INTRODUCTION. Am J Clin Nutr 2014;99: Printed in USA. Ó 2014 American Society for Nutrition 479

Flavonoid-rich fruit and vegetables improve microvascular reactivity and inflammatory status in men at risk of cardiovascular disease FLAVURS: a randomized controlled trial 1 5 Anna L Macready, Trevor W George, Mary F Chong, Dauren S Alimbetov, Yannan Jin, Alberto Vidal, Jeremy PE Spencer, Orla B Kennedy, Kieran M Tuohy, Anne-Marie Minihane, Michael H Gordon, and Julie A Lovegrove for the FLAVURS Study Group ABSTRACT Background: Observed associations between increased fruit and vegetable (F&V) consumption, particularly those F&Vs that are rich in flavonoids, and vascular health improvements require confirmation in adequately powered randomized controlled trials. Objective: This study was designed to measure the dose-response relation between high-flavonoid (HF), low-flavonoid (LF), and habitual F&V intakes and vascular function and other cardiovascular disease (CVD) risk indicators. Design: A single-blind, dose-dependent, parallel randomized controlled dietary intervention study was conducted. Male and female low-f&v consumers who had a $1.5-fold increased risk of CVD (n = 174) were randomly assigned to receive an HF F&V, an LF F&V, or a habitual diet, with HF and LF F&V amounts sequentially increasing by 2, 4, and 6 (12, 14, and 16) portions/d every 6 wk over habitual intakes. Microvascular reactivity (laser Doppler imaging with iontophoresis), arterial stiffness [pulse wave velocity, pulse wave analysis (PWA)], 24-h ambulatory blood pressure, and biomarkers of nitric oxide (NO), vascular function, and inflammation were determined at baseline and at 6, 12, and 18 wk. Results: In men, the HF F&V diet increased endothelium-dependent microvascular reactivity (P = 0.017) with 12 portions/d (at 6 wk) and reduced C-reactive protein (P = 0.001), E-selectin (P = 0.0005), and vascular cell adhesion molecule (P = 0.0468) with 14 portions/d (at 12 wk). HF F&Vs increased plasma NO (P = 0.0243) with 14 portions/d (at 12 wk) in the group as a whole. An increase in F&Vs, regardless of flavonoid content in the groups as a whole, mitigated increases in vascular stiffness measured by PWA (P = 0.0065) and reductions in NO (P = 0.0299) in the control group. Conclusion: These data support recommendations to increase F&V intake to $6 portions daily, with additional benefit from F&Vs that are rich in flavonoids, particularly in men with an increased risk of CVD. This trial was registered at www.controlled-trials.com as ISRCTN47748735. Am J Clin Nutr 2014;99:479 89. INTRODUCTION Epidemiologic studies consistently report that consuming fruit and vegetables (F&Vs) 6 lowers cardiovascular disease (CVD) risk (1, 2). Although increased intakes of F&Vs are accepted as being cardioprotective (3, 4), clear dose-response relations are poorly defined, which has resulted in major inconsistencies in recommended F&V intakes between individual countries (5). F&Vs contain a wide range of potentially cardioprotective components such as fiber, folate, nitrate, vitamins, and a large number of nonnutrient phytochemicals. These include flavonoids, such as anthocyanins, flavonols, and flavanones, that are found in high concentrations in commonly consumed F&Vs such as berries, citrus fruit, apples, grapes, peppers, onions, broccoli, and herbs (6). Benefits with respect to lipid, hemostatic, and inflammatory biomarkers and vascular-related biomarkers such as blood pressure (BP) and vascular reactivity have been observed with increased flavonoid intakes from phytochemical isolates (7, 8) and from individual flavonoid-rich and nitrate-rich F&Vs (9, 10). However, there is a paucity of information regarding the relative biopotency of flavonoid bioactives, dose responsiveness with CVD risk biomarkers, and the 1 From the Hugh Sinclair Unit of Human Nutrition (ALM, TWG, MFC, DSA, YJ, AV, JPES, OBK, KMT, A-MM, MHG, and JAL), and the Institute for Cardiovascular and Metabolic Research, Department of Food and Nutritional Sciences (ALM and JAL), University of Reading, Reading, Berkshire, United Kingdom. 2 ALM and TWG are joint first authors. 3 Present address: for TWG, Northumbria University, Faculty of Health and Life Sciences, Newcastle upon Tyne, United Kingdom; for KMT, Research and Innovation Centre, Fondazione Edmund Mach, San Michele all Adige, Trento, Italy; for A-MM, Norwich Medical School, University of East Anglia, Norwich, United Kingdom; for MFC, Clinical Nutrition Research Centre, Singapore Institute for Clinical Sciences, Singapore. 4 Supported by the Food Standards Agency, United Kingdom (project no. N02039/F5234012). Sainsbury s PLC (supermarket retailer) provided the study foods. 5 Address correspondence and requests for reprints to JA Lovegrove, Hugh Sinclair Unit of Human Nutrition, Department of Food and Nutritional Sciences, University of Reading, Reading, Berkshire RG6 6AP, United Kingdom. E-mail: j.a.lovegrove@reading.ac.uk. 6 Abbreviations used: ABP, ambulatory blood pressure; AIx, augmentation index; AUC, area under the flux versus time curve; BP, blood pressure; CRP, C-reactive protein; CVD, cardiovascular disease; DBP, diastolic blood pressure; DVP, digital volume pulse; enos, endothelial nitric oxide synthase; FLAVURS, FLAvonoids and Vascular function at the University of Reading Study; F&V, fruit and vegetable; HF, high-flavonoid; LDI, laser Doppler imaging with iontophoresis; LF, low-flavonoid; NO, nitric oxide; PWA, pulse wave analysis; PWV, pulse wave velocity; SBP, systolic blood pressure; SNP, sodium nitroprusside; VCAM, vascular cell adhesion molecule. Received August 22, 2013. Accepted for publication December 20, 2013. First published online January 22, 2014; doi: 10.3945/ajcn.113.074237. Am J Clin Nutr 2014;99:479 89. Printed in USA. Ó 2014 American Society for Nutrition 479

480 MACREADY ET AL impact of flavonoids when consumed in whole foods rather than in isolated compounds or extracts. Therefore, on the basis of available data, more specific guidelines regarding the types of F&Vs, which should be recommended and consumed, cannot currently be made with any degree of confidence. Endothelial dysfunction is recognized as a critical, early, modifiable event in the development of atherosclerosis (11). Numerous studies have highlighted the prognostic value of in vivo measures of vascular reactivity (of both the coronary and peripheral arteries) in predicting future coronary events (12). Moreover, the impact of many of the well-recognized CVD risk markers, such as hypercholesterolemia and oxidative stress, on atherogenesis are believed to be, in part, mediated by their effects on vascular function (13). In 2 randomized controlled trials, F&V consumption improved small artery elasticity (14) and dosedependent microvascular function in hypertensive individuals (15), although these beneficial effects could not be attributed to any one component of F&Vs. Over the past decade, there has been accumulating and relatively consistent evidence to suggest that dietary flavonoids may explain, in large part, the CVD and specifically vascular benefits of increased F&V intake. Epidemiologic studies have reported a greater CVD risk reduction in individuals with higher flavonoid intakes compared with those with lower intakes (16). For example, anthocyanin intake has been associated with less arterial stiffness and central BP in women from the Twins UK registry (17). In support of these data, intervention studies have reported that consumption of individual flavonoid-rich foods or extracts was associated with a reduction in vascular dysfunction, inflammation, and oxidative stress (18 22). Despite this, and with some exceptions (15, 23, 24), well-powered studies using commonly consumed flavonoid-rich F&Vs are very limited in number and generally low in power with relatively short intervention periods (9, 25 27). Furthermore, although a number of potential physiologic mechanisms have been studied in isolation, the relative impact of flavonoids on key established and emerging cardiovascular risk indicators is unclear (28). The FLAvonoids and Vascular function at the University of Reading Study (FLAVURS), a randomized, controlled, suitably powered, dose-dependent dietary intervention, was performed in low-f&v consumers at risk of CVD. The efficacy of the dietary strategy used has been reported previously (28). The essential research question addressed the optimal quantity and types of F&Vs required to promote vascular health and modify other established and putative cardiovascular