Green tea (Camellia sinensis) for the prevention of cancer

Tommaso Filippini; Marcella Malavolti; Francesca Borrelli; Angelo A Izzo; Susan J Fairweather-Tait; Markus Horneber; Marco Vinceti

doi:10.1002/14651858.CD005004.pub3

. 2020 Mar 2;2020(3):CD005004. doi: 10.1002/14651858.CD005004.pub3

Green tea (Camellia sinensis) for the prevention of cancer

Tommaso Filippini ¹, Marcella Malavolti ¹, Francesca Borrelli ², Angelo A Izzo ², Susan J Fairweather-Tait ³, Markus Horneber ⁴, Marco Vinceti ^1,^5,^✉

Editor: Cochrane Gynaecological, Neuro-oncology and Orphan Cancer Group

PMCID: PMC7059963 PMID: 32118296

Abstract

Background

This review is an update of a previously published review in the Cochrane Database of Systematic Reviews (2009, Issue 3).Tea is one of the most commonly consumed beverages worldwide. Teas from the plant Camellia sinensis can be grouped into green, black and oolong tea, and drinking habits vary cross‐culturally. C sinensis contains polyphenols, one subgroup being catechins. Catechins are powerful antioxidants, and laboratory studies have suggested that these compounds may inhibit cancer cell proliferation. Some experimental and nonexperimental epidemiological studies have suggested that green tea may have cancer‐preventative effects.

Objectives

To assess possible associations between green tea consumption and the risk of cancer incidence and mortality as primary outcomes, and safety data and quality of life as secondary outcomes.

Search methods

We searched eligible studies up to January 2019 in CENTRAL, MEDLINE, Embase, ClinicalTrials.gov, and reference lists of previous reviews and included studies.

Selection criteria

We included all epidemiological studies, experimental (i.e. randomised controlled trials (RCTs)) and nonexperimental (non‐randomised studies, i.e. observational studies with both cohort and case‐control design) that investigated the association of green tea consumption with cancer risk or quality of life, or both.

Data collection and analysis

Two or more review authors independently applied the study criteria, extracted data and assessed methodological quality of studies. We summarised the results according to diagnosis of cancer type.

Main results

In this review update, we included in total 142 completed studies (11 experimental and 131 nonexperimental) and two ongoing studies. This is an additional 10 experimental and 85 nonexperimental studies from those included in the previous version of the review.

Eleven experimental studies allocated a total of 1795 participants to either green tea extract or placebo, all demonstrating an overall high methodological quality based on 'Risk of bias' assessment. For incident prostate cancer, the summary risk ratio (RR) in the green tea‐supplemented participants was 0.50 (95% confidence interval (CI) 0.18 to 1.36), based on three studies and involving 201 participants (low‐certainty evidence). The summary RR for gynaecological cancer was 1.50 (95% CI 0.41 to 5.48; 2 studies, 1157 participants; low‐certainty evidence). No evidence of effect of non‐melanoma skin cancer emerged (summary RR 1.00, 95% CI 0.06 to 15.92; 1 study, 1075 participants; low‐certainty evidence). In addition, adverse effects of green tea extract intake were reported, including gastrointestinal disorders, elevation of liver enzymes, and, more rarely, insomnia, raised blood pressure and skin/subcutaneous reactions. Consumption of green tea extracts induced a slight improvement in quality of life, compared with placebo, based on three experimental studies.

In nonexperimental studies, we included over 1,100,000 participants from 46 cohort studies and 85 case‐control studies, which were on average of intermediate to high methodological quality based on Newcastle‐Ottawa Scale 'Risk of bias' assessment. When comparing the highest intake of green tea with the lowest, we found a lower overall cancer incidence (summary RR 0.83, 95% CI 0.65 to 1.07), based on three studies, involving 52,479 participants (low‐certainty evidence). Conversely, we found no association between green tea consumption and cancer‐related mortality (summary RR 0.99, 95% CI 0.91 to 1.07), based on eight studies and 504,366 participants (low‐certainty evidence).

For most of the site‐specific cancers we observed a decreased RR in the highest category of green tea consumption compared with the lowest one. After stratifying the analysis according to study design, we found strongly conflicting results for some cancer sites: oesophageal, prostate and urinary tract cancer, and leukaemia showed an increased RR in cohort studies and a decreased RR or no difference in case‐control studies.

Authors' conclusions

Overall, findings from experimental and nonexperimental epidemiological studies yielded inconsistent results, thus providing limited evidence for the beneficial effect of green tea consumption on the overall risk of cancer or on specific cancer sites.

Some evidence of a beneficial effect of green tea at some cancer sites emerged from the RCTs and from case‐control studies, but their methodological limitations, such as the low number and size of the studies, and the inconsistencies with the results of cohort studies, limit the interpretability of the RR estimates. The studies also indicated the occurrence of several side effects associated with high intakes of green tea. In addition, the majority of included studies were carried out in Asian populations characterised by a high intake of green tea, thus limiting the generalisability of the findings to other populations. Well conducted and adequately powered RCTs would be needed to draw conclusions on the possible beneficial effects of green tea consumption on cancer risk.

Plain language summary

Green tea for the prevention of cancer

Background There is a high consumption worldwide of green tea (Camellia sinensis), that contains polyphenols which have a powerful antioxidant activity that can prevent the formation of free radicals that may cause damage and cell death. Therefore it has been suggested that green tea might reduce cancer risk, a theory that has been tested through a number of studies on human populations, which examined the link between green tea consumption and cancer.

The aim of the review We assessed the association between green tea consumption and the risk of developing cancer in epidemiologic studies.

Main findings In this review we included 142 studies with more than 1.1 million participants looking for an association between green tea consumption and cancers of the digestive tract and the female reproductive system, breast, prostate, kidney and urinary tract, nasopharynx, lung, blood, skin, thyroid and brain. The majority of the studies were of medium to high quality in terms of how they were conducted. Overall, the evidence from the studies showed that the consumption of green tea to reduce the risk of cancer was inconsistent.

Some studies suggested a beneficial effect on cancer risk, while others indicated no effect, and even suggested a slightly increased cancer risk. In particular, results from experimental studies suggested that green tea extract supplementation yielded a decreased risk for prostate cancer, but increased risk for gynaecological cancers. For non‐melanoma skin cancer no difference in cancer cases emerged. Green tea supplementation seemed to slightly improve quality of life compared with placebo, although it was associated with some adverse effects including gastrointestinal disorders, higher levels of liver enzymes, and, more rarely, insomnia, raised blood pressure and skin reactions.

In nonexperimental studies, comparing people consuming the highest amount of green tea to those in the lowest category of consumption, we found an indication of a lower occurrence of new cases of overall types of cancer, while no difference emerged for lethal cases. However, results according to the type of cancer and study design were inconsistent. 

What are the conclusions?  A beneficial effect of green tea consumption on cancer prevention remains unproven so far. Caution is advised regarding supplementation with high‐dose green tea extracts due to the possible adverse effects.

Summary of findings

Background

This review is an update of a previously published Cochrane review (Boehm 2009).

Description of the intervention

Tea (Camellia sinensis) is the most highly consumed manufactured drink in the world (FAO 2018). Between 2007 and 2016, world tea production grew by an average annual rate of 4.4%. Global tea consumption was 5.53 million tonnes in 2016 with an annual growth rate of 4.5% between 2007 and 2016. Three‐quarters of global production is consumed locally, driven particularly by China, India and other emerging economies (FAO 2018). In high‐income countries, consumption is much lower, being generally one‐fifth of that found in low‐ and middle‐income countries. Tea consumption has stabilised in recent years, with a few exceptions (FAO 2015), for example between 1990 and 2014, total tea consumption increased in the USA by about 38% (USDA 2018).

Brewed tea is obtained from the infusion of leaves and buds of Camellia sinensis. The most commonly consumed types of tea are green and black tea. Approximately 20% of the world's Camellia sinensis consumption is in the form of green tea; the other 80% is black (FAO 2015). Tea is characterised by the manufacturing process that the leaves undergo after harvesting. Green tea is made by processing fresh leaves using heat or hot steam immediately after collection, thus minimising any oxidation processes. Conversely, in black tea, the leaves undergo several treatments, including withering by blowing air, preconditioning, 'cut‐tear‐curl', fermentation and final drying, which result in an oxidised tea (Preedy 2014). Depending on these processes, the degree of oxidation may vary greatly, thus influencing the content of antioxidant compounds (Preedy 2014).

Due to the high content of antioxidant compounds, a great deal of attention has been given to green tea with regard to the possible prevention of chronic diseases and cancer (Eisenstein 2019; Yang 2019), as well as possible beneficial effects on cardiovascular disease, insulin sensitivity and lipid profiles (Liu 2013b; Yang 2019; Yu 2017).

How the intervention might work

Pharmacology of Camellia sinensis

The active ingredients of green tea include polyphenols most of which are flavonols, commonly known as catechins. These account for 30% to 40% of the extractable solids of dried green tea leaves. Other active ingredients are alkaloids, such as caffeine and theobromine, carbohydrates, and minerals and other trace elements, such as fluoride and aluminium (Coppock 2016; Filippini 2019; Milani 2019; Yang 2019). Green tea contains higher amounts of catechins than black tea (Peluso 2017), and green tea processing prevents oxidation (Chen 2007). After fermentation from green to black tea, about 15% of catechins remain unchanged while the rest of the catechins are converted to theaflavins, which are polyphenol pigments and thearubigins (Blumenthal 2003). Brewing conditions, including water temperature and infusion time, influence the antioxidant capacity of green tea (Sharpe 2016).

