Divergent Anti-Inflammatory Effects of Different Oil Acute Consumption on Healthy Individuals

ORIGINAL ARTICLE Divergent anti-inflammatory effects of different oil acute consumption on healthy individuals

N Papageorgiou1, D Tousoulis1, T Psaltopoulou, A Giolis, C Antoniades, E Tsiamis, A Miliou, K Toutouzas, G Siasos, C Stefanadis

1st Cardiology Unit, Athens University Medical School, Hippokration Hospital, Athens, Greece

Background/Objectives: Inter-cellular adhesion molecule 1 (ICAM-1), vascular cell adhesion molecule 1 (VCAM-1) and tumor necrosis factor-a (TNF-a), are implicated in atherogenesis. In addition, several types of oil as part of different types of diet are associated with the initiation of atherosclerosis and consequently with the risk of cardiovascular disease. However, the exact role of oil consumption on vascular inflammation remains unknown. In this parallel study, we assessed the acute effects of extra virgin olive oil, soy oil, corn oil and cod liver oil on circulating soluble(s) forms of adhesion molecules and TNF-a. Subjects/Methods: In all, 67 healthy volunteers were randomized to receive 50 ml of oil. Soluble forms of VCAM-1, ICAM-1 and TNF-a were measured by enzyme-linked immunosorbent assay at baseline and at 3 h post oil consumption. Results: All types of oil had no significant effect on soluble VCAM-1 levels (P ¼ nonsignificant (NS) for all). On the contrary, all oil types decreased ICAM-1 levels (Po0.01). Olive oil (Po0.05), soy oil and cod liver oil (Po0.01 for both) reduced TNF-a levels significantly, in contrast to corn oil, which induced a nonsignificant decrease (P ¼ NS). Moreover, there was a significant correlation between the absolute change in ICAM-1 and TNF-a levels (r ¼ 0.379, Po0.05), but not between the absolute changes in VCAM-1 and TNF-a levels (r ¼ 0.019, P ¼ NS). Conclusions: Acute consumption of all types of oil decreased significantly ICAM-1 levels. In addition, olive oil, soy oil and cod liver oil decreased significantly TNF-a levels. Moreover, the absolute change in TNF-a levels was correlated with the absolute change in ICAM-1 levels. These findings indicate that acute consumption of specific types of oil is associated with specific significant anti-inflammatory effects. European Journal of Clinical Nutrition (2011) 65, 514–519; doi:10.1038/ejcn.2011.8; published online 16 February 2011

Keywords: oil; ICAM-1; VCAM-1; TNF-a; inflammation

Introduction the vascular wall (Klein et al., 1995). Migration to the arterial wall, rolling and tethering of leukocytes to the endothelium Evidence suggests that inflammatory process contributes to (adhesion) has a crucial role in atherogenesis. Adhesion is a the development of atherosclerosis (Libby et al., 2009). The complex process depending on interactions between mole- role of several inflammatory biomarkers, such as acute phase cules such as the vascular adhesion molecule-1 (VCAM-1) proteins, proinflammatory cytokines and adhesion mole- and the intercellular molecule-1 (ICAM-1) on the endothe- cules, is nowadays under investigation as regards the relation lium and the integrins on neutrophils (Meydani, 2003). with cardiovascular disease (Biasucci, 2004). Interestingly, studies have shown that despite the similar Tumor necrosis factor-a (TNF-a) is known to be released by structure and function of VCAM-1 and ICAM-1, these two macrophages, as well as vascular endothelial and smooth molecules may participate at different stages of atherogenesis muscle cells, promoting the expression of cell adhesion (Blann and Lip, 2000). molecules and the migration of leukocytes into and through Recently, it is widely reported that dietary habits may modify cardiovascular disease rates in diverse populations. Adhesion molecules and proinflammatory cytokines are Correspondence: Dr D Tousoulis, Cardiology Unit, Athens University Medical implicated as important mediating pathways for cardiovas- School, Hippokration Hospital, Vasilissis Sofias 114, Athens 115 28, Greece. cular diseases, activated though different intakes of food E-mail: [email protected] groups, such as alcohol (Bau et al., 2007), as well as obesity 1These authors contributed equally to this work. Received 1 August 2010; revised 13 December 2010; accepted 9 January (Forsythe et al., 2008). Accordingly, despite the fact that it is 2011; published online 16 February 2011 not easy to specify substances with anti-atherosclerotic Oil consumption and inflammation N Papageorgiou et al 515 properties comprising diets, studies have shown that com- activities without modifications. Briefly, all measurements ponents such as olive oil or n-3 fatty acids may have a key were carried out in the morning in a dark quiet room under role in cardiovascular protection, by especially inhibiting controlled temperature (22–24 1C), and after patients were at inflammatory process (Massaro et al., 2008; Perez-Martinez a rest for at least 10 min. All subjects abstained from food, et al., 2010). However, the affect of acute consumption of tobacco, alcohol and caffeine-containing drinks or anti- different types of oil on proinflammatory cytokines and oxidant supplements such as juices for at least 8 h before oil adhesion molecules remains unknown. ingestion (baseline measurement). In addition, all subjects Thus, the aim of this study was to examine the acute abstained from the aforementioned parameters during the effects of extra virgin olive oil, soy oil, corn oil and cod-liver 3 h (before the second measurement) post oil consumption. oil on circulating levels of soluble VCAM-1 (sVCAM-1), The study protocol was approved by the Institutional Ethics soluble ICAM-1 (sICAM-1) and TNF-a in young, healthy Committee and written informed consent was obtained individuals. from all subjects. In addition, the study was carried out in accordance with the Declaration of Helsinki II and European Union guidelines for good clinical practice. Materials and methods

