The study protocol and consent form were approved by the Griffin Hospital (Derby, CT) Institutional Review Board and the Yale University (New Haven, CT) Human Investigation Committee and conducted in accordance with the Declaration of Helsinki. Written informed consent was obtained, and all subjects received nominal monetary compensation for their participation.

Subjects were female or male, 18 years of age or older with metabolic syndrome 18. Eligible subjects were nonsmokers (as smoking impairs endothelial function 19), not taking any other vitamins or dietary supplements, and able to have blood pressure measured bilaterally at the brachial artery. Subjects were excluded with any unstable medical condition that would limit the ability to participate fully in the trial. Subjects were also excluded if using insulin, glucose sensitizing medication, and vasoactive medications (including glucocorticoids, antineoplastic agents, psychoactive agents, or bronchodilators). Other exclusion criteria included rheumatologic disease requiring regular use of NSAIDs or alternative medications, preexisting cardiovascular disease, or a diagnosed eating disorder.

Metabolic syndrome was defined using the International Diabetes Federation criteria 18 of: waist circumference of ≥ 94 cm (males) or ≥ 80 cm (females) plus any two of the following: (a) blood pressure ≥ 130/85 or taking antihypertensive medication, (b) fasting plasma glucose (FPG) > 100 mg/dL, (c) serum triglycerides (TG) > 150 mg/dL, (d) high-density lipoprotein (HDL) < 40 mg/dL in men, and < 50 mg/dL in women. Subjects on lipid lowering medication were assumed to meet criteria for (c) and (d) above, though required to have a stable dose for at least three months and willing to refrain from taking medication for 12 hours prior to endothelial function assessment.

This study was a randomized, double blind, placebo controlled crossover clinical trial with three treatment arms designed to assess the effects of 8 weeks of daily ingestion of two encapsulated fruit and vegetable juice powder concentrates (Juice Plus+) vs. placebo. The two blends were chosen to further assess whether the addition of berry powders (Blend 1) would result in differential effects compared to a general blend of fruits and vegetables (Blend 2). Blend 1 consisted of a combination of fruit, vegetable, and berry mixtures containing 7.5 mg β-carotene, 276 mg vitamin C, 72 mg vitamin E (RRR-α-tocopherol), 780 μg folate, and 80 mg calcium 2021. Blend 2 consisted of a combination of fruit and vegetable mixtures containing 7.5 mg β-carotene, 234 mg vitamin C, 30 mg vitamin E (RRR-α-tocopherol), 420 μg folate, and 60 mg calcium 21. A description of the blends used in the study are in Table 1.

For the initial assignment, subjects were randomized to Blend 1, Blend 2, or placebo. As the pharmacokinetics of the tested blends are unknown, our design incorporated an 8-week washout period with no intervention after completing the initial treatment assignment. Subjects were then crossed-over to one of the two remaining assignments. Following 8-weeks of sustained intervention in the second assignment and an 8-week washout period, subjects then crossed-over to the remaining treatment assignment for 8-weeks followed by a final 8-week washout period. Subjects were evaluated on six occasions during the study: baseline, at the completion of each of the three sustained treatment assignments, and following two washout periods (see Figure 1)

Subjects were instructed to take three capsules twice daily for a total of six capsules per day (4.5 grams). Placebo capsules were provided by the study sponsor and were identical in appearance to active treatments to ensure blinding. Subjects continued consuming their usual diet; no dietary advice or guidance was provided during the course of the study.

Recruitment: Screening commenced in December 2004. Of the 632 persons screened for eligibility, 534 did not meet eligibility criteria, 5 refused to participate, and 29 were not randomized for other reasons (see Figure 1). Enrollment continued until October 2006. 64 subjects (52% female, mean age 56.9 years) were randomized to sequences of Blend 1, Blend 2, and placebo. Subjects received the intervention between July 2005 and October 2007. Of the enrolled subjects, 48 completed at least 1 time point and 44 completed the entire trial (see Figure 1). Pill count data was available for 75% of all supplement bottles dispensed; 84% of these demonstrated compliance of over 80%.

