Dietary inflammatory index and anthropometric measures of obesity in a population sample at high cardiovascular risk from the PREDIMED(PREvención con DIeta MEDiterranea) trial

Dietary inflammatory index and anthropometric measures of obesity in a population sample at high cardiovascular risk from the PREDIMED (PREvencio ´n con DIeta MEDiterra ´nea) trial M. Ruiz-Canela 1,2 *, I. Zazpe 2,3 , N. Shivappa 4,5 ,J.R.He ´bert 4,5 ,A.Sa ´nchez-Tainta 2 , D. Corella 2,6 , J. Salas-Salvado ´ 2,7 , M. Fito ´ 2,8 , R. M. Lamuela-Ravento ´s 2,9 , J. Rekondo 2,10 , J. Ferna ´ndez-Crehuet 2,11 , M. Fiol 2,12 , J. M. Santos-Lozano 2,13 , L. Serra-Majem 2,14 , X. Pinto 2,15 , J. A. Martı ´nez 2,3 , E. Ros 2,16 , R. Estruch 2,17 and M. A. Martı ´nez-Gonza ´lez 1,2 1 Department of Preventive Medicine and Public Health, Facultad de Medicina-Clı ´nica Universidad de Navarra, School of Medicine, Universidad de Navarra, Irunlarrea 1, 31080 Pamplona, Navarra, Spain 2 CIBER Fisiopatologı ´a de la Obesidad y Nutricio ´n (CIBER obn), Instituto de Salud Carlos III, Madrid, Spain 3 Department of Nutrition and Food Sciences, University of Navarra, Pamplona, Spain 4 Cancer Prevention and Control Program, University of South Carolina, Columbia, SC, USA 5 Department of Epidemiology and Biostatistics, Arnold School of Public Health, University of South Carolina, Columbia, SC, USA 6 Department of Preventive Medicine, University of Valencia, Valencia, Spain 7 Human Nutrition Unit, Sant Joan University Hospital, IISPV, Universitat Rovira i Virgili, Reus, Spain 8 Inflammatory and Cardiovascular Disease Programme (RICAD), IMIM-Hospital del Mar Medical Research Institute, Barcelona, Spain 9 Department of Nutrition and Food Science, School of Pharmacy, Instituto de Investigacio ´n en Nutricio ´n y Seguridad Alimentaria, University of Barcelona, Barcelona, Spain 10 Department of Cardiology, University Hospital of Alava, Vitoria, Spain 11 Department of Preventive Medicine, University of Malaga, Malaga, Spain 12 Institute of Health Sciences (IUNICS), University of Balearic Islands, Palma de Mallorca, Spain 13 Department of Family Medicine, San Pablo Health Center, Primary Care Division of Sevilla, Andalusian Health Service & Department of Medicine, University of Sevilla, Spain 14 Research Institute of Biomedical and Health Sciences, University of Las Palmas de Gran Canaria, Las Palmas de Gran Canaria, Spain 15 Lipids and Vascular Risk Unit, Internal Medicine, Hospital Universitario de Bellvitge, Hospitalet de Llobregat, Barcelona, Spain 16 Lipid Clinic, Department of Endocrinology and Nutrition, Institut d’Investigacions Biome `diques August Pi Sunyer (IDIBAPS), Hospital Clinic, Barcelona, Spain 17 Department of Internal Medicine, IDIBAPS, Hospital Clinic, University of Barcelona, Barcelona, Spain (Submitted 26 July 2014 – Final revision received 3 December 2014 – Accepted 5 December 2014 – First published online 27 February 2015) Abstract The dietary inflammatory index (DII) is a new tool to assess the inflammatory potential of the diet. In the present study, we aimed to determine the association between the DII and BMI, waist circumference and waist:height ratio (WHtR). We conducted a cross-sectional study of 7236 participants recruited into the PREvencio ´n con DIeta MEDiterra ´nea trial. Information from a validated 137-item FFQ was used to calculate energy, food and nutrient intakes. A fourteen-item dietary screener was used to assess adherence to the Mediterranean diet (MeDiet). Sex-specific multivariable linear regression models were fitted to estimate differences (and 95 % CI) in BMI, waist circumference and WHtR across the quintiles of the DII. All nutrient intakes, healthy foods and adherence to the MeDiet were higher in the quintile with the lowest DII score (more anti-inflammatory values) except for intakes of animal protein, saturated fat and monounsaturated fat. Although an inverse association between the DII and total energy was apparent, the DII was associated with higher average BMI, waist circumference and WHtR after adjusting for known risk factors. The adjusted difference in the WHtR for women and men between the highest and *Corresponding author: M. Ruiz-Canela, fax: þ34 948 42 56 49, email [email protected] Abbreviations: CRP, C-reactive protein; DII, dietary inflammatory index; MeDiet, Mediterranean diet; PREDIMED, PREvencio ´n con DIeta MEDiterra ´nea; WHtR, waist:height ratio. British Journal of Nutrition (2015), 113, 984–995 doi:10.1017/S0007114514004401 qThe Authors 2015 British Journal of Nutrition https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0007114514004401 Downloaded from https://www.cambridge.org/core. ULPGC. Biblioteca Universitaria, on 25 Oct 2017 at 11:02:34, subject to the Cambridge Core terms of use, available at lowest quintiles of the DII was 1·60 % (95 % CI 0·87, 2·33) and 1·04 % (95 % CI 0·35, 1·74), respectively. Pro-inflammatory scores remained associated with obesity after controlling for the effect that adherence to a MeDiet had on inflammation. In conclusion, the present study shows a direct association between the DII and indices of obesity, and supports the hypothesis that diet may have a role in the development of obesity through inflammatory modulation mechanisms. Key words: