Cholesterol 4

7/29/2019 Cholesterol 4

1/5

ELSEVIER

Green Tea Intake in Relation to Serum Lipid Levels in Middle-AgedJapanese Men and WomenYOSHITAKA TSUBONO, MD AND SHOICHIRO TSUGANE, MD

PURPOSE: The relationship between green tea intake and serum concentrations of total cholesterol,triglycerides, and high density lipoprotein cholesterol was examined.METHODS: The subjects were 630 middle-aged men and their 370 wives sampled from five regionsof Japan during 1989-1991. Consumption frequency of 38 foods, including green tea, was determinedby interview. Three-day food records were collected from 207 of the men and 164 of the wives. Themean serum concentrations of the three lipids were compared according to the three levels of dailygreen tea intake (< 1 cup, l-4 cups, and > 4 cups), with adjustments for various health habits, foodfrequencies, and nutrient intakes.RESULTS: After extensive multivariate adjustmentsor nondietary and dietary covariates, green teawas not associated with any of the three lipid levels.CONCLUSIONS: The results of this cross-sectional study do not support the beneficial effects ofgreen tea on serum lipid levels.Ann [email protected] 1997; 7:280-284. 0 199 7 by Elsevier Science Inc.KEY WORDS: Cholesterol, High-Density Lipoprotein Cholesterol, Tea, Triglycerides, Cross-SectionalStudy.

INTRODUCTIONThere has been a growing interest in the hypothesis thatgreen tea has a preventive effect against cardiovascular dis-eases. Studies in laboratory animals have suggested thatadministration of green tea extracts suppresses he absorp-tion of cholesterol from the digestive tract (1 ,2) and reducesthe serum level of total choles terol (2-4). Two cross-sec-tional epidemiologic studies of Japanese men have addressedthis issue. Imai and colleagues (5) reported a significantinverse relationship between green tea consumption andserum levels of total cholesterol and triglycerides, as well as apositive association between green tea drinking and high-density lipoprotein (HDL) cholesterol levels. Kono and col-leagues (6) reported a nonsignificant inverse relationshipbetween consumption of green tea and total cholesterol andno association with triglyceride or HDL-cholesterol levels.However, these studies did not control thoroughly for poten-tial confounding by other d ietary fac tors, and no data havebeen available for women. We therefore analyzed cross-sectional data collected from a population-based sample ofmiddle-aged Japanese men and women in order to examine

From the Epidemiology and Biostatistics Division, National CancerCenter Research Institute,. East, Kashiwa, Japan (Y.T., ST.); and Depart-ment of Public Health, Tohoku University School of Medicine, Sendai,Japan (Y.T.).Address reprint requests to: Shoichiro Tsugane, M.D., Epidemiologyand Biostatistics Division, National Cancer Center Research Institu te,East, 6-5-l Kashiwanoha, Kashiwa 277, Japan.

Received May 21, 1996; accepted Janua ry 28, 1997.

the association between green tea intake and serum lipidlevels, with extensive controls for other health habits anddietary variables.

MATERIALS AND METHODSThe subjects were participants in the Japan Ecological Can-cer Study, a cross-sectional study conducted in five regionsof Japan that have different mortality rates from gastriccancer. Details of the study have been described elsewhere(7, 8). Briefly, in the winters of 1989-1991, 880 men aged40-49 years were randomly sampled from the general popu-lations living in five Public Health Center districts; 634 ofthe men (72%) responded. The study also included 373wives of some of these men (aged 28-52 years).The subjec ts were invited to the local community centersto provide blood samples after fasting for at least 5 hours.After whole blood samples were left at room temperaturefor 1 hour to facilitate clotting, the serum was separated bycentrifugation, frozen within 30 min of collection, andstored at -80C until assay. The concentrations of totalcholesterol (9), triglycerides (10) and HDL cholesterol (11)were measured by enzymatic methods at a commercial labo-ratory (Special Reference Laboratory, Tokyo) with an auto-analyzer (model 736, Hitachi, Tokyo). Within-run coeffi-cients of variation for the measurement of total cholesterol,triglycerides, and HDL cholesterol were 0.6%, 0.9%, and1.2%, respectively, and between-run coefficients of variationwere 0.7%, 1.7%, and OS%, respectively.

