Protective effect of green tea against prostate cancer: A case-control study in southeast China
To investigate whether green tea consumption has an etiological association with prostate cancer, a case-control study was conducted in Hangzhou, southeast China during 2001–2002. The cases were 130 incident patients with histologically confirmed adenocarcinoma of the prostate. The controls were 274 hospital inpatients without prostate cancer or any other malignant diseases, and matched to the age of cases. Information on duration, quantity and frequency of usual tea consumption, as well as the number of new batches brewed per day, were collected by face-to-face interview using a structured questionnaire. The risk of prostate cancer for tea consumption was assessed using multivariate logistic regression adjusting for age, locality, education, income, body mass index, physical activity, alcohol consumption, tobacco smoking, total fat intake, marital status, age at marriage, number of children, history of vasectomy and family history of prostate cancer. Among the cases, 55.4% were tea drinkers compared to 79.9% for the controls. Almost all the tea consumed was green tea. The prostate cancer risk declined with increasing frequency, duration and quantity of green tea consumption. The adjusted odds ratio (OR), relative to non-tea drinkers, were 0.28 (95% CI = 0.17–0.47) for tea drinking, 0.12 (95% CI = 0.06–0.26) for drinking tea over 40 years, 0.09 (95% CI = 0.04–0.21) for those consuming more than 1.5 kg of tea leaves yearly, and 0.27 (95% CI = 0.15–0.48) for those drinking more than 3 cups (1 litre) daily. The dose response relationships were also significant, suggesting that green tea is protective against prostate cancer. © 2003 Wiley-Liss, Inc.
Prostate cancer accounts for about 200,000 deaths annually worldwide. It is the most commonly diagnosed and the second leading cause of male cancer deaths in western countries. The incidence rate of prostate cancer per 100,000 is 104.33 in the United States, 75.97 in Australia but only 1.74 in China.1, 2 Such large differences between countries suggest that dietary, lifestyle and environmental factors may be influential in the etiology of prostate cancer.3
About 75% of prostate cancer cases worldwide occur in men 65 years of age or older, and both incidence and mortality rates tend to increase with age.2, 4, 5 Although there are no dietary risk factors established currently, a high fat intake may increase the prostate cancer risk,2, 6 whereas diets with relatively high levels of phyto-estrogens such as lignins, isoflavones and genistein seem to be protective.7 Hormones (androgens) play an important role too,8, 9 whereas men with family history of prostate cancer are at high-risk of the disease.10, 11 The effects of these variables in low incidence populations, however, are unclear. A history of vasectomy is also linked to the risk of prostate cancer in China, although the evidence is not well established in western countries.12
Because prostate cancer has few symptoms during its initial stages, it is typically diagnosed at an advanced stage in China. Patients often present with symptoms of urinary outflow obstruction or metastatic tumors in bone, the latter often associated with chronic pain. PSA testing has become popular in China for diagnosis, but its use in screening is not widespread. Treatment of prostate cancer is difficult and somewhat controversial.13, 14, 15 Therefore, strategies that could enhance primary prevention are invaluable.
There has been considerable interest in the protective effect of tea, particularly green tea, both in vitro and in vivo.16, 17, 18 In animal studies of cancers of the skin, esophagus, lung, mammary glands and colon, intake of green or black tea as the sole drinking fluid has been shown to lower the incidence, multiplicity, and volume of induced tumors, compared to animals restricted to water as the only beverage. Moreover, epidemiological studies have found inverse association between tea and several lifestyle-related cancers, including cancers of the esophagus, lung, colon, breast, pancreas, stomach and skin. Current research suggests that tea drinking should be part of a healthy diet to lower the incidence of certain cancers.19, 20, 21, 22
China is the second largest producer of tea and has the largest area in the world for tea cultivation.23 Green tea accounts for 75% of the tea production and is the most widely consumed beverage by Chinese residents. Zhejiang Province, located in southeast China, is the home of the famous green tea Long Jing. The annual output of green tea from Zhejiang exceeds 138,000 tones, which amounts to almost 20% of the annual tea production in China.24 Despite the popularity of tea, there has been no epidemiological study undertaken in China documenting its association with prostate cancer risk, although the few findings from North America have been inconsistent.25, 26, 27 Furthermore, the effect of quantity of tea leaves consumed has never been thoroughly investigated, so that the apparent dose-response relationship cannot be firmly established in the literature. The present case-control study, conducted during 2001–02 in Hangzhou, capital of Zhejiang Province, attempts to ascertain the relationship between tea consumption and prostate cancer.