Inflammation: Diet: Obesity: BMI: Waist circumference: Waist:height ratio The obesity pandemic constitutes a major public health problem in most high-income countries, and it is emerging as a threat in more affluent sectors of developing countries (1) . In 2008, more than 10 % of the World’s adult population, i.e. about 500 million people, were obese according to the WHO (2) . It was estimated that 3·4 million adult deaths worldwide were, in 2010, attributable to obesity or overweight (1) . This is a global crisis because 65 % of the world’s population live in countries where overweight and obesity kill more people than being underweight (2) . Obesity usually is the result of the accumulation of excess body fat, and it is often characterised as a state of low-grade chronic inflammation (3) . This obesity-induced inflammation has multi-organ metabolic effects affecting the adipose tissue, liver, muscle, pancreas and brain (4) . Metabolic differences exist according to the location of fat cells. For example, excessive deposition of fat in visceral adipose tissue (i.e. intra-abdominal fat) is associated with higher health risks than subcutaneous fat accumulation in the extremities (5) .In fact, different anthropometric adiposity measures including waist circumference or waist:height ratio (WHtR) are used to assess the role of adiposity in CVD risk (6,7) . A number of studies have shown an association between diet and inflammatory biomarkers, and how this translates into increased or decreased risk of chronic metabolic diseases (8–15) . Part of the preventive role of healthy dietary patterns, such as the Mediterranean diet (MeDiet), could be attributed to the anti-inflammatory properties of some of their main components (15–19) . This anti-inflammatory effect may decrease the low-grade inflammation usually found in obese patients (20,21) . However, a MeDiet may also attenuate inflammation in the absence of weight loss (22) . A recent hypothesis is that obesity could also be partly the consequence of a previous chronic low-grade inflammation; therefore, a bidirectional association between inflammation and obesity may exist (23) . Consequently, it can be useful to characterise an individual’s diet according to its inflammatory properties in order to investigate the inflammatory links between obesity and diet (24) .The dietary inflammatory index (DII) is a new tool to assess this inflammatory potential of the diet (25) . In the present study, we examine the relationships between nutrient intake or food group consumption and the DII, as well as the association between the DII and indices of both general and abdominal obesity in the PREvencio ´n con DIeta MEDiterra ´nea (PREDIMED) trial. Methods Ethics statement The protocol was approved by the Research Ethics Committees at all recruiting centres: University of Navarra; University of Valencia; University Rovira i Virgili; IMIM-Hospital del Mar Medical Research Institute; University of Barcelona; University Hospital of Alava; University of Malaga; University of the Balearic Islands; University of Las Palmas de Gran Canaria; University Hospital of Bellvitge; Hospital Clinic. Participants signed a written informed consent form. Study design and participants The ‘PREDIMED’ study was a parallel-group, multi-centre, clinical trial that aimed to assess the effects of the traditional MeDiet on the primary prevention of CVD (protocol available at http://www.predimed.es). A detailed description of methods and patients has been published elsewhere (26,27) . The study was conducted between October 2003 and December 2010 by eleven recruiting centres in Spain. Eligible participants were men aged 55–80 years and women aged 60–80 years with no previous CVD. At baseline, participants should have a diagnosis of type 2 diabetes mellitus or at least three of the following major cardiovascular risk factors: smoking (more than one cigarette per d during the last month); hypertension (systolic blood pressure $140 mmHg or diastolic blood pressure $90 mmHg or antihypertensive medication); elevated LDL-cholesterol levels ($1600 mg/l); low HDL-cholesterol levels (#400 mg/l in men or #500 mg/l in women, independently of lipid-lowering therapy); BMI $25 kg/m 2 ; family history of premature CHD. A total of 7447 participants were randomised in a 1:1:1 ratio to a parallel-design intervention trial of dietary advice: (1) a MeDiet supplemented with extra-virgin olive oil; (2) a MeDiet supplemented with nuts; (3) a low-fat diet (control group). Medical conditions and risk factors related to eligibility were collected using a questionnaire during the first screening visit. Participants, with the assistance of trained dietitians, completed an FFQ. This FFQ was adapted from the Willett questionnaire and validated in Spain (28) . It includes 137 items plus vitamin/mineral supplements, and specific questions for patterns of alcohol consumption. Energy and nutrient intakes were calculated from Spanish food composition tables (29) . Participants also completed the Spanish validated version of the Minnesota physical activity questionnaire (30) , and a fourteen-item dietary screener to assess the adherence to the MeDiet (31) . PREDIMED dietitians were responsible for the accurate completion of the