.

Assessment of biomarkers

Fasting blood samples were collected for measurement of serum concentration of insulin, blood glucose, and adhesion molecules including E-selectin, soluble intercellular adhesion molecule (sICAM-1), and soluble vascular adhesion molecule 1 (sVCAM-1). These blood samples were centrifuged for 30–45 min after collection. Then, serums were kept at − 80 to be used for the analysis. We used available commercial kits by ELISA method (Biosource International and Bender Med Systems) for assessment of sICAM-1 (nearest to 0.6 mg/dL), sVCAM-1 (nearest to 2.3 mg/dL), and E-selectin (nearest to 0.3 mg/dL). We measured fasting blood glucose (FBG) through the use of an enzymatic calorimetric (a method that assesses FBG through glucose oxidase activity). Serum insulin was also estimated through the ELISA method (Bender Med System). Then, we assessed insulin resistance and insulin sensitivity, through the following formulas:

HOMA-IR = FBS (mmol/L) × Insulin (µmol/mL)/22.524.

HOMA-β = (20 × insulin in mIU/mL)/ (FBG in mmol/L − 3.5).

QUICKY = 1/(log (fasting insulin (µU/mL) + log (fasting blood glucose (mg/dL))34.

Assessment of other variables

Socioeconomic variables including the number of family members, educational level, residual status, number of bedrooms in their house, being a house owner, number and types of their cars, salary, and other sociodemographic properties such as age, marital status, menopause status, previous history of diseases, habits of taking medications or supplementations and smoking were assessed by using a self-administrated questionnaire. Body weight was measured by a digital scale (nearest to 0.1 kg), while subjects were shoeless and wearing light clothes. A tape measure was applied for evaluating standing status height. Then, body mass index (BMI) was calculated through the following formula: weight (in kilograms)/height (in meters) squared. The short form of the International Physical Activity Questionnaire (IPAQ)35 was used for estimating daily physical activity in MET-hour per week.

Statistical analysis

Major nutrient patterns were extracted by performing factor analysis and entering 35 macro- and micro-nutrients in the analysis; these 35 nutrients were determined based on some previous publications in this regard24,36,37. Kaiser–Meyer–Olkin (KMO) test was applied to find out if the distribution of nutrients could be strong enough to use principal components. Factors with eigenvalues > 2 were considered as significant to extract major nutrient patterns. Scree plot was also used to identify the main nutrient patterns. Varimax rotation was conducted to extract independent nutrient patterns. Continuous and categorical characteristics of subjects were classified across tertiles of each nutrient pattern through the use of one-way ANOVA and chi-square tests, respectively. Mean dietary intakes of energy, food groups, and nutrients of participants across tertiles of nutrient patterns were obtained by ANCOVA. Mean values of glycemic factors and markers of insulin resistance and endothelial function across tertiles of nutrient patterns were estimated through ANCOVA in four models. This relationship was controlled for age and energy intake in the first model. Physical activity (MET-h/week), current corticosteroids and OCP intake (yes/no), marriage status (categorical), menopausal status (yes/no), systolic blood pressure (SBP), diastolic blood pressure (DBP), and socioeconomic status (categorical) were additionally controlled in the second model. Additional adjustment for BMI was conducted in the third model. In model 4 for association of nutrient patterns and glycemic factors and insulin resistance, additional adjustment was done for endothelial indices (E-selectin, sICAM-1, and sVCAM-1). While for association of nutrient patterns and endothelial markers, further adjustment was done for blood glucose and lipid profiles including serum triglyceride, serum total cholesterol, HDL-c, and LDL-c, in model 4. P values < 0.05 were assumed as statistically significant. Linear association between tertiles of nutrient patterns and indices of insulin resistance and endothelial function was assessed by linear regression analysis in both crude and adjusted models. Version 26 of SPSS was applied to perform all analysis.

Ethical approval and consent to participate

All participants provided an informed written consent. The study protocol was approved by the local Ethics Committee of Isfahan University of Medical Sciences in 2022 (IR.TUMS.MEDICINE.REC.1400.178).

