Effect of Black Tea Intake on Blood Cholesterol Concentrations in Individuals with Mild Hypercholesterolemia: A Diet-Controlled Randomized Trial

Abstract

Habitual intake of black tea has been associated with relatively lower serum cholesterol concentrations in observational studies. However, clinical trial results evaluating the effects of black tea on serum cholesterol have been inconsistent. Several factors could explain these mixed results, in particular, uncontrolled confounding caused by lifestyle factors, e.g. diet. This diet-controlled clinical trial estimates the effect of black tea flavonoid consumption on cholesterol concentrations in 57 borderline hypercholesterolemic individuals (total cholesterol concentrations between 190 and 260 mg/dl (4.9 and 6.7 mmol/L)). A double blind, randomized crossover trial was conducted in Minneapolis, MN from April 2002 through April 2004, wherein key conditions were tightly controlled to minimize possible confounding. Participants consumed a controlled low-flavonoid diet plus 5 cups per day of black tea or tea-like placebo over two 4-week treatment periods. The flavonoid-free caffeinated placebo matched the tea in color and taste. Differences in cholesterol concentrations at the end of each treatment period were evaluated via linear mixed models. Differences (95% CI) in mg/dl among those treated with tea versus placebo were 3.43 (−7.08, 13.94) for total cholesterol, −1.02 (−11.34, 9.30) for low-density lipoprotein cholesterol (LDL-C), 0.58 (−2.98, 4.14) for high-density lipoprotein cholesterol (HDL-C), 15.22 (−40.91, 71.35) for triglycerides, and −0.39 (−11.16, 10.38) for LDL plus HDL cholesterol fraction. The LCL-C/HDL-C ratio decreased by −0.1 units (95% CI −0.41, 0.21). No results were statistically or clinically significant. Thus, the intake of 5 cups of black tea per day did not significantly alter the lipid profile of borderline hypercholesterolemic subjects.

INTRODUCTION

Tea brewed from Camellia sinensis is the most commonly consumed beverage in the world after water.^1,2 It is rich in polyphenolic flavonoids that possess antioxidant, anti-mutagenic, anti-inflammatory, and antiallergenic properties.³ These flavonoids may also be hypocholesterolemic,^3–5 as shown in cell culture,^6–8 animal,^4,6,9–21 and observational^22–32 studies of black tea, the most commonly consumed tea in the United States. Clinical trial results, however, have been inconsistent.^33–39 Some randomized trials have found significant hypocholesterolemic associations between black tea consumption and lipid profiles,^33,36,39 while others have reported no effect on lipid profiles.^34,35,37,38 A review on the subject found limited evidence that tea has favorable effects on cardiovascular disease risk factors, including hypercholesterolemia.⁴⁰ It urged cautious interpretation of the results due to the small number of potentially biased trials, and emphasized the need for further high quality trials with longer-term follow-up.

Several factors may contribute to the mixed results of previous black tea and cholesterol trials, such as varying study duration, strength and brewing method of tea, average habitual tea consumption, and differences in participants’ dietary habits. Prior studies, with the exception of one clinical trial, have not rigorously controlled all of these factors.³⁶ The trial presented herein addresses prior deficits in the black tea and serum lipid literature by examining the effect of black tea beverage intake (5 cups/day, 700 mg tea solids/cup) on serum lipid concentrations using a standardized tea treatment and appropriate and consistent placebo in the presence of a completely controlled diet.

METHODS

Participants

Between April 2002 and April 2004, 1500 individuals were recruited via TV, radio, and print advertisements in Minneapolis and St. Paul, MN. They were screened via telephone for eligibility prior to secondary screening at the University of Minnesota General Clinical Research Center (GCRC, currently the Clinical and Translational Science Institute), where a screening blood draw occurred. Basic eligibility criteria included age 45–65 years, borderline hypercholesterolemia (total cholesterol concentrations 190–260 mg/dl or 4.9–6.7 mmol/L), 35–65 mg/dl (0.9–1.6 mmol/L) of high-density lipoprotein cholesterol (HDL-C), and triglyceride (TG) concentrations <600 mg/dl (6.8 mmol/L). A slightly wider range than standard borderline hypercholesterolemia was utilized.⁴¹ See on-line supplemental materials for additional inclusion/exclusion criteria.