MATERIAL AND METHODS
Study design and participants
A hospital based 1:2 case-control design was used to investigate the association between tea consumption and prostate cancer. Cases were identified by daily searching of all inpatient records, including those at the urology wards of the 8 public hospitals in Hangzhou, between July 2001 and June 2002. Inclusion criteria for cases were defined to be men over 45 years of age with a confirmed histopathological report of adenocarcinoma of the prostate, who had been residents of Zhejiang Province for at least 10 years and were capable of being interviewed (not too ill to participate). To ensure complete ascertainment of cases, all medical records and laboratory pathology reports were reviewed. Of the 143 cases identified during the data collection period, 133 (93%) were interviewed and 10 (7%) refused participation. Three patients were later excluded because their dates of diagnosis were more than 3 years ago. Most of the final 130 cases (84%) were recent patients interviewed within 12 months from diagnosis. The distribution of cases by their cancer stage is: A (12), B (28), C (37) and D (52). The stage classification for 1 patient was missing. All 130 cases were histopathologically confirmed, including 5 asymptomatic cases who were diagnosed initially by PSA test. These 5 cases were subsequently biopsied and their diagnosis confirmed by histology.
During the same period 274 inpatient controls were recruited from the 8 hospitals and interviewed. Men recruited for controls were matched with cases by age and geographical area. Subjects were eligible if they had no prostate cancer or any other diagnosed malignant diseases confirmed by physical examination, X-ray, operation or histopathological reports within the past 12 months. Potential controls with a diagnosis of Alzheimer's disease, a history of stroke, or have been on long-term modifications of diet for medical reasons were excluded. The same exclusion criteria also applied to cases. Subjects with a history of stroke were excluded because the effect of stroke on memory was uncertain. The majority of the eligible controls were recruited from the urology wards (65.3%), orthopedic and injury wards (19.3%), hemorrhoid and rectum wards after their diagnoses were confirmed. Of the 284 eligible controls identified, 274 (96.5%) participated in the study, 8 (2.8%) declined the interview and 2 persons (0.7%) later withdrew for personal reasons.
A structured questionnaire that measured consumption of tea and other beverages was used. Information collected also included demographic characteristics, height and weight, usual diet and physical activity, lifestyle habits such as alcohol drinking and smoking, medical history and family history of prostate cancer, occupation, and factors related to hormonal, marital and reproductive status. A reference recall period was set at the year that was 5 years before diagnosis for cases or 5 years before interview for controls. The questionnaire was modified from that used in our recent ovarian cancer study,28 which in turn had included components from the Hawaii Cancer Research Survey,29 Australian Health Survey 199530 and the United States food survey 1992.31 These 3 questionnaires have been validated in studies of large multi-ethnic populations including Chinese immigrants whom are comparable to our study population of Chinese men.
Ethics and interview
The study was approved by the Human Research Ethics Committee of the researchers' institution and by the Zhejiang hospital administration and doctors-in-charge of the relevant wards. An appointment for face-to-face interview was made after obtaining formal consent from the participants. Confidentiality and anonymity issues were also explained. The first author conducted all interviews, usually in the presence of the participant's next-of-kin to minimize recall bias. Each interview usually took an hour to complete. Histopathological records were obtained from the pathology department or retrieved from inpatient medical records of the eight hospitals.