Results

The current study was conducted on 368 female nurses working in Iran hospitals. The mean age and BMI of participants were respectively 35.21 years and 24.04 kg/m2. Three nutrient dietary patterns have been extracted through factor analysis (Fig. 1). Factor loadings of each single nutrient in each nutrient pattern are provided in Table 1. Overall, 78.5% of all dietary changes have been explained through these three nutrient patterns. Nutrient pattern 1 was associated with greater amounts of potassium, folate, vitamin A, vitamin C, magnesium, beta carotene, pantothenic acid, sugar, phosphorus, riboflavin, biotin, vitamin K, calcium, and carbohydrate. This pattern has been supposed to be rich in dairy products, fruits, and vegetables. The second nutrient pattern was correlated with higher intakes of chromium, selenium, copper, vitamin B6, monounsaturated fatty acid (MUFA), thiamin, polyunsaturated fatty acid (PUFA), vitamin D, iron, and dietary fiber. This nutrient pattern was considered to be full of legumes, nuts, and protein foods. The third nutrient pattern was related to higher values of saturated fatty acid (SFA), cholesterol, vitamin E, sodium, vitamin B12, zinc, and protein. Therefore, this NP seemed to be correlated with higher consumption of animal fat and meat + vitamin E.

Figure 1
figure 1

Scree plot for identifying major nutrient patterns in Iranian women.

Full size image
Table 1 Factor loadings and explained variances for major nutrient patterns (NPs).
Full size table

General features of the study subjects across tertiles of nutrient patterns are shown in Table 2. There was no significant difference in socio-demographic characteristics across tertiles of nutrient patterns 1 and 2. However, a marginally lower BMI (23.4 vs. 24.4, P = 0.05) and waist circumferences (79.1 vs. 82.1, P = 0.05) have been observed among subjects in the highest tertiles in comparison to those in the lowest tertile of NP3. Participants with menopause status were lower in the highest tertile compared to the lowest tertile of NP3 (2.2% vs. 10.7%, P = 0.01). Other socio-demographic characteristics were not significantly different between tertiles of NP3.

Table 2 General characteristics of study population across categories of nutrient pattern scores.
Full size table

Usual dietary intakes of individuals across tertile of NPs are presented in Table 3. Consumption of total energy intake (P = 0.001), vegetables (P < 0.001), fruits (P < 0.001), low-fat dairy (P < 0.001), legumes and nuts (P < 0.001) and total dietary fiber (P < 0.001) were significantly higher among subjects in the highest vs. lowest tertile of NP1. Lower intakes of refined grains (P ˂ 0.001), oils (P = 0.002), protein (P ˂ 0.001), total fat (P = 0.001), SFA (P ˂ 0.001), MUFA (P = 0.01) and PUFA (P  ˂ 0.001) have also been observed in tertile 3 in comparison to tertile 1 of NP1. Participants in the highest tertile compared with the lowest tertile of the second nutrient pattern had higher intakes of energy (P = 0.01), vegetables (P  ˂ 0.001), fruits (P  ˂ 0.001), SFA (P = 0.03), protein (P  ˂ 0.001), carbohydrate (P  ˂ 0.001) and total dietary fiber (P  ˂ 0.001), and lower intakes of white meat (P = 0.04), refined grains (P = 0.04), cholesterol (P  ˂ 0.001) and sodium (P  ˂ 0.001). The third vs. first tertile of nutrient pattern 3 was associated with higher consumptions of energy intake (P = 0.001), white meat (P = 0.001), refined grains (P = 0.03), oils (P  ˂ 0.001), protein (P  ˂ 0.001), fats (P  ˂ 0.001), SFA (P  ˂ 0.001), cholesterol (P  ˂ 0.001) and sodium (P  ˂ 0.001), and lower intakes of vegetables (P  ˂ 0.001), fruits (P  ˂ 0.001), carbohydrate (P  ˂ 0.001) and dietary fiber (P  ˂ 0.001).

Table 3 Dietary intakes of study participants across tertiles of nutrient patterns.
Full size table

Multivariable-adjusted mean ± SE of glycemic indices and insulin resistance markers across tertiles of nutrient patterns are reported in Table 4. The indices of glycemic profile and insulin resistance were not significantly different across tertiles of NP1. Subjects in the highest tertile of NP2 had significantly lower insulin levels (6.8 ± 1.1 vs. 8.4 ± 1.1, P = 0.006) in comparison to the lowest tertile in fully-adjusted model. Participants in the top tertile of NP2 compared with the bottom tertile had lower levels of HOMA-IR (1.3 ± 0.2 vs. 1.7 ± 0.2, P = 0.02), in the fully-adjusted model. Other glycemic indices were not significantly different across tertiles of NP2. Subjects in the highest tertile of NP3 in comparison to the lowest tertile, had higher levels of HOMA-β (542.0 ± 176.0 vs. 44.1 ± 175.0, P = 0.03), in the second model. This association was significant even after adjustment for all potential covariates (531.3 ± 176.2 vs. 48.7 ± 179.8, P = 0.03).