Study population and protocol Laboratory measurements A total number of 66 healthy young subjects were recruited Venous blood samples were obtained at the beginning of randomly for this parallel study according to their age, each protocol and at 3 h post intervention. Routine chemical gender and body mass index. Exclusion criteria were the methods were used to determine serum concentrations of presence of clinical evidence for atherosclerotic peripheral total cholesterol, triglycerides, high-density lipoprotein and vascular disease or coronary atherosclerosis, the presence of low-density lipoprotein cholesterol. After centrifugation any acute or chronic inflammatory disease (for example, at 1150 g at 4 1C for 10 min, serum was collected and stored liver disease). Subjects with any classic risk factor for at À801 C until assayed. Quantitative sandwich enzyme atherosclerosis such as smoking, as well as those receiving immunoassay technique was used to determine serum any systemic medication, were also excluded from the study. sVCAM-1 (ng/ml; Quantikine human sVCAM-1 ELISA kit, Apart from the questionnaire given asking clinical and R&D Systems, Minneapolis, MN, USA), sICAM-1 (ng/ml; medical variables, anthropometric and blood pressure assess- Quantikine human sICAM-1/CD54 ELISA kit, R&D Systems), ments were carried out. as well as TNF-a levels (pg/ml; Quantikine HS, human TNF-a/ All the participants were randomly allocated into four TNFSF1A ELISA kit, R&D systems). groups to receive four different types of oil to receive an equivalent amount of 50 ml (and almost equal kilocalories; Vogel et al., 2000 and Tousoulis et al., 2010). Each type of oil Statistical analysis had a known macronutrients composition commercially All variables were tested for normal distribution by available. Thus, the composition (provided by the manufac- Kolmogorov–Smirnov test. Normally distributed variables turer) per 100 ml of each type of oil was extra virgin olive oil are presented as mean values with standard errors (824 kcal or 3449.9 kJ, 91.6g fatty acids, 12.8 g saturated, (mean±s.d.), whereas not normally distributed data are 70.5 g monounsaturated, 8.3 g polyunsaturated), corn oil presented as median (25–75th percentiles values). Correla- (825 kcal or 3454.1 kJ, 92 g fatty acids, 12 g saturated, 25 g tions were assessed by Spearman’s correlation coefficient (r) monounsaturated, 55 g polyunsaturated), soy oil (825 kcal or because of not normally distributed data. Between groups 3454.1 kJ, 92 g fatty acids, 14 g saturated, 23 g monounsatu- comparisons were carried out by w2-test for categorical rated, 55 g polyunsaturated) and cod liver oil (830 kcal or variables or by one-way analysis of variance, followed by 3475.0 kJ, 92 g cod oil, 0.00828 g eicosapentaenoic acid, Bonferoni post hoc correction for multiple comparisons of 0.00736 g docosahexaenoic acid, 4 6g monounsaturated fatty normally distributed continuous variables or by Kruskal– acids, 30 g polyunsaturated fatty acids, 0.0112 g vitamin A, Wallis followed by Dunn’s post test for not normally 0.0001 g vitamin D, 0.068 g vitamin E). It is worth to distributed variables. Moreover, the effect of oil consump- mention that extra virgin olive oil was administered, as tion on each variable was examined by performing paired minor components (for example, b-carotene, a-tocopherol, t-test or Wilcoxon signed-rank test as appropriate. Our study b-sitosterol, tyrosol, oleuropein and hydroxytyrosol) of was powered based on the data regarding to sICAM-1: pilot virgin olive oil are thought to be responsible for much of data suggested that the difference in the response of its beneficial effects. Oils were administered directly from matched pairs is normally distributed with standard devia- their commercial form (liquid) and not as part of a meal. tion 13.0. If the true difference in the mean response of Measurements were taken at baseline (0 h) and at 3 h post oil matched pairs is 66.77, we would need to study four pairs of consumption. None of the participants followed any specific subjects to be able to reject the null hypothesis that this recommendation regarding diet or physical activity before response difference is zero with probability (power) 0.99. the study. All subjects participated normally to all their daily The type I error probability associated with this test of this

