Study design and subjects
This study was conducted exclusively at the National Institutes of Health (NIH) Clinical Center, Bethesda, MD. The study protocol was approved by the Institutional Review Board of the National Heart, Lung, and Blood Institute, and all procedures followed were in accordance with institutional guidelines. Adults between 21 and 65 y of age, in good general health except for mild-to-moderate hypertension (systolic blood pressure of 140–170 mm Hg and/or diastolic blood pressure of 95–110 mm Hg without antihypertensive medication), and not taking any medication or nutritional supplements except for antihypertensive agents were recruited from the local community through newspaper advertisements. Subjects were specifically excluded from study enrollment if they were taking any medications other than antihypertensive drugs; if they were pregnant or had diabetes, liver disease, pulmonary disease, renal insufficiency, coronary heart disease, heart failure, peripheral vascular disease, coagulopathy, or any other severe systemic diseases; or if they were allergic to cocoa or perflutren lipid microspheres in microbubble contrast (Definity; Bristol Myers Squibb Medical Imaging Inc, North Billerica, MA). Subjects were also excluded if they had actively smoked within the past 2 y, were receiving treatment for or had a history of any form of cancer, or if they had positive blood tests for HIV, hepatitis B, or hepatitis C.
Of 65 individuals screened, 45 were deemed eligible for study run-in (). Informed consent was obtained from each subject. During the initial 1 wk run-in period, study participants met with a dietitian and were instructed to maintain their usual physical activity and to begin a low-flavanol diet. All subjects were specifically provided with a list of foods rich in flavanols, which they were to avoid eating. In addition, antihypertensive medications, if taken, were discontinued. In some cases, the antihypertensive drug regimen was tapered off over the course of the run-in period for safety reasons. During the run-in period, and for the duration of the entire study, blood pressure was monitored thrice weekly in the NIH outpatient clinic. Blood pressure was measured in the dominant arm in the sitting position with a standard sphygmomanometer with appropriately sized cuff. Blood pressure recorded was the mean of 3 successive readings 5 min apart (subjects were seated for ≥15 min before measurement). Participants whose blood pressure during the run-in phase did not meet the inclusion criteria were excluded from further study participation. Systolic or diastolic blood pressure measured at the screening visit (for participants not taking antihypertensive medication) or at the end of the run-in period (for participants taken off antihypertensive medication at the beginning of the run-in period) was used to determine study eligibility. In addition to monitoring blood pressure in the outpatient clinic, participants were required to self-monitor their blood pressure at home using a portable blood pressure–measuring device (OMRON Digital Blood Pressure Monitor, Bannockburn, IL). If a participant’s blood pressure exceeded 160/100 mm Hg during measurements on 3 consecutive days or >170/110 mm Hg on 3 determinations over a period of ≥15 min at any time, they were withdrawn from the study and appropriate antihypertensive therapy resumed. Of the 29 subjects who were randomly assigned, 20 subjects completed the entire study (). Nine subjects failed to complete the entire study for a variety of reasons, including transient flank pain during administration of the microbubble contrast, allergic reaction to the microbubble contrast agent, family emergencies, other personal problems, difficulties with intravenous access, loss to follow-up, or exacerbation of chronic back pain.
Participant flow throughout the trial.
After the 1-wk run-in period, enrolled subjects were randomly assigned in a double-blind fashion (block randomization by NIH Clinical Center Pharmacy) to the initial arm of the study: either a cocoa drink (≈450 mg total flavanols twice a day) or a matching placebo drink (≈14 mg total flavanols twice a day) for 2 wk. This was followed by a 1-wk washout period. Subjects were then crossed over to the other treatment arm for an additional 2 wk. Each enrolled subject underwent a hyperinsulinemic isoglycemic glucose clamp study and forearm vascular studies at baseline (at the end of 1-wk run-in period) and after each 2-wk treatment period. In addition, during the beginning of the study day (baseline and at the end of each 2-wk treatment period), blood pressure was measured and blood samples were drawn after the cocoa (≈450 mg total flavanols) or placebo (≈14 mg total flavanols) drink to estimate the pharmacokinetics of flavanols and their metabolites in plasma. A research nurse counted placebo or cocoa powder packets at the end of each treatment period to help monitor subject compliance. Study investigators and participants were blinded to treatment assignment, and assignment codes were not available to investigators until 20 participants completed the entire study and the database had been completed and secured. Participant blinding was assessed by a questionnaire administered at the end of 6 wk that asked patients to indicate which treatment they believed they received during each of the 2 phases (cocoa, placebo, or uncertain).