Table 4 Multivariable-adjusted glycemic profile and insulin resistance across tertiles of nutrient pattern scores.
Full size table

Table 5 shows the multivariable-adjusted mean ± SE of endothelial function markers across tertiles of nutrient patterns. Individuals in the highest tertile in comparison to those in the lowest tertile of NP1 had higher levels of sICAM-1 in the crude model (223.7 ± 8.5 vs. 201.1 ± 6.4, P = 0.03). This significant difference disappeared after adjustment for all covariates in model 4. In the crude model, levels of E-selectin were lower in the highest tertile compared with the lowest tertile of NP2 (79.6 ± 3.1 vs. 98.6 ± 7.8, P = 0.01). However, there was no significant difference in E-selectin levels across tertiles of NP2, after controlling for potential covariates (84.9 ± 6.4 vs. 82.0 ± 6.3, P = 0.94). Individuals in the highest tertile of NP2 had also lower levels of sVCAM-1 in comparison to the lowest tertile, after adjusting for all potential variables (444.2 ± 27.9 vs. 475.8 ± 28.4, P = 0.03). Indices of endothelial function were not significantly different across tertiles of NP3, in both crude and fully-adjusted model.

Table 5 Multivariable-adjusted association between markers of endothelial function and tertiles of nutrient pattern scores.
Full size table

The linear associations of dietary nutrient patterns with insulin resistance and endothelial function indices are reported in Table 6. A significant increase in values of sICAM-1 was seen along with each one increase in tertiles of NP1, in the crude model (B = 11.16, 0.95% CI 1.45, 20.87). This association was also significant in model 1, after adjustment for age and energy intake (B = 21.61, 0.95% CI 9.76, 33.45). However, this association disappeared after further adjustment for other potential variables. There was no linear association between NP2 and markers of insulin resistance and endothelial function. Furthermore, each increase in tertiles of NP3 was associated with a marginal increase in HOMA-IR values in model 3 (B = 0.42, 0.95% CI 0.00, 0.84). This association was removed after adjustment for endothelial function markers in model 4 (B = 0.40, 95% CI − 0.02, 0.83).

Table 6 Linear association of nutrient dietary patterns1 with insulin resistance and endothelial function indexes.
Full size table

Since no significant consistent association was observed between nutrient patterns and most of the indexes of both insulin resistance and endothelial dysfunction, the pathway analysis was not conducted in the current study.

Discussion

In the current cross-sectional study, we illustrated that following two nutrient patterns was associated with insulin resistance and endothelial function indices. Such that, higher adherence to NP2, which consisted of chromium, selenium, copper, vitamin B6, MUFA, thiamin, vitamin D, and iron, considered as “legumes, nuts and protein foods nutrient pattern”, was associated with lower values of Insulin, HOMA-IR, and VCAM-1. Moreover, higher adherence to NP3 consisting of SFA, cholesterol, vitamin E, sodium, vitamin B12, zinc, and protein, named as “animal fat and meat + vitamin E nutrient pattern”, was associated with higher values of HOMA-β. Although HOMA-β is considered as an index of beta-cell function, its increased levels have shown to be associated with impaired glucose tolerance, type 2 diabetes, and insulin resistance38,39. Adhering to the third nutrient pattern in the current investigation has led to higher values of HOMA-β, but resulted in a reduction in QUICKY levels, a definite indicator of insulin resistance40,41, although this association was not statistically significant (P = 0.19). In addition, no linear association has been observed between tertiles of nutrient patterns and levels of glycemic and endothelial indices after considering all potential variables.

Obesity is known as an important risk factor for insulin resistance and prevalent around the world42. It has been declared that during recent years, a significant rise in prevalence of type 2 diabetes was concerning in some countries, despite lower numbers of obesity42,43. On the other hand, metabolic disorders such as hypertension and abdominal obesity44,45 are drastically associated with increased endothelial dysfunction and consequently coronary artery diseases46. So, it can be very important to find an effective way for managing these conditions. According to our study, following a diet rich in unsaturated fatty acids, copper, selenium, manganese, chromium, zinc, vitamin B6, thiamin, vitamin D, and dietary fiber, along with lower consumption of SFA, cholesterol, vitamin E, sodium, potassium, and vitamin B12 might help reduce risks of insulin resistance and endothelial dysfunction. More clinical trials are necessary to confirm these observations.