questionnaires. For the present study, 133 participants were excluded from the analyses because they reported values for total energy intake outside of the predefined limits (,3347 kJ (,800 kcal)/d or .17 573 kJ (.4200 kcal)/d for men; ,2510 kJ (,600 kcal)/d or .14 644 kJ (.3500 kcal)/d for women). These limits were set in accordance with those recommended by Willett in Nutritional Epidemiology (32) . Another seventy-eight participants were excluded Dietary inflammatory index and obesity 985 British Journal of Nutrition https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0007114514004401 Downloaded from https://www.cambridge.org/core. ULPGC. Biblioteca Universitaria, on 25 Oct 2017 at 11:02:34, subject to the Cambridge Core terms of use, available at because of lack of information on the FFQ needed to calculate the DII. This study was registered as an International Standard Randomised Controlled Trial, number ISRCTN35739639. Dietary inflammatory index The design and development of the DII has been described elsewhere (25) . Briefly, the DII is a scoring algorithm based on an extensive review of the literature published from 1950 to 2010, linking 1943 articles to a total of forty-five food parameters and including various macronutrients, micronutrients, flavonoids and food items (Fig. 1). These dietary parameters were scored according to whether they increased (þ1), decreased (21) or had no effect (0) on six inflammatory biomarkers (IL-1b, IL-4, IL-6, IL-10, TNF-aand C-reactive protein (CRP)). An overall food parameter-specific inflammatory effect score was calculated and multiplied by a centred percentile value for each food. This percentile was calculated by first linking the dietary data from a study to the regionally representative world database intake, which was based on actual human consumption in eleven populations from different parts of the world that provided a robust estimate of a mean and standard deviation for each parameter. These values then become the multipliers to express an individual’s exposure, relative to the ‘standard global mean’ as a z-score. This was achieved by subtracting the ‘standard global mean’ from the amount reported, and dividing this value by the standard deviation. To minimise the effect of ‘right skewing’, this value was then converted to a centred percentile score. The centred percentile score for each food parameter for each individual was multiplied by the respective food parameter effect score that was derived from the literature review in order to obtain a food parameter-specific DII score for an individual. All of the food parameter-specific DII scores were then summed to create the overall DII score for every participant in the study. The greater the DII score, the more pro-inflammatory the diet, and more negative values represent more anti-inflammatory diets. The DII score could take on values ranging from 7·98 (maximally pro-inflammatory) to 28·87 (maximally anti-inflammatory) (25) . Construct validation of the DII was performed using data derived from two different sources of dietary intake information, and serum high-sensitivity CRP as the construct validator (33) . Outcome Trained and certified PREDIMED nurses performed all baseline anthropometric adiposity measures including weight and height (from which BMI (kg/m 2 ) was computed), waist circumference (cm) and WHtR (%) following validated procedures. A waist:height ratio equal to 1 was taken as 100 %. Baseline weight was measured using a calibrated balance beam scale with the subjects barefoot and wearing light clothes. The nurse measured height using a wall-mounted calibrated stadiometer. Waist circumference was measured using an anthropometric measuring tape, at a horizontal plane midway between the lowest rib and the iliac crest. Statistical analyses Statistical analyses were stratified by sex. Comparisons of quantitative variables across the quintiles of the DII were made using a one-way ANOVA. The compared variables included total energy intake, physical activity and nutrient and food consumption. Intakes of carbohydrate, protein and fat (and fat subtypes) are expressed as a percentage of total energy intake (Table 1). Categorical variables were compared using the Pearson x 2 test. Sex-specific least-squared means of BMI, waist circumference and WHtR were estimated across the quintiles of the DII. Pearson’s correlation coefficients (95 % CI) between these anthropometric adipose measures and the DII were also calculated. Sex-specific multiple linear regression models were used to estimate the differences (and 95 % CI) in the indices of general obesity and abdominal obesity according to the quintiles of the DII. Covariates included in these models were age (years), smoking status (never, current or former smoker), diabetes (yes or no), hypertension (yes or no), leisure-time physical activity (metabolic equivalents-min/d), educational level (illiterate/elementary education, secondary education or university), marital status (married, widowed, single or other), total energy intake (kJ/d) and study centre. In addition, tests of linear trend across the successive quintiles of the DII were conducted using the median value for each quintile category as a continuous variable, and after adjusting for the