The catechins found in green tea include epigallocatechin‐3‐gallate (EGCG), epigallocatechin, epicatechin‐3‐gallate and epicatechin, gallocatechins and gallocatechin gallate. EGCG is the predominant and most studied catechin in green tea (Peluso 2017; Yang 2019), as it is a powerful antioxidant believed to be an important determinant of the therapeutic qualities of green tea (Chen 2019; Gao 2016; Peluso 2017). It is suggested that EGCG works by suppressing the formation of new blood vessels (angiogenesis) and regulating their permeability, thereby cutting off the blood supply to cancerous cells (Demeule 2002; Diniz 2017; Maiti 2003; Rashidi 2017; Yang 2019). In vitro studies and in vivo animal models have shown EGCG to be a potent chemo‐preventative agent (Liao 2001; Shirakami 2018; Xu 2019).

Green tea catechins have also been shown to decrease plasma lipid peroxide and malondialdehyde concentrations, to increase plasma ascorbate concentrations, to decrease non‐haem iron absorption, and increase the resistance of low‐density lipoproteins to oxidation (Williamson 2005). It is recognised that most classes of catechins are sufficiently well absorbed to have the potential to induce biological effects, since they cross the intestinal barrier and reach concentrations in the blood stream that have been shown in vitro to exert effects (Liao 2001; Manach 2005; Scalbert 2000). They are reported to be rapidly absorbed and eliminated in humans. Peak plasma concentrations were observed between one to three hours after oral administration and reached total catechin concentrations in the sub‐ or low‐μM range, and with a half‐life of two to four hours. Parent flavonoids are deglycosylated during digestion, are absorbed in the small intestine, and appear in the blood as phase II metabolites (Williamson 2018). Pharmacokinetic studies show that the flavonoid epicatechin is absorbed in the small intestine with a number of structural‐related epicatechin metabolites (SREM) attaining sub‐μmol/L peak plasma concentrations (Cmax) approximately one hour after ingestion (Borges 2018). The SREMs are excreted in urine over a 24‐hour period in amounts corresponding to 20% of epicatechin intake. If unabsorbed along the small intestine epicatechin undergoes microbiota‐mediated conversions in the colon, which, being absorbed, appear in plasma as phase II metabolites with a Cmax of 5.8 hours after consumption, and they are excreted in quantities equivalent to 42% of the ingested epicatechin (Borges 2018).

Possible mechanisms of action of Camellia sinensis polyphenols

Green tea polyphenols inhibit cell proliferation and viability, and have been shown (primarily in in vitro and ex vivo studies) to exert a powerful antioxidant activity (Ahmad 1999; Romano 2013; Schröder 2019; Yang 1993; Yang 1997). Several mechanisms have been proposed for the potential anticancer activity of green tea catechins (Yang 2019). Their polyphenolic structure allows electron delocalisation, conferring the ability to quench free radicals. EGCG, has been shown to reduce reactive oxygen species, such as superoxide radical, singlet oxygen, hydroxyl radical, peroxyl radical, nitric oxide, nitrogen dioxide and peroxynitrite (Sang 2011). Tea polyphenols are also strong chelators of metal ions, thus hampering the formation of reactive oxygen species. Several hypotheses have been put forward to explain a possible cancer‐preventive activity of catechins (Fujiki 1999; Yang 2019), including counteraction of tumour growth, invasion, metastasis and cell transformation, as well as inhibiting the interaction of tumour promoters, hormones and various growth factors with their receptors (Beltz 2006; Peluso 2017; Rahmani 2015; Rashidi 2017; Yang 2019). However, although in vitro and animal mechanistic studies indicate that flavonoids have anticancer properties, much of the evidence is derived from culture studies using unmetabolised flavonoids, and the simple antioxidant hypothesis is no longer an acceptable explanation (Kerimi 2018). In animal models, where high doses of green tea extracts and constituents have been used, strong evidence for the cancer‐preventive activity of tea constituents has been noted (Yang 2009; Yang 2011b). However, because of differences in endogenous metabolism and gut microflora, animal studies produce data that may not be necessarily be extrapolated to humans (Borges 2016).

Why it is important to do this review

Many reviews have been undertaken in recent years to examine the association between green tea consumption and cancer risk. Examples include a meta‐analysis that concluded that there is a dose‐response relationship between green tea consumption and prevention of prostate cancer when more than seven cups of tea are consumed per day (Guo 2017). Another recent dose‐response meta‐analysis reported a relationship between green tea drinking and prevention of liver cancer, with the downward trend being most obvious when the consumption was increased to four cups per day (Ni 2017). Similarly, dose‐response analysis of green tea consumption and biliary tract cancer suggested that the risk decreased by 4% with each additional cup of tea per day, especially in women (Xiong 2017). A systematic review examining the effect of green tea on risk of breast cancer suggested a protective effect (Gianfredi 2018), whereas another was inconclusive (Najaf 2018). With regard to gastric cancer, the effect of green tea may be temperature‐dependent, with high‐dose, long‐term consumption reducing the risk, whereas very high‐temperature green tea may possibly increase the risk of gastric cancer (Huang 2017).

Since several additional experimental and nonexperimental epidemiological studies have become available since the previous version of this Cochrane Review, we undertook an update to assess the relationship between peoples' green tea or EGCG consumption and cancer risk.

Objectives

To assess possible associations between green tea consumption and the risk of cancer incidence and mortality as primary outcomes, and safety data and quality of life as secondary outcomes.

Methods

Criteria for considering studies for this review

Types of studies

We included studies in which participants consumed green tea orally, either as drinkable tea or as extracts. Studies used one of the following designs.

Experimental studies: randomised controlled trials (RCTs)
Nonexperimental studies: both cohort and case‐control observational studies

We did not consider case‐series, case reports and other studies without a comparator, editorials, reviews, animal studies and in vitro studies for this review.

Types of participants

Adult participants (18 years of age and older).

Types of interventions

We were interested in studies that focused on the consumption of green tea, either as part of an intervention (experimental) study or measured in a nonexperimental study. The exposure variable was the consumption of green tea or green tea extract (only monotherapy preparations for oral consumption in liquid, powder or tablet form). We excluded studies that used green tea extract supplementation as part of a multi‐component preparation if they did not include a study arm using green tea extracts in monotherapy.

We defined green tea as non‐fermented tea leaves, and studies had to mention that green tea, non‐fermented tea or 'matsu‐cha', as it is called in Asia, had been consumed. We considered any method of quantifying this variable (e.g. direct recording, recall questionnaire) as potentially valid. We excluded studies that did not distinguish the type of tea (e.g. black tea versus green tea) or did not report quantitative data of at least two different amounts or frequencies of green tea consumption.

We excluded pharmacokinetic‐type studies because they were unlikely to contribute useful data on long‐term effects of green tea.

Only studies reporting the duration of green tea consumption in their summary were included.

Types of outcome measures

Primary outcomes

The primary outcome measures were:

the number of participants developing cancer (incidence);
the number of participants dying from cancer (mortality).

Results from nonexperimental epidemiological studies had to include an estimate of the risk ratio (RR), or sufficient data for us to calculate it.

We used the following categories to combine and analyse different types of cancer.

Gastrointestinal cancers: including oral cancer, pharyngeal cancer, laryngeal cancer, oesophageal cancer, stomach cancer, liver cancer, pancreatic cancer, biliary tract cancer, and colorectal cancer
Respiratory tract cancer: including nasopharyngeal cancer, lung cancer, and mesothelioma
Breast cancer
Urogenital tract cancers: including prostate cancer, endometrial cancer, ovarian cancer, renal cancer, and urinary tract cancer
Haematological cancers: including haematopoietic cancer, leukaemia, lymphoma, and multiple myeloma
All other types of cancer

Secondary outcomes

Safety data and data on quality of life

Search methods for identification of studies

Electronic searches

For the original review we searched the following electronic databases from inception to January 2009 to retrieve studies for potential inclusion: the Cochrane Central Register of Controlled Trials (CENTRAL), MEDLINE (Ovid), Embase (Ovid), Amed, CancerLit, PsychInfo and Phytobase.

For this update we searched the following electronic databases up to January 2019:

the Cochrane Central Register of Controlled Trials (CENTRAL; 2019, Issue 1) in the Cochrane Library (Appendix 1);
MEDLINE via Ovid (January 2009 to January week 1, 2019) (Appendix 2);
Embase via Ovid (January 2009 to 2019 Week 1) (Appendix 3).

Searching other resources

We systematically checked references from published studies for further studies. We specifically screened the reference list from studies retrieved in full text, and also from previous systematic reviews and meta‐analysis on the topic, including non‐English papers, though we assumed that some of the articles from Asian countries would not be retrievable via Western medical databases. We obtained all relevant non‐English articles and a Japanese/Chinese Cochrane collaborator acted as a filter for study selection. Publications in languages other than English were translated in‐house or by using relevant services. Finally, we also checked for other relevant studies in the clinical trials registry ClinicalTrials.gov up to January 2019.