risk markers. We hypothesized that flavonoid-rich F&Vs have a greater protective effect than flavonoid-poor F&Vs on micro- and macrovascular function. SUBJECTS AND METHODS Subjects Two cohorts of men and women aged 26 70 y were recruited to the FLAVURS trial from the population in and around Reading, United Kingdom. Recruitment was carried out between December 2007 March 2008 and November 2008 January 2009, with baseline start dates for participants in each cohort occurring between January and August of each year. Key exclusion criteria included the following: previous stroke or myocardial infarction; clinically diagnosed diabetes, liver, or renal disease; drug treatment for lipid reduction or hypertension; antiinflammatory medication; and pregnancy or lactation. In addition, a minimum of 2 structured 24-h dietary recalls were randomly conducted to determine habitual daily intake of F&Vs, and individuals consuming less than the UK population average of 4.4 portions F&Vs/d (one portion is equivalent to 80 g fresh weight) (29) were eligible to join the study. Of the eligible participants, only those with an RR of CVD.1.5, established by using a methodology adapted from the Framingham CVD risk scoring tool, were recruited and randomly assigned to 1 of 3 dietary groups by the project research manager using a minimization procedure matching for age, sex ratio, BMI, red wine and tea consumption, and smoking (30). To be eligible, the participant was required to score a minimum of 2 points. Risk markers were scored as follows: BP [either systolic BP (SBP) of 130 139 mm Hg (1 point) or $140 mm Hg (2 points) or diastolic BP (DBP) $90 mm Hg (2 points)]; plasma total cholesterol [5.2 6.1 mmol/l (1 point) or 6.2 7.2 mmol/l (2 points)]; HDL cholesterol [in men: 1.0 1.1 (1 point) or,0.9 mmol/l (2 points); in women: 1.2 1.3 mmol/l (1 point) or,0.9 1.1 mmol/l (2 points)]; obesity/adiposity [either waist circumference in women of 80 87 cm (1 point) or.88 cm (2 points) and waist circumference in men of 94 101 cm (1 point) or.102 cm (2 points) or a BMI (in kg/m 2 ) of 25 30 (1 point) or.30 (2 points); BMI for Asian populations: 23 27.5 (1 point) or.27.5 (2 points)]; and smoking status [.10 cigarettes/d (2 points)] (28). Before baseline measures, all 3 treatment groups were blinded to the interventions after assignment; during the study, only the dietary intervention groups continued to be blinded. Researchers performing sample and vascular analyses were blinded to group assignment. In total, 307 individuals were screened; after identification of CVD risk, 221 were recruited into the study, 174 were randomly assigned to 1 of the 3 intervention groups, and 154 completed all study visits. The participant flowchart shown in Figure 1 was also reported in full by Chong et al (28). Study design The study was a randomized, controlled, dose-dependent parallel design with 2 sequentially increasing dose-response dietary treatments containing high-flavonoid (HF) and low-flavonoid (LF) F&Vs and a control group in which participants maintained their habitual diet throughout the study. FLAVURS was registered as a randomized clinical trial (ISRCTN47748735) and conducted according to the guidelines laid down in the Declaration of Helsinki. Ethical approval for the study was obtained from the Local Research Ethics Committee of the Isle of Wight, Portsmouth and South East Hampshire (REC: 07/H0501/81), and the University of Reading s Research (REC: 07/22) Ethics Committee. Informed consent was obtained before participation. After screening and random assignment, participants were invited to the university for a familiarization visit to acquaint them with the clinical setting and measurements that would be performed during their subsequent study visits, followed by a 2-wk run-in during which they were asked to maintain their current diet and lifestyle. At the end of the study run-in, participants attended the baseline (week 0) visit. The HF and LF participants target intake of F&Vs was increased over and above habitual intake by 2, 4, and 6 (12, 14, and 16) 80-g

F&V FLAVONOIDS IMPROVE VASCULAR FUNCTION IN MEN 481 FIGURE 1. Flow of participants through the FLAVURS trial. Reproduced with permission from reference 28. CT, control (habitual diet); f, females; FLAVURS, FLAvonoids and Vascular function at the University of Reading Study; HF, high-flavonoid; LF, low-flavonoid; m, males; #, no data. portions/d over 3 consecutive 6-wk periods. Participants attended 4 clinical visits at the Hugh Sinclair Unit of Human Nutrition (weeks 0, 6, 12, and 18), where measures of vascular function, fasting blood samples, and 24-h urine samples were collected. Twentyfour-hour ambulatory blood pressure and anthropometric measures were determined at each time point. Throughout the study, participants were asked to maintain constant weight, minimize other health and lifestyle changes, and to immediately report changes in medication status or other influencers of CVD risk profile. Dietary intervention At the baseline visit (week 0), participants were provided with information on the diet to which they had been assigned, and close contact with the study dietitian/nutritionist was maintained during the 18 wk to provide dietary advice and support. Portions of F&Vs were defined as 80 g for fresh, frozen, or canned items or 40 g for dried items and $150 ml fresh juice. With the use of the USDA flavonoids database (31), HF and LF foods were defined as.15 mg/100 g and as,5 mg/100 g of total flavonoids, respectively, with adjustments made to account for fresh, dry, or canned F&V weight. F&Vs were provided to participants, and considerable effort was devoted to matching the intake of other potentially bioactive components of the F&Vs such as carotenoids, vitamin C, folate, and nonstarch polysaccharides (28). A variety of food sources were provided, including fresh, canned, frozen, and dried F&Vs and juices and composite foods such as fruit smoothies and pasta sauces. To aid compliance, all F&Vs were provided to those in the LF and HF groups and delivered to participants homes on a weekly basis. Details of the dietary intervention have been reported previously (28). At least 2 structured 24-h dietary recalls were conducted randomly during each 6-wk intervention phase, a photographic atlas was used for portion size determination (32), and nutrient intake was determined by Dietplan6 (Forestfield Software Ltd), which was supplemented with flavonoid amounts in various foods listed in the USDA flavonoid database (28, 31). In addition to the two 24-h recall and daily record sheets, compliance was assessed at each visit by biomarkers of F&V intake, which included plasma vitamin C, folate, and carotenoids and urinary flavonoids and potassium. Details of their determination were described previously (28). Assessment of vascular health The primary outcome for the study was vascular function. All measurements were completed with participants in a supine position in a quiet room at 22 6 18C. Microvascular reactivity was the primary vascular outcome, measured by laser Doppler imaging with iontophoresis (LDI) (33). Solutions of 1% acetylcholine chloride for measurements of endothelium-dependent reactivity as the experimental measure and of 1% sodium nitroprusside (SNP) for measurements of endothelium-independent reactivity as a control were delivered transdermally on the forearm via anodal and cathodal chambers, respectively. Repeated scans gave a total charge of 8 millicoulombs. Microvascular response was determined by using the area under the flux versus time curve (AUC). Measurements were conducted by using a moorldi2 laser Doppler imager (Moor Instruments Ltd), as previously described (34). Carotid-femoral pulse wave velocity (PWV; m/s) and radial pulse wave analysis (PWA) augmentation index (AIx; %) determined the stiffness of larger conduit and smaller peripheral vessels, respectively, with correction to a heart rate of 75 beats/min (PWA AIx HR75, %) (35, 36). Ambulatory blood pressure (ABP) was determined by using ABP monitors (A&D Instruments Ltd) during the week preceding the study day (37). Ambulatory SBP and DBP were recorded at 30-min intervals from 0700 to 2200 and at 60-min intervals from 2200 to 0700. Subjects were supplied with a diary card to note activities performed and sleep throughout the recording period. Digital volume pulse (DVP) photoplethysmography (Pulse Trace; Micro Medial) was used to calculate the stiffness index (m/s) and the reflection index (%) and to derive the heart rate (beats/min).