aforementioned confounding variables. Residuals of the DII were obtained in a linear regression analysis of the association between the DII and a previously validated fourteen-item PREDIMED screener of adherence to the MeDiet (31) . These residuals represent the information provided by the DII, which is not explained at all by adherence to the MeDiet (i.e. they exhibit zero correlation with the MeDiet score). They were included as an independent variable after transformation into quintiles in a multivariable regression model with the same covariates listed previously (residual model). All Pvalues presented are two-tailed, and differences were considered statistically significant at P#0·05. All statistical analyses were performed using STATA w version 12.0 (Stata Corp). Results Of the 7447 initially randomised subjects in the PREDIMED trial, 7236 were included in the present study. The remaining participants (n211, 2·8 %) were excluded because of incomplete data on their FFQ (n78) or baseline energy intake outside of the predefined values (n133). Among the 7236 participants, 57 % were women. The mean age of the participants was 68 (SD 5·8) years for women and 66 (SD 6·6) years for men. The median DII score for women was 20·78 (24·90 to 3·68) and 20·91 (25·23 to 3·69) for men. Table 1 shows the main characteristics of the participants according to the categories of the DII score by sex. All differences between the quintiles of this index were statistically M. Ruiz-Canela et al.986 British Journal of Nutrition https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0007114514004401 Downloaded from https://www.cambridge.org/core. ULPGC. Biblioteca Universitaria, on 25 Oct 2017 at 11:02:34, subject to the Cambridge Core terms of use, available at Review of articles published from 1950 to 2010, resulting in 1943 studies linking a total of forty-five food parameters with inflammatory biomarkers +1 to each article if the effects were pro-inflammatory (significantly increased IL-1β, IL-6, TNF-α or CRP, or decreased IL-4 or IL-10), –1 if the effects were anti-inflammatory (significantly decreased IL-1β, IL-6, TNF-α or CRP, or increased IL-4 or IL-10), 0 if the food parameter did not produce any significant change in the inflammatory marker The score for each food parameter was weighted according to the study design. The weights were 10 (experimental design), 8 (observational), 7 (case–control), 6 (cross-sectional), 5 (experimental with animals), 3 (cell culture) A food parameter-specific overall inflammatory effect score was calculated by substracting the anti-inflammatory fraction from the pro-inflammatory fraction. This score was corrected if the total weighted number of articles was <236. In these cases the raw overall inflammatory score is multiplied by the total weighted number of articles divided by 236 z-Score and centred percentiles for each of the thirty-three food parameters* for each participant of the PREDIMED trial were calculated based on the average and standard deviation for each food parameter obtained from the global database that was created from the consumption of the original forty-five food parameters fron eleven countries from around the world The centred percentile for each food parameter is multiplied by the respective ‘overall food parameter-specific inflammatory effect score’ to obtain the ‘food parameter-specific DII score’ Twelve food parameters were excluded because they could not be measured with the FFQ used in the PREDIMED trial All of the ‘food parameter-specific DII scores’ are summed to create the ‘overall DII score’ for each individual A score for each food parameter was calculated giving: Fig. 1. Sequence of steps in creating the dietary inflammatory index in the PREvencio ´n con DIeta MEDiterra ´nea (PREDIMED) trial. *Alcohol, b-carotene, caffeine, carbohydrate, cholesterol, energy, iron, fibre, folic acid, garlic, green/black tea, magnesium, MUFA, n-3 fatty acids, n-6 fatty acids, niacin, onion, pepper, protein, PUFA, riboflavin, saturated fat, Se, thiamin, total fat, trans-fat, vitamin A, vitamin B 12 , vitamin B 6 , vitamin C, vitamin D, vitamin E and Zn. CRP, C-reactive protein; DII, dietary inflammatory index. (A colour version of this figure can be found online at http://www.journals.cambridge.org/bjn). Dietary inflammatory index and obesity 987 British Journal of Nutrition https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0007114514004401 Downloaded from https://www.cambridge.org/core. ULPGC. Biblioteca Universitaria, on 25 Oct 2017 at 11:02:34, subject to the Cambridge Core terms of use, available at Table 1. Description of the main characteristics of the participants according to quintiles (Q) of the dietary inflammatory index (DII) score in the PREvencio ´n con DIeta MEDiterra ´nea (PREDIMED) Trial, 2003–9 (Mean values and standard deviations; median values, minimum and maximum values, and percentages) Quintiles of the DII Women Men Q1* Q2 Q3 Q4 Q5† PQ1* Q2 Q3 Q4 Q5† P DII Median 22·6 21·6 20·8 0·1 1·5 22·7 21·7 20·9 20·07 1·3 Minimum, maximum 24·9, 2222, 21·2 21·2, 20·3 20·3, 0·7 0·7, 3·7 25·2, 22·2 22·2, 21·4 21·4, 20·5 20·5, 0,5 0·5, 3·7 n829 829 829 829 829 619 618 618 618 618 Age (years) ,0·001 0·180 Mean 6767686869 6666666666 SD 66 6 66 6 6 7 77 