Data collection and analysis

Selection of studies

To be included, studies had to report on the consumption of green tea, non‐fermented tea or 'matsu‐cha'. Two review authors checked studies identified by the searches and included articles on initial screen only if they could determine from the abstract that the article was a report of either an experimental intervention or a nonexperimental study. When we could not reject with certainty a title or abstract, we assessed the full text.

Two review authors independently analysed the full text of all potentially relevant eligible studies. Reasons for excluding any study are stated in Criteria for considering studies for this review. All disagreements were resolved by discussion between the two review authors. If any data were missing from the study reports, we attempted to obtain the data by contacting the study authors.

Data extraction and management

Two review authors independently performed data extraction using pre‐defined and pre‐tested data extraction forms. We resolved discrepancies by discussion. We categorised studies into experimental (RCTs) and nonexperimental studies (i.e. cohort studies, including cohort‐nested studies, and case‐control studies). We also grouped data according to study design and cancer type. We entered the extracted data into Review Manager 5 (Review Manager 2014), and two review authors double‐checked the entries.

Assessment of risk of bias in included studies

Four review authors independently assessed the risk of bias of the included studies.

Experimental studies

We used the Cochrane 'Risk of bias' tool to assess risk of bias in the included RCTs (Higgins 2017). The criteria relate to the following domains:

Selection bias: random sequence generation and allocation concealment
Performance bias: blinding of participants and personnel (i.e. treatment providers)
Detection bias: blinding of outcome assessment
Attrition bias: incomplete outcome data
Reporting bias: selective reporting of outcomes
other possible sources of bias (please specify)

We considered studies that we assessed as 'adequate' in all main domains to be at low risk of bias. Studies in which there was no clear judgement concerning the procedures in one or more key domains we considered to be at least of medium risk of bias. Studies with clearly inadequate procedures in one or more of the key domains we considered to be at high risk of bias.

Nonexperimental studies

We used the Newcastle‐Ottawa Scale (NOS) to assess the methodological quality of epidemiologic studies (Wells 2001). The NOS is based on a 'star' system in which a study is assessed on three broad perspectives:

selection of study groups;
comparability of the groups;
ascertainment of outcome or exposure of interest for cohort or case‐control studies, respectively.

High‐quality answers to each NOS question are identified with a star/asterisk. Details used during the evaluation are reported in two templates, one for cohort (Appendix 4), and one for case‐control studies (Appendix 5). Both cohort and case‐control studies can receive a maximum of nine stars or points. We considered studies with six or fewer points as low quality, with seven to eight points as medium quality, and with nine points as high quality.

Measures of treatment effect

We used the following measures of the effect of treatment or exposure.

For dichotomous outcomes (i.e. cancer risk), we used the risk ratio (RR) for both experimental and nonexperimental studies.
For continuous outcomes (evaluation of scores for quality of life), we used the mean difference between treatment arms in experimental studies.

Unit of analysis issues

We did not note any unit of analysis issues.

Dealing with missing data

When a study had missing data in the level of exposure assessment, risk estimates or confidence intervals, we attempted to obtain the data by contacting the study authors. Nevertheless, we reported the available data in Characteristics of included studies. We did not impute missing data for any of the outcomes for data analysis.

Assessment of heterogeneity

We used the Chi² test for heterogeneity and the I² statistic (Higgins 2003), to quantify heterogeneity of study results. We interpreted the I² statistic as per guidance provided in the Cochrane Handbook for Systematic Reviews of Interventions (Deeks 2017): 0% to 40% might not be important; 30% to 60% represented moderate heterogeneity; 50% to 90% represented substantial heterogeneity; and 75% to 100% represented considerable heterogeneity.

Assessment of reporting biases

We followed the recommendations for testing for funnel plot asymmetry as described in the Cochrane Handbook for Systematic Reviews of Interventions (Sterne 2017). Funnel plot asymmetry may be due to reporting bias. We produced funnel plots to assess the potential for small‐study effects when at least five studies reported results for the same type of cancer (Egger 1997).

Data synthesis

We carried out a meta‐analysis of the included studies when the study results reported estimate for cancer risk or we could compute it from raw data. When studies reported more than one estimated risk, we used the results generated by the most adjusted model. We used a random‐effects model for all analyses. We carried out and reported overall analyses and analyses stratified by study design.

Subgroup analysis and investigation of heterogeneity

We performed subgroup analysis for any cancer incidence and mortality and, whenever possible, for all different types of cancer site according to categories identified in Primary outcomes. In order to investigate possible sources of heterogeneity, we also performed stratified analyses according to study design of nonexperimental studies (hospital‐based case‐control design, population‐based case‐control design, cohort design).

Sensitivity analysis

We performed the following sensitivity analyses:

study design of nonexperimental studies (hospital‐based case‐control design, population‐based case‐control design, cohort design).

Summary of findings and assessment of the certainty of the evidence

We presented the overall certainty of evidence of all outcomes investigated in the experimental studies, namely prostate cancer, gynaecological cancer and non‐melanoma skin cancer. We also presented the certainty of evidence for the primary outcomes of nonexperimental studies and for which it was possible to evaluate publication bias, that is, when at least five studies reported results for the same type of cancer (Egger 1997).

We evaluated the overall certainty of evidence according to the GRADE approach (Atkins 2004), which takes into account issues related not only to internal validity (risk of bias, inconsistency, imprecision, publication bias) but also to external validity, such as directness of results (Langendam 2013). We created three 'Summary of findings' tables (Table 1; Table 2; Table 3), adhering to the methods described in the Cochrane Handbook for Systematic Reviews of Interventions (Schünemann 2017), and using GRADEpro GDT. We used the GRADE checklist and GRADE Working Group certainty of evidence definitions (Meader 2014), as follows.

Summary of findings 1. Green tea extract supplementation compared with placebo for preventing cancer: experimental studies.

Green tea extract supplementation compared with placebo for cancer prevention: experimental studies
Patient or population: adults (aged at least 18 years) Settings: outpatients Intervention: green tea extract supplementation Comparison: placebo
Outcomes	Relative effect (95% CI)	*Anticipated absolute effects (95% CI) (studies)**			No of participants (studies)	Certainty of the evidence (GRADE)	Comments/explanations
Outcomes	Relative effect (95% CI)	Placebo	Green tea extract supplementation	Difference	No of participants (studies)	Certainty of the evidence (GRADE)	Comments/explanations
Prostate cancer incidence	RR 0.50 (0.18 to 1.36)	22.0%	11.7% (4.4 to 28.7)	10.3% fewer (17.6 fewer to 6.7 more)	201 (3 RCTs)	⊕⊕⊝⊝ Low	Very large effects, but all participants were at high risk of prostate cancer at the time of recruitment, with high‐grade prostatic intraepithelial neoplasia and/or atypical small acinar proliferation less than 3 months before, thus the indirectness in transferring the results to the general population, high imprecision of the summary estimates based on only 201 participants and 32 cases, and high inconsistency between study results
Gynaecological cancer incidence	RR 1.50 (0.41 to 5.48)	0.9%	1.3% (0.4 to 4.6)	0.4% more (0.5 fewer to 3.7 more)	1157 (2 RCTs)	⊕⊕⊝⊝ Low	Large effects, but high imprecision of the summary RR and high inconsistency of results due to contradictory findings from two available studies.
Non‐melanoma skin cancer incidence	RR 1.00 (0.06 to 15.92)	0.2%	0.2% (0.0 to 2.9)	0.0% fewer (0.0 fewer to 2.7 more)	1075 (1 RCT)	⊕⊕⊝⊝ Low	Very high imprecision based on only one study and no large effect detected.
Quality of life	Not estimable	‐	‐	‐	‐	Not assessed	Due to the large number of different scales used it was not possible to the overall certainty of evidence.
The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; RCT: randomised controlled trial; RR: risk ratio
GRADE Working Group grades of evidence High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate; the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited; the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate; the true effect is likely to be substantially different from the estimate of effect.

Highest compared with lowest green tea exposure for preventing cancer in nonexperimental studies
Patient or population: adults (aged at least 18 years) Setting: outpatient Intervention: highest green tea exposure Comparison: lowest green tea exposure
Outcomes (number of studies)	Relative effect (95% CI)	Number of participants (number of cases)	Certainty of the evidence (GRADE)	Comments
Any cancer incidence (3 studies)	RR 0.83 (0.65 to 1.07)	52,479  (4962 cases)	⊕⊕⊝⊝ Low	Large^a but imprecise effect. Similar but imprecise effect from the 2 cohort studies (RR 0.81, 95% CI 0.50 to 1.32)
Any cancer mortality (8 studies)	RR 0.99 (0.91 to 1.07)	504,366  (21,439 cases)	⊕⊕⊝⊝ Low	Not a large effect. All cohort studies
The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; RR: risk ratio
GRADE Working Group grades of evidence High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate; the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited; the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate; the true effect is likely to be substantially different from the estimate of effect.