Of the 1500 people initially screened, 400 (26.7%) were eligible and provided consent for secondary screening at the GCRC. Of these 400, 57 (14.3%) volunteers met all entry criteria, consented to participate, and were enrolled. Informed consent was obtained via a structured interview with the study coordinator prior to both screening and enrollment. The University of Minnesota, Twin Cities Institutional Review Board approved this study. The trial is registered at ClinicalTrials.gov (NCT01882283).

Study Design

At the end of a one-week run-in period, 57 participants were block-randomized by sex using a computer-generated list to initial consumption of either 5 cups per day of black tea or placebo (tea-like) beverage. Block randomization was used to equally distribute potential differences within sexes, e.g. metabolism, across randomization arms. Participants switched treatment assignments at the beginning of the second treatment period. Serum lipid values were measured at baseline, and the beginning and end of each treatment period. Research team members, participants and analysts remained blinded until the final analysis was completed. Only the study statistician had access to randomization codes.

Intervention

The intervention consisted of 5 cups per day of black tea or a tea-like placebo for 4 weeks, plus a provided, low-flavonoid diet. The trial included two 41-day treatment periods and a 3-week washout period, which occurred between the treatment periods. Each treatment period was comprised of 13 days of run-in time and 28 days of tea or placebo treatment. All participants consumed tea-like placebo during run-in periods. The isocaloric study diet was consumed throughout run-in and treatment periods, which allowed for a flavonoid washout from the self-selected diet consumed during the washout period. An overview of the entire study period and biological sample collection are described in Figure 1.

Figure 1.

Crossover study design overview, including timing of biological sample collection points, in a study of the effect of 5 cups per day of black tea on serum cholesterol concentrations (n=57)

Black tea was selected for this intervention because it is the most commonly consumed form of tea in U.S. The selection of five cups of tea per day was based on 1) its therapeutic potential, and 2), that if therapeutic, this volume could be reasonably incorporated into a patient’s daily routine. Significant effects on plasma cholesterol concentration, platelet aggregation, brachial artery reactivity, and oxidative damage, given response times, could occur within a one-month period of tea intake, hence the 28-day treatment period.^3,42–45 The washout period duration allowed for re-equilibration of parent catechin and catechin metabolite concentrations to self-selected diet concentrations following tea intake.^46–50

Tea Treatment and Placebo

Black tea and tea-like preparations arrived in identical individual serving packets from the Lipton Tea Company (currently Unilever-Best Foods NA), and were coded with randomly generated numbers that linked their content to the treatment assignment in a blinded fashion. They were pre-brewed, and matched in terms of color and taste, caffeine, aspartame, malic acid and fruit flavor content (On-line Supplemental Table 1). The flavonoid composition of the black tea treatment is described in the On-line Supplemental Table 2.

Participants were given specific treatment preparation instructions, and the tea/placebo was prepared for drinking and consumed off-site. If desired, the addition of sweetener to the beverage was allowed, but not milk. Participants were encouraged to consume the treatment at equal intervals throughout the day, but no mandatory times were specified. No specific limit was set on the amount of water that could be added to the preparation or the cup size.

Controlled Low-Flavonoid Diet

Participants were to consume the entire controlled low-flavonoid diet and daily water-soluble vitamin supplements (100% RDA), which were provided throughout each treatment period. Due to the low naturally occurring vitamin content of the controlled diet, vitamin supplementation was given for ethical reasons. With the exception of calorie-free and decaffeinated soft drinks, consumption of non-study provided foods was not permitted. All pain relievers (besides acetaminophen) were restricted.