482 MACREADY ET AL Plasma collection and analysis Blood samples were collected by using an evacuated-tube system and transferred to citrate, EDTA, and lithium heparin evacuated tubes; temporarily stored on ice; and centrifuged at 48C at 1700 g within 30 min. Plasma was then separated into aliquots in cryogenic vials for storage at 2808C. All samples from each subject were analyzed within one batch to reduce interbatch variation after study completion. An Instrument Laboratory ILAB 600 autoanalyzer using standard kits with appropriate sero-normal and low- and highquality control standards (Instrumentation Laboratory Ltd) included in all batches was used to analyze C-reactive protein (CRP) (Quantex CRP ultrasensitive kits supplied by Instrumentation Laboratory Ltd). ELISA kits were used to determine intercellular soluble adhesion molecule, vascular cell adhesion molecule (VCAM), E-selectin (R&D Systems Europe Ltd), von Willebrand factor antigen (Corgenix UK Ltd), plasminogen activator inhibitor-1, plasma TNF-a, IL-6, total nitrate/ nitrite (Active Motif; Rixensart) as a biomarker of nitric oxide (NO) production, and fibrinogen. Study power Because the primary aim of the study was to investigate microvascular reactivity measured by LDI, it was calculated that a total of 153 participants (n = 51/group) were required. This would enable detection of a 15% difference in LDI vascular reactivity, with an estimated SD of 225 flux units, with P, 0.05 and 80% power. A total of 180 participants were required to be recruited to allow for a 15% dropout rate. Statistical analysis Data were analyzed by using the MIXED procedure for linear mixed models (LMM, PROC MIXED) (38) in SAS, version 9.1.3 (SAS Institute). Within-subject variability was explained by using random effects throughout the study. The variance-covariance matrix used a compound symmetry variance-covariance matrix, and the Kenward-Rogers methodology (39) was applied to estimate the df in the mixed models. After adjustment for baseline and cohort effects, the fixed-effects model was then adjusted for treatment group and the remaining covariates including visit, age, sex, baseline BMI, tea/red wine consumption, and smoking. Normality was assessed by using the Kolmogorov-Smirnov test (40), and nonnormal data were subjected to natural log transformation. To evaluate and validate each model, predicted values were plotted against standardized residuals. A sensitivity analysis of outliers was performed, and those that were.3 standardized residuals from the mean were excluded from the model. Effects found to be significant were subjected to further analysis, with orthogonal contrasts being carried out where possible, and estimates of contrasts were calculated to assess the differences. Where variables presented a significant number of zero values recorded below the detection limit of the assays, the probability of detection was calculated by using a generalized linear mixed model (GLMM, PROC GLIMMIX). A logit link was then used to assess the detectable values, with adjustments made for potential overdispersion. A cutoff significance value of P = 0.05 was used throughout (28). RESULTS Compliance to FLAVURS dietary intervention Recruitment occurred in 2 cohorts (December 2007 March 2008 and November 2008 January 2009); the participant flow is shown in Figure 1 and has been fully reported elsewhere (28). The baseline characteristics of the participants are shown in Table 1. There were no significant differences between the 3 intervention groups at baseline. The dietary macro- and TABLE 1 Group baseline characteristics: FLAVURS 1 HF group LF group Control group All Women Men All Women Men All Women Men Subjects (n) 58 22 36 59 25 34 57 21 36 Age (y) 50 6 1 2 52 6 1 47 6 1 51 6 1 52 6 2 49 6 1 52 6 1 53 6 1 50 6 2 Nonsmokers (n) 50 32 18 51 30 21 51 34 17 Anthropometric measures BMI (kg/m 2 ) 27.6 6 0.3 28.2 6 0.6 26.8 6 0.6 28.0 6 0.3 28.4 6 0.5 27.4 6 0.5 27.3 6 0.4 27.0 6 0.6 27.9 6 0.6 Waist circumference (cm) 93.4 6 0.8 92.4 6 1.3 95.2 6 1.4 93.9 6 0.7 92.9 6 1.3 95.4 6 1.1 92.3 6 1.0 88.8 6 1.4 98.3 6 1.9 BP (mm Hg) Mean 24-h SBP 126 6 2 122 6 2 131 6 1 128 6 2 129 6 2 126 6 1 125 6 2 122 6 2 130 6 2 Mean 24-h DBP 77 6 1 73 6 1 81 6 1 77 6 1 77 6 1 78 6 1 76 6 1 75 6 1 79 6 1 Glucose (mmol/l) 5.6 6 0.0 5.5 6 0.0 5.8 6 0.1 5.7 6 0.0 5.6 6 0.1 6.2 6 0.2 5.5 6 0.0 5.4 6 0.1 5.6 6 0.1 Lipids (mmol/l) Total cholesterol 5.7 6 0.2 5.8 6 0.2 5.8 6 0.2 5.6 6 0.1 5.7 6 0.1 5.8 6 0.1 5.2 6 0.2 5.2 6 0.2 5.1 6 0.1 Triacylglycerol 1.3 6 0.1 1.0 6 0.0 1.6 6 0.1 1.4 6 0.0 1.2 6 0.1 1.8 6 0.1 1.3 6 0.0 1.2 6 0.1 1.6 6 0.1 HDL cholesterol 1.5 6 0.0 1.6 6 0.1 1.4 6 0.0 1.6 6 0.0 1.7 6 0.0 1.4 6 0.0 1.5 6 0.0 1.7 6 0.1 1.2 6 0.1 LDL cholesterol 3.9 6 0.1 3.7 6 0.1 3.9 6 0.1 3.7 6 0.1 3.8 6 0.1 3.9 6 0.1 3.4 6 0.1 3.2 6 0.2 3.2 6 0.1 1 Values are based on baseline measures for 3 treatment groups: HF F&Vs, LF F&Vs, and habitual diet (control). n = 174. The MIXED procedure for linear mixed models (LMM, PROC MIXED; SAS Institute) was used to test for baseline differences in sex, age, lifestyle, anthropometric variables, BP, glucose, and lipids between treatment groups. None of the characteristics were significantly different between the 3 groups. BP, blood pressure; DBP, diastolic blood pressure; F&V, fruit and vegetable; FLAVURS, FLAvonoids and Vascular function at the University of Reading Study; HF, high-flavonoid; LF, lowflavonoid; SBP, systolic blood pressure. 2 Mean 6 SEM (all such values).