Study	Country	Target cancer	Outcomes	Participants	Intervention	Duration
Bettuzzi 2006	Italy	Prostate cancer	Prostate cancer incidence LUTS PSA values QoL Safety data	60 men	Green tea total catechins 600 mg/day = EGCG: ~300 mg/day	12 months
Dostal 2015	USA	Breast cancer	Breast cancer biomarkers Circulating F2‐isoprostane levels Oestrogen metabolite levels Non‐melanoma skin cancer Uterine cancer Safety data	1075 women	Green tea total catechins: 1315 (± 116) mg/day = EGCG: 843 (± 44) mg/day	12 months
Dryden 2013	USA	‐	Ulcerative disease activity QoL Safety data	20 men and women	Green tea extracts: Polyphenon E = EGCG: 200 mg or 400 mg/day	56 days
Garcia 2014	USA	Cervical cancer	Oncogenic HPV clearance CIN1 clearance Safety data	98 women	Green tea extracts: Polyphenon E = EGCG: 800 mg/day	4 months
Garland 2006	USA	Lung cancer	Biomarkers of oxidative stress Safety data	178 (89 men and 89 women)	Green tea extracts: Polyphenon E = ECGC 800 mg/day	6 months
Kumar 2015	USA	Prostate cancer	Prostate cancer incidence Safety data	97 men	Green tea extracts: Polyphenon E = EGCG: 400 mg/day	12 months
Lane 2018	UK	Prostate cancer	PSA levels Clinical outcome (weight and blood pressure) Safety data	88 men	Green tea extracts = EGCG: 600 mg/day	6 months
Micali 2017	Italy	Prostate cancer	Prostate cancer incidence PSA levels QoL Safety data	60 men	Green tea extracts 600 mg = EGCG: 300 mg/day	12 months
Roshdy 2013	Egypt	Uterine fibroids	Severity of symptoms QoL Safety data	39 women	Green tea extracts = EGCG: ~400 mg/day	4 months
Sinicrope 2017	USA	Colon cancer	Change in rectal aberrant crypt foci Safety data	39 (14 men and 25 women)	Green tea extracts = EGCG: 400 mg/day	6 months
Tsao 2009	Japan	Oral cancer	Histological response Safety data	41 (19 men and 22 women)	Green tea extracts 500, 750 or 1000 mg/day	12 weeks
Shannon 2010	USA	Prostate cancer	Prostate cancer incidence Immunoistochemical response	67 men out of 120 planned	Green tea polyphenols = EGCG ~600 mg/day)	12 weeks
NCT01496521	China	Oesophageal cancer	Oesophageal cancer incidence Occurrence of high grade dysplasia Invasive oesophageal squamous cell carcinoma	Not reported	Tea polyphenols 600 mg/day	12 months
EGCG: (–)‐epigallocatechin‐3‐gallate; CIN: cervical intraepithelial neoplasia; HPV: human papillomavirus LUTS: lower urinary tract symptoms; PSA: prostate‐specific antigens; QoL: quality of life

Study	Country	Cohort	Cancer	Outcome	Participants	Cases
Allen 2004	Japan	Life Span Study	Prostate	Incidence	18,115 men	193
Chyou 1993	USA	Honolulu Hearth Program	Bladder	Incidence	7991 men	96
Dai 2010	China	Shangai Women's Health Study	Breast	Incidence	72,861	614
Fujino 2002 in: Inoue 2009a	Japan	JACC Study	Stomach	Incidence	44,930	379
Galanis 1998	USA	Hawaii Health Surveillance Program	Stomach	Incidence	11,907	108
Hoshiyama 2002 in: Inoue 2009a	Japan	JACC Study	Stomach	Mortality	44,930	359
Ide 2007	Japan	JACC Study	Oral	Incidence	50,221	37
Inoue 2009a	Japan	JACC, JPHC‐I, JPHC‐II, MIYAGI, 3‐pref MIYAGI, and 3‐pref AICHI Studies	Stomach	Incidence	219,080	3577
Inoue 2009b	Japan	JPHC‐II Study	Liver	Incidence	18,815	110
Ishikawa 2006	Japan	MIYAGI and 3‐pref MIYAGI Studies	Oesophageal	Incidence	26,723	78
Iwai 2002	Japan	JACC Study	Any cancer	Mortality	2855	31
Iwasaki 2010a	Japan	JPHC‐I and JPHC‐II Studies	Breast	Incidence	67,422	581
Iwasaki 2010b in: Iwasaki 2010a	Japan	JPHC‐I and JPHC‐II Studies	Breast	Incidence	67,422	144
Key 1999	Japan	Life Span Study	Breast	Incidence	34,765	405
Khan 2004	Japan	Public Health Centers in Hokkaido Prefecture	Any cancer Lung Stomach Colorectal Pancreatic	Mortality	3158	243 51 51 29 25
Kikuchi 2006	Japan	Ohsaki Cohort Study	Prostate	Incidence	18,961	110
Koizumi 2003 in: Inoue 2009a	Japan	MIYAGI and 3‐pref MIYAGI Studies	Stomach	Incidence	65,915	733
Kurahashi 2007	Japan	JPHC‐I and JPHC‐II Studies	Prostate	Incidence	49,920	404
Kurahashi 2009	Japan	JPHC‐I and JPHC‐II Studies	Bladder	Incidence	104,440	206
Kuriyama 2006	Japan	Ohsaki Cohort Study	Any cancer Lung Stomach Colorectal	Mortality	40,530	1134 218 193 132
Lee 2007	Japan	JPHC‐I and JPHC‐II Studies	Colorectal	Incidence	96,162	1158
Li 2008	Japan	Ohsaki Cohort Study	Lung	Incidence	41,440	302
Li 2018	China	Kailuan Cohort	Lung	Incidence	103,010	964
Lin 2008	Japan	JACC Study	Pancreatic	Mortality	77,850	292
Liu 2016	China	Chinese Prospective Smoking Study	Any cancer	Mortality	164,681	7002
Luo 2007	Japan	JPHC‐I and JPHC‐II Studies	Pancreatic	Incidence	102,137	233
Makiuchi 2016	Japan	JPHC‐I and JPHC‐II Studies	Biliary tract	Incidence	140,420	271
Michikawa 2011	Japan	JPHC‐I and JPHC‐II Studies	Thyroid	Incidence	100,507	159
Montague 2012	China	Singapore Chinese Health Study	Prostate	Incidence	27,293	298
Nagano 2001	Japan	Life Span Study	Any cancer and several specific types	Incidence	38,540	4049
Naganuma 2009	Japan	Ohsaki Cohort Study	Hematopoitic	Incidence	41,761	157
Nakachi 2000	Japan	Saitama Prefecture	Any cancer	Incidence	8552	488
Nakamura 2011	Japan	Takayama and Gigu Prefectures	Pancreatic	Mortality	30,826	52
Nechuta 2012	China	Shangai Women's Health Study	Digestive system	Incidence	67,230	1239
Oba 2006	Japan	Takayama and Gigu Prefectures	Colon	Incidence	30,836	213
Odegaard 2015	China	Singapore Chinese Health Study	Any cancer	Mortality	52,584	4092
Ogawa 2016	Japan	JPHC‐I and JPHC‐II Studies	Brain	Incidence	106,324	155
Saito 2015	Japan	JPHC‐I and JPHC‐II Studies	Any cancer	Mortality	90,914	5327
Sasazuki 2004 in: Inoue 2009a	Japan	JPHC‐I and JPHC‐II Studies	Stomach	Incidence	72,943	892
Sauvaget 2005 in: Nagano 2001	Japan	Life Span Study	Stomach	Incidence	38,576	1270
Severson 1989	USA	Honolulu Hearth Program	Prostate	Incidence	7821	174
Shimazu 2008	Japan	JPHC‐I and JPHC‐II Studies	Endometrial	Incidence	53,724	117
Sun 2007	China	Singapore Chinese Health Study	Colorectal	Incidence	61,320	845
Suzuki 2004	Japan	MIYAGI and 3‐pref MIYAGI Studies	Breast	Incidence	35,004	222
Suzuki 2005	Japan	MIYAGI and 3‐pref MIYAGI Studies	Colorectal	Incidence	65,915	516
Suzuki 2009	Japan	Prospective Shizuoka Elderly Cohort	Any cancer Stomach Lung Colorectal	Mortality	12,251	400 68 88 43
Tamura 2018	Japan	Takayama and Gigu Prefectures	Liver	Incidence	30,824	172
Tsubono 2001 in: Inoue 2009a	Japan	3‐pref MIYAGI Study	Stomach	Incidence	26,311	419
Ugai 2017	Japan	JPHC‐I and JPHC‐II Studies	Lymphoma Multiple myeloma	Incidence	95,807	411 138
Ugai 2018	Japan	JPHC‐I and JPHC‐II Studies	Acute myeloid leukaemia	Incidence	95,807	85
Ui 2009	Japan	Ohsaki Cohort Study	Liver	Incidence	41,761	247
Yang 2007 in: Nechuta 2012	China	Shangai Women's Health Study	Colorectal	Incidence	69,710	256
Yang 2011a	China	Shangai Men's Health Study	Colorectal	Incidence	60,567	243
Zhao 2017	China	Shangai Women's Health Study and Shangai Men's Health Study	Any cancer	Incidence	115,954	3210