The Nutrition Data System for Research Software for commercial entrees (Version 4.04_32, 2001, Nutrition Coordinating Center, University of Minnesota, Minneapolis, MN) was used to determine the food composition of the study diet entrees, and to identify entree components that may contain flavonoids. Entrees with a possible significant flavonoid content were not used. Prior to the study, the nutritional composition of the entire study diet was calculated using Nutritionist V nutrient analysis software (Version 2.3, 2000, First Data Bank, San Bruno, CA). The macronutrient composition of the study diet was 15% protein, 51% carbohydrate, and 34% fat. The ratio of saturated, monounsaturated, and polyunsaturated fatty acids was 1:1:0.6 to meet usual intake values as reported in the National Health and Nutrition Examination Survey III.⁵³

Nutritional composition was later re-analyzed using the Grand Forks Research Analysis of Nutrient Data software (Release 24, 2011, Grand Forks Human Nutrition Research Center, Agricultural Research Service, U.S. Department of Agriculture, Grand Forks, ND), to determine the additional macronutrient values reported here.^51,52 The diet contained 12.8 g saturated fat, 14.1 g monounsaturated fat, 7.8 g polyunsaturated fat and 6.3 g fiber per 1000 kcals. The diet, which contained no fruits and very small amounts of vegetables, was analyzed for multiple flavonoids. No flavonoids were detected. A detailed description of this analysis is found in the on-line supplemental materials.

Participants were assigned energy intake estimates congruent with weight maintenance. The study diet was comprised of a 5-day menu rotation, and calculated at five different energy levels: 2000, 2500, 3000, 3500, and 4000 kcal. Energy intake estimates were calculated using Harris Benedict Equation, multiplied by an estimated activity factor.^54–56 The activity factors, ranging from 1.4-sedentary to 2.0-very active, were estimated using subjects’ self-reported physical activity levels.⁵⁶ A unit-muffin of 100 kcal, which matched the diet’s exact macronutrient composition, was used to ensure weight maintenance. The results of daily weighs informed the need for energy intake modification.

Measures

Participants visited the GCRC Monday through Friday throughout the study for daily weighs, and to collect meals. Once per week, GCRC nurses measured each subject’s weight, blood pressure, and pulse, and completed blood draws and urine collections. During these nurse evaluations, participants were interviewed to evaluate pharmaceutical usage, dietary intake, and motivation to maintain study requirements. The number of tea packets used was also counted to evaluate treatment compliance.

All compliance to treatment, provided diet, non-study beverage restriction, non-study food consumption, multivitamin intake, and pharmaceutical use was self-reported on daily compliance records, either at GCRC (weekdays) or at home (weekend days). Participants were asked to describe the type of deviation and its magnitude, e.g. mg of restricted medication. No biologic analyses were conducted to verify compliance.

Biological Samples

Morning measurements from Day 1 of the first tea/placebo treatment period established baseline data for weight, blood pressure, and serum lipids. Thereafter, fasting serum samples were collected on days 7, 14, 21 and 28 of each treatment period. Blood was collected in 2.5 ml Serum Separation Tubes (Becton Dickinson, Franklin Lanes, NJ), and the resulting serum sample was analyzed for TC, HDL-C, and TG at the University of Minnesota Medical Center, Fairview Collaborative Studies Clinical Laboratory (Minneapolis, MN), using the COBAS FARA instrument (Roche Analytical Instruments, Inc. Nutley, NJ). Serum total cholesterol, HDL-C and TG were measured enzymatically.⁵⁷ HDL-C was determined after precipitation of LDL-containing lipoproteins with dextran sulfate/magnesium.⁵⁸ LDL-C was calculated in participants with TG concentrations <400 mg/dl (<4.5 mmol/L) by the Friedewald equation.⁵⁹

Analyses

At an α level of 0.05, the trial had 95% power to detect differences of 6.69 mg/dl of TC (0.2 mmol/L), 4.57 mg/dl of LDL-C (0.1 mmol/L), 1.37 mg/dl of HDL-C (0.04 mmol/L), and 22.79 mg/dl of TG (0.3 mmol/L) between treatment groups. These values are similar to, but more conservative than, what was found in Davies et al trial.³⁶ Equivalence of mean baseline characteristic values was analyzed using one-way ANOVA to ensure that important covariates were equally distributed across randomization groups. A paired t-test was used to evaluate treatment carry-over effects between baseline and Day 1 lipid concentrations.