European Journal of Clinical Nutrition Oil consumption and inflammation N Papageorgiou et al 516 null hypothesis was 0.01. All analyses were carried out by using cantly. Controversially to the neutral/slightly negative effect the SPSS v18.0 (SPSS, Chicago, IL, USA) statistical package. of corn oil on sVCAM-1, the latter had also a significant effect on sICAM-1 levels (Po0.01; Figure 2).

Results Effect of oil consumption on TNF-a The demographic characteristics and lipid profile of the Extra virgin olive (Po0.05), soy oil (Po0.01) and cod liver oil participants are presented in Table 1. There were no intakes (Po0.01) reduced significantly TNF-a levels, whereas significant differences between the four groups at baseline. corn oil induced a nonsignificant decrease in TNF-a levels (P ¼ NS; Figure 3). Interestingly, the absolute change in TNF-a levels Effect of oil consumption on sVCAM-1 and sICAM-1 (39 pairs) was significantly correlated with the absolute Acute consumption of olive oil, soy oil, cod liver oil as well as change in ICAM-1 levels (66 pairs) after oil consumption corn oil had no significant effect on sVCAM-1 levels (r ¼ 0.379, Po0.05; Figure 4a). However, there was no (P ¼ nonsignificant (NS) for all; Figure 1). significant correlation between absolute changes in TNF-a Interestingly, cod liver oil (Po0.01), olive oil (Po0.01), as levels and VCAM-1 levels (59 pairs; r ¼ 0.019, P ¼ NS; well as soy oil (Po0.01) decreased sICAM-1 levels signifi- Figure 4b).

Table 1 Demographic characteristics of the participants

Olive oil Soy oil Corn oil Cod liver oil P-value

N (males/females) 19 (12/7) 17 (10/7) 15 (7/8) 15 (9/6) 0.796 Age (years) 25.8±3.09 26.2±6.39 24.7±2.35 29.0±8.62 0.196 BMI (kg/m2) 23.0±2.83 23.2±4.18 23.1±3.04 22.6±2.74 0.960 Systolic blood pressure 110±11.7 114±11.9 113±10.8 111±9.80 0.714 Diastolic blood pressure 72±6.80 76±5.00 76±4.94 73±6.87 0.129 Cholesterol (mg/dl) 172±33.1 182±25.9 168±29.4 178±35.3 0.597 Triglycerides (mg/dl)a 79 (48–127) 84 (31–191) 66 (42–138) 76 (45–134) 0.773 HDL (mg/dl) 50±7.76 51±9.02 55±7.82 54±14.7 0.437 LDL (mg/dl) 109±27.3 108±20.9 107±27.5 102±28.5 0.884

Abbreviations: BMI, body mass index; HDL, high-density lipoprotein cholesterol; LDL, low-density lipoprotein cholesterol. aValues expressed as median (25–75th percentile). All values are expressed as mean±s.d., no significant differences were observed for all variables among groups.

Figure 1 All types of oil had no significant effect on sVCAM-1 levels (ng/ml; P ¼ NS for all). Values are expressed as mean±s.d.