Study	Country	Study type	Cancer	Cases/ Controls	Sex
Bandera 2010	USA	PCC	Endometrial	397/373	Women
Berroukche 2012	Algeria	HCC	Prostate	160/160	Men
Bonner 2005	China	PCC	Lung	122/121	Both
Chen 2011	China	HCC	Oesophageal	150/300	Both
Chen 2015 in: Chen 2017a	China	HCC	Oral	203/572	Both
Chen 2016 in: Chen 2017a	China	HCC	Oral	207/480	Women
Chen 2017a	China	HCC	Oral	586/1024	Both
Fu 2013	China	HCC	Oral	723/857	Both
Gao 1994	China	PCC	Oesophageal	902/1312	Both
Gao 2005	China	PCC	Endometrial	955/1087	Women
Gavrilas 2018	Romania	HCC	Colorectal	151/151	Both
Goodman 2003	USA	PCC	Ovarian	164/194	Women
Goto 1990	Japan	PCC	Pancreatic	71/142	Both
Green 2014	Australia	PCC	Colorectal	854/948	Both
Hakim 2000	USA	PCC	Skin	243/216	Both
Han 2008	China	PCC	Lung	523/1924	Both
Hemelt 2010	China	HCC	Bladder	419/392	Both
Hoshiyama 1992	Japan	PCC	Stomach	251/483	Men
Hoshiyama 2004 in: Inoue 2009a	Japan	Nested case‐cohort	Stomach	151/256	Both
Hsu 2012	China	PCC	Nasopharyngeal	368/317	Both
Huang 1999	Japan	HCC	Stomach	887/28,619	Both
Inoue 1994 in: Huang 1999	Japan	HCC	Stomach	668/668	Both
Inoue 1998	Japan	HCC	Oesophageal Stomach Colorectal	1706/21,128	Both
Inoue 2008	China	PCC	Breast	380/662	Women
Islami 2009	Iran	PCC	Oesophageal	266/571	Both
Iwasaki 2014	Japan	HCC	Breast	369/405	Both
Ji 1996	China	PCC	Stomach	1029/1347	Both
Ji 1997	China	PCC	Colorectal Pancreatic	2156/1552	Both
Jia 2016	China	CC	Lung Mesothelioma	53/106	Both
Jian 2004	China	HCC	Prostate	130/274	Men
Jin 2013	China	PCC	Lung	799/2020	Both
Kakuta 2009	Japan	PCC	Endometrial	152/285	Women
Kato 1990	Japan	HCC	Colorectal	221/578	Both
Kato 1990a in: Huang 1999	Japan	HCC	Stomach	427/3014	Both
Kono 1988	Japan	PCC/HCC	Stomach	139/278 139/2575	Both
Kubik 2004 in: Kubik 2008	Czech Republic	HCC	Lung	435/1710	Women
Kubik 2008	Czech Republic	HCC	Lung	1096/2966	Both
Kuo 2009	China	PCC	Leukaemia	93/223	Both
Lassed 2016	Algeria	HCC	Prostate	90/190	Men
Lee 2017	China	HCC	Prostate	404/395	Men
Lei 1994	China	HCC	Lung	792/792	Both
Le Marchand 2000	USA	PCC	Lung	582/582	Both
Leung 2016	China	PCC	Ovarian	104/471	Women
Li 2011a	China	PCC/HCC	Any cancer Breast Colorectal Leukaemia	425/540 425/540	Both
Li 2011b in: Mu 2003	China	PCC	Liver	204/415	Both
Li 2014	China	PCC	Prostate	250/500	Men
Li 2016	China	HCC	Breast	756/789	Women
Lin 2012	China	HCC	Lung	170/340	Both
Liu 2010	China	PCC	Stomach	641/1847	Both
Liu 2017	China	HCC	Leukaemia	442/442	Both
Mao 2011	China	HCC	Stomach	200/200	Both
Mizoo 2013	Japan	HCC	Breast	472/464	Women
Mizuno 1992	Japan	HCC	Pancreatic	124/124	Both
Mu 2003	China	HCC	Stomach Liver Oespphagus	628/415	Both
Mu 2005 in: Mu 2003	China	PCC	Stomach	206/415	Both
Nagle 2010	Australia	PCC	Ovarian	1368/1462	Women
Oze 2014	Japan	PCC	Upper digestive system	922/2883	Both
Peng 2013	China	PCC	Colorectal	672/672	Both
Peng 2015	China	PCC	Oesphageal	285/570	Both
Ruan 2010	China	HCC	Nasopharyngeal	1355/1459	Both
Setiawan 2001	China	PCC	Stomach	133/433	Both
Shrubsole 2009	China	PCC	Breast	3371/3380	Women
Song 2008	USA	PCC	Ovarian	781/1263	Women
Sonoda 2004	Japan	HCC	Prostate	140/140	Men
Tajima 1985	Japan	HCC	Stomach	93/93	Both
Takezaki 2000	Japan	HCC	Pharynx Oesophageal	346/11,936	Both
Takezaki 2001	Japan	HCC	Lung	945/4153	Both
Tewes 1990	China	HCC	Lung	200/200	Women
Tse 2017	China	HCC	Prostate	431/402	Men
Wakai 2004	Japan	HCC	Bladder	124/620	Both
Wang 1999	China	PCC	Oesophageal and stomach	209/209	Both
Wang 2006	China	PCC	Oesophageal	107/107	Both
Wang 2007	China	PCC	Oesophageal	355/209	Both
Wang 2012a	China	HCC	Renal	250/299	Both
Wang 2012b	China	HCC	Multiple myeloma	220/220	Both
Wang 2012c	China	PCC	Pancreatic	908/1067	Both
Wang 2013a	China	HCC	Breast	157/314	Women
Wang 2013b	USA	HCC	Bladder	1007/1299	Both
Wang 2015	China	HCC	Stomach	160/320	Both
Wilkens 1996	USA	PCC	Bladder	261/522	Both
Wu 2003	USA	PCC	Breast	501/594	Women
Wu 2009a	China	HCC	Prostate	142/142	Men
Wu 2009b	China	PCC	Oesophageal	1,502/3,879	Both
Xu 2007	China	PCC	Endometrial	1204/1212	Women
Xu 2013	China	PCC	Lung	1225/1234	Both
Yan 2016	China	PCC	Oral	593/1128	Both
Ye 1998	China	PCC	Stomach	272/544	Both
Yu 1995	China	PCC	Stomach	711/711	Both
Zhang 2002	China	PCC/HCC	Ovarian	254/261 254/340	Women
Zhang 2007	China	HCC	Breast	1009/1009	Women
Zhang 2008	China	HCC	Leukaemia	107/110	Both
Zheng 1993	China	HCC	Oral	404/404	Both
Zhong 2001	China	PCC	Lung	649/675	Both
HCC: hospital‐based case‐control study; PCC: population‐based case‐control study