Differences in lipid concentrations across treatment groups were calculated using the Day 28 serum measurement taken at the end of each treatment period. The effect of tea consumption on each lipid concentration was analyzed as a repeated measures regression using the PROC MIXED procedure in SAS (Version 8.2, released 2001, SAS Institute, Inc., Cary, NC). Analyses were adjusted for treatment period via the inclusion of a period covariate. Treatment-period interactions were also tested. All data analyses followed intention-to-treat (ITT) principles. Results were considered statistically significant at an alpha level of 0.05.

RESULTS AND DISCUSSION

Participants and Compliance

The study sample consisted of 32 men and 25 women, who had a mean age of 52.4 years (range 45–65). Baseline characteristics of participants by treatment-order assignment are summarized in Table 1. Thirty participants were randomized to consume black tea treatment and 27 were randomized to placebo in the first treatment period. No statistically significant difference was observed in the baseline characteristics across treatment groups, except for HDL-C concentration, which was slightly lower among participants starting the tea treatment first (p=0.04).

Table 1.

Baseline characteristics by primary treatment assignment of participants in a study to examine the effect of 5 cups per day of black tea on serum cholesterol concentrations (n=57)

Characteristic	Randomization Assignment
	Black Tea First (n=30) Mean (SD)	Placebo First (n=27) Mean (SD)
% Female	46.7	40.7
Age (years)	51.7 (5.1)	53.2 (4.9)
BMI	31.4 (5.8)	30.0 (6.1)
Systolic Blood Pressure (mm Hg)	130.1 (15.2)	129.2 (12.0)
Total Cholesterol (mg/dl)	207.2a (27.9)	206.7 (22.8)
HDL-C (mg/dl)*	38.6a (8.3)	43.3 (9.2)
LDL-C (mg/dl)	128.0b (20.0)	133.6c (25.9)
VLDL-C (mg/dl)	44.5a (34.9)	34.8 (35.7)
TG (mg/dl)	222.7a (174.4)	174.1 (178.5)
AST (U/L)d	22.6a (7.4)	24.9 (7.7)
Glucose (mg/dl)d	100.9a (15.1)	99.1 (8.0)
Creatinine (mg/dl)d	0.9a (0.2)	1.0 (0.1)
Tea (cups per day)d	1.3 (1.5)	1.0 (1.4)

^*p-value<0.05

^aSerum TC, HDL-C, VLDL-C, TG, AST, glucose, and creatinine value not available for 1 participant

^bLDL-C value calculated using Friedewald equation. Cannot be estimated for 5 participants with >400 mg/dl TG (>4.5 mmol/L)

^cLDL-C calculated using Friedewald equation. Cannot be estimated for 2 participants with >400 mg/dl TG (>4.5 mmol/L)

^dValues from eligibility visit.

Post-randomization, two participants left the study for reasons unrelated to treatment during the first treatment period. One participant moved out of state, and another demonstrated willful non-participation in the study. No significant changes in body weight occurred during the study period. No compliance violations were found regarding drug intake, with the exception of one subject who took aspirin on two occasions. Compliance rates on all other indicators were above 96% (On-line Supplemental Table 3).