Cocoa and Placebo Drink Preparations
Thirty one grams of cocoa or placebo beverage powder (CocoaPro; Mars Incorporated, Hackettstown, NJ) mixed in 150 mL warm water was consumed twice daily during the treatment phases of the study. The cocoa and placebo drinks were isocaloric and matched for fat, macronutrients, mineral content, theobromine, and caffeine (). In addition, the cocoa and placebo drinks were similar in color, taste, and packaging. Packets of cocoa and placebo powder were stored by the NIH Clinical Center pharmacy. To test the stability and variability of the cocoa and placebo powders, the polyphenol, macronutrient, and mineral contents were reassessed after the end of the study.
Composition of cocoa and placebo powders used to make the study drinks
Plasma flavanol and metabolite measurements and pharmacokinetics
Circulating plasma flavanols and their metabolites were measured after oral ingestion of the placebo drink (≈14 mg total flavanols) or cocoa drink (≈450 mg total flavanols) at the beginning of each glucose clamp study (at baseline and at the end of each 2-wk treatment period). After subjects were given the placebo or cocoa drink (time 0), peripheral blood samples were collected in EDTA-containing tubes containing ascorbate (1 mg/mL) at time 0, 0.5, 1, 2, and 3 h after oral administration. Plasma was obtained from blood samples by centrifugation and immediately frozen on dry ice for storage at − 80 °C. Plasma samples were treated and analyzed by using methods described previously (29
). Ingested flavanols in cocoa or placebo drinks are rapidly metabolized in phase I and II biotransformations to various O
-methylated, and O
-sulfated forms (35
). Therefore, plasma samples were treated with β
-glucuronidase according to procedures detailed previously (36
). Resulting metabolites (epicatechin, catechin, 4′-O
-methyl-catechin, and 3′-O
-methyl-catechin) were separated and analyzed by using reversed-phase HPLC coupled with fluorescence detection as described previously (29
). Concentrations of individual metabolites were quantified by using external calibration curves generated with the use of authentic standards. The analytic chemists were blinded with respect to sample information until after the end of the study, when the database was completed and locked. Pharmacokinetic parameters for total plasma flavanols (sum of epicatechin, catechin, 4′-O
-methyl-catechin, and 3′-O
-methyl-catechin concentrations) were derived after oral dosing assuming first-order kinetics (37
). The elimination constant, Ke
, was estimated from the slope of the linear regression of log-transformed concentration values plotted versus time in the terminal phase, assuming first-order kinetics and instantaneous mixing. The apparent elimination half-time (t1/2
) was calculated as t1/2
= log (2)/Ke
(assumes instantaneous mixing and no significant endogenous production). Time to peak plasma concentration (Tmax
) and peak plasma concentration (Cmax
) were estimated from the observed concentration versus time data assuming instantaneous mixing. The area under the curve to the last measurable concentration (AUC0–3 h
) was calculated by using the trapezoidal rule for the observed values from 0 h to the last measured time point (3 h).