F&V FLAVONOIDS IMPROVE VASCULAR FUNCTION IN MEN 483 micronutrient intakes and urinary and plasma F&V biomarkers throughout the study are reported in Supplemental Table 1 (under Supplemental data in the online issue) and described in full elsewhere (28). In brief, compliance to the target numbers and types of F&Vs was broadly met and verified by dietary records and plasma and urinary biomarkers. Mean portion numbers of F&Vs consumed by the 3 groups were similar at baseline, and the HF and LF groups increased consumption in a dose-dependent manner throughout the intervention (P = 0.015). However, it is of note that the target numbers of additional F&V portions were not met precisely. For example, in the HF group, 2.4, 3.1, and 3.7 additional HF F&V portions were consumed after 6, 12, and 18 wk, respectively, whereas 2.2, 3.4, and 4.7 additional LF F&V portions were consumed after 6, 12, and 18 wk, respectively; the possible reasons for this were discussed previously (28). There was a dose-dependent increase in dietary and urinary flavonoids in the HF group, with no change in other groups (P = 0.0001). Significantly higher dietary intakes of folate (P = 0.035), nonstarch polysaccharides (P = 0.001), vitamin C (P = 0.0001), and carotenoids (P = 0.0001) were observed in both intervention groups compared with the control group, and these intake data were broadly supported by nutrient biomarker analysis (28). Impact of FLAVURS intervention on microvascular reactivity, arterial stiffness, and BP There were no significant effects of treatment on the primary outcome of endothelium-dependent microvascular reactivity (LDI-acetylcholine) to acetylcholine administration or endothelium-independent microvascular reactivity (LDI-SNP) to SNP administration in the groups as a whole (Table 2). A significant dose 3 treatment 3 sex interaction (P = 0.0291) for the endothelium-dependent microvascular reactivity as measured by LDI-acetylcholine (Table 2) was observed. Post hoc analysis emerging from the model showed that men in the HF group had increased endothelium-dependent vasodilation with 12 target portions/d, which remained elevated with 14 and 16 target portions/d (P = 0.017) (Table 2). Furthermore, HF men had significantly greater endothelium-dependent vasodilation at 12 wk compared with the LF and control groups (P = 0.0428) (Table 2). There was no significant effect of HF treatment in women. In LF women, a significant increase in endotheliumdependent vasodilation was observed with 12 portions/d (P = 0.0107), which showed a tendency to decrease thereafter, although this did not reach statistical significance (Table 2). Lower endothelial vasodilation was observed in HF women consuming 12 (P = 0.0004) and 14 (P = 0.0252) target portions/d compared with LF and control groups. The endothelium-independent microvascular reactivity, LDI-SNP AUC, decreased in HF women consuming 12 portions/d (P = 0.0002) and then increased with consumption of 16 portions/d (P = 0.0309), whereas it increased in LF women consuming 12 portions/d (P = 0.0035). LDI-SNP AUC showed lower endothelial-independent vasodilation in HF women with 12 (P = 0.0032) and 14 (P = 0.0007) portions/d compared with LF and control groups, whereas LF men showed lower endothelial-independent vasodilation with 14 portions/d (P = 0.0056). No significant interactions were observed for PWV; however, a significant dose 3 treatment interaction was observed for PWA AIx (P = 0.0129), although this was not maintained after correction to a heart rate of 75 beats/min. The increase in PWA AIx observed in the control group, which reached significance at 18 wk (P = 0.0065), was not observed in either of the F&V intervention groups (Table 2). It is of note that there was a significant sex effect for PWV (P, 0.0001), and PWA AIx (P = 0.0129) before correction for heart rate, where women had a 2.2 6.4% greater PWV and PWA AIx compared with men over the course of the intervention period. There was no significant difference in SBP or DBP measurements determined by ABP monitoring of mean 24-h SBP and DBP or nocturnal dipping, and no differences were shown for DVP-SI, DVP-RI, or heart rate. Impact of FLAVURS intervention on biomarkers of NO, endothelial function, and inflammation A significant dose 3 treatment 3 sex interaction was observed (P = 0.0072) for CRP. Only the men in the HF group had a significant reduction in CRP with 14 and 16 target portions/ d compared with baseline and 12 target portions/d (P = 0.001), whereas control men had a significant increase in CRP from baseline to 12 target portions/d (6 wk) (P = 0.0044). Men in the HF and LF groups had significantly lower CRP at 12 (P = 0.0126) and 14 target portions (P = 0.001) compared with control men. A significant dose 3 treatment interaction was shown for VCAM (P = 0.0468), with lower concentrations in F&V groups than in the control group with 14 target portions/ d (P = 0.0468), largely driven by a significant reduction in concentrations for the HF group over time (P = 0.0041). Furthermore, a significant dose 3 treatment 3 sex interaction in E-selectin was observed (P = 0.0378). Reductions in E-selectin were observed over the intervention period in HF men, reaching significance (P = 0.0005) with 14 target portions, and in LF men reaching significance at 12 target portions (P = 0.0006) and women at 14 target portions (P = 0.0047) (Table 3). There was a significant dose 3 treatment interaction for NO (P = 0.0293), with the HF group having a significantly higher concentration than the LF and control groups with 14 target portions (P = 0.0243). A reduction in NO over the intervention period was observed in the control group, which reached significance between 12 and 14 target portions/d (between 