References	Detailed results
Bettuzzi 2006	Prostate cancer incidence Bettuzzi 2006: one prostate cancer case in the treatment group and 9 cases in the placebo group Brausi 2008 reported a longer follow‐up on a subset of participants: 13 cases in the intervention group and 9 in the placebo group, all after the suspension of the treatment. Further 3 cases of prostate cancer were diagnosed, 1 in the treatment group and 2 cases in the placebo group LUTS: a slightly higher decrease in IPSS was found in the treatment group from 11.12 to 9.12, than in the placebo group, from 8.27 to 7.00 PSA levels: no substantial difference in the PSA levels between groups QoL: QoL score assessed after 3 months decreased in the intervention group (from 2.06 to 1.76), while slightly increased in the placebo group (from 1.30 to 1.47) Safety data: 2 cases of diarrhoea in each arm, reported as very mild disorders
Dostal 2015	Dostal 2015 Follow‐up: 59 participants (39 in the intervention and 20 in the control group) stopped taking study product but remained in the study. Participants stopped mainly due to adverse symptoms (N = 50). Dropout: 138 participants withdrew from the study due to request (N = 93), adverse event (N = 22), protocol violation (N = 10), lost to follow‐up (N = 10), investigator judgment (N = 3) and death (N = 1). 18 of 22 (82%) who withdrew to adverse events were in the intervention group. Cancer incidence: 2 participants in the placebo group were diagnosed with uterine cancer after randomisation, 1 woman 1 day after beginning study product (for this reason not included in the risk analysis) and the other during the last month of participation. 2 women (1 in the treatment and 1 in the placebo group) reported a diagnosis of non‐melanoma skin cancer during the study. Safety data: total 1141 adverse events documented in the intervention group and 1031 in the placebo group. The most common events were infections; gastrointestinal disorders (nausea, indigestion, diarrhoea, constipation, vomiting, increased gassiness, abdominal pain, increased acid reflux); vascular disorders; respiratory, thoracic and mediastinal disorders; general disorders and administration site condition; musculoskeletal and connective tissue disorders, mainly of grade 1 and 2 of severity. Higher adverse effects in intervention groups were skin and subcutaneous tissue disorders (mainly rash or allergic skin reaction), ALT elevations and nausea and partially indigestion and constipation. Groups did not differ in severity of adverse effects. Yu 2017 in: Dostal 2015 Follow‐up: in the present report data on 513 men in the intervention group and 508 women in the control group are reported, due to missing data of ALT at baseline (N = 3) or during follow‐up (N = 51). Treatment increased both ALT and AST, whereas no increase was reported in the control group. AKP and total bilirubin did not increase in both treatment and control group. Webster 2018 in: Dostal 2015 QoL: data on overall QoL of recruited women. QoL assessed using MENQOL scale showed overall scores higher in women aged 50‐54.9 years
Dryden 2013	Follow‐up: 10 participants randomised in cohort 1:8 to treatment group (low‐dose) and 2 to placebo: 10 participants randomised in cohort 2: 8 to treatment group (high‐dose) and 2 in placebo Safety data: no participants required termination due to serious adverse event. 1 participant in the treatment group required hospitalisation due to disease progression. Higher incidence of heartburn, increased thirst and increased diarrhoea were found in treatment group. QoL: measured with the Inflammatory Bowel Disease Questionnaire, QoL generally inversely correlated with the activity of the disease.
Garcia 2014	Follow‐up: 82 participants (41 in the intervention and 41 in the control group) analysed for primary outcome Primary outcome Complete response (negative high‐risk HPV and normal histopathology) was noted in 7 (17.1%) and 6 (14.6%) for the treatment and control group, respectively Partial response (negative high‐risk HPV and evidence of low‐grade CIN) occurred more frequently in the control group, 1 (2.4%) vs. 614.6%) Progression/negative response (persistence of high‐risk HPV+ and worsening of CIN or invasive cancer) was more common in the treatment group, 6 (14.6%) vs 3 (7.3%) No response (persistence of high‐risk HPV+ with no progression similar to intervention and control group, 27 (65.9%) vs. 26 (63.4%) Secondary outcome Safety data: 163 and 136 adverse events in the control and in the intervention group. 2 serious events occurred in 1 participant in the placebo group. 1 and 2 participants in the control and in the intervention group respectively discontinued the study due to adverse events Adverse events were all Grade 1 and Grade 2, except 1 Grade 3 ALT elevation and one Grade 3 back pain in the Polyphenon E arm and one Grade 3 ALT and AST elevation in the placebo arm. Nausea was reported more frequently by participants receiving Polyphenon E compared to those receiving placebo, 32.0% vs 18.8%, respectively, as well as elevated ALT/AST, 10.0%/8.0% and 2.1%/2.1%, respectively, as well as was incidence of dizziness (14.0% vs 6.3%)
Garland 2006	Primary outcome Urinary 8‐hydroxydeoxyguanosine levels increased in Group A (+2.36) and B (+5.20), while decreased in Group C (−1.08) Urinary 8‐F2‐isoprostanes levels decreased in both Group A (−39.87) and B (−35.80), with no change in Group C (0.71) Secondary outcome Safety data: no significant adverse events were reported, including no liver toxicity, but higher frequency of nausea, constipation and gastrointestinal reflux disease detected
Kumar 2015	Prostate cancer incidence: total 14 cases at the end of the study (1 year): 5 prostate cancer cases in the treatment group and 9 cases in the control group. Prostate cancer + ASAP: total 13 (8 + 5) cases at the end of the study (1 year): 3 (3 + 0) prostate cancer cases in the treatment group and 10 (5 + 5) cases in the control group. No significant differences between the treatment and placebo arms were observed in LUTS and QoL scores from baseline to end of study. Safety data: 26 adverse effects (14 in the treatment group and 12 in control group censored at 6 months due to adverse effects), 2 (1 + 1) between 6 and 12 months and 55 (29 + 26) adverse effects at 12 months. Higher possible and probable events in the treatment arm, all but 1 of grade 1 and 2. Total 381 adverse effects, 212 in the treatment group and 169 in the control group. 11 and 7 in treatment and control groups respectively met off‐study criteria due to adverse effects. Higher incidence of coagulation toxicity, gastrointestinal and pain reported in the treatment group as well as toxicity to skin and musculoskeletal/soft tissue. Data on adverse effects also reported in Kumar 2016. Data on body weight or abdominal obesity in Kumar 2017
Lane 2018	Primary outcome Risk for prostate cancer NR Secondary outcomes PSA levels did not differ between green tea groups after 6 months of treatment Safety data: most incident adverse events were nocturia and insomnia that were higher in green tea drink (46% and 22%) and green tea capsules (50% and 21%) compared to placebo (33% and 8%). Hypertension was similar across groups, while fatigue was higher in green tea capsules (24%) and placebo (18%) than green tea drink (5%). Less frequent adverse events were cramp, shortness of breath, heartburn, headache and diarrhoea.
Micali 2017	Follow‐up: 44 (22 in the intervention and 22 in the control group) completed the study. 16 participants (8 in the intervention and 8 in the control group) chose to withdraw from the study. Primary outcome Prostate cancer incidence After 6 months, 2/22 (9%) and 4/22 (18%) prostate cancer occurred in the treatment and control group, respectively. After 1 year total prostate cancer cases were 4/22 (18%) in both groups Secondary outcomes Safety data: only Grade 1 and 2 side effects occurred, including nausea, emesis, abdominal pain, insomnia, fatigue and diarrhoea Mean (SD) PSA levels decreased from 5.9 (2.3) ng/mL to 3.8 (1.8) in treatment group and increased from 4.7 (2.5) to 5.8 (2.6) in control group LUTS and QoL scores reported to be improved after 1‐year study but results were NR.
Roshdy 2013	Primary outcome Total fibroid volume decreased by an average of 32.6% in the treatment group and increased by an average of 24.2% in the placebo group Secondary outcomes HRQoL Scale 1: mean change in score for the intervention group of −25.28 (SD ± 17.38) mean change of 7.1 (SD ± 15.5) in placebo group Scale 2: overall increase of 20.7 (SD ± 21) in the percentile scores for HRQoL in the treatment group and 2.19 (SD ± 17.4) in the placebo group Safety data: none of the participants reported any adverse events of any grade throughout the whole study period.
Sinicrope 2017	Percent change in rectal ACF did not differ between the study arms after 6 months of treatment Safety data: similar adverse events between the study arms. 1 participant in placebo arm reported grade 3 adverse event and 2 participants (1 in each arm) reported grade 2+ adverse events, namely elevated AST, ALT in the treatment group
Tsao 2009	Histological response Higher response rate in the 3 combined GTE arms (50%) versus placebo (18.2%), with dose‐dependent effect: 58% in the combined higher‐dose GTE arms (750 and 1000 mg/m2) versus 36.4% (GTE at 500 mg/m2) and 18.2% (placebo) Higher histological response rate in the 3 combined GTE arms (21.4%) vs placebo (9.1%), not dose‐dependent Safety data: adverse events reported by 28 of the 30 (93.3%) participants in treatment groups and 9 of the 11 (81.8%) participants of control group The most frequently reported adverse effects are grade 1 to 2, including insomnia, headache, nausea and nervousness. The grade 3 toxicity was reported by 2 participants in Group B, namely insomnia, diarrhoea and oral/neck pain. Insomnia was the most frequent adverse event.

Study	Random sequence generation	Allocationconcealment	Blinding of participants and personnel	Blinding of outcome assessment	Incomplete outcome data	Selective reporting	Other bias
Bettuzzi 2006	Low risk	Unclear risk	Low risk (all outcomes)	Low risk: PSA levels Unclear risk: prostate cancer and LUTS	Low risk: prostate cancer and PSA levels Unclear risk: LUTS	Unclear risk	Low risk
Dostal 2015	Low risk	Low risk	Low risk (all outcomes)	Low risk (all outcomes)	Low risk (all outcomes)	Low risk	Unclear risk
Dryden 2013	Low risk	Low risk	Low risk (all outcomes)	Low risk (all outcomes)	Low risk (all outcomes)	Low risk	High risk
Garcia 2014	Low risk	Low risk	Low risk (all outcomes)	Low risk (all outcomes)	Low risk (all outcomes)	Low risk	Low risk
Garland 2006	Low risk	Unclear risk	Low risk (all outcomes)	Low risk (all outcomes)	High risk	Low risk	Unclear risk
Kumar 2015	Low risk	Low risk	Low risk (all outcomes)	Low risk (all outcomes)	Low risk: prostate cancer and PSA levels Unclear risk: LUTS	Low risk	Unclear risk
Lane 2018	Low risk	Low risk	Low risk (all outcomes)	Low risk (all outcomes)	Unclear risk (all outcomes)	High risk	Unclear risk
Micali 2017	Low risk	Unclear risk	Low risk (all outcomes)	Low risk (all outcomes)	Low risk (all outcomes)	Unclear risk	High risk
Roshdy 2013	Low risk	Low risk	Low risk (all outcomes)	Low risk (all outcomes)	Unclear risk (all outcomes)	Low risk	Unclear risk
Sinicrope 2017	Low risk	Unclear risk	Low risk (al outcomes)	Unclear risk (all outcomes)	Low risk (all outcomes)	Low risk	Low risk
Tsao 2009	Low risk	Low risk	Low risk (al outcomes)	Low risk: other outcomes Unclear risk: oral lesions	Unclear risk (all outcomes)	Unclear risk	Low risk
LUTS: lower urinary tract symptoms; PSA: prostate‐specific antigens