Measurement of tea exposure
Questions on habitual tea consumption were taken from our previous ovarian cancer study28 and the Arizona skin cancer study.21 Participants were first classified as either “ever” or “never” tea drinkers (<once per month). The ever-drinkers were then asked for information on usual consumption pattern namely types of tea drank, frequency (number of cups consumed per day), number of new batches brewed each day, quantity (kg) of tea leaves consumed per year, quantity (g) of tea leaves used per batch, and years of regular tea drinking. Frequency of liquid tea consumed was measured by a standard container (350 ml cup) during the interview. The use of teapots and small cups in other parts of China is not popular among Zhejiang residents (whose average cup size being 300–400 ml). The common method of preparation is to brew dry tea leaves in a cup using hot water without milk and sugar. Although calendars were used to assist the participants to recall their tea exposure along with other data 5 years ago, they were blinded to the purpose of the study.
The feasibility and validity of the questionnaire used to measure tea exposure was evaluated in a preliminary pilot trial involving 21 Zhejiang residents. The pretest led to a few minor modifications of the initial questionnaire. The internal reliability and reproducibility of the questionnaire were then assessed by a test–retest study of 40 subjects conducted in Hangzhou. Intraclass correlation coefficients were found to be high (0.82–0.95) for tea consumption variables, confirming that it was an appropriate instrument to measure the required habitual information for the target population.
All data were checked for completeness at the end of each interview. The data were coded and analyzed using the SPSS package. To assess potential survival bias, data for the 109 cancer patients (interviewed within one year from diagnosis) and data for all cases (interviewed within 3 years from diagnosis) were analyzed separately. Demographic characteristics and potential risk factors between cases and controls were compared by t-test for continuous variables and chi-square test for categorical variables. To facilitate analysis, tea consumption variables, except “tea drinking (ever vs. never)”, were categorized into 3 or 4 levels according to the distributions of controls; see Table III. Crude and adjusted OR and associated 95% CI of prostate cancer risk for tea consumption variables were obtained from fitting separate unconditional multivariate logistic regression models, with non-tea drinking being the reference category. Each fitted regression equation included terms for adjusting age at interview, locality of residence, education, family income, body mass index (BMI), physical activity level, total fat intake (g/day), alcohol consumption, tobacco smoking, marital status, age at marriage, number of children, family history of prostate cancer and history of vasectomy. These variables were included in the multivariate models because they were either plausible risk factors of prostate cancer from the literature or potential confounders with tea consumption according to the univariate analysis. The overall physical activity was measured in terms of metabolic equivalent tasks (MET), with scores 1.5, 3.0 and 6.0 assigned for sedentary, moderate and vigorous activities respectively to generate weighted estimates of time spent each week in physical activities.32, 33 In estimating alcohol intake, alcohol concentration in beer was set at 5%, whereas wine and rice wine were 12% and liquor 47%, according to reference values recorded in the Chinese food ingredients database.34 Men consuming more than 25 drinks per week were classified as heavy drinkers.35 Each quantitative tea consumption variable was subjected to a linear trend test in prostate cancer risk. Finally, model adequacy was assessed using the Hosmer and Lemeshow goodness-of-fit statistic.
