Dietary Inflammatory Index is Associated with Excessive Body Weight and Dietary Patterns in Subjects with Cardiometabolic Risk

Unhealthy eating habits can trigger chronic inflammation in organs and tissues, and subsequent cardiovascular risk. Thus, this study aimed to evaluate the potential association of Dietary Inflammatory Index (DIITM) score with cardiometabolic risk factors and a priori and a posteriori dietary patterns in adults and elderly Brazilians (42 ± 16 years) from a health care program. This cross-sectional study was carried out with 248 individuals (138 women and 110 men) from a Cardiovascular Health Care Program of the Universidade Federal de Viçosa (PROCARDIO-UFV, ReBEC ID number: RBR-5n4y2g). Sociodemographic and clinical characteristics, lifestyle and anthropometric data were assessed by standardized protocols. DII scores were calculated from a 24 h-recall, and dietary patterns were determined by a posteriori method, as Principal Component Analysis (PAC), and a priori method, as Healthy Eating Index (HEI). The mean DII was 0.15 ± 0.84. The subjects included in the most pro-inflammatory category(3rd tertile) were more likely to be individuals with excessive body weight, increased waist-to-hip ratio and lower quality diet assessed by PAC and HEI. The most pro-inflammatory diet was associated with excessive body weight as well as other worse dietary patterns.

In turn, dietary pattern analysis allow the combination of foods consumed by a population and the identification of both protective and harmful dietary factors for the development of CNCD [3,7,8]. Dietary pattern analysis can be performed a priori or a posteriori. The first uses indexes, which can evaluate the adequacy of food intake based on guidelines recommendations previously determined, such as Dietary Inflammatory Index (DII™) [9,10], Healthy Eating Index (HEI) [11] and Dietary Quality Index (DQI) [12]. The second one is based on statistical analysis of food consumption data to develop new dietary patterns, such as factor analysis and Principal Component Analysis (PCA) [3,13].
Moreover, a dietary pattern may be anti-inflammatory, as in the case of the Mediterranean diet, or be pro-inflammatory, such as Western diet, and therefore contribute to the prevention or development of CNCD [1,3,6,[14][15][16][17]. The DII was developed and validated in 2009 [8] and updated in 2014 [9] to evaluate the inflammatory effect of diet on health and disease. The index is based on extensive literature review and resulted in 45 dietary parameters (macronutrients, micronutrients and bioactive compounds such as teas, flavonoids and spices) based on their influence on the inflammatory markers including interleukin (IL)-1β, IL-4, IL-6, IL-10, tumor necrosis factor (TNF)-α and C-reactive protein (CRP) [18].
Some studies have identified positive associations between DII scores and overweight or obesity in Poland, Spain, USA and Mexico populations [18][19][20][21]. However, the association of this index in subjects at high cardiometabolic risk with dietary patterns has rarely been reported [18,21], especially in the Brazilian population [22].
Therefore, the hypothesis of this study is that a more pro-inflammatory diet is associated with cardiometabolic risk factors.
This study aimed to evaluate the potential association of Dietary Inflammatory Index (DII™) score with cardiometabolic risk factors and a priori and a posteriori dietary patterns in adults and elderly Brazilians from a Cardiovascular Health Care Program.

Subjects
The Cardiovascular Health Care Program (PROCARDIO-UFV) performs nutritional intervention in the academic community of the Universidade Federal de Viçosa (UFV). The inclusion criteria are: age > 20 years; diagnosis of cardiovascular disease or have cardiometabolic risk factors, as previously described [23][24][25].
The Ethics Committee in Human Research of the Universidade Federal de Viçosa approved the protocol of the present study (Of Ref. No. 066/2012/CEPH). All subjects provided written informed consent, according to the general recommendations of the Declaration of Helsinki. The PROCARDIO-UFV is registered in the Brazilian Registry of Clinical Trials (ReBEC: RBR-5n4y2g).
In this cross-sectional study, we included data from 248 individuals (138 women and 110 men), adults and elderly (42 ± 16 years), participants in the PROCARDIO-UFV from March 2012 until July 2017, who presented complete medical records to study analysis.