Study	Selection				Comparability		Outcome			Total (out of 9)
Study	S1	S2	S3	S4	C1	C2	O1	O2	O3	Total (out of 9)
Allen 2004	c(0)	a(1)	b(1)	a(1)	1	0	b(1)	a(1)	a(1)	7
Chyou 1993	b(1)	a(1)	b(1)	a(1)	1	1	b(1)	a(1)	a(1)	9
Dai 2010	a(1)	a(1)	b(1)	a(1)	1	1	b(1)	b(0)	b(1)	8
Fujino 2002 in: Inoue 2009a	a(1)	a(1)	b(1)	a(1)	1	0	b(1)	a(1)	a(1)	8
Galanis 1998	b(1)	a(1)	b(1)	a(1)	1	0	b(1)	b(0)	b(1)	7
Hoshiyama 2002 in: Inoue 2009a	a(1)	a(1)	b(1)	a(1)	1	1	b(1)	a(1)	c(0)	8
Hoshiyama 2004^a in: Inoue 2009a	a(1)	a(1)	a(1)	a(1)	1	1	b(1)	a(1)	c(0)	8
Ide 2007	a(1)	a(1)	b(1)	a(1)	1	1	b(1)	a(1)	a(1)	9
Inoue 2009a	a(1)	a(1)	b(1)	a(1)	1	1	b(1)	a(1)	b(1)	9
Inoue 2009b	a(1)	a(1)	b(1)	a(1)	1	1	b(1)	a(1)	a(1)	9
Ishikawa 2006	a(1)	a(1)	b(1)	a(1)	1	1	b(1)	a(1)	c(0)	8
Iwai 2002	a(1)	a(1)	b(1)	a(1)	1	0	b(1)	a(1)	b(1)	8
Iwasaki 2010a	a(1)	a(1)	b(1)	a(1)	1	1	b(1)	a(1)	b(1)	9
Iwasaki 2010b in: Iwasaki 2010a	a(1)	a(1)	b(1)	a(1)	1	0	b(1)	a(1)	b(1)	8
Key 1999	c(0)	a(1)	b(1)	a(1)	1	0	b(1)	a(1)	a(1)	7
Khan 2004	a(1)	a(1)	b(1)	a(1)	1	1	b(1)	a(1)	a(1)	9
Kikuchi 2006	a(1)	a(1)	b(1)	a(1)	1	1	b(1)	a(1)	b(1)	9
Kurahashi 2007	a(1)	a(1)	b(1)	a(1)	1	1	b(1)	a(1)	b(1)	9
Kurahashi 2009	a(1)	a(1)	b(1)	a(1)	1	1	b(1)	a(1)	b(1)	9
Kuriyama 2006	a(1)	a(1)	b(1)	a(1)	1	1	b(1)	a(1)	a(1)	9
Lee 2007	a(1)	a(1)	b(1)	a(1)	1	1	b(1)	a(1)	b(1)	9
Li 2008	a(1)	a(1)	b(1)	a(1)	1	1	b(1)	a(1)	b(1)	9
Li 2018	c(0)	a(1)	b(1)	a(1)	1	1	b(1)	a(1)	b(1)	8
Lin 2008	a(1)	a(1)	b(1)	a(1)	1	1	b(1)	a(1)	a(1)	9
Liu 2016	a(1)	a(1)	b(1)	a(1)	1	1	b(1)	a(1)	a(1)	9
Luo 2007	a(1)	a(1)	b(1)	a(1)	1	1	b(1)	a(1)	b(1)	9
Makiuchi 2016	a(1)	a(1)	b(1)	a(1)	1	1	b(1)	a(1)	b(1)	9
Michikawa 2011	a(1)	a(1)	b(1)	a(1)	1	1	b(1)	a(1)	b(1)	9
Montague 2012	a(1)	a(1)	b(1)	a(1)	1	1	b(1)	a(1)	b(1)	9
Nagano 2001	c(0)	a(1)	b(1)	a(1)	1	1	b(1)	a(1)	a(1)	8
Naganuma 2009	a(1)	a(1)	b(1)	a(1)	1	1	b(1)	a(1)	a(1)	9
Nakachi 2000	a(1)	a(1)	b(1)	a(1)	1	1	b(1)	a(1)	a(1)	9
Nakamura 2011	a(1)	a(1)	b(1)	a(1)	1	1	b(1)	a(1)	b(1)	9
Nechuta 2012	a(1)	a(1)	b(1)	a(1)	1	1	b(1)	a(1)	a(1)	9
Oba 2006	a(1)	a(1)	b(1)	a(1)	1	1	b(1)	a(1)	b(1)	9
Odegaard 2015	a(1)	a(1)	b(1)	a(1)	1	1	b(1)	a(1)	b(1)	9
Ogawa 2016	a(1)	a(1)	b(1)	a(1)	1	1	b(1)	a(1)	b(1)	9
Saito 2015	a(1)	a(1)	b(1)	a(1)	1	1	b(1)	a(1)	b(1)	9
Sasazuki 2004 in: Inoue 2009a	a(1)	a(1)	b(1)	a(1)	1	1	b(1)	a(1)	b(1)	9
Sauvaget 2005 in: Nagano 2001	c(0)	a(1)	b(1)	a(1)	1	1	b(1)	a(1)	a(1)	8
Severson 1989	a(1)	a(1)	b(1)	a(1)	1	0	b(1)	a(1)	b(1)	8
Shimazu 2008	a(1)	a(1)	b(1)	a(1)	1	1	b(1)	a(1)	b(1)	9
Sun 2007	a(1)	a(1)	b(1)	a(1)	1	1	b(1)	a(1)	b(1)	9
Suzuki 2004	a(1)	a(1)	b(1)	a(1)	1	1	b(1)	a(1)	b(1)	9
Suzuki 2005	a(1)	a(1)	b(1)	a(1)	1	1	b(1)	a(1)	b(1)	9
Suzuki 2009	a(1)	a(1)	b(1)	a(1)	1	1	b(1)	a(1)	b(1)	9
Tamura 2018	a(1)	a(1)	b(1)	a(1)	1	1	b(1)	a(1)	b(1)	9
Tsubono 2001 in: Inoue 2009a	a(1)	a(1)	b(1)	a(1)	1	1	b(1)	a(1)	b(1)	9
Ugai 2017	a(1)	a(1)	b(1)	a(1)	1	1	b(1)	a(1)	b(1)	9
Ugai 2018	a(1)	a(1)	b(1)	a(1)	1	1	b(1)	a(1)	b(1)	9
Ui 2009	a(1)	a(1)	b(1)	a(1)	1	1	b(1)	a(1)	b(1)	9
Yang 2007 in: Nechuta 2012	a(1)	a(1)	b(1)	a(1)	1	1	b(1)	a(1)	b(1)	9
Yang 2011a	a(1)	a(1)	b(1)	a(1)	1	1	b(1)	a(1)	b(1)	9
Zhao 2017	a(1)	a(1)	b(1)	a(1)	1	1	b(1)	a(1)	b(1)	9

Date	Event	Description
18 July 2019	New citation required but conclusions have not changed	We included an additional 130 references relating to 93 studies. However the conclusions overall remain unchanged.
18 July 2019	New search has been performed	An updated search from January 2009 to January 2019 was conducted and 93 new studies were added to 52 in the review.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1.1 Prostate cancer incidence	3	201	Risk Ratio (IV, Random, 95% CI)	0.50 [0.18, 1.36]
1.2 Gynaecological cancer incidence	2	1157	Risk Ratio (M‐H, Random, 95% CI)	1.50 [0.41, 5.48]
1.2.1 Endometrial cancer	1	1075	Risk Ratio (M‐H, Random, 95% CI)	0.33 [0.01, 8.15]
1.2.2 Cervical cancer	1	82	Risk Ratio (M‐H, Random, 95% CI)	2.00 [0.54, 7.46]
1.3 Non‐melanoma skin cancer incidence	1	1075	Risk Ratio (M‐H, Random, 95% CI)	1.00 [0.06, 15.92]