Previous studies have estimated the association between various nutrients and IR markers. For example, a prospective cohort study on 995 subjects has suggested a reduction in IR and hyperinsulinemia by following a nutrient pattern rich in potassium, vitamins B6, C, and A24. Moreover, significant inverse associations were observed between adherence to the nutrient pattern rich in vitamin B and dietary fiber, and another pattern, called zinc, thiamin, and plant proteins with the values of glycated hemoglobin and fasting glucose in a prospective cohort study in South Africa47. Furthermore, another observational study among Iranian overweight and obese adolescents has reported an increased risk of metabolically unhealthy obesity as well as an increment in HOMA-IR levels through following a “high fat and sodium” nutrient pattern48. Moreover, it has been reported that following a diet with a higher Mediterranean-style score, rich in MUFA, PUFA, nuts, and seeds in children, might be associated with lower levels of HOMA-IR, fat mass index (FMI), and cardiometabolic risk in their adulthood49. Another 3-year prospective cohort study has found an inverse association between higher dietary approaches to stop hypertension (DASH) score and IR. DASH score was defined by higher intakes of legumes, nuts, fruits, and vegetables, and lower intakes of sodium, red and processed meat, and sweetened beverages in the mentioned study50. A meta-analysis of 44 trial and prospective cohort studies on patients with diabetes has also demonstrated a reduction in HbA1C and HOMA-IR levels in higher intakes of dietary fiber51. These investigations might confirm the favorable effect of NP2 in the current study (named legumes, nuts, and protein food) on levels of serum insulin and HOMA-IR. On the other hand, saturated fatty acids have been proven to increase the risk of insulin resistance52. Higher meat consumption was associated with an increase in HOMA and insulin levels in a population of non-diabetic women53. It has also been claimed that diets rich in animal protein might increase insulin resistance regardless of weight54. So, the increments in levels of HOMA-β in the present study across tertiles of NP3 (described as the meat and animal fat pattern) could be supported by these evaluations.

Several mechanisms might explain the association of nutrients with insulin resistance and endothelial dysfunction. Interventional studies have suggested that supplementation of zinc, selenium, and chromium might improve insulin resistance by reducing oxidative stress which can impair insulin secretion from β cells55,56. Additionally, it has been suggested that chromium might be able to increase insulin binding through increasing the number of insulin receptors and their phosphorylation57. The protective role of selenium against insulin resistance and type 2 diabetes might be associated with its ability to enhance the activity of glutathione peroxidase (GPx), which defends against reactive oxygen species (ROS)58. A combination of vitamin D3 and chromium has also shown to decrease HOMA-IR levels by regulation of inflammatory markers like TNF-α16. On the other hand, MUFA consumption has a favorable effect on sVCAM-1 through the reduction in NF-kB, another marker of oxidative stress59,60. Co-supplementation of omega 3 fatty acids and chromium could also enhance endothelial function by preventing the activity of phospholipase A2, a prooxidant enzyme, and provoking antioxidant enzymes61. A randomized control trial on 124 children with type 1 diabetes documented that folate and vitamin B6 supplementation had a positive effect on endothelial function, because folate supplementation could enhance levels of tetrahydrobiopterin, a substantial cofactor for NO synthesis62. Furthermore, vitamin B6 could regulate the inflammatory response63. Vitamin D and its receptors (VDRs) could also enhance endothelial function by increasing NO synthesis, through a positive regulation in the activity of endothelial Nitric Oxide Synthase (eNOS)64.

As far as we know, this is the first study investigating the association of various NPs with insulin resistance and endothelial dysfunction. Moreover, validated questionnaires were used to assess dietary intakes and covariates. Nevertheless, some limitations can be acknowledged in our study. Considering cross-sectional design of the study, causal relationships could not be confirmed. Since the current investigation was conducted on a population of nurses living in Isfahan, generalizing the results to all Iranian women might not be totally possible. Although was controlled for several confounders in the analyses, the effect of residual confounders might not be avoided. In addition, misclassification and measurement errors are unavoidable due to the self-reported design of questionnaires. Finally, the study was carried out on a particular group of people (female nurses working in hospitals) and its findings could not be generalized to the whole adult population.

In conclusion, in the current cross-sectional study higher adherence to the second nutrient pattern, associated with higher intakes of chromium, selenium, copper, vitamin B6, MUFA, PUFA, vitamin D, and iron was associated with lower Insulin, HOMA-IR, and VCAM-1 values. However, higher adherence to the third nutrient pattern, rich in SFA, cholesterol, vitamin E, sodium, and vitamin B12 was associated with higher HOMA-β values. Considering the findings of the current study, adhering to a nutrient dietary pattern, rich in selenium, copper, iron, vitamin B6, vitamin D, and unsaturated fatty acids (including PUFAs and MUFAs) with lower intakes of cholesterol, sodium, vitamins E and B12, and saturated fatty acids can reduce the risk of insulin resistance and endothelial dysfunction in female population. However, further prospective investigations are required to affirm these associations.