Food Consumption, HEI, PCA and DII
The food intake was estimated using a 24-hour recall (24HR) using a 5-step multiple-pass method carried out by a trained dietitian, for better data collection quality [26]. Portion size was estimated using standard food quantities (grams) including food photographs and a standard manual on portions of food and household measures.
The dietary patterns were determined by PCA, as previously described [28]. Five patterns were identified as "Traditional", "Caloric", "Unhealthy", "Healthy" and "Healthy Snacks". Scores greater than or equal to zero were considered positive for the dietary patterns. Each pattern was characterized by a group of foods as rice and tubers, beans, vegetable oils, non-leafy vegetables, meats, fish and eggs (grilled, cooked or roasted) for the "Traditional"; meat, offal and eggs (fried), processed meat, sweets and sugar, soft beverages and artificial juices for the "Caloric", "Unhealthy" was represented by fast food and pasta; "Healthy" was composed by whole grain food and nuts, milk, dairy, fruits and natural juices and finally "Healthy Snacks" by leafy vegetables, chicken salad sandwich.
For the Healthy Eating Index (HEI) analysis, the revised index for the Brazilian population was used, and the median (71 points) was considered to classify the diet as adequate (greater than 71 points) and inadequate (lower than 71 points) [29].
The design and development of DII has been described previously [8,9]. Briefly, the DII was calculated using a scoring algorithm based on a review of 1,943 articles that showed the association of 45 food parameters with six inflammatory biomarkers (IL-1b, IL-4, IL-6, IL-10, TNF-α and CRP). Food parameters evaluated in the articles were positively marked (+1) if the effect was considered pro-inflammatory, negatively (-1) in the case of an anti-inflammatory effect, or zero, if the parameters did not produce a significant change in the biomarkers. The 24HR-derived food and nutrient consumption was first adjusted for total energy (i.e., per 1000 kilocalories) and then standardized to a globally representative energy-adjusted dietary database, which was constructed based on dietary intake from 11 populations living in different regions of the world. From this, the energy-adjusted standardized dietary intake (expressed as z-scores of the referent database and then as centered proportions) was then multiplied by the literature-derived inflammatory effect score for each DII component. Individual scores were then summed to determine the overall DII (DII™) score for everyone, with positive scores, representing a more pro-inflammatory diet, and negative values, representing more anti-inflammatory diets.
The DII score in this study was based in the following 19 food parameters, obtained using the 24HR: n-6 fatty acids, iron, fiber, magnesium, zinc, selenium, vitamin C, vitamin A, vitamin B12, carbohydrates, proteins, lipids, cholesterol, trans, saturated, mono and polyunsaturated fats, energy (kcal) and alcohol. In the present study, the DII was categorized into tertiles, whereas first (< -0.22), second (≥ -0.22 and <0.45), and third (≥ 0.45) tertiles were the most anti-inflammatory, neutral and the most pro-inflammatory tertile, respectively.

Cardiometabolic Risk Factors Assessment
Weight, height and waist circumference (WC) were measured according to the protocol standardized by PROCARDIO-UFV previously described [30]. The body mass index (BMI) was calculated and classified as excessive body weight to BMI ≥ 25.0 kg/m² for adults [31] and BMI ≥ 28.0 kg/m² for elderly [32]. Abdominal obesity was considered as WC ≥ 90 for men and ≥ 80 cm for women [33]. Waist-to-height ratio (WHtR) > 0.5 and waist-to-hip ratio (WHR) > 1.0 for men and > 0.85 for women were considered a cardiometabolic risk [31]. Percentage of body fat (%BF) was estimated by horizontal tetrapolar electric bioimpedance (Biodynamics® 310 model, Washington, USA), according to the protocol proposed [34], and values > 20% for men and > 30% for women were considered as excessive body fat [35].
CRP was categorized as recommended by the US Centers for Disease Control and Prevention (CDC) and the American Heart Association, with values ≤ 3.0 mg/dL being considered adequate and those >10 mg/dL (presented by four individuals) excluded, since they may indicate acute inflammation [41]. For analysis of the ferritin and the TyG index, the 75th percentile of the sample was used as the cut-off point, being greater than or equal to 207 and 5.01, respectively. For HOMA-IR and fasting glucose, high values were considered above 2.71 and ≥ 100 mg/dL, respectively [42].
Blood pressure was measured using a mechanical mercury sphygmomanometer (Missouri®, São Paulo, Brazil) with approximately 02 mmHg [39] considering high blood pressure values ≥ 140/90 mmHg [43]. Moreover, the participants were the ones to report medical diagnosis of diabetes, hypertension and dyslipidemias and the use of medications.