Blood Lipids

No period effects or treatment-period interactions were detected. At the end of the study period, mean differences in absolute values were +1.64% for total cholesterol (mean difference 3.43; 95%CI: −7.08, 13.94; p-value=0.2), −0.77% for low-density lipoprotein cholesterol (LDL-C) (mean difference −1.02; 95% CI: −11.34, 9.30; p-value=0.7), +1.40% for high-density lipoprotein cholesterol (HDL-C) (mean difference 0.58; 95%CI: −2.98, 4.14; p-value=0.3), +7.81% for triglycerides (mean difference 15.22; 95%CI: −40.91, 71.35; p-value=0.1), −0.22% for LDL plus HDL cholesterol fraction (mean difference −0.39 ; 95% CI: −11.16, 10.38; p-value=0.9) in those treated with tea versus placebo. There was a −3.08% difference in mean LDL-C/HDL-C ratio values across treatment periods (mean difference −0.1; 95%CI −0.41, 0.21; p-value=0.2). Mean absolute values (± standard error) across treatment groups, mean differences, and p-values for all lipid categories are shown in Table 2. The trial results indicated that consumption of 5 cups of tea per day for four weeks did not significantly change TC, LDL-C plus HDL-C, LDL-C, HDL-C, or TG concentrations.

Table 2.

Fasting cholesterol concentrations at the end of treatment period in a randomized cross-over study to examine the effect of 5 cups per day of black tea on serum cholesterol concentrations (n=57)

Serum Lipid Type (mean±SE)	Tea (n=57)	Placebo (n=57)	Mean difference (95% CI)	% Mean Difference	p-value
Total cholesterol (mg/dl)	212.30±3.76	208.87±3.74	3.43 (−7.08, 13.94)	1.64	0.20
LCL-C (mg/dl)	130.80±3.43	131.82±3.40	−1.02 (−11.34, 9.30)	−0.77	0.70
HDL-C (mg/dl)	41.90±1.27	41.32±1.27	0.58 (−2.98, 4.14)	1.40	0.30
Triglycerides (mg/dl)	210.00±20.05	194.78±20.01	15.22 (−40.91, 71.35)	7.81	0.10
LDL-C + HDL-C (mg/dl)	173.68±3.86	174.07±3.83	−0.39 (−11.16, 10.38)	−0.22	0.90
LDL-C/HDL-C ratio	3.15±0.11	3.25±0.11	−0.1 (−0.41, 0.21)	−3.08	0.20

The current study is the only clinical trial to date to isolate the effect of consuming the polyphenolic flavonoids found in black tea beverage on cholesterol concentrations, while strictly controlling dietary intake. This trial excluded current smokers, included stringent inclusion criteria, controlled caffeinated beverage and pharmaceutical use, and used participants as their own controls to maximize its internal validity. Critically, this study compared black tea beverage to an appropriate placebo (matched to tea treatment with the exception of flavonoids) according to a pre-specified protocol.

The results of this study contradict the majority of observational research and animal trials, and three clinical trials, one of which was also diet-controlled.^{9,10,17,19–21,23–27,33,36,39,60} The strong cholesterol lowering effects of black tea found in animal and cell culture studies may be due to the high doses (mg/kg) used in those studies versus most human research. Seven observational studies (six cross-sectional and one case-control) have reported that black tea consumption is correlated with lower total cholesterol in humans.^{23–27,31,32} However, other observational studies (three cross-sectional and one prospective) found no association between tea drinking and total cholesterol.^22,28–30 Tea consumption may be a surrogate marker for lifestyle factors that could serve as confounders, e.g. exercise habits, smoking patterns, lower coffee consumption, or dietary differences. Incomplete adjustment for these factors, differences of consumed tea quantity, and/or differences in types (e.g. blended, decaffeinated, instant) and preparation (e.g. brewing time, amount used, brewing temperature) of tea consumed may explain the inconsistent results of observational studies.