Family history of early CHD (%) 27 28 27 26 24 0·362 21 18 17 16 15 0·039 Hypertension (%) 84 86 87 87 87 0·359 76 78 79 79 77 0·436 Dyslipidaemia (%) 79 79 76 75 74 0·016 71 66 66 65 63 0·033 Diabetes (%) 44 40 44 44 49 0·011 53 53 53 56 56 0·625 Smoking (%) 0·250 0·176 Current smoker 4 5 6 6 6 24 24 25 28 27 Former smoker 7 8 8 8 6 48 52 46 46 49 Total energy intake (kcal/d) ,0·001 ,0·001 Mean 2442 2281 2136 1990 1767 2769 2577 2412 2252 2038 SD 478 449 435 414 400 585 558 511 459 463 Total energy intake (kJ/d) Mean 10 217 9544 8937 8326 7393 11 586 10 782 10 092 9422 8527 SD 2000 1879 1820 1732 1674 2448 2335 2138 1920 1937 Physical activity (MET-h/d) ,0·001 ,0·001 Mean 3·4 2·9 3·0 2·8 2·4 6·1 5·3 5·1 4·8 4·3 SD 3·2 2·8 2·8 2·7 2·4 6·0 4·7 4·9 4·4 3·9 Alcohol intake (g/d) 0·001 ,0·001 Mean 3·8 3·2 2·7 3·3 2·7 18·4 15·2 15·5 14·5 13·3 SD 7·6 6·1 5·1 6·1 5·7 18·1 17·5 19·0 18·2 17·9 Marital status (%) 0·004 0·043 Married 72 66 66 67 62 92 91 89 89 86 Widowed 20 27 27 26 30 3 3 3 3 6 Single/other 8 7 7 7 8 5 6 8 8 8 Educational level (%) ,0·001 0·066 Primary education or less 80 86 87 86 88 64 66 71 69 67 Secondary education 13 11 9 11 9 21 23 19 22 22 Any college 7 3 4 3 3 15 12 11 10 11 MET, metabolic equivalents. * Highest anti-inflammatory values of the DII. † Highest pro-inflammatory values of the DII. M. Ruiz-Canela et al.988 British Journal of Nutrition https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0007114514004401 Downloaded from https://www.cambridge.org/core. ULPGC. Biblioteca Universitaria, on 25 Oct 2017 at 11:02:34, subject to the Cambridge Core terms of use, available at Table 2. Nutrient and food consumption according to quintiles (Q) of the dietary inflammatory index (DII) in the PREvencio ´n con DIeta MEDiterra ´nea (PREDIMED) trial, 2003–9 (Mean values and standard deviations) Quintiles of the DII Women Men Q1* Q2–Q4 Q5† Q1* Q2–Q4 Q5† Mean SD Mean SD Mean SD PMean SD Mean SD Mean SD P Carbohydrate intake (%E) 43·4 7·0 42·6 6·7 40·8 6·8 ,0·001 41·2 7·0 41·5 7·3 39·2 7·8 ,0·001 Protein intake (%E) 17·2 2·6 17·1 2·7 17·2 3·1 0·452 16·0 2·6 15·9 2·7 16·0 2·8 0·542 Vegetal protein intake (%E) 6·1 1·1 5·4 1·0 4·9 1·0 ,0·001 5·8 1·1 5·4 1·0 4·9 1·0 ,0·001 Animal protein intake (%E) 11·1 2·7 11·6 2·8 12·3 3·2 ,0·001 10·2 2·7 10·5 2·8 11·1 2·9 ,0·001 Total fat intake (%E) 38·4 6·8 39·4 6·6 40·9 6·9 ,0·001 38·2 6·5 38·5 6·7 40·4 7·2 ,0·001 Saturated fat (%E) 9·2 2·0 10·0 2·1 10·8 2·4 ,0·001 9·3 2·1 9·9 2·2 10·8 2·4 ,0·001 Monounsaturated fat (%E) 18·2 4·5 19·7 4·5 21·0 4·8 ,0·001 18·3 4·2 19·2 4·4 20·9 4·6 ,0·001 Polyunsaturated fat (%E) 7·2 2·2 6·2 2·1 5·6 1·7 ,0·001 7·1 2·2 6·2 2·0 5·6 1·6 ,0·001 Alcohol consumption (g/d) 3·8 7·4 3·0 5·8 2·8 5·8 0·002 18·5 18·1 15·0 18·2 13·4 17·8 ,0·001 Fibre (g/d) 34·6 9·3 24·7 6·1 16·1 3·6 ,0·001 34·4 8·8 25·8 6·5 17·1 3·8 ,0·001 Vitamin E (mg/d) 12·5 3·7 9·7 3·7 7·0 2·6 ,0·001 13·4 4·8 10·5 4·0 7·5 2·5 ,0·001 Vitamin C (mg/d) 289 99 201 70 123 42 ,0·001 271 98 195 72 122 44 ,0·001 Vitamin A (mg/d) 1757 784 1235 640 847 485 ,0·001 1870 907 1325 1095 849 500 ,0·001 Vegetables (g/d) 483 168 327 107 213 78 ,0·001 482 179 324 111 212 80 ,0·001 Fruits (g/d) 487 214 379 186 231 118 ,0·001 462 231 377 195 237 138 ,0·001 Cereals (g/d) 143 78 132 76 111 63 ,0·001 175 96 165 95 138 86 ,0·001 Potatoes (g/d) 96 58 80 48 62 39 ,0·001 102 58 86 46 70 40 ,0·001 Legumes (g/d) 25·8 16·8 19·7 12·9 14·8 7·8 ,0·00126·215·121·514·016·3 8·4,0·001 Nuts (g/d) 18·6 18·4 8·2 10·8 3·3 5·6 ,0·001 20·6 19·6 10·3 12·9 4·9 8·1 ,0·001 Fish and seafoods (g/d) 123·9 60·6 95·2 41·7 74·8 36·5 ,0·001 130·7 56·2 101·0 53·8 79·7 39·3 ,0·001 Meat and meat products (g/d) 125·0 57·3 127·2 50·9 114·5 49·1 ,0·001 144·7 68·8 141·1 58·3 136·0 61·9 0·044 Dairy products (g/d) 435·5 238·4 408·1 221·0 381·4 218·9 ,0·001 369·1 222·4 342·4 204·7 319 205·9 ,0·001 Mediterranean diet score (0–14) 9·5 1·9 8·6 1·8 7·7 1·7 ,0·001 9·8 1·9 8·7 1·9 8·0 1·7 ,0·001 %E, percentage of energy. * Highest anti-inflammatory values of the DII. † Highest pro-inflammatory values of the DII. Dietary inflammatory index and obesity 989 British Journal of Nutrition https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0007114514004401 Downloaded from https://www.cambridge.org/core. ULPGC. Biblioteca Universitaria, on 25 Oct 2017 at 11:02:34, subject to the Cambridge Core terms of use, available at significant among women, except for the percentage of subjects with a family history of early CHD, the presence of hypertension and smoking status. Among men, differences between the quintiles of the DII according to age, hypertension, diabetes and smoking status were not statistically significant. In both sexes, the level of physical activity was inversely associated with the DII, as was total energy intake and alcohol intake. All macroand micronutrient intakes were higher in the quintile with the lowest DII score (anti-inflammatory dietary pattern), except for intakes of animal protein, saturated fat and monounsaturated fat, both among women and men (Table 2). Better adherence to a MeDiet also was associated with lower DII scores. Table 3 shows the adjusted indices of obesity based on BMI, waist circumference and WHtR, according to the DII score stratified by sex. The lower and upper limits of this score are shown for