6 and 12 wk) (P = 0.0299) (Table 3). No other biomarkers showed significant changes, and there were no significant effects of treatment on any anthropometric measures for the group as a whole (28). DISCUSSION The FLAVURS randomized controlled trial was designed to help refine current dietary guidelines regarding the amount and type of F&Vs that should be recommended, because current national public health F&V recommendations are based largely on epidemiologic data (5, 41, 42). It is the first intervention, to our knowledge, to test the hypothesis that increased consumption of F&Vs, particularly those rich in flavonoids, improves vascular function and reduces other CVD risk markers in a dosedependent manner. Whereas results for the group as a whole remain inconclusive, a sex interaction was observed. Post hoc observations that emerged from the model showed that in men at

484 MACREADY ET AL TABLE 2 Vascular reactivity, ambulatory blood pressure, DVP, and HR between groups for habitual diet and 12, 14, and 16 target F&V portions/d measured at weeks 0, 6, 12, and 18, respectively: FLAVURS 1 HF (n = 58) LF (n = 59) Control (n = 57) P value Variable All F M All F M All F M D 3 T D 3 T 3 S LDI-Ach AUC NS 0.0291 Habitual diet 1190 6 96 1287 6 146 1069 6 111 a 960 6 71 910 6 84 a 1034 6 127 1180 6 124 1216 6 162 1118 6 196 12 portions/d 1176 6 109 1053 6 106 c 1325 6 201 b 1215 6 101 1255 6 132 b,d 1155 6 159 1261 6 104 1346 6 136 d 1114 6 158 14 portions/d 1140 6 87 976 6 94 c 1356 6 147 b,c 1067 6 82 1102 6 121 b,d 1015 6 97 d 1188 6 94 1125 6 100 d 1310 6 198 d 16 portions/d 1179 6 82 1063 6 85 1332 6 148 b 1031 6 86 998 6 114 b 1079 6 133 1133 6 88 1155 6 121 1094 6 122 LDI-SNP AUC NS 0.0053 Habitual diet 1470 6 155 1554 6 236 a 1367 6 189 975 6 78 831 6 70 a 1192 6 153 1209 6 117 1195 6 137 1233 6 219 12 portions/d 1217 6 114 1036 6 99 b,d 1433 6 215 1180 6 101 1152 6 121 b,c 1222 6 179 1316 6 118 1261 6 113 c 1413 6 259 14 portions/d 1247 6 112 922 6 91 b,d 1652 6 192 d 1046 6 79 1059 6 98 b,c 1027 6 133 c 1261 6 112 1201 6 128 c 1378 6 218 d 16 portions/d 1261 6 95 1210 6 116 a 1325 6 159 1187 6 108 1172 6 129 b 1209 6 191 1129 6 88 1054 6 103 1258 6 162 PWV (m/s) NS NS Habitual diet 8.4 6 0.2 8.8 6 0.3 8.0 6 0.4 8.5 6 0.3 9.0 6 0.3 7.8 6 0.6 8.2 6 0.2 8.7 6 0.2 7.4 6 0.4 12 portions/d 8.2 6 0.2 8.6 6 0.3 7.7 6 0.3 8.6 6 0.3 9.2 6 0.4 7.7 6 0.5 8.0 6 0.1 8.6 6 0.2 7.1 6 0.3 14 portions/d 8.4 6 0.2 8.8 6 0.3 7.9 6 0.4 8.4 6 0.2 9.0 6 0.3 7.6 6 0.4 8.2 6 0.2 8.6 6 0.2 7.4 6 0.3 16 portions/d 8.5 6 0.2 8.9 6 0.4 8.0 6 0.4 8.4 6 0.2 9.0 6 0.4 7.7 6 0.4 8.1 6 0.1 8.6 6 0.2 7.3 6 0.3 PWA AIx (%) 0.0129 NS Habitual diet 24.9 6 1.6 30.4 6 1.7 17.1 6 2.6 25.1 6 1.7 29.8 6 2.1 18.2 6 2.5 25.1 6 1.8 a 30.5 6 1.8 15.9 6 3.1 12 portions/d 23.7 6 1.8 29.9 6 2.0 15.1 6 2.4 24.7 6 1.8 29.7 6 2.2 17.0 6 2.5 25.0 6 1.8 a 31.2 6 1.6 14.4 6 3.0 14 portions/d 24.7 61.8 30.1 6 2.1 16.7 6 2.6 24.7 6 1.7 29.4 6 2.1 17.9 6 2.5 26.1 6 1.8 a 30.8 6 1.9 17.5 6 3.2 16 portions/d 24.6 6 1.6 29.2 6 2.0 18.2 6 2.2 25.5 6 1.7 30.6 6 2.0 18.4 6 2.6 27.7 6 1.9 b 33.1 6 1.7 17.5 6 3.4 PWA AIx HR75 (%) NS NS Habitual diet 18.1 6 1.7 24.0 6 1.6 9.8 6 2.7 20.3 6 1.7 26.3 6 1.8 12.1 6 2.6 18.2 6 1.8 24.2 6 1.8 8.0 6 2.7 12 portions/d 17.9 6 1.7 24.4 6 1.8 8.9 6 2.3 19.7 6 1.8 26.6 6 1.7 10.5 6 2.6 18.4 6 1.7 24.5 6 1.5 7.9 6 2.5 14 portions/d 17.7 6 1.7 22.9 6 2.0 10.1 6 2.4 20.7 6 1.7 27.1 6 1.6 11.5 6 2.5 19.0 6 1.8 24.5 6 1.8 9.2 6 2.8 16 portions/d 17.9 6 1.4 22.5 6 1.7 11.6 6 1.9 20.4 6 1.8 27.2 6 1.6 11.2 6 2.6 20.6 6 1.8 26.6 6 1.7 9.3 6 3.0 Mean 24-h SBP (mm Hg) NS NS Habitual diet 126 6 2 122 6 2 131 6 2 128 6 2 129 6 3 125 6 2 125 6 2 122 6 2 130 6 3 12 portions/d 126 6 2 122 6 2 132 6 3 129 6 2 131 6 3 127 6 3 126 6 2 124 6 2 129 6 3 14 portions/d 126 6 2 123 6 3 131 6 3 127 6 2 125 6 2 128 6 2 124 6 2 121 6 2 128 6 3 16 portions/d 125 6 2 119 6 3 133 6 4 130 6 2 129 6 3 130 6 2 124 6 2 121 6 2 129 6 2 Mean 24-h DBP (mm Hg) NS NS Habitual diet 77 6 1 74 6 2 81 6 1 77 6 1 77 6 1 78 6 1 76 6 1 75 6 1 79 6 2 12 portions/d 76 6 1 73 6 1 80 6 2 77 6 1 75 6 1 78 6 1 76 6 1 75 6 1 78 6 2 14 portions/d 76 6 1 74 6 2 80 6 2 77 6 1 76 6 2 78 6 1 76 6 1 74 6 1 80 6 1 16 portions/d 76 6 1 72 6 2 81 6 2 79 6 1 79 6 2 80 6 2 76 6 1 74 6 1 78 6 1 DVP-SI (m/s) NS NS Habitual diet 8.0 6 0.3 7.2 6 0.3 8.8 6 0.6 7.9 6 0.2 7.8 6 0.4 8.0 6 0.4 8.2 6 0.3 8.1 6 0.5 8.3 6 0.6 12 portions/d 7.9 6 0.3 7.4 6 0.4 8.4 6 0.5 7.7 6 0.3 7.2 6 0.4 8.2 6 0.5 7.7 6 0.3 7.6 6 0.4 7.9 6 0.6 14 portions/d 7.7 6 0.3 7.2 6 0.4 8.3 6 0.5 7.8 6 0.3 7.6 6 0.5 8.0 6 0.5 7.6 6 0.3 7.6 6 0.4 7.7 6 0.7 16 portions/d 7.9 6 0.3 7.4 6 0.5 8.4 6 0.5 7.5 6 0.2 6.5 6 0.3 8.4 6 0.4 8.1 6 0.4 7.6 6 0.5 8.7 6 0.8 DVP-RI (%) NS NS Habitual diet 69.2 6 2.0 63.2 6 2.7 76.2 6 2.2 69.7 6 1.8 64.0 6 2.3 76.6 6 2.4 72.9 6 1.9 68.6 6 2.4 79.1 6 2.6 12 portions/d 71.6 6 1.5 67.4 6 2.1 76.1 6 1.7 69.0 6 1.8 64.2 6 2.4 73.8 6 2.4 72.3 6 1.7 68.6 6 2.3 77.1 6 2.3 (Continued)