Other Characteristics of Sample
The information about age (years), income (did not report; up to 2 minimum wages; 2 to 4 minimum wages; 4 to 10 minimum wages and more than 10 minimum wages), smoking (never smoked or smoked), physical activity and diagnosed diseases as well as information about medicine use (statins, antihypertensive and oral hypoglycemic) was obtained by semi-structured questionnaire elaborated specially for the PROCARDIO-UFV study.

Statistical Analysis
Descriptive analysis were performed for the main variables of interest. The distribution of the variables was verified by means of boxplot graphs, histogram, kurtosis and skewness values. The analysis of variance (ANOVA) was used to compare averages between DII tertiles, followed by Tukey and Bonferroni post hoc tests. The Pearson's chi-squared test was used to evaluate the distribution of the qualitative variables among the DII tertiles. Poisson regression models were performed to analyze the association of DII (independent variable) with cardiometabolic risk factors and dietary patterns (dependent variables), using the second tertile (neutral inflammation) as reference. Multiple Poisson regression models were adjusted for age, sex, income, smoking, physical activity and statin, antihypertensive and oral hypoglycemic use. The Hosmer & Lemeshow test was used to verify the fit of the final model. The prevalence ratio (PR) with 95% confidence interval (95% CI) was used as an effect measure. All statistical analysis were performed using SPSS software version 22.0 and Stata version 13.0, considering a significance level of 5%.

Results
The mean age of the participants was 42 years, ranging from 20 to 80 years, 55.6% were female and 69.8% had excessive body weight. The mean DII was 0.15 ± 0.84. Those subjects with most pro-inflammatory diet (third tertile) were more likely to be women, non smokers, and non-diabetics (Table 1). They also had lower mean age and WHR, compared to the most anti-inflammatory diet (first tertile).
Regarding diet composition, subjects included in the third tertile of DII had higher mean values of energy, lipid, cholesterol, monounsaturated fat and saturated fat, while those included in the first tertile had higher intakes of carbohydrates, omega 6, iron, fiber, magnesium, vitamin C, zinc and polyunsaturated fatty acids (Table 2).
In addition, the lowest tertile of DII (the most anti-inflammatory diet) was inversely associated with inadequate diet (HEI below 71 points) (PR: 0.57; 95%CI: 0.38-0.88), while the highest tertile (the most pro-inflammatory) was directly associated with inadequate diet (PR: 1.52; 95%CI: 1.15-2.01). From the five dietary patterns obtained by the PCA method, DII was associated with three. After all adjustments, the lowest DII tertile was associated with "Healthy" (PR: 3.71; 95%CI: 1.54-8.90) and "Healthy snacks" (PR: 3.05; 95%CI: 1.12-8.32), while the highest DII tertile was associated with more "Unhealthy" diets, (PR: 2.94; 95%CI: 1.03-8.35) ( Table 4).   polyunsaturated fatty acid. Energy is in kcal; Carbohydrate is in g; Protein is in g; Lipid is in g; Cholesterol is in g; Saturated fat is in g; Trans fat is in g; MUFA is in g; PUFA is in g; Omega 6 is in g; Iron is in mg; Fiber is in g; Magnesium is in mg; Vitamin C is in mg; Zinc is in mg; Selenium is in mg; Vitamin A is in mg; Vitamin B12 is in mg; Alcohol is in %v/v. Variables expressed as mean ± standard deviation. Different letters indicate statistical significance on the same line. Values are determined using ANOVA test.