Table III. Crude and Adjusted or and Associated 95% CI of Prostate Cancer Risk for Tea Drinking
|Drink tea|| || || || || || |
| No||58 (44.6)||55 (20.1)|| || || || |
| Yes||72 (55.4)||219 (79.9)||0.31||0.20–0.49||0.28||0.17–0.47|
|Years of tea drinking|| || || || || || |
| 0||58 (44.6)||55 (20.1)||1.00|| ||1.00|| |
| 1–19||22 (16.9)||42 (15.3)||0.50||0.26–0.94||0.50||0.25–1.01|
| 20–40||38 (29.2)||101 (36.9)||0.36||0.21–0.60||0.32||0.18–0.59|
| 40||12 (9.2)||76 (27.7)||0.15||0.07–0.31||0.12||0.06–0.26|
| χ2 for trend|| || || ||38.162|| || |
|Total tea leaves consumed (kg/year)|| || || || || || |
| 0||58 (44.6)||55 (20.1)||1.00|| ||1.00|| |
| <1||32 (24.6)||68 (24.8)||0.45||0.26–0.78||0.43||0.23–0.80|
| 1–1.5||31 (23.8)||79 (28.8)||0.37||0.21–0.65||0.33||0.17–0.61|
| >1.5||9 (6.9)||72 (26.3)||0.12||0.05–0.26||0.09||0.04–0.21|
| χ2 for trend|| || || ||31.852|| || |
|Tea leaves used (g/batch)|| || || || || || |
| 0||58 (44.6)||55 (20.1)||1.00|| ||1.00|| |
| 0.3–1.9||34 (26.2)||71 (25.9)||0.45||0.26–0.79||0.43||0.24–0.79|
| 2–3||23 (17.7)||62 (22.6)||0.35||0.19–0.64||0.28||0.14–0.55|
| >3||15 (11.5)||86 (31.4)||0.17||0.09–0.32||0.14||0.07–0.29|
| χ2 for trend|| || || ||36.802|| || |
|Number of cups/day|| || || || || || |
| <1||63 (48.5)||76 (27.7)||1.00|| ||1.00|| |
| 1–3||39 (30.0)||80 (29.2)||0.59||0.35–0.98||0.53||0.30–0.94|
| >3||28 (21.5)||118 (43.1)||0.29||0.17–0.49||0.27||0.15–0.48|
|χ2 for trend|| || || ||25.212|| || |
|Number of new batches brewed/day|| || || || || || |
| 0||58 (44.6)||55 (20.1)||1.00|| ||1.00|| |
| 1||55 (47.3)||169 (61.7)||0.31||0.19–0.50||0.28||0.17–0.48|
| ≥2||17 (13.1)||50 (18.2)||0.32||0.17–0.63||0.24||0.11–0.51|
|χ2 for trend|| || || ||18.632|| || |
|Quantity of green tea leaves (g/day)|| || || || || || |
| 0||58 (44.6)||55 (20.1)||1.00|| ||1.00|| |
| 0.3–2.9||50 (38.5)||108 (39.4)||0.44||0.27–0.72||0.41||0.24–0.72|
| 3–4.9||13 (10.0)||48 (17.5)||0.26||0.13–0.53||0.24||0.11–0.53|
| ≥5||9 (6.9)||63 (23.0)||0.14||0.06–0.30||0.10||0.04–0.23|
| χ2 for trend|| || || ||36.472|| || |
Table I contrasts the sample characteristics of men with and without prostate cancer for tea drinkers and non-drinkers (<once per month). There were no significant differences between cases and controls in mean age at interview, locality (urban or rural areas), education, family income, alcohol consumption, tobacco smoking, marital status, and history of vasectomy. The two groups were also similar in BMI (cases: mean = 23.4, SD = 3.1; controls: mean = 22.7, SD = 3.1), physical activity MET score (cases: mean = 157.4, SD = 59.2; controls: mean = 156.5, SD = 55.4), total fat intake (cases: mean = 60.3 g/day, SD = 24.6; controls: mean = 56.2 g/day, SD = 20.7) and age at marriage (cases: mean = 26.2 years SD = 3.9; controls: mean = 27.0 years SD = 4.9). Compared to controls, the cases tended to have slightly more children and had apparent family susceptibility.