F&V FLAVONOIDS IMPROVE VASCULAR FUNCTION IN MEN 485 TABLE 2 (Continued) HF (n = 58) LF (n = 59) Control (n = 57) P value Variable All F M All F M All F M D 3 T D 3 T 3 S 14 portions/d 69.7 6 1.7 67.9 6 2.4 71.6 6 2.4 69.1 6 2.1 62.5 6 2.7 76.7 6 2.5 69.4 6 2.0 66.1 6 2.6 74.4 6 2.7 16 portions/d 71.2 6 2.0 65.1 6 3.0 74.8 6 2.4 67.8 6 2.6 54.1 6 2.3 78.8 6 2.4 70.9 6 2.9 62.6 6 4.1 80.4 6 2.7 HR (beats/min) NS NS Habitual diet 61 6 1 61 6 2 60 6 1 63 6 1 66 6 1 60 6 1 61 6 1 61 6 1 60 6 2 12 portions/d 61 6 1 61 6 1 61 6 1 63 6 1 65 6 1 61 6 2 60 6 1 61 6 1 59 6 2 14 portions/d 61 6 1 61 6 1 62 6 2 63 6 1 66 6 2 59 6 1 60 6 1 60 6 1 59 6 2 16 portions/d 62 6 1 63 6 2 61 6 2 63 6 1 65 6 2 60 6 2 61 6 1 64 6 2 58 6 2 1 All values are means 6 SEMs based on 4 repeated measures (0, 6, 12, and 18 wk) for 3 treatment groups [HF F&Vs, LF F&Vs, and habitual diet (control)] over 18 wk. n = 174. The target additional F&V portions were for HF and LF groups only. The MIXED procedure for linear mixed models (LMM, PROC MIXED; SAS Institute) was used to test for dose-dependent changes over time in vascular reactivity, ambulatory blood pressure, DVP, and HR indexes between HF F&V, LF F&V, and habitual diet (control) treatment groups. Superscript letters represent significant inter- and intratreatment interactions, respectively. a,b LDI-Ach AUC increased in HF men (P = 0.017) and in LF women (P = 0.0107) with 12 portions/d (6 wk); LDI-SNP AUC decreased in HF women with 12 portions/d (P = 0.0002) and then increased at 16 portions/d (P = 0.0309), whereas it increased in LF women with 12 portions/d (P = 0.0035); PWA AIx showed an increase over time for habitual diet (control) participants from 18 wk (P = 0.0065). c,d LDI-Ach AUC showed higher endothelial vasodilation in HF men with 14 portions/d (P = 0.0428) and lower endothelial vasodilation in HF women with 12 (P = 0.004) and 14 (P = 0.0252) portions/d compared with LF and control groups. LDI-SNP AUC showed lower endothelium-independent vasodilation in HF women with 12 (P = 0.0032) and 14 (P = 0.0007) portions/d compared with LF and control groups, whereas LF men showed lower endothelium-independent vasodilation with 14 portions/d (P = 0.0056). AUC, area under the flux versus time curve; DBP, diastolic blood pressure; DVP-RI, digital volume pulse reflection index; DVP-SI, digital volume pulse stiffness index; D 3 T, dose 3 treatment; D 3 T 3 S, dose 3 treatment 3 sex; F&V, fruit and vegetable; FLAVURS, FLAvonoids and Vascular function at the University of Reading Study; HF, high-flavonoid; HR, heart rate; LDI-Ach, laser Doppler imaging with acetylcholine; LDI-SNP, laser Doppler imaging with sodium nitroprusside; LF, low-flavonoid; PWA AIx, pulse wave analysis augmentation index; PWA AIx HR75, pulse wave analysis augmentation index with correction to a heart rate of 75 beats/min; PWV, pulse wave velocity; SBP, systolic blood pressure. increased risk of CVD, consumption of HF F&Vs was associated with a significant increase in endothelium-dependent microvascular reactivity after consuming an additional 2.4 portions HF F&Vs ($6 portions total F&Vs). This improvement in microvascular function was associated with lower plasma CRP, VCAM, and E-selectin and with higher plasma NO concentrations. In most cases, these effects persisted with consumption of additional portions of F&Vs. F&V consumption, irrespective of its flavonoid content, was also associated with mitigation of the increase in AIx (a marker of vascular stiffness) and reduction in NO, observed in the control group. In the current study, the significant increase in endotheliumdependent microvascular reactivity in men, after an increased intake of flavonoids (derived from F&Vs) from 36 6 5 to 140 6 14 mg/d (consistent with 2.4 additional portions of HF F&Vs) in the FLAVURS participants, is supportive of the vascular improvements reported in a previous study investigating a single dose of flavonoid-rich foods (20). In this previous study, Dohadwala et al (20) reported a favorable effect on arterial stiffness (determined by PWV) after cranberry juice consumption (94 mg anthocyanins daily) for 4 wk in a small group of men (n = 15) and women (n = 7) with coronary artery disease, although no effect on AIx (determined by PWA) was observed. In addition, a previous study from our group reported a near significant effect (P = 0.06) of consumption of a flavonoid-rich F&V purée drink over 6 wk (equivalent to 296 mg anthocyanins daily) on endothelium-dependent microvascular reactivity measured by LDI (9). However, other studies, often with short intervention periods (1 and 6 wk) and low subject numbers (n = 8 and n = 24), investigating flavonoid-rich fruit and orange and cranberry juices reported no vascular improvements (43, 44). Within FLAVURS, the improvement in endothelium-dependent microvascular reactivity in men was not observed in the female participants, despite the fact that they had significantly greater mean arterial stiffness (determined by PWV and PWA) compared with the men. The reason for the differential effects of sex is unclear. However, the women had differences in their baseline microvascular reactivity between groups that were not observed in men. Women in the LF group had particularly low baseline values compared with the HF and control groups, although these values increased to the same levels as the other 2 groups by the end of the intervention. The inconsistencies in the baseline values in the women would have acted as a significant confounder and may have negated identification of any potential beneficial impact of the intervention. It should also be noted here that no additional adjustment was included in the model to account for false-positive results in multiple comparison testing. For these reasons, it would seem prudent for these sex differences to be investigated in further well-controlled studies. McCall et al (45) reported a dose-dependent increase in endothelium-dependent microvascular function determined by venous occlusion plethysmography in the brachial circulation after consuming 1, 3, or 6 mixed F&V portions daily for 8 wk in a parallel design. Our data on endothelium-dependent microvascular reactivity after chronic HF F&V consumption supports this improvement in microvascular function from a total of 3.9 to 6.3 portions of F&Vs daily, although we did not observe any further benefit with increased consumption. The data therefore suggest a maximum dose effect of w6 portions of F&Vs daily. This is an interesting concept and requires investigation.