Discussion
The present study demonstrated that the most pro-inflammatory DII scores were associated with excess adiposity in adults and elderly individuals with cardiometabolic risk factors. We emphasize that this is one of the first studies that investigated the DII with cardiovascular risk factors and dietary patterns of the Brazilian population. Previous study carried out with multiple sclerosis patients in Brazil verified that the DII score was directly correlated with BMI in those patients in the progressive stage of the disease (r = 0.556, p = 0.025) [44], though the other study carried out in Brazil did not find an association between DII with insulin resistance and metabolic syndrome [22].
Our findings are corroborated by other studies in Spain and the USA [17,45,46]. Altogether, results suggest diet-induced inflammation may contribute to increasing or maintaining excessive body weight. The relationship between inflammation and adiposity excess may be bidirectional -overweight may induce inflammation, whereas a pro-inflammatory diet may lead to increasing or maintaining excess adiposity [47]. One possible mechanism that explains this relationship is the activation of molecular processes associated with Toll-like receptors and Nod-type receptors, which in turn induce the activation of inflammatory markers in adipose tissue [48]. Furthermore, a meta-analysis has identified significant association between pro-inflammatory diet and risk for CVD events and mortality [49].
Although this study found no association between DII with other cardiometabolic risk factors as we expected, other studies also did not observe such an association [20,[50][51][52][53]. Noteworthy, the mean of the DII (0.15) of our sample did not characterize a pro-inflammatory diet in general, but in the ORISCAV-LUX study conducted with 1352 people they also did not present a pro-inflammatory diet in general [20]. Our results are different from those observed in other studies in which the DII was much higher. Cross-sectional analysis from Polish-Norwegian Study (PONS) and National Health and Nutrition Examination Survey (NHANES) presented a DII mean of 0.93 ± 1.44 [47] and 0.87 ± 1.08 [54], respectively. In a case control study with women that aborted, the DII was 1.51± 0.63 in cases and 1.22± 0.63 in controls [55]. Based on that, this may have contributed to the lack of association with some anthropometric and biochemical markers related to cardiometabolic risk in our study.
In addition, higher percentage of diabetics, smokers or ex-smokers, i.e.,individuals who are at elevated risk for the development of CVD were in the first tertile of the DII (more anti-inflammatory diet). These individuals may have received medical and nutritional guidance at some point, and had become more concerned about their health, and therefore began consuming a more healthy diet. This previous guidance in life-course of subjects with cardiometabolic risk may have contributed to the lack of association with other cardiometabolic markers. The same was observed in other studies in which people with metabolic syndrome, particularly with abdominal obesity and high blood pressure, had a diet with an antiinflammatory profile [20].
We also presented a positive association between most anti-inflammatory diet and healthier dietary patterns, such as "Healthy" and "Healthy snacks" from PCA (a posteriori method), as well as positive association between most pro-inflammatory diet and inadequate diet (a priori method) and "Unhealthy" pattern (a posteriori method). We highlight that this is one of the few studies that evaluated the association between DII and dietary patterns. The dietary pattern may present anti-inflammatory characteristics, as in the case of the Mediterranean diet, or pro-inflammatory, such as Western diet, contributing to the prevention or development of CNCD [1,3,4,13]. This difference is due to the nutritional composition of these food patterns, the first one characterized by the high content of anti-inflammatory nutrients and antioxidants, and the second by excessive consumption of foods of animal origin, sources of saturated fat and cholesterol, responsible for changes in the lipid profile [56]. The pro-inflammatory diet activates the immune system (with secretion of pro-inflammatory cytokines) and increases the expression of adhesion molecules, with consequent inflammatory process in vascular tissue [18,49,58]. On the other hand, studies have shown that the Vegetarian and Mediterranean diets reduce inflammatory markers such as CRP and IL-6 and improve endothelial function [5,13,59]. This is the first study to associate DII with anthropometric and biochemical variables, and dietary patterns in Brazilians with cardiovascular risk. An additional strength is the use of Dietary Inflammatory Index to evaluate these associations, which eliminates any influence of energy intake. Therefore, we suggest that other studies should be performed with different age groups and with longitudinal design. As limitations, we used a 24-hour recall, which does not reflect a habitual intake. However, this method has been used thoroughly with success in other epidemiologic studies [60,61], including DII analyses [1,45,46,54,58,62,63]. Although we were able to compute the DII from only 19 of the 45 potential items of food and nutrients that could be used to calculate this index, other published studies also derive DII scores from a sub-optimal number of items, and the ability to still detect significant associations suggests that this has only caused to a potential underestimated of the associations [1,52,58,64,65].

Conclusions
In this population of Brazilian middle-aged and elderly adults, the most pro-inflammatory diet was observed in individuals with excessive body weight and in those with a worse dietary pattern, diagnosed both by a priori and a posteriori method. These results contribute to a better understanding of the role of diet on inflammatory processes related to excess adiposity in a population with cardiovascular risk, a common situation among Brazilians. More studies with longitudinal design are required to discern the association between diet associated inflammation and cardio-metabolic risk factors in other south American populations of high-risk adults.