0 1997 by Elsevier Science Inc.655 Avenue of the Americas, New York, NY 10010 1047s2797/97/$17.00PII SlO47-2797(97)00005-7


2/5

Selected Abbreviations and AcronymsHIlL = high cienrity lipoproteinsRMI = ldy nxw in&*.x

On the same day as blood samples were collected, traineddietitians or public health nurses interviewed each subjectabout their consumption frequency of 38 food items (includ-ing green tea), their level of alcohol consumption, cigarettesmoking, leisure time physical activity, and education. Thefoods selected for inclusion in the questionnaire representedthe general patterns of diets in the country as closely aspossible considering the limited number o f questionnaireitems. The subjects were asked the average number of daysper week during the past month they consumed each fooditem. Four frequency categories ranged from less than oncea week to almost every day. For green tea and several otherbeverages, subjects who consumed them almost every daywere also asked the average number of cups drunk per day.Information on portion size was not specified o r sought forother f&d items; thus nutrient intakes were not estimatedfrom the responses on food frequencies. Each subjectsweight and height were measured, and body mass index(BMI) was computed as weight divided by the square of height(kg/m). After exclusion of five subjects with incompleteresponses and two subjects who indicated the use of choles-terol-lowering medications, 1000 subjects remained for anal-ysis (630 men and 370 women).

Within 2 weeks after blood drawing and interview byquestionnaire, records of weighed food intake over threeconsecutive weekdays were collected from a subgroup of thesubjects who volunteered (207 men and 164 women). Therecords were obtained by a method used in the NationalNutrition Survey (IZ), with some modifications. Researchdietitian5 instrucred the subjects to record all foods andbeverages prepared and consumed in a specially designedhooklet. Levels of daily nutrient consumption were com-puted from these records using the Standard Food Composi-tion Tables published by the Science and TechnologyAgency of Japan (13). Pearson correlations between foodrecord measurements and questionnaire estimates for theconsumption of meat, fish, vegetables, and fruit were 0.23,d.24, 0.12, 0.40, respectively.

Intake of green tea was trichotomized as < 1 cup per day,I -4 cups per day, and > 4 cups per day in order to representthe tertiles as closely as possible. The mean levels of threeserum lIpids were compared according to the categories ofgreen tea hy analysis of covariance using the GLM Procedurem the SAS program package (14).

Three sets of covariates, namely, health habits, foodfrequencies, and nutrients were controlled for as potentialconfc,unders both separately and in combination. Healthhabits included BMI, weekly alcohol intake (contin-

uous variables), smoking (current smoktr/7:o1Ismoker), ler-sure time physical activity (one or more times per month/less often), education (lower than high s&oi/high schoolor higher), consumption of oolong rea XX! coffee (numberof cups per day), age in years, and five \tudy districts (usingfour indicator variahles). Food frequencies referred to thecalculated monthly consumption frequencie\ of six majorfood sources of animal lipids as continuoils v:u-lahles {beef,pork, chicken, bacon/ham/sausages, eggs, and milk/otherdairy products). Nutrients included the dailv Intake of totalenergy, saturated fatty acids, polyunsaturatcii fatty acids andtotal cho lesterol as continuous variables, I.\ uch were avail-able from rhe subjects who provided t hc t;,X~d reco& (207men and 164 women).