486 MACREADY ET AL TABLE 3 Markers of inflammation and endothelial function between groups for habitual diet and 12, 14, and 16 F&V portions/d measured at weeks 0, 6, 12, and 18, respectively: FLAVURS 1 HF (n = 58) LF (n = 59) Control (n = 57) P value Variable All F M All F M All F M D 3 T D 3 T 3 S CRP (mg/ml) NS 0.0072 Habitual diet 1.9 6 0.2 2.0 6 0.4 1.8 6 0.3 a 1.8 6 0.2 2.1 6 0.3 1.3 6 0.3 2.0 6 0.3 2.3 6 0.5 a 1.5 6 0.3 a 12 portions/d 2.2 6 0.3 2.3 6 0.4 2.0 6 0.4 a,c 1.9 6 0.3 2.3 6 0.4 1.1 6 0.3 c 2.5 6 0.5 2.0 6 0.4 b 3.3 6 1.2 b,d 14 portions/d 1.6 6 0.2 2.0 6 0.4 c 1.2 6 0.2 b,c 2.1 6 0.3 2.5 6 0.5 d 1.2 6 0.4 c 1.9 6 0.3 2.0 6 0.4 b,c 1.9 6 0.3 a,d 16 portions/d 2.0 6 0.3 2.5 6 0.5 1.4 6 0.3 b 1.8 6 0.3 2.3 6 0.3 1.1 6 0.3 2.1 6 0.3 2.1 6 0.4 a 2.0 6 0.5 a ICAM (ng/ml) NS NS Habitual diet 916 6 34 933 6 44 893 6 55 903 6 36 889 6 34 923 6 74 932 6 31 937 6 41 922 6 46 12 portions/d 934 6 34 958 6 44 896 6 53 888 6 36 896 6 38 877 6 68 960 6 30 960 6 42 959 6 43 14 portions/d 942 6 35 946 6 45 937 6 57 922 6 40 929 6 38 911 6 87 994 6 37 986 6 46 1009 6 67 16 portions/d 963 6 32 953 6 42 975 6 52 950 6 42 957 6 47 940 6 82 993 6 33 999 6 44 983 6 49 VCAM (ng/ml) 0.0468 NS Habitual diet 645 6 23 a 638 6 31 656 6 35 654 6 24 646 6 24 666 6 47 641 6 24 633 6 30 655 6 41 12 portions/d 635 6 23 a 627 6 30 647 6 35 639 6 25 634 6 25 648 6 51 644 6 22 640 6 26 651 6 39 14 portions/d 625 6 19 b,c 620 6 25 633 6 30 624 6 18 c 636 6 23 604 6 28 659 6 24 c 670 6 30 639 6 43 16 portions/d 636 6 22 a 625 6 31 651 6 29 633 6 19 649 6 25 608 6 27 639 6 21 643 6 26 632 6 37 E-selectin (ng/ml) NS 0.0378 Habitual diet 33.4 6 1.7 31.0 6 2.1 37.1 6 2.8 a 36.0 6 1.9 34.3 6 2.6 a 38.6 6 2.9 a 34.8 6 1.4 32.2 6 1.6 39.1 6 2.2 12 portions/d 33.8 6 1.8 30.2 6 2.1 39.2 6 2.8 a,c 34.0 6 1.8 33.2 6 2.4 a 35.4 6 2.9 b,d 34.3 6 1.4 31.3 6 1.7 39.0 6 2.1 c 14 portions/d 32.6 6 1.7 30.1 6 2.0 36.3 6 2.8 b 33.4 6 1.6 31.4 6 2.0 b 36.8 6 2.6 b 33.9 6 1.4 31.2 6 1.7 38.8 6 2.2 16 portions/d 32.0 6 1.9 28.9 6 1.9 36.5 6 3.3 b 33.6 6 1.8 31.8 6 2.4 b 36.5 6 2.8 b 34.5 6 1.4 31.8 6 1.7 39.2 6 2.1 vwf (% of normal) NS NS Habitual diet 84.4 6 5.5 85.7 6 8.0 82.5 6 6.9 92.6 6 4.9 91.5 6 5.8 94.4 6 8.7 75.4 6 5.6 80.1 6 7.7 67.5 6 7.6 12 portions/d 84.4 6 4.8 90.2 6 6.0 75.0 6 7.5 82.0 6 5.2 73.9 6 6.1 94.9 6 8.9 74.9 6 5.0 77.2 6 7.1 71.0 6 6.0 14 portions/d 81.5 6 6.3 89.8 6 8.4 69.8 6 9.0 86.4 6 4.4 88.4 6 5.9 83.0 6 6.2 79.4 6 5.0 79.6 6 6.4 78.9 6 8.0 16 portions/d 85.0 6 4.2 88.5 6 6.3 80.3 6 4.9 84.4 6 3.7 86.0 6 4.8 82.0 6 6.1 82.3 6 5.7 88.4 6 7.6 72.4 6 7.9 PAI-1 (ng/ml) NS NS Habitual diet 3.1 6 0.3 3.0 6 0.4 3.2 6 0.6 3.3 6 0.4 3.2 6 0.4 3.4 6 0.7 3.4 6 0.4 2.8 6 0.4 4.3 6 0.7 12 portions/d 3.8 6 0.5 3.1 6 0.5 4.9 6 0.9 3.6 6 0.4 3.8 6 0.6 3.4 6 0.4 4.1 6 0.5 4.0 6 0.6 4.3 6 0.8 14 portions/d 4.0 6 0.5 3.8 6 0.6 4.2 6 0.9 4.1 6 0.4 3.7 6 0.7 4.7 6 0.7 4.2 6 0.5 3.4 6 0.4 5.8 6 1.0 16 portions/d 3.8 6 0.5 3.2 6 0.3 4.7 6 1.1 3.8 6 0.4 3.8 6 0.7 3.9 6 0.5 3.7 6 0.3 3.4 6 0.3 4.4 6 0.9 TNF-a (pg/ml) NS NS Habitual diet 1.3 6 0.2 1.0 6 0.2 1.6 6 0.4 1.1 6 0.1 1.2 6 0.2 0.8 6 0.2 1.8 6 0.4 2.3 6 0.6 0.8 6 0.2 12 portions/d 1.0 6 0.2 1.1 6 0.3 0.9 6 0.1 1.1 6 0.1 1.2 6 0.2 0.8 6 0.1 1.2 6 0.2 1.4 6 0.3 0.8 6 0.2 14 portions/d 0.9 6 0.2 0.9 6 0.3 0.9 6 0.1 1.0 6 0.1 1.1 6 0.2 0.9 6 0.1 1.7 6 0.4 2.1 6 0.6 0.8 6 0.2 16 portions/d 0.9 6 0.1 0.9 6 0.1 0.9 6 0.2 1.1 6 0.1 1.1 6 0.2 1.0 6 0.1 1.6 6 0.3 2.0 6 0.5 0.9 6 0.2 IL-6 (pg/ml) NS NS Habitual diet 1.6 6 0.2 1.6 6 0.3 1.6 6 0.4 1.3 6 0.1 1.5 6 0.2 1.1 6 0.1 1.3 6 0.1 1.3 6 0.1 1.2 6 0.1 12 portions/d 1.6 6 0.2 1.6 6 0.3 1.6 6 0.4 1.7 6 0.2 2.0 6 0.3 1.2 6 0.1 1.2 6 0.1 1.2 6 0.1 1.3 6 0.1 14 portions/d 1.7 6 0.2 1.6 6 0.3 1.7 6 0.4 1.6 6 0.2 1.7 6 0.3 1.3 6 0.2 1.4 6 0.1 1.2 6 0.1 1.7 6 0.2 16 portions/d 1.9 6 0.3 1.9 6 0.4 1.8 6 0.4 1.6 6 0.2 1.9 6 0.2 1.1 6 0.2 1.4 6 0.1 1.3 6 0.1 1.4 6 0.2 NO (mmol/l) 0.0293 NS Habitual diet 10.8 6 0.5 11.2 6 0.7 10.3 6 0.8 10.4 6 0.3 10.6 6 0.4 10.1 6 0.5 10.6 6 0.3 a 10.8 6 0.5 10.1 6 0.4 12 portions/d 11.7 6 0.5 12.0 6 0.6 11.3 6 0.8 10.3 6 0.4 10.5 6 0.4 10.1 6 0.6 11.4 6 0.5 a 11.0 6 0.6 11.9 6 0.7 (Continued)

F&V FLAVONOIDS IMPROVE VASCULAR FUNCTION IN MEN 487 TABLE 3 (Continued) HF (n = 58) LF (n = 59) Control (n = 57) P value Variable All F M All F M All F M D 3 T D 3 T 3 S 14 portions/d 11.7 6 0.6 c 11.6 6 0.8 12.0 6 0.8 11.0 6 0.5 d 10.5 6 0.5 11.7 6 0.8 10.2 6 0.4 b,d 10.3 6 0.5 10.1 6 0.6 16 portions/d 11.4 6 0.5 11.8 6 0.7 10.9 6 0.6 11.2 6 0.5 10.9 6 0.7 11.8 6 0.6 10.4 6 0.4 b 10.5 6 0.6 10.1 6 0.6 Fibrinogen (g/l) NS NS Habitual diet 3.4 6 0.1 3.5 6 0.1 3.3 6 0.2 3.2 6 0.1 3.4 6 0.1 3.1 6 0.1 3.2 6 0.1 3.4 6 0.1 3.0 6 0.2 12 portions/d 3.5 6 0.1 3.6 6 0.2 3.4 6 0.1 3.3 6 0.1 3.5 6 0.2 3.0 6 0.1 3.3 6 0.1 3.4 6 0.1 3.2 6 0.1 14 portions/d 3.2 6 0.1 3.3 6 0.1 3.1 6 0.1 3.2 6 0.1 3.4 6 0.2 3.0 6 0.1 3.2 6 0.1 3.3 6 0.1 3.1 6 0.2 16 portions/d 3.4 6 0.1 3.5 6 0.1 3.3 6 0.2 3.4 6 0.1 3.5 6 0.1 3.2 6 0.1 3.2 6 0.1 3.4 6 0.1 2.8 6 0.1 1 Values are means 6 SEMs based on 4 repeated measures (0, 6, 12, and 18 wk) for 3 treatment groups [HF F&Vs, LF F&Vs, and habitual diet (control)] over 18 wk. n = 174. The target additional F&V portions were for HF and LF groups only. The MIXED procedure for linear mixed models (LMM, PROC MIXED; SAS Institute) was used to test for dose-dependent changes over time in markers of