European Journal of Clinical Nutrition Oil consumption and inflammation N Papageorgiou et al 517

Figure 2 Cod liver oil, olive oil, soy oil and corn oil decreased sICAM-1 levels (ng/ml) significantly (Po0.01 for all). Values are expressed as mean±s.d.

Figure 3 Extra virgin olive (Po0.05), soy oil and cod liver oil consumption (Po0.01 for both) reduced significantly TNF-a levels (pg/ml). No effect was observed for corn oil (P ¼ NS). Values are expressed as mean±s.d.

Discussion healthy individuals. We found that all types of oil induced a significant reduction in ICAM-1 levels. In addition, extra In this study, we assessed the acute effect of extra virgin olive virgin olive oil, soy oil and cod liver oil reduced also TNF-a oil, soy oil, corn oil and cod liver oil consumption on levels. Moreover, we observed a significant correlation adhesion molecules and proinflammatory cytokines in between the absolute change of TNF-a and ICAM-1 levels.

European Journal of Clinical Nutrition Oil consumption and inflammation N Papageorgiou et al 518 250 rho=0.379, p<0.05 600 rho=0.147, p=NS

500 200

400 150 300 100 200 dICAM-1 (ng/ml) dVCAM-1 (ng/ml) 50 100

0 0

0.0 0.1 0.2 0.3 0.4 0.0 0.1 0.2 0.3 0.4 dTNF-a (pg/ml) dTNF-a (pg/ml) Figure 4 (a) Acute oil consumption resulted in a significant correlation between the absolute changes(d) in TNF-a levels (pg/ml) and ICAM-1 levels (ng/ml; b; r ¼ 0.379, Po0.05). (b) There was no significant correlation between the absolute changes(d) in TNF-a levels and VCAM-1 levels (r ¼ 0.147, P ¼ NS).

These findings indicate that specific types of oil exert ICAM-1 levels, but had had no effect on VCAM-1 levels significant anti-inflammatory effects by affecting proinflam- (Paulo et al., 2008). matory cytokines and specific adhesion molecules. In this study, we have shown that VCAM-1 levels are unaffected after ingestion of all types of oil, a finding which is in line with our previous observation in a smaller cohort Effect of oil acute consumption on adhesion molecules (Tousoulis et al., 2010). Importantly, we have demonstrated a It is a fact that adhesion molecules participate in the significant nonspecific acute effect of olive, soy, corn and development of atherosclerosis. In addition, it has been cod liver oil on circulating levels of sICAM-1. It is likely that shown that dietary habits or modifications may prevent a common ingredient in all types of oil may modify the atherogenesis by affecting inflammatory molecules such as cleavage rate of sICAM-1 on cell membranes leading to rapid VCAM-1 and ICAM-1. changes of its circulating levels. Mena et al. (2009) while assessing the relationship between Mediterranean diet (supplemented with virgin olive oil) and adhesion molecules, showed a significant inverse correlation Effect of oil acute consumption on TNF-a of this dietary pattern with ICAM-1 and VCAM serum levels. TNF-a has a crucial role in the inflammatory process Interestingly, in the PREDIMED study (Salas-Salvado´ et al., implicated in atherogenesis. Thus, dietary modification 2008) among middle-aged individuals at high cardiovascular targeting TNF-a could be beneficial in preventing athero- risk, the highest consumption of nuts and virgin olive oil was sclerosis, as suggested by studies focusing on olive oil, soy oil associated with the lowest concentrations of VCAM-1, and polyunsaturated lipids. ICAM-1, although only for ICAM-1 was this difference Consumption of olive oil as a single component reduced statistically significant in the case of nuts and for VCAM-1 significantly not only VCAM-1, but also TNF-a serum levels in the case of virgin olive oil. The results were not in in coronary venous blood, as reported by Serrano-Martinez the same direction when analyzing n-3 fatty acids; in et al. (2005). In addition, not only extra virgin olive oil, but a prospective, randomized, double-blind study enrolling also refined olive oil has been found to downregulate TNF-a patients with myocardial infarction to receive highly con- (Marquez-Martin et al., 2006). Moreover, dietary interven- centrated n-3 polyunsaturated fatty acids or corn oil, tion with n-3 polyunsaturated fatty acid and plant sterols inflammatory markers, such as E-selectin and ICAM-1 were reduced systemic inflammation in hyperlipidemic indivi- unaffected at 1 year later (Grundt et al., 2004). Importantly, duals, reflected in a decrease of TNF-a levels by 10% (Micallef Fuentes et al. (2008) showed that different types of fat and Garg, 2009). Interesting results have been reported by composition meals affected differentially (in 4 h) VCAM-1 studies of soy consumption. Thus, TNF-a levels were levels in healthy men, as monounsaturated fatty acids significantly lower in the soy nut group compared with the seemed to be more beneficial than a-linoleic by reducing soy protein and control groups as this was demonstrated by VCAM-1 levels. Data seem more conflicting, when referring Azadbakht et al. (2008) implying a role in the soya lipid to healthy adults. Studies such as this by Eschen et al. fraction. In contrast to these findings, Jenkins et al. (2002) (2004) have shown a dose dependent effect as it was showed that there was no significant effect on interleukin-6 observed a significant decrease in ICAM-1 in the polyun- or TNF-a in middle-aged men and women on the high- saturated fatty acid 2.0 g group and a significant increase in isoflavone versus the low-isoflavone soy food diet. VCAM-1 in the polyunsaturated fatty acid 6.6 g group. In this study, we have shown that acute consumption of However, other studies conducted in young healthy subjects, extra virgin olive oil, soy oil and cod liver oil reduced showed that diets rich in lean fish significantly decreased significantly TNF-a levels, whereas corn oil induced a