Results of the current study are in agreement with four other clinical trials.^34,35,37,38 While the direction of the results was similar, these trials differed in methodology from the current trial (treatment type, duration, target population). Some had methodological weaknesses that may have had an important effect on their results, e.g. allowing participants to consume variable amounts of tea and coffee, small sample size.^35,37 Two other randomized clinical trials, which also did not control background diet, found that black tea had hypocholesterolemic effects.^33,39 These two trials differed methodologically from the present study in multiple ways. Neither study compared within-participant serum lipid values. Bahorun et al. compared black tea intake to a hot water placebo in healthy adults. The study experienced differential loss to follow-up, totaling approximately 12% of their study population, and no ITT analysis was reported. Fujita et al. compared the effects of black tea extract capsules (333 mg x 3/day) to a placebo for a three-month period, and also measured serum lipids at 1 month post-intervention. These results indicated an 8.5% decrease in TC, a 8.4% decrease in TG and 11.8% decrease in LDL-C after 3 months, which were all statistically significant at p-value of 0.01. The difference in results from Fujita et al. and this study may be driven primarily by treatment type (extract capsules from fermented fenhai broad-leaf tea vs. powdered tea drink), due to the highly concentrated nature of the extract capsules. Also, analyses compared treatment groups to changes over time within each group, and did not present the results of comparisons between groups.

The current trial most resembles the randomized double-blind crossover trial conducted by Davies et al., which provided a standardized placebo and a controlled diet to participants. The study included similar treatment duration, tea dosages, and population as the current trial. The work by Davies et al., however, differed from the current trial in that they provided the National Cholesterol Education Program Step I-type diet and, initially, a placebo without caffeine. Furthermore, the study added a post hoc treatment arm with a caffeinated placebo. Their results showed that 15 mildly hypercholesterolemic adults experienced declines in TC (3.8%; p=0.06, trend) and LDL-C (7.5%; p=0.01) concentrations after consumption of black tea compared to placebo without caffeine. The comparison of tea consumption to a caffeinated placebo during a post hoc treatment period resulted in differences of 6.5% TC (p=<0.001) and 11.1% LDL-C (p=0.002) across treatment groups, larger differences than found when comparing tea to the non-caffeinated placebo. These results contradict the results of multiple clinical trials published on this topic to date, including the current trial. The incorporation of a post hoc third treatment period with 12 of 15 original trial participants, during which all participants consumed a caffeinated placebo drink, may have contributed to the large effects observed in this trial.³⁶

Participants in the current study served as their own controls, which reduced error variance and the potential for observation bias. Participants were highly compliant with study procedures, and there was minimal loss-to-follow-up. We estimated that the respective lengths of the treatment and washout periods were appropriate to observe the effect of interest and eliminate carry over effects of treatment due to the short time required for re-equilibration of the parent catechin and catechin metabolite concentrations after a high intake of tea.^{42,46–48,50} Conversely, lipophilic metabolites of catechins may need a longer period of time for re-equilibration, but no carry over effect was seen as lipid concentrations returned to baseline during the washout period. The use of absolute cholesterol concentrations in the primary outcome analyses, versus measures of difference, was appropriate, as cholesterol concentration equilibration occurs within two weeks.⁴²

Limitations of the present trial include an inability to test for dose response relationships between tea and lipid concentrations, and to evaluate the effects of tea consumption against different background diets. Also, while study staff carefully monitored participant compliance via multiple mechanisms, no biological analyses were conducted to validate self-reported compliance, a potential weakness shared with most other trials on this subject.

CONCLUSIONS

In summary, the intake of 5 cups of tea in combination with a low-flavonoid typical American diet did not significantly alter the lipid profile of borderline hypercholesterolemic participants. While tea may have very small beneficial effects on lipid profiles, this trial indicates that there is little therapeutic benefit to using black tea beverage to prevent hypercholesterolemia progression. Further research is needed to explore other tea-consumption-related mechanisms of potential importance in cardiovascular disease etiology, e.g. improvement of endothelial function, inflammation reduction, platelet hyperactivity, and oxidative damage.