inflammation and endothelial function between HF F&V, LF F&V, and habitual diet (control) treatment groups. Superscript letters represent significant inter- and intratreatment interactions, respectively. a,b CRP decreased in HF men, reaching significance with 14 and 16 portions/d (P = 0.0010); in control men, CRP increased between baseline and 6 wk (P = 0.0044); in control women, CRP increased over time between 12 (P = 0.0238) and 18 (P = 0.0078) wk; VCAM decreased in the HF group between baseline and 14 portions/d (P = 0.0041) but increased with 16 portions/d (P = 0.0317); E-selectin decreased in HF men between 12 and 14 portions/d (P = 0.0005), in LF men from baseline to 12 portions/d (P = 0.0006), and in LF women between 12 and 14 portions/d (P = 0.0047); NO decreased in the control group, reaching significance between 6 and 12 wk (P = 0.0299). c,d HF and LF men had lower CRP concentrations after 12 (P = 0.0126) and 14 (P = 0.001) portions/d compared with control men; LF women at 14 portions/d had higher CRP than did HF and control women (P = 0.0219); VCAM in HF and LF groups was lower than in the control group with 14 portions/d (P = 0.0468); LF men had lower E-selectin than did HF and control men with 12 portions/d (P = 0.0378); HF groups had higher NO than did LF and control groups with 14 portions/d (P = 0.0293). CRP, C-reactive protein; D 3 T, dose 3 treatment; D 3 T 3 S, dose 3 treatment 3 sex; F&V, fruit and vegetable; FLAVURS, FLAvonoids and Vascular function at the University of Reading Study; HF, high-flavonoid; ICAM, intercellular adhesion molecule; LF, lowflavonoid; NO, nitric oxide; PAI-1, plasminogen activator inhibitor-1; VCAM, vascular cell adhesion molecule; vwf, von Willebrand factor. Interestingly, in line with this, a study in a large cohort of Swedish men and women (71,706) with a 13-y follow-up reported that consumption of,5 portions F&Vs/d was associated with progressively shorter survival and higher mortality rates (46); however, neither study reported sex effects. Consumption of F&Vs was not associated with significant changes in measures of arterial stiffness measured by PWV and DVP. Yet, a significant impact of F&V consumption, irrespective of flavonoid content, on arterial stiffness measured by PWA AIx was observed in the group as a whole. However, this significance was lost after normalization, suggesting that the response may be influenced by BP. The significant impact of F&V consumption on arterial stiffness observed with PWA appeared to be a mitigation of the increase in Alx observed over time in the control group and was not reflective of a dose-response improvement in measures of arterial stiffness. Of note was the significant concomitant reduction in NO in the control group, suggesting that the lower NO availability may, at least in part, have contributed to the increased arterial stiffness in this group. The lack of effect on PWV was not unexpected, because previous research has supported a dissociation between PWV and PWA AIx (47). Possible mechanisms by which flavonoids can improve vascular function include increasing NO bioavailability, a key modulator of vascular function (48). In FLAVURS, higher plasma nitrite concentrations, a surrogate marker of NO (49), were observed in the HF group compared with the LF and control groups, reaching significance after 4 additional target portions of HF F&Vs/d. One mechanism by which NO is synthesized is by the action of endothelial NO synthase (enos) on arginine (49). It has been reported that flavonoids may enhance endogenous NO availability via increased gene expression or activation of enos, facilitating vasodilation and improved vascular function, or by preserving the intracellular arginine pool (50). Our data are consistent with a possible increase in NO bioavailability, yet it is unclear whether this occurred through increased NO synthesis, via activation of enos, or via decreased NO decomposition through its reaction with thiols and/or superoxide. With regard to the latter, the reaction of flavonoid/phenolic acid metabolites with NADPH oxidase (which generates superoxide) might also be expected to increase NO bioavailability through a decrease in its decomposition on reaction with superoxide (51). Although dietary nitrate intake was not quantified in FLAVURS, the F&Vs supplied may have contributed to this, which may have influenced the increased NO concentration. A proinflammatory phenotype has been related to endothelial dysfunction with CRP as an independent CVD risk factor (52). The general reductions in CRP, E-selectin, and VCAM in the male participants suggest that a reduced endothelial inflammatory status may have contributed to the improvements in microvascular reactivity in male participants in FLAVURS in the HF group. Numerous cell and animal studies have reported a beneficial impact of selected flavonoids on inflammatory cytokines (53, 54). However, flavonoid intervention studies in humans do not consistently report significant effects on inflammatory markers (20, 21, 55), although, again, previous studies have had relatively small groups, short study durations, and different subject groups. No effect was observed for increasing doses of either HF or LF F&Vs on 24-h ABP measures. Although some studies report a reduction in BP after F&Vs (56), this is not consistently observed (20, 45). Results may be