European Journal of Clinical Nutrition Oil consumption and inflammation N Papageorgiou et al 519 nonsignificant decrease in TNF-a levels. It is likely that the response study using n-3 fatty acids. Nutr Metab Cardiovasc Dis different types of oils may exert different intracellular 14, 180–185. ‘antioxidant’ effects, which are then reflected to a variable Forsythe LK, Wallace JM, Livingstone MB (2008). Obesity and inflammation: the effects of weight loss. Nutr Res Rev 21, 117–133. activation of intracellular redox sensitive pathways (such as Fuentes F, Lo´pez-Miranda J, Pe´rez-Martıńez P, Jimeńez Y, MarıńC, nuclear factor-kB and activator protein-1). This different Go´mez P et al. (2008). Chronic effects of a high-fat diet enriched effect on intracellular redox signaling could perfectly explain with virgin olive oil and a low-fat diet enriched with alpha- linolenic acid on postprandial endothelial function in healthy the differences in the circulating levels of soluble adhesion men. Br J Nutr 100, 159–165. molecules, which represent a marker of the expression of Grundt H, Nilsen DW, Hetland Ø, Mansoor MA (2004). Clinical these molecules on the cellular surface, a process regulated outcome and atherothrombogenic risk profile after prolonged by these redox-sensitive transcriptional pathways. We have wash-out following long-term treatment with high doses of n-3 PUFAs in patients with an acute myocardial infarction. Clin Nutr also found a significant association between the absolute 23, 491–500. changes (d) in TNF-a levels and ICAM-1 levels. These Jenkins DJ, Kendall CW, Jackson CJ, Connelly PW, Parker T, Faulkner D findings indicate that the expression of the proinflammatory et al. (2002). Effects of high- and low-isoflavone soyfoods on blood cytokine TNF-a, which was modified by oil consumption, lipids, oxidized LDL, homocysteine, and blood pressure in hyperlipidemic men and women. Am J Clin Nutr 76, 365–372. may regulate the subsequent expression of ICAM-1. There- Klein CL, Bittinger F, Ko¨hler H, Wagner M, Otto M, Hermanns I et al. fore, a potential pathway linking these molecules may be (1995). Comparative studies on vascular endothelium in vitro.3. affected by the anti-inflammatory effects of specific oils. Effects of cytokines on the expression of E-selectin, ICAM-1 and Although our findings point out a significant role for oil VCAM-1 by cultured human endothelial cells obtained from different passages. Pathobiology 63, 83–92. acute consumption in inflammatory markers, it is difficult to Libby P, Ridker PM, Hansson GK (2009). Leducq transatlantic network extrapolate our results to clinical practice, as our study was on atherothrombosis. Inflammation in atherosclerosis: from mainly a pathophysiological study designed to assess the pathophysiology to practice. J Am Coll Cardiol 54, 2129–2138. potential pathways mediating the effect of different oil acute Marquez-Martin A, De La Puerta R, Fernandez-Arche A, Ruiz- Gutierrez V, Yaqoob P (2006). Modulation of cytokine secretion consumption on inflammatory process. by pentacyclic triterpenes from olive pomace oil in human In conclusion, acute consumption of all types of oil mononuclear cells. Cytokine 36, 211–217. decreased significantly ICAM-1 levels. In addition, olive oil, Massaro M, Scoditti E, Carluccio MA, De Caterina R (2008). Basic soy oil and cod liver oil decreased significantly TNF-a levels. mechanisms behind the effects of n-3 fatty acids on cardiovascular