Hyperinsulinemic isoglycemic glucose clamp
Insulin sensitivity was evaluated by glucose clamp as previously described (7
). The steady state period of the clamp was defined as a period of ≥60 min (1–2 h after the beginning of the insulin infusion) during which the CV for blood glucose, plasma insulin, and glucose infusion rate was <5%. The glucose clamp–derived index of insulin sensitivity (SIClamp
) was defined as M
) corrected for body weight, where M
is the steady state glucose infusion rate (mg/min), G
is the steady state blood glucose concentrations (mg/dL), and ΔI
is the difference between basal and steady state plasma insulin concentrations (μ
Quantitative insulin-sensitivity check index
Quantitative insulin-sensitivity check index (QUICKI) was calculated as previously defined (38
). QUICKI is calculated as 1/[log(I0
) + log(G0
)], where I0
is fasting insulin (μ
U/mL) and G0
is fasting glucose (mg/dL). Because QUICKI is the reciprocal of the log-transformed product of fasting glucose and insulin, it is a dimensionless index without units.
Brachial artery blood flow
Insulin is known to increase brachial artery diameter (BAD) and brachial artery blood flow (BAF). These vascular actions of insulin help to couple regulation of metabolic and hemodynamic homeostasis (1
). Therefore, as a measure of endothelial function relevant to insulin action, we assessed insulin-stimulated increases in BAD and BAF by using Doppler ultrasound. Measurements were performed at the beginning of each glucose clamp study (fasting insulin concentrations) and 2 h after initiation of the insulin infusion (during steady state hyperinsulinemia), as described previously (7
). Briefly, the right brachial artery was visualized on the anterior aspect of the arm, 2–15 cm proximal to the antecubital fossa with a high-resolution ultrasound probe (HDI-5000 ultrasound machine with a 12-MHz linear array transducer; Philips Ultrasound, Bothell, WA). The position of the transducer on the arm was marked to facilitate visualization of the same portion of the artery throughout the study. BAD was measured from the anterior to the posterior “m” line (interface between media and adventitia) by using video calipers at end-diastole, coincident with the R wave on the electrocardiogram. BAF was estimated from Doppler flow velocity time integral (VTI), BAD, and HR measurements by using the equation BAF = π
× VTI × HR. Insulin-induced changes in BAD and BAF were expressed as a percentage of preclamp baseline values.
Forearm skeletal muscle capillary recruitment
Insulin-stimulated capillary recruitment in the deep flexor muscles of the forearm was assessed by using microbubble contrast-enhanced ultrasonography as described previously by others (39
). We compared estimates of microvascular blood flow before insulin infusion and at a period of steady state hyperinsulinemia during the glucose clamp. Immediately after each BAF determination, skeletal muscle capillary recruitment was estimated by using a power Doppler imaging technique. Ultrasound imaging of deep flexor muscles of the forearm was performed in a transaxial plane 5 cm distal to the antecubital fossa (P4-2 phased array transducer, HDI-5000; Phillips Ultrasound). Gain settings were kept constant throughout each study. A suspension of echogenic microbubbles with a similar size and rheology to red blood cells (Definity; Bristol Myers Squibb Medical Imaging Inc) was infused intravenously at a constant rate with an infusion pump (model A-99; Razel Industries, Stamford, CT). Two minutes after initiating the infusion of microbubbles (sufficient to achieve systemic steady state distribution of microbubbles), microbubbles were destroyed in a chosen region of interest (ROI) by using high-energy ultrasound (high mechanical index). Subsequently, a pulsing interval (time) versus video-intensity curve was generated to evaluate microbubble replenishment kinetics. This data reflects the refilling of microbubbles from outside the ROI where microbubble contrast agent is still intact. As the pulsing interval becomes longer, the acoustic intensity increases because of replenishment of microbubble contrast in the ROI. Images were recorded onto SVHS videotape and analyzed by using MCE software (University of Virginia). Plots of video contrast intensity (y
) versus pulsing interval (t
) were fit to the first-order exponential equation y
(1 − e−βt
). In this equation, the parameter α
represents the maximal signal intensity measured after complete refilling and is proportional to capillary blood volume (CBV) in the ROI. The parameter β
is proportional to the initial capillary blood flow velocity (CFV). Capillary blood flow (CBF) was calculated as the product of CBV and CFV.