Endothelial function was assessed as flow-mediated dilatation (FMD) of the brachial artery using standard procedures 22. In brief, FMD was measured noninvasively in the right brachial artery with a high-frequency ultrasound (Sonos 4500; Phillips Medical Systems, Andover, MA) in accordance with published guidelines 12. Measures of vessel diameter and flow velocity were obtained by a single dedicated vascular clinical research specialist (YM) blinded to subject treatment status. Repeat scans were obtained at 15, 60, and 120 seconds after deflation. At each scanning interval, both cross-sectional vessel diameter and flow velocity were recorded. In addition to brachial diameter at 60 seconds after cuff release, flow after cuff deflation within the first 15 seconds was used as an indicator of stimulus strength; hyperemic flow being the stimulus for endothelial reactivity. To account for potential variability in stimulus strength, FMD was divided by flow at 15 seconds after cuff deflation to create a stimulus-adjusted response measure.

Normal FMD in young, healthy subjects is typically > 10% 23. Impaired FMD is generally < 10%; Su et al. reported mean FMD values of 10.68 in those with normal glucose tolerance, 8.86 in patients with impaired glucose tolerance, and 5.27 in type 2 diabetes mellitus 24.

The primary outcome measures were the difference in FMD between the two blends of supplements and placebo following a 2-hour oral glucose tolerance test (OGTT) after 8 weeks of treatment.

Change in the fasting (unchallenged) FMD after 8 weeks on active treatment and placebo assignments, the difference in FMD between Blend 1 and Blend 2, weight, lipid panel, and serum insulin. Acute (single dose) effects on endothelial function and plasma glucose were assessed by consuming the supplements concurrently with a 75 g glucose challenge load and assessing FMD after two hours.

Outcomes were assessed and laboratory measures were collected and analyzed at Griffin Hospital (Derby, CT) as routine clinical samples following each visit.

Subjects meeting eligibility criteria and enrolling in the study were randomized by the study sponsor to treatment sequence using the web site Randomization.com (http://www.randomization.com). The sponsor provided coded bottles for each subject containing their daily capsules. Subjects were enrolled and assigned to interventions by the study coordinator. All study personnel and subjects were blinded to treatment assignment throughout the duration of the intervention.

Statistical analysis was conducted using SAS software (Version 9.1, SAS Institute, Cary, NC, USA). Repeated measures ANOVA were performed to compare the means of both dependent and independent variables at baseline and after each washout period between the three treatment groups. To adjust for any potential carryover effect following washout, treatment sequence (timing of each treatment) was entered as a control variable in multivariable analysis.

Analysis was by intention-to-treat. Missing individual data was addressed by the last observation carried forward method. Data were assessed for normality prior to analysis. Consistent with guidelines from the International Brachial Artery Reactivity Task Force 12, a sample size of 48 individuals, allowing for 20% attrition and non-adherence, was predicted to provide 80% power to detect a minimal clinically important difference of 3.0% in post-glucose load FMD between intervention and placebo groups following 8 weeks of daily treatment, as well as adjusting for multiple comparisons for three pair-wise comparisons (two-tailed α = 0.05).

Per-protocol analysis included subjects that completed at least one treatment assignment. Paired t-tests were used to compare baseline measures of FMD and plasma analyses to the values obtained at the end of each washout period (within-group analyses).

Compared to placebo, no significant differences (P > 0.05) were seen in FMD change between baseline and 8-weeks in subjects consuming Blend 1 or Blend 2 (Table 3). Findings persisted after controlling for the variability of the strength of the stimulus that determines vasodilatation (Table 2). No significant deterioration in endothelial function was seen in any group after consuming the glucose load (Table 3). A random sample of 20 FMD measurements were reanalyzed; the correlation (Pearson's r) between the initial and second assessment values was 99%.

No significant changes in serum insulin were seen in either intervention group between baseline and 8-weeks compared to placebo (P>0.05). Furthermore, no significant differences were seen between Blend 1, Blend, 2, and placebo in FPG and total cholesterol (all P>0.05) (Table 3).

Body weight did not change in any group after eight weeks.

Per-protocol analyses consisting of subjects that completed the entire intervention (n = 27-37) did not reveal any major differences from the intention-to-treat analysis. Furthermore, analyses of highly compliant subjects (>80% compliance) and the combination of compliant subjects completing the trial did not find any significant treatment effects of either blend (Table 4). Acute (single dose) effects of Blend 1 trended towards a significant improvement in endothelial function two hours after consumption of a 75-g glucose load when compared to baseline (Blend 1: 1.1 ± 3.7%, P = 0.0549). No significant between-group differences were found. No effects of gender or age were found.

No adverse effects related to the intervention were seen during the course of the study.