Table I. Selected Characteristics of Participants With and Without Prostate Cancer
|Age at interview, mean years (SD)||71.7 (7.1)||74.0 (6.9)||70.9 (7.5)||73.7 (5.5)|
|Fat, mean g/day (SD)||62.0 (21.2)||58.1 (28.2)||57.1 (20.2)||52.6 (22.1)|
|MET, mean (SD)||159.4 (57.6)||154.9 (61.7)||155.9 (52.9)||158.7 (64.9)|
|Age at marriage, mean years (SD)||26.4 (3.8)||25.8 (4.0)||27.1 (4.9)||26.6 (5.1)|
|BMI, mean (SD)||23.7 (3.3)||23.1 (2.9)||22.9 (3.2)||22.2 (2.8)|
|Locality of residence, n (%)|| || || || |
| Urban||56 (77.8)||41 (70.7)||168 (76.7)||39 (70.9)|
| Rural||16 (22.2)||17 (29.3)||51 (23.2)||16 (29.1)|
|Education, n (%)|| || || || |
| None||8 (11.1)||10 (17.2)||20 (9.1)||9 (16.4)|
| Primary||19 (26.4)||18 (31.0)||63 (28.8)||15 (27.3)|
| Secondary||28 (38.9)||16 (27.6)||88 (40.2)||22 (40.0)|
| Tertiary||17 (23.6)||14 (24.1)||48 (21.9)||9 (16.4)|
|Family income per month, n (%)|| || || || |
| ≤500 RMB||11 (15.3)||13 (22.4)||40 (18.3)||10 (18.2)|
| 501–1000 RMB||27 (37.5)||28 (48.3)||90 (41.1)||23 (41.8)|
| 1001–2000 RMB||28 (38.9)||15 (25.9)||79 (36.1)||20 (36.4)|
| >2001 RMB||6 (8.3)||2 (3.4)||10 (4.6)||2 (3.6)|
|Alcohol consumption, n (%)|| || || || |
| Never||28 (38.9)||32 (55.2)||104 (47.5)||38 (69.1)|
| Moderate||27 (37.5)||11 (19.0)||65 (29.7)||5 (9.1)|
| Heavy||17 (23.6)||15 (25.9)||50 (22.8)||12 (21.8)|
|Tobacco smoking, n (%)|| || || || |
| Never||21 (29.2)||23 (39.7)||76 (34.7)||27 (49.1)|
| Former||17 (23.6)||3 (5.2)||56 (25.6)||8 (14.5)|
| Current||34 (47.2)||32 (55.2)||87 (39.7)||20 (36.4)|
|Marital status, n (%)|| || || || |
| Married||67 (93.1)||47 (81.0)||195 (89.0)||50 (90.9)|
| Widowed/divorced/separated||5 (6.9)||11 (19.0)||24 (11.0)||5 (9.1)|
|History of vasectomy, n (%)|| || || || |
| No||71 (98.6)||56 (96.6)||212 (96.8)||54 (98.2)|
| Yes||1 (1.4)||2 (3.4)||7 (3.2)||1 (1.8)|
|Prostate cancer in first degree relatives1, n (%)|| || || || |
| No||58 (86.6)||53 (91.4)||195 (89.0)||46 (83.6)|
| Yes||2 (2.8)||1 (1.7)||0||0|
| Unclear||12 (16.7)||4 (6.9)||24 (11.0)||9 (16.4)|
|Number of children1, n (%)|| || || || |
| ≤2||26 (36.1)||13 (22.4)||91 (41.6)||19 (34.5)|
| 3–4||33 (45.8)||27 (46.6)||98 (44.7)||26 (47.3)|
| >4||13 (18.1)||18 (31.0)||30 (13.7)||10 (18.2)|
Among the 130 prostate cancer patients, 55% were classified as tea drinkers compared to 80% of the controls. Table II summarizes the tea consumption patterns for the subset of participants who drank tea. It is evident that the 219 controls had longer duration of tea drinking and drank significantly more tea (in terms of frequency and quantity of tea leaves consumed per year and per batch) than the 72 cases. On average, the controls consumed a larger quantity of green tea leaves per day than the prostate cancer patients. The 2 groups of ever tea drinkers, however, were not significantly different in the types of tea drank and number of new batches brewed each day.