The covariates were selected a priori .JI,J ~verc grL)upettinto the three categories above fL)r the Ioi owrn;: l-c;ison.Variables in the health habits category wee: ielccted hecausethey have been cc>ntrolled for in ml 1st ot tl:e ~&~vation;~tstudies of diet and health outcomes. StutTv districts wereincluded in order to adjust for the region,J differences indiet and other varlahtes. Consumption frequ~n r;sther that1in the f&td-frequency category, because ttht: I~O beveragesin themselves cwld have health effects 1.~ c~;u&c \*aaculardiseases and should be included in both ;m,llyies Jjustingfor other food frequencies and nutrients. Io~.\d frequenciesand nutrients were considered separately lx~:;-tusv he latterwere available only for the subset of rhc .t~t>lv~ .t\ ctho pro-vided the ?-day food records.

The levels o f serum lipids were cmpiu.eJ at tirst withoutcontrolling for the covariates, then with adlu~tment forhealth habits only, fo r health habits and !(~od frequencies,and tinally for health habits and nutrients. ,211 analyses wereconducted sepamtely for rnen and women, Ttls rrlglycerideconcentrations were lc,g-transforrlle~l ;I: k..i the fc~ometriimeans arc presented.

_____ --.--.____-RESULTSTable 1 presents descriptive characteriqtic% (tt thr subjects.On average, the men and women consulted 3.5 and 3.1cups of green tea per day, respectively, and 69% of the menand 68% of the women drank one or mL cholesterolaccording to the three categories of grecli tea intake. F


3/5

282 Tsubono and Tsugane AEP Vol. 7, No. 4GREEN TEA AND SERUM LIPID LEVELS iMay 1997:280-284

TABLE 1. Descriptive characteristics of 630 men and 370women participating in the Japan Ecological Cancer Study,1989-1991

Men WomenMeasurement Mean (SD) Mean (SD)No. of subjectsAge (years)BMI (kg/m)Alcohol intake (g/wk)Cigarette smoking (% current

smokers)Education (% high school

or higher)Leisure time physicalactivity (% 2 doing 1 times/mo)Green tea intake (cups/d)

Serum lipid levels (mg/dL)Total cholesterolTriglyceridesHDL-cholesterol

630 37044.4 (3.0) 41.4 (4.2)23.9 (2.9) 23.3 (3.2)178 (182) 12 (41)

67%

47%3.5197.7 (34.5) 184.6 (33.6)119.2 (1.8) 74.4 (1.5)49.6 (12.7) 55.0 (11.9)

8%72%

31%(3.9) 3.1 (3.0)

a SD = standard deviation

and both total cholesterol (P for trend, 0 .018) and HDLcholesterol (P for trend, O .OOl), but no association withtriglycerides. The trends for total cholesterol and HDL cho-lesterol diminished after controlling for the covariates.

All analyses were repeated after dividing the upper cate-gory of green tea intake into two levels (5-8 and 2 9cups per day) in order to examine the effect of very highconsumption, but the results did not change materially.Analyses based on the quintiles of green tea intake revealedessentially the sam e findings. The findings did not differappreciably in the analyses with further adjustment for in-take frequencies o f green and yellow vegetables and fruits,or consumption of carotene, vitamins C and E, or in analysescombining the frequencies of oolong tea and green tea.When the responders and nonresponders to the 3-day foodrecords were analyzed separate ly, total cholesterol showedmarginally significant inverse association with green teaamong male nonresponders (P for trend, 0.040), but thetrend was not observed in male responders (P for trend,0.612), female responders (P for trend, 0.222) or femalenonresponders (P for trend, 0.676). The resu lts with theother two lipids were comparable for the responders andnonresponders and were basically the sam e as the findingsobserved in all sub jects. The subjec ts with a history of isch-emit heart diesase tended to have a larger consumption ofgreen tea than did other subjects (4.6 vs 3.4 cups per day),but the small number of such subjects (n = 11) precludedany meaningful inference.