each quintile. Mean values of all three adiposity indices increased linearly across the successive quintiles of DII scores (from anti-inflammatory to pro-inflammatory levels). A significant positive correlation was observed between these obesity indices and the DII score. Among women, the DII was directly associated with BMI, after adjusting for multiple factors related to obesity (Table 4). Being in the highest quintile of the DII was associated with an increase in BMI of 0·79 kg/m 2 (95 % CI 0·35, 1·23) compared with the lowest quintile (Pfor trend¼0·001). This association was not statistically significant for men. Table 4 further shows that waist circumference and WHtR increased progressively across quintiles 2–4 and 5 compared with the lowest quintile of the DII, both in women and men (Pfor trend being statistically significant in all comparisons). Table 5 shows the association of the DII with the anthropometric indices, after considering the possible contribution of the MeDiet elements to the antior pro-inflammatory capacity of the diet. A higher pro-inflammatory level of the diet (beyond the effect of lower adherence to the MeDiet) was associated with higher adjusted means of BMI, waist circumference and WHtR (Pfor trend ,0·05 in all comparisons except for BMI among men). The predicted increase in anthropometric measures was statistically significant in women, except for the increase in BMI and WHtR, when the intermediate DII quintiles (2–4) were compared with the lowest category. On the contrary, the results were not statistically significant among men, except for waist circumference and WHtR, when comparing the highest v. the lowest quintile of the residuals of the DII. Discussion In the present study, we used the dietary inflammatory index (DII) score to assess the capacity of the overall dietary pattern to promote inflammation. Higher values of the DII represent a higher inflammatory potential of the diet. As expected, we observed that the DII was inversely associated with the intake of healthy foods, nutrients and adherence to the MeDiet. A pro-inflammatory DII was directly associated with the indices of general and abdominal obesity, independent Table 3. General obesity and abdominal obesity according to quintiles (Q) of the dietary inflammatory index in the PREvencio ´n con DIeta MEDiterra ´nea (PREDIMED) trial, 2003–9 (Adjusted average indices and 95 % confidence intervals) Quintiles of the dietary inflammatory index Q1 (highest anti-inflammatory) Q2 Q3 Q4 Q5 (highest pro-inflammatory) Adjusted average indices 95 % CI Adjusted average indices 95 % CI Adjusted average indices 95 % CI Adjusted average indices 95 % CI Adjusted average indices 95 % CI Pearson coefficient (r)95%CI Women (n) 829 829 829 829 829 Range 24·9 to 22·0 22·0 to 21·2 21·2 to 20·3 20·3 to 0·7 0·7 to 3·7 BMI (kg/m 2 )* 29·9 29·6, 30·2 30·3 30·0, 30·5 30·5 30·3, 30. 8 30·7 30·4, 31·0 30·8 30·5, 31·1 0·06 0·03, 0·09 Waist circumference (cm)* 96·5 95·8, 97·3 97·9 97·2, 98·6 99·0 98·3, 99·7 99·1 98·4, 99·8 99·5 98·8, 100·3 0·05 0·02, 0·08 Waist:height ratio (%)*† 62·7 62·2, 63·1 63·6 63·9, 64·8 64·0 63·5, 64·4 64·1 63·7, 64·6 64·3 63·9, 64·8 0·06 0·03, 0·09 Men (n) 619 618 618 618 618 Range 25·2 to 22·2 22·2 to 21·4 21·4 to 20·5 20·5 to 0·5 0·5 to 3·7 BMI (kg/m 2 )* 29·2 28·9, 29·5 29·1 28·8, 29·3 29·2 29·0, 29·5 29·5 29·3, 29·8 29·6 29·3, 29·9 0·05 0·01, 0·08 Waist circumference (cm)* 102·2 101·4, 103·0 102·4 101·6, 103·2 103·0 102·3, 103·8 104·4 103·7, 105·2 104·2 103·4, 105·0 0·08 0·05, 0·20 Waist:height ratio (%)*† 61·2 60·8, 61·7 61·2 60·8, 61·7 61·7 61·3, 62·1 62·3 61·9, 62·7 62·4 61·9, 62·9 0·09 0·06, 0·13 * Adjusted for age, smoking status, diabetes, hypertension, marital status, educational level, physical activity, total energy intake and study centre. † A waist:height ratio equal to 1 was taken as 100 %. M. Ruiz-Canela et al.990 British Journal of Nutrition https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0007114514004401 Downloaded from https://www.cambridge.org/core. ULPGC. Biblioteca Universitaria, on 25 Oct 2017 at 11:02:34, subject to the Cambridge Core terms of use, available at of established risk factors for obesity including total energy intake, age, smoking status, diabetes, hypertension, physical activity, educational level and marital status. These results were consistent for both sexes except for BMI in men. In the residual model (after removing the variability explained by the MeDiet), the association between the inflammatory potential of the diet and higher adiposity indices remained apparent; however, there was a clear association between the DII and the abdominal indices of obesity for women compared with men. The associations observed between nutrient intake or food consumption and the DII are consistent with previous research. Several studies have shown an inverse association between healthy diets and markers of inflammation, as well as a direct association with ‘Western-like’ dietary patterns (8–14) . Specifically, a lower CRP concentration has been associated with a higher intake of fruits and vegetables (34–36) , legumes (37) , nuts (38) , and low-fat dairy consumption (39) . Previous studies have also observed associations of specific nutrients such as total dietary fibre intake (40) , moderate alcohol consumption (41) , and vitamin E and vitamin C intake (42) with lower levels of inflammation markers. On the contrary, animal protein seems to increase the inflammatory status of obese individuals (43) . We also found that a higher consumption of dairy products and meat (or meat products) was less frequent in the highest DII quintile. A systematic review has found no impact of dairy product consumption on biomarkers of inflammation in overweight and obese adults (44) . However, only one out of eight trials included in this review defined inflammation as its primary outcome, and there were some methodological limitations in them such as insufficient statistical power (44) . Concerning the consumption of meat, a cross-sectional analysis of data from 3690 diabetes-free female participants found that a higher intake of meat protein was associated with higher plasma levels of inflammatory markers (45) . However, a cross-sectional study has shown that the association between red meat intake and inflammatory markers was no Table 4. Multivariable-adjusted* differences in the indices of general obesity and abdominal obesity according to quintiles (Q) of the dietary inflammatory index in the PREvencio ´n con DIeta MEDiterra ´nea (PREDIMED) trial, 2003–9 (Adjusted differences and 95 % confidence intervals) Quintiles of the dietary inflammatory index Q1 (highest anti-inflammatory) Q2–Q4 Q5 (highest pro-inflammatory) Adjusted difference 95 % CI Adjusted difference 95 % CI Adjusted difference 95 % CI Pfor trend Women BMI (kg/m 2 )* 0 Reference 0·56 0·23, 0·89 0·79 0·35, 1·23 0·001 Waist circumference (cm)* 0 Reference 2·03 1·17, 2·90 2·79 1·64, 3·93 ,0·001 Waist:height ratio (%)*† 0 Reference 1·19 0·64, 1·74 1·60 0·87, 2·33 ,0·001 Men BMI (kg/m 2 )* 0 Reference 0·05 20·26, 0·37 0·33 20·08, 0·74 0·100 Waist circumference (cm)* 0 Reference 0·98 0·07, 1·89 1·74 0·53, 2·94 0·005 Waist:height ratio (%)*† 0 Reference 0·44 20·09, 0·97 1·04 0·35, 1·74 0·003 * Adjusted for age, smoking status, diabetes, hypertensive, physical activity, total energy intake, educational level, marital status and study centre. † A waist:height ratio equal to 1 was taken as 100 %. Table 5. Multivariable-adjusted* differences in the indices of general obesity and abdominal obesity according to adherence to the residuals of the dietary inflammatory index on the fourteen-item PREvencio ´n con DIeta MEDiterra ´nea (PREDIMED) score of adherence to the Mediterranean diet in the PREDIMED trial 2003–9 (Adjusted differences and 95 % confidence intervals) Quintiles of the dietary inflammatory index (adjusted for adherence to the Mediterranean diet) Q1 (highest anti-inflammatory) Q2–Q4 Q5 (highest pro-inflammatory) Adjusted difference 95 % CI Adjusted difference 95 % CI Adjusted difference 95 % CI Pfor trend* Women BMI (kg/m 2 )* 0 Reference 0·14 20·19, 0·48 0·36 20·08, 0·80 0·041 Waist circumference (cm)* 0 Reference 1·01 0·14, 2·81 1·67 0·52, 2·82 ,0·001 Waist:height ratio (%)*† 0 Reference 0·34 20·20, 0·91 0·76 0·04, 1·49 0·007 Men BMI (kg/m 2 )* 0 Reference 0·06 20·26, 0·36 0·31 20·10, 0·73 0·057 Waist circumference (cm)* 0 Reference 0·77 20·14, 1·69 1·46 0·24, 2·68 0·004 Waist:height ratio (%)*† 0 Reference 0·44 20·09, 0·96 0·96 0·26, 1·67 0·005 * Adjusted for age, smoking status, diabetes status, hypertensive status, physical activity, energy intake, educational level, marital status and study centre. † A waist:height ratio equal to 1 was taken as 100 %. Dietary inflammatory index and obesity 991 British Journal of Nutrition https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0007114514004401 Downloaded from https://www.cambridge.org/core. ULPGC. Biblioteca Universitaria, on 25 Oct 2017 at 11:02:34, subject to the Cambridge Core terms of use, available at longer observed after adjustment for BMI (46) . Therefore, it is suggested that the association between red meat intake and inflammation is probably mediated by obesity. In the present study, a higher pro-inflammatory diet was observed in participants with higher BMI, waist circumference and WHtR. This result suggests the hypothesis that a dietinduced inflammation might contribute to increasing or maintaining obesity, especially abdominal obesity, in a population that is mostly overweight or obese. The origin of inflammation during obesity is not yet fully understood. It is acknowledged that inflammation is induced by adiposity (4,5) , but this relationship can be bidirectional (i.e. a pro-inflammatory diet can increase or maintain adiposity), thus creating a vicious cycle, because nutrient excess and some specific foods or nutrients also have been associated with inflammation (47) . The potential mechanisms underlying this association is the activation of pathogen-associated molecular patterns, such as Toll-like receptors and Nod-like receptors, which induce the activation of inflammatory markers in several tissues including the adipose tissue (48) . Moreover, dietary patterns (e.g. highfat/low-fibre or low-fat/high-fibre diet) and single specific nutrients (e.g. dietary fibre) appear to have important consequences in the gut microbiota, which is also involved in low-grade inflammation associated with obesity (49–52) . The residuals of the DII (from a regression model