Table II. Tea Consumption Patterns of Tea Drinking Participants1
|Years of tea drinking||30.35 (12.62)||36.68 (14.88)||<0.01|
|Total tea leaves consumed (kg/year)||1.09 (0.98)||1.54 (1.30)||<0.01|
|Tea leaves used (g/batch)||2.27 (1.57)||3.38 (2.86)||<0.001|
|Quantity of green tea leaves (g/day)||2.78 (2.26)||4.13 (3.49)||<0.001|
|Number of cups/day, n (%)|| || || |
| <1||5 (6.9)||21 (9.6)||0.031|
| 1–3||39 (54.2)||80 (36.5)|| |
| >3||28 (38.9)||118 (53.9)|| |
|Number of new batches brewed/day, n (%)|| || || |
| 1||55 (76.4)||169 (77.2)|| |
| ≥2||17 (23.6)||50 (22.8)||0.892|
|Types of tea drunk, n (%)|| || || |
| Black and green tea||6 (8.3)||12 (5.5)|| |
| Green tea only||66 (91.7)||207 (94.5)||0.383|
Table III presents the results from separate multivariate logistic regression fits for each tea consumption measure. The adjusted OR was 0.28 (95% CI = 0.17–0.47) for tea-drinkers relative to non-drinkers. The risk of prostate cancer tended to decline with increasing frequency and duration of tea drinking, with adjusted OR = 0.27 (95% CI = 0.15–0.48) and 0.12 (95% CI = 0.06–0.26) for those drinking more than 3 cups per day and over 40 years, respectively, compared to never or seldom tea-drinkers. The corresponding linear trend was significant. Inverse associations were also observed for quantity of tea leaves consumed per year and per batch, and particularly across all levels of green tea consumption, with significant dose-response relationships. Moreover, increasing the number of new batches brewed per day to 2 or more was associated with a 76% reduced risk of prostate cancer. Results from the 109 recent patients were also similar and omitted for brevity. Finally, the Hosmer and Lemeshow goodness-of-fit statistic ranged between 3.98 (p-value = 0.86) and 11.23 (p-value = 0.19), indicating no lack of fit for the logistic regression models.
It should be remarked that for the 291 participants that drank tea, they all consumed green tea but 18 of them (6 cases and 12 controls) drank both green tea and black tea. Unlike green tea, black and Oolong tea are not commonly consumed by Zhejiang residents. Therefore, the effect of different tea types on prostate cancer risk cannot be further analyzed owing to the small number of participants involved in drinking black tea.
In this epidemiological study, we documented in detail the quantitative effects of green tea consumption with respect to prostate cancer risk for Chinese men. Some progress has been made recently in elucidating the molecular mechanisms of cancer chemo-prevention by tea and tea polyphenols. The principle active constituent of tea polyphenols, epigallocatechin gallate, can offer protection against all stages of carcinogenesis (tumor initiation, promotion, and progression) by suppressing tumor promotion and inflammation, because of its antioxidant properties against free radicals, blocking signal transduction and nuclear oncogene expression, trapping of ultimate carcinogens, and inducing apoptosis and cell cycle arrest.36, 37, 38 In addition, it can repress the transcription of the androgen receptor gene and thus inhibit androgen action.16
Despite the inhibitory effects of tea against carcinogenesis have been consistently demonstrated in laboratory experiments, the relationship between tea drinking and human prostate cancer remains unclear. A lack of detailed and specific information on tea consumption has limited the conclusion drawn by previous studies. According to a recent Canadian case-control study, tea seemed to be protective among “heavy tea drinkers” (exceeding 500 ml/day).26 Their quantity of tea intake was estimated based on the single year pre-diagnosis period, lifetime tea consumption was not considered, whereas the type of tea and the quantity of tea leaf consumed were not specified and no adjustment was made for the consumption of other beverages.