DISCUSSIONWhen the health effects of green tea are examined by obser,vational studies, potential confounding and effect modificaption by other d ietary factors need to be controlled for. Wehave previously found, using the same data set, that greentea intake is associated with the consumption of foods andnutrients that could modify the risks of cardiovascular dis,eases and cancer, especially in men (Y. Tsubono, unpub,lished manuscript). For instance, the men with higher in takeof green tea tended to have larger consumption of green,yellow, and pickled vegetables, and smaller consumption oforange juice, total lipids, cholesterol, and carotene.Previous studies of green tea intake and serum lipids didnot control thoroughly for the influence of other dietaryvariables in the analyses. For instance, Imai and colleagues(5) adjusted the data only for age, smoking, alcohol con-sumption, and relative body weight, but they did not controlfor any foods or nutrients in the study that reported a signifi-cant inverse association between green tea intake and totalcholesterol and triglycerides, as well as a positive associationbetween green tea and HDL cholesterol. Kono and col-leagues (6) reported a significant inverse association withtotal cholestero l when no covariates were controlled for (Pfor trend, O.Ol), and when adjustments were made onlyfor nondietary variables including the subjects rank (Self-Defense Forces officials), smoking, alcohol intake, physicalactivity, and BMI (P for trend, 0.03). The relation wasweakened and fell short of statistical significance (P fortrend, 0.09) after further adjustment for the intake frequen-cies of four foods (rice, soybean-paste soup, milk, and breadfor breakfast) as proxies of adherence to the Japanese andWestern diets.

We observed in the crude analysis that green tea intakewas associa ted inversely with triglycerides and positivelywith HDL cholesterol in men, and was related positivelywith total and HDL cholesterol in women. However, thesetrends disappeared after adjustments for dietary and nondi-etary covariates. Our results indicate that uncontrolled con-founding by the consumption of other dietary componentscan seriously distort the associations. Although green teashowed an inverse association of borderline significancewith total cho lesterol in male nonresponders to the foodrecords, this association appears to be a chance finding sincethe trend was not replicated in ma le responders, femaleresponders, or nonresponders. We therefore conclude that,unlike the findings of other cross-sectional studies (5, 6),green tea intake was not associated with serum lipid profilein this study.

Our results do not conflict with a potential role for greentea in preventing cardiovascular disease through mecha-nisms other than a direct effect on serum lipid levels. Greentea is rich in known antioxidants such as vitamin C (13)and polyphenols (15) and is consumed frequently in some


4/5

TABLE 2. The mean serum levels of lipids (mg/dL) according to consumption of green tea among 63 0 men and ?;C womenparticipating in the Japan Ecological Cancer Study, 1989-1991 -__.-II______Men Women -.---___Total HDL Total kiDI.

Mean levels n cholesterol Triglycerides cholesterol n cholesterol Trlplyccriiics ch&ster


5/5

284 Tsubono and TsugmeC;REEN TEA AND SERUM LIPID LEVELS

10. Spayd RW, Bruschi B, Burdick BA, et al. Multilayer film elementsfor clinical analysis: Applications to representative chemical determ i-nants. Clin Chem. 1978;24:1343-1350.11. Lopes-Virella MF, Stone P, Ellis S, Colwell JA. Cholesterol determina-tion in high-density lipoprotems separated by three different methods.Clin Chem. 1977;23:882-884.12. Ministry of Health and Welfare. Report of the National NutritionSurvey in 1992 (in Japanese). Tokyo: Daiichi-Shuppan; 1994.

AEI Col. 7. No. 4May 1997: 280-284

13. Science and Technology Agency. Srandard Tables of Food Compost-tton in Japan (in Japanese). Fourth revised edition. Tokyo: PrintingBureau, Ministry of Finance; 1 982.14. SAS Institute, Inc. SAS/STATR Users Guide, version 6.4. Vol 2.Gary, NC: SAS Institute, Inc; 1989.15. Yang CS, Wang Z. Tea and cancer. J Nat1 Cancer Inst. 1993;85:1038-1049.

Cholesterol 4

Documents

Transcript of Cholesterol 4