Circulating endothelial adhesion molecules and adipocytokines
Soluble E-selectin, intercellular adhesion molecule 1 (sICAM-1), and vascular cell adhesion molecule 1 (sVCAM-1) were measured in serum in duplicate in the same assay by using the Human CVD Panel 1 Lincoplex kit (a multiplex assay kit based on Luminex xMAP technology; Linco Research, Inc, St. Charles, MI). The detection limits for these assays were 79, 9, and 16 pg/mL, respectively. The intraassay CVs were between 4.5% and 11.2%, and the interassay CVs were between 8.5% and 13.4%. Circulating adiponectin was also measured by using the Human CVD Panel 1 Lincoplex kit. The detection limit for this assay was 56 pg/mL, and the interassay and intraassay CVs were 16% and 9.2%, respectively. Plasma leptin, tumor necrosis factor-α (TNF-α), interleukin-6 (IL-6), and monocyte chemoattractant protein-1 (MCP-1) concentrations were measured in duplicate in the same assay from a fasting plasma sample with the use of the Human Cytokine Lincoplex kit (Linco Research, Inc, St. Charles, MI). The detection limits for these assays were 85, 0.14, 1.6, and 0.14 pg/mL, respectively. The intraassay CVs were between 1.4% and 7.9%, and the interassay CVs were <20%. All assays were carried out on the same day to minimize assay variability.
Routine assays for serum lipids, plasma glucose and insulin, and hemoglobin A1c were performed in the Department of Laboratory Medicine at the Clinical Center, NIH.
Data from participants who completed all phases of the protocol (n
= 20) were analyzed according to a preestablished statistical analysis plan. Changes in blood pressure and changes in insulin sensitivity as measured by glucose clamp were prospectively designated as the primary endpoints for the study. All other comparisons were considered secondary. The presence of skewed data were evaluated by visual inspection of Q-Q plots, stem and leaf plots, or box plots and verified by the Shapiro-Wilk test for normal distribution. After testing data for normality, we used Student’s paired t
test or Wilcoxon’s signed-rank test to evaluate differences between the outcome measures SIClamp
, blood pressure, BAF, capillary recruitment, and adipocytokine concentration with treatment (cocoa compared with placebo). A one-sample sign test was used to assess the effect of hyperinsulinemia during the glucose clamp on vascular variables. The primary outcomes in this study were change in blood pressure and change in insulin sensitivity as measured by glucose clamp. Therefore, power analysis was calculated for these outcomes in this 2-treatment crossover study. On the basis of mean SIClamp
values that we obtained for hypertensive subjects in previous studies using the glucose clamp method at an insulin infusion rate of 120 mU/m2
· min (SIClamp
= 3.70 ± 0.19 10−4
dL · kg−1
U/mL), a sample size of n
= 20 is sufficient to detect a 10% difference between the treatment and placebo in subjects with >90% power (41
). A previous study reported that dark chocolate consumption (500 mg flavanols daily for 2 wk) reduced systolic blood pressure by 5.1 ± 2.4 mm Hg (mean ± SD) and diastolic blood pressure by 1.8 ± 2.0 mm Hg in hypertensive subjects (n
= 13) (31
). On the basis of this study, a sample size of 20 is sufficient to detect a 5-mm Hg change in systolic blood pressure with >90% power and a 2-mm Hg change in diastolic blood pressure with >90% power and a 2-sided α
= 0.05. To evaluate whether subjects were effectively blinded, Fisher’s exact test was used to compare correct with incorrect treatment guesses by each subject at the end of the trial. P
values were not adjusted for multiple comparisons, and P
values <0.05 were considered to represent statistical significance. The statistical software programs SPSS (version 10.0; SPSS Inc, Chicago, IL) and SAS (version 9.0; SAS Institute Inc, Cary, NC) was used for the data analysis.