The contradictory evidence in the literature may also be attributed to the differences in tea consumption patterns, drinking habit and other confounding factors among the various populations. In particular, the type of tea consumed and the method of preparation vary across countries and geographical regions. The effect of mixing beverages is complex and difficult to evaluate. A recent study on the influence of tea beverage types showed that hot black tea but not iced tea was associated with a lower risk of squamous cell carcinoma of the skin. Moreover, the black tea flavonoids, theaflavins and thearubigins can form insoluble complexes with caffeine, the so-called “tea cream,” when tea is cooled, which in turn can reduce the intake of active compounds.21 In southeast China, green tea is typically the only non-alcoholic beverage consumed by men, especially older men, throughout their lifetime. The common method of preparation is to brew dry green tea leaves in a cup using hot water without milk and sugar. It is known that the first 2 cups brewed from a new batch of tea contain almost equal amounts of epigallocatechin gallate, but its level can be substantially decreased in the third cup.39 It has also been reported that most of the antioxidants in tea can enter the liquid phase within 5 min of brewing,40 and the total amount of polyphenols in 600 ml or less of liquid tea is insufficient to provide a significant protective effect.17 Therefore, in addition to the number of cups of liquid tea consumed, it is important to record the number of new batches brewed each day and the quantity of tea leaves consumed per year and per batch, to accurately assess tea exposure of the participants. Alcohol consumption was also adjusted for in our study.
Although a case-control design for the present study was appropriate due to the low incidence of prostate cancer in China, several issues should be taken into consideration when interpreting the findings. A major feature of our study was that extensive information on tea consumption patterns were collected using a validated and reliable instrument specifically targeting Chinese men. Test–retest results conducted in Hangzhou have confirmed the reproducibility of the questionnaire. There was little variation in the method of green tea preparation by Zhejiang residents. Our previous study that employed a similar recruitment strategy also found no differences in tea consumption pattern between hospital-based controls and community controls,28 so that the participants recruited would be representative of the Zhejiang population. It may be argued that tea drinkers may have healthier dietary patterns and that green tea could be a marker of another aspect of diet that is protective against prostate cancer. Although there are no established dietary risk factors for Chinese men at present, the possibility of residual confounding by other (as yet unidentified) dietary habits cannot be ruled out. With regard to potential sources of biases that could affect the outcome, selection or recruitment bias seemed to be minimal in view of the low refusal rate of participation. The majority of cases were recently diagnosed, whereas the identification procedure ensured that ascertainment of cases was complete. Survival bias was found to be minimal in our study. Because the inverse association between tea consumption and prostate cancer has not been established at the time of interview and that participants were blinded to the purpose of the study, information bias concerning personal habits such as tea drinking was unlikely. In addition, a reference recall period (5 years before diagnosis for cases and 5 years before interview for controls) was adopted to avoid possible change in tea exposure since the onset of the disease. To minimize recall bias, a standard container (350 ml cup) was used to measure the frequency of habitual tea intake and most of the interviews were conducted in the presence of participant's next-of-kin. Finally, a single investigator (first author) conducted all interviews following the same procedure for both cases and controls to avoid intra- and inter- interviewer biases.
In conclusion, the present study of Chinese men suggests that increasing the frequency, duration and quantity of green tea consumption can lead to a lower risk of adenocarcinoma of the prostate. This is the first study providing comprehensive evidence of the protective effect of green tea against prostate cancer. The finding is well supported by animal and cell experiments on green tea and tea polyphenols both in vitro and in vivo. The preventive role of other types of tea needs to be further investigated because black and Oolong teas are not commonly drunk in our study population of Zhejiang residents.
We are indebted to the following persons for their kind assistance and permission to interview patients in their hospitals: Professor Du Chuan Jun, The Second Affiliated Hospital of Zhejiang University; Professor Shen Zhou Jun, The First Affiliated Hospital of Zhejiang University; Dr. Zhang Da Hong, Sir Run Run Shaw Hospital; Dr. Mei Fang, Hangzhou Fifth Hospital; Dr. Hou Yu Guo, Hangzhou Traditional Chinese Medicine Hospital; Dr. Li Zheng Yong, Hangzhou Fourth Hospital; Dr. Pan Yu Min, Hangzhou First Hospital; Dr. Zhou Qiang, Zhejiang Province Peoples Hospital.