disease. Prostaglandins Leukot Essent Fatty Acids 79, 109–115. Moreover, the degree of change in TNF-a levels was Mena MP, Sacanella E, Vazquez-Agell M, Morales M, Fito´ M, Escoda R correlated with the degree of change in ICAM-1 levels, et al. (2009). Inhibition of circulating immune cell activation: a indicating that the expression of the proinflammatory molecular antiinflammatory effect of the Mediterranean diet. Am J cytokine TNF-a after oil consumption, may regulate the Clin Nutr 89, 248–256. Meydani M (2003). Soluble adhesion molecules: surrogate markers of subsequent expression of ICAM-1 level. These findings cardiovascular disease? Nutr Rev 61, 63–68. suggest that soy oil, cod liver oil and virgin olive oil exert Micallef MA, Garg ML (2009). Anti-inflammatory and cardioprotec- specific significant anti-inflammatory effects. tive effects of n-3 polyunsaturated fatty acids and plant sterols in hyperlipidemic individuals. Atherosclerosis 204, 476–482. Paulo MC, Andrade AM, Andrade ML, Morais MG, Kiely M, Parra D Conflict of interest et al. (2008). Influence of n-3 polyunsaturated fatty acids on soluble cellular adhesion molecules as biomarkers of cardiovascular risk in young healthy subjects. Nutr Metab Cardiovasc Dis 18, The authors declare no conflict of interest. 664–670. Perez-Martinez P, Moreno-Conde M, Cruz-Teno C, Ruano J, Fuentes F, Delgado-Lista J et al. (2010). Dietary fat differentially influences References regulatory endothelial function during the postprandial state in patients with metabolic syndrome: From the LIPGENE study. Azadbakht L, Atabak S, Esmaillzadeh A (2008). Soy protein intake, Atherosclerosis 209, 533–538. cardiorenal indices, and C-reactive protein in type 2 diabetes with Salas-Salvado´ J, Garcia-Arellano A, Estruch R, Marquez-Sandoval F, nephropathy: a longitudinal randomized clinical trial. Diab Care Corella D, Fiol M et al. (2008). Components of the mediterranean- 31, 648–654. type food pattern and serum inflammatory markers among Bau PF, Bau CH, Rosito GA, Manfroi WC, Fuchs FD (2007). Alcohol patients at high risk for cardiovascular disease. Eur J Clin Nutr consumption, cardiovascular health, and endothelial function 62, 651–659. markers. Alcohol 41, 479–488. Serrano-Martinez M, Palacios M, Martinez-Losa E, Lezaun R, Biasucci LM, CDC; AHA (2004). CDC/AHA workshop on markers of Maravi C, Prado M et al. (2005). A Mediterranean dietary style inflammation and cardiovascular disease: application to clinical influences TNF-alpha and VCAM-1 coronary blood levels in and public health practice: clinical use of inflammatory markers unstable angina patients. Eur J Nutr 44, 348–354. in patients with cardiovascular diseases: a background paper. Tousoulis D, Papageorgiou N, Antoniades C, Giolis A, Bouras G, Circulation 110, e560–e567. Gounari P et al. (2010). Acute effects of different types of oil Blann AD, Lip GY (2000). Cell adhesion molecules in cardiovascular consumption on endothelial function, oxidative stress status and disease and its risk factors—what can soluble levels tell us? J Clin vascular inflammation in healthy volunteers. Br J Nutr 103, 43–49. Endocrinol Metab 85, 1745–1747. Vogel RA, Corretti MC, Plotnick GD (2000). The postprandial effect Eschen O, Christensen JH, De Caterina R, Schmidt EB (2004). of components of the mediterranean diet on endothelial function. Soluble adhesion molecules in healthy subjects: a dose- J Am Coll Cardiol 36, 1455–1460.

European Journal of Clinical Nutrition