on adherence to the MeDiet) were also associated with obesity indices. These residuals represent the information provided by the DII about the antior pro-inflammatory capacity of a diet, which could not be explained by adherence to the MeDiet. The most pro-inflammatory diet showed a stronger association with waist circumference than with other anthropometric indices, both among women and men. These results are in close agreement with previous findings, which showed that central adiposity-related indices are more strongly correlated with plasma pro-inflammatory markers than indices assessing total adiposity in healthy young adults (5) . Moreover, abdominal adiposity has been associated with elevated CRP levels, independent of BMI in older adults (53) . As a consequence, the present results reinforce the usefulness of the DII to assess the inflammatory properties of a diet, and the association between inflammation and central obesity indices. The present results are also consistent with those of studies reporting a stronger association between CRP and BMI in women than in men (54,55) . This between-sex difference could be partially explained by a greater accumulation of subcutaneous fat in women than in men, and higher lean mass in men (55) . Sex differences in the metabolic activity of adipose tissue, as well as in the association between leptin and CRP, may also explain these differences (56,57) . The strengths of the present study include the following: large sample size; use of a validated instrument to measure the inflammatory potential of the diet; adjustment for a large number of factors associated with obesity; detailed measures of obesity indices; validation of all assessment instruments including the MeDiet screener, the FFQ and the physical activity questionnaire. The present study also has its limitations, the main one being the cross-sectional nature of our analyses. It is, therefore, unclear whether obese individuals are more likely to choose pro-inflammatory diets, or if pro-inflammatory diets contribute to promoting or maintaining obesity. Both weight reduction and an overall healthy dietary pattern have the capacity to reduce inflammatory markers. Thus, the association between the DII and obesity indices remains to be confirmed in prospective analyses. Another limitation is that anthropometric measures are surrogate markers of abdominal obesity. Waist circumference and WHtR do not differentiate between visceral adipose tissue and subcutaneous abdominal adipose tissue (58) . Therefore, we cannot determine whether the DII is more strongly associated with visceral, subcutaneous or both types of abdominal fat mass. Finally, the DII is limited by the existing knowledge of the inflammatory factors involved in obesity. However, the DII has been found to be associated with the following factors: inflammatory cytokines including CRP and IL-6 (33,59,60) ; glucose intolerance component of the metabolic syndrome (59) ; odds of asthma and of reduced FEV 1 (forced expiratory volume in 1 s) in an Australian population (60) . It has also been reportedly associated with a higher risk of colorectal cancer (61) , prostate cancer (62) and pancreatic cancer (63) . In conclusion, the present findings indicate an association between anti-inflammatory values of the DII and intake of healthy foods and nutrients and higher adherence to the MeDiet. A pro-inflammatory diet is associated with elevated indices of central and abdominal obesity. This association suggests that the DII may have the capacity to help elucidate the role that diet plays in the development of obesity through inflammatory processes. Supplementary material To view supplementary material for this article, please visit http://dx.doi.org/10.1017/S0007114514004401 Acknowledgements The authors are very grateful to all the participants of the study for their enthusiastic collaboration; the PREDIMED personnel for their excellent assistance; and the personnel of all affiliated primary care centres. The PREDIMED trial was supported by the Instituto de Salud Carlos III (ISCIII), the official funding agency for Biomedical Research of the Spanish Government, through grants provided to research networks specifically developed for the trial: RTIC G03/140 (to R. E.), CIBERobn and RTIC RD 06/0045 (to M. A. M.-G.). J. R. H. was supported by an Established Investigator Award in Cancer Prevention and Control from the Cancer Training Branch of the National Cancer Institute (K05 CA136975). The authors also acknowledge grants received from Centro Nacional de Investigaciones Cardiovasculares CNIC 06/2007, Fondo de Investigacio ´n Sanitaria – Fondo Europeo de Desarrollo Regional (PI04-2239, PI05/ 2584, CP06/00100, PI07/0240, PI07/1138, PI07/0954, PI07/ 0473, PI10/01407, PI11/01647 and PI1102505), Ministerio de Ciencia e Innovacio ´n (AGL-2009-13906-C02, AGL2010-22319C03), Fundacio ´n Mapfre 2010, Public Health Division of the Department of Health of the Autonomous Government of M. Ruiz-Canela et al.992 British Journal of Nutrition https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0007114514004401 Downloaded from https://www.cambridge.org/core. ULPGC. Biblioteca Universitaria, on 25 Oct 2017 at 11:02:34, subject to the Cambridge Core terms of use, available at