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
2.1 Any cancer incidence	3	52479	Risk Ratio (IV, Random, 95% CI)	0.83 [0.65, 1.07]
2.1.1 Cohort studies	2	51629	Risk Ratio (IV, Random, 95% CI)	0.81 [0.50, 1.32]
2.1.2 Population‐based case‐control studies	1	850	Risk Ratio (IV, Random, 95% CI)	0.78 [0.60, 1.01]
2.2 Any cancer mortality	8	504366	Risk Ratio (IV, Random, 95% CI)	0.99 [0.91, 1.07]
2.3 Oral cancer	5	55977	Risk Ratio (IV, Fixed, 95% CI)	0.71 [0.62, 0.82]
2.3.1 Cohort studies	1	50258	Risk Ratio (IV, Fixed, 95% CI)	0.44 [0.19, 1.04]
2.3.2 Population‐based case‐control studies	1	1721	Risk Ratio (IV, Fixed, 95% CI)	0.58 [0.42, 0.79]
2.3.3 Hospital‐based case‐control studies	3	3998	Risk Ratio (IV, Fixed, 95% CI)	0.77 [0.65, 0.90]
2.4 Oral, pharyngeal and laryngeal cancer	1	2040	Risk Ratio (IV, Random, 95% CI)	1.47 [1.12, 1.93]
2.4.1 Cohort studies	1	2040	Risk Ratio (IV, Random, 95% CI)	1.47 [1.12, 1.93]
2.5 Pharyngeal cancer	1	12282	Risk Ratio (IV, Random, 95% CI)	0.83 [0.30, 2.30]
2.5.1 Hospital‐based case‐control study	1	12282	Risk Ratio (IV, Random, 95% CI)	0.83 [0.30, 2.30]
2.6 Any gut cancer	7	70299	Risk Ratio (IV, Random, 95% CI)	0.78 [0.59, 1.02]
2.6.1 Cohort studies	2	52298	Risk Ratio (IV, Random, 95% CI)	0.86 [0.27, 2.79]
2.6.2 Population‐based case‐control studies	1	1721	Risk Ratio (IV, Random, 95% CI)	0.58 [0.42, 0.79]
2.6.3 Hospital‐based case‐control studies	4	16280	Risk Ratio (IV, Random, 95% CI)	0.77 [0.65, 0.90]
2.7 Oesophageal cancer	13	74895	Risk Ratio (IV, Random, 95% CI)	0.81 [0.64, 1.04]
2.7.1 Cohort studies	1	26801	Risk Ratio (IV, Random, 95% CI)	1.67 [0.88, 3.16]
2.7.2 Population‐based case‐control studies	9	14111	Risk Ratio (IV, Random, 95% CI)	0.74 [0.55, 1.00]
2.7.3 Hospital‐based case‐control studies	3	33983	Risk Ratio (IV, Random, 95% CI)	0.86 [0.57, 1.27]
2.8 Stomach cancer	18	438595	Risk Ratio (IV, Random, 95% CI)	0.86 [0.74, 1.01]
2.8.1 Cohort studies	7	398286	Risk Ratio (IV, Random, 95% CI)	0.99 [0.85, 1.14]
2.8.2 Population‐based case‐control studies	8	9923	Risk Ratio (IV, Random, 95% CI)	0.74 [0.53, 1.02]
2.8.3 Hospital‐based case‐control studies	3	30386	Risk Ratio (IV, Random, 95% CI)	0.90 [0.74, 1.09]
2.9 Liver cancer	6	198885	Risk Ratio (IV, Random, 95% CI)	0.88 [0.68, 1.14]
2.9.1 Cohort studies	5	198266	Risk Ratio (IV, Random, 95% CI)	0.93 [0.71, 1.20]
2.9.2 Population‐based case‐control studies	1	619	Risk Ratio (IV, Random, 95% CI)	0.55 [0.28, 1.09]
2.10 Pancreatic cancer	10	326564	Risk Ratio (IV, Random, 95% CI)	0.88 [0.70, 1.10]
2.10.1 Cohort studies	6	320596	Risk Ratio (IV, Random, 95% CI)	1.04 [0.84, 1.30]
2.10.2 Population‐based case‐control studies	3	5720	Risk Ratio (IV, Random, 95% CI)	0.67 [0.48, 0.96]
2.10.3 Hospital‐based case‐control studies	1	248	Risk Ratio (IV, Random, 95% CI)	1.94 [1.06, 3.55]
2.11 Biliary tract cancer	3	195800	Risk Ratio (IV, Random, 95% CI)	0.79 [0.57, 1.11]
2.11.1 Cohort studies	3	195800	Risk Ratio (IV, Random, 95% CI)	0.79 [0.57, 1.11]
2.12 Colorectal cancer	16	610295	Risk Ratio (IV, Random, 95% CI)	0.84 [0.74, 0.96]
2.12.1 Cohort studies	9	554298	Risk Ratio (IV, Random, 95% CI)	1.00 [0.92, 1.08]
2.12.2 Population‐based case‐control studies	5	12811	Risk Ratio (IV, Random, 95% CI)	0.74 [0.61, 0.90]
2.12.3 Hospital‐based case‐control studies	2	43186	Risk Ratio (IV, Random, 95% CI)	0.53 [0.17, 1.60]
2.13 Colon cancer	10	389974	Risk Ratio (IV, Random, 95% CI)	0.89 [0.80, 0.98]
2.13.1 Cohort studies	6	361348	Risk Ratio (IV, Random, 95% CI)	0.93 [0.82, 1.05]
2.13.2 Population‐based case‐control studies	3	7136	Risk Ratio (IV, Random, 95% CI)	0.83 [0.69, 1.00]
2.13.3 Hospital‐based case‐control studies	1	21490	Risk Ratio (IV, Random, 95% CI)	0.77 [0.47, 1.26]
2.14 Rectal cancer	9	356851	Risk Ratio (IV, Random, 95% CI)	0.89 [0.75, 1.05]
2.14.1 Cohort studies	5	329570	Risk Ratio (IV, Random, 95% CI)	0.92 [0.77, 1.09]
2.14.2 Population‐based case‐control studies	3	5887	Risk Ratio (IV, Random, 95% CI)	0.83 [0.56, 1.24]
2.14.3 Hospital‐based case‐control studies	1	21394	Risk Ratio (IV, Random, 95% CI)	1.25 [0.62, 2.51]
2.15 Nasopharyngeal carcinoma	2	2290	Risk Ratio (IV, Random, 95% CI)	0.49 [0.36, 0.67]
2.15.1 Population‐based case‐control studies	1	685	Risk Ratio (IV, Random, 95% CI)	0.61 [0.41, 0.91]
2.15.2 Hospital‐based case‐control studiesCohort studies	1	1605	Risk Ratio (IV, Random, 95% CI)	0.44 [0.36, 0.54]
2.16 Lung cancer	17	269565	Risk Ratio (IV, Random, 95% CI)	0.88 [0.76, 1.02]
2.16.1 Cohort studies	6	240987	Risk Ratio (IV, Random, 95% CI)	1.02 [0.79, 1.31]
2.16.2 Population‐based case‐control studies	5	9703	Risk Ratio (IV, Random, 95% CI)	0.73 [0.61, 0.87]
2.16.3 Hospital‐based case‐control studies	6	18875	Risk Ratio (IV, Random, 95% CI)	0.90 [0.69, 1.17]
2.17 Breast cancer	14	250822	Risk Ratio (IV, Random, 95% CI)	0.88 [0.75, 1.02]
2.17.1 Cohort studies	5	235706	Risk Ratio (IV, Random, 95% CI)	1.01 [0.86, 1.19]
2.17.2 Population‐based case‐control studies	4	9336	Risk Ratio (IV, Random, 95% CI)	0.87 [0.70, 1.08]
2.17.3 Hospital‐based case‐control studies	5	5780	Risk Ratio (IV, Random, 95% CI)	0.81 [0.58, 1.13]
2.18 Gynaecological cancer	10	66738	Risk Ratio (IV, Random, 95% CI)	0.69 [0.57, 0.83]
2.18.1 Cohort studies	1	53841	Risk Ratio (IV, Random, 95% CI)	0.75 [0.43, 1.30]
2.18.2 Population‐based case‐control studies	9	12897	Risk Ratio (IV, Random, 95% CI)	0.68 [0.56, 0.84]
2.19 Endometrial cancer	5	60416	Risk Ratio (IV, Random, 95% CI)	0.77 [0.65, 0.91]
2.19.1 Cohort studies	1	53841	Risk Ratio (IV, Random, 95% CI)	0.75 [0.43, 1.30]
2.19.2 Population‐based case‐control studies	4	6575	Risk Ratio (IV, Random, 95% CI)	0.75 [0.61, 0.94]
2.20 Ovarian cancer	5	6322	Risk Ratio (IV, Random, 95% CI)	0.64 [0.45, 0.90]
2.20.1 Population‐based case‐control studies	5	6322	Risk Ratio (IV, Random, 95% CI)	0.64 [0.45, 0.90]
2.21 Prostate cancer	13	127239	Risk Ratio (IV, Random, 95% CI)	0.73 [0.56, 0.94]
2.21.1 Cohort studies	5	123289	Risk Ratio (IV, Random, 95% CI)	1.09 [0.89, 1.32]
2.21.2 Population‐based case‐control studies	1	750	Risk Ratio (IV, Random, 95% CI)	0.59 [0.40, 0.87]
2.21.3 Hospital‐based case‐control studies	7	3200	Risk Ratio (IV, Random, 95% CI)	0.50 [0.39, 0.63]
2.22 Renal cancer	1	549	Risk Ratio (IV, Random, 95% CI)	0.34 [0.21, 0.55]
2.22.1 Hospital‐based case‐control studies	1	549	Risk Ratio (IV, Random, 95% CI)	0.34 [0.21, 0.55]
2.23 Urinary tract cancer	7	156039	Risk Ratio (IV, Random, 95% CI)	1.04 [0.79, 1.37]
2.23.1 Cohort studies	3	151395	Risk Ratio (IV, Random, 95% CI)	1.24 [0.87, 1.76]
2.23.2 Population‐based case‐control studies	1	783	Risk Ratio (IV, Random, 95% CI)	1.08 [0.61, 1.92]
2.23.3 Hospital‐based case‐control studies	3	3861	Risk Ratio (IV, Random, 95% CI)	0.84 [0.53, 1.32]
2.24 Hematopoietic cancer	2	80646	Risk Ratio (IV, Random, 95% CI)	0.75 [0.45, 1.27]
2.24.1 Cohort studies	2	80646	Risk Ratio (IV, Random, 95% CI)	0.75 [0.45, 1.27]
2.25 Leukaemia	5	97778	Risk Ratio (IV, Random, 95% CI)	0.81 [0.57, 1.15]
2.25.1 Cohort studies ‐ acute myeloid leukaemia	1	95892	Risk Ratio (IV, Random, 95% CI)	1.20 [0.62, 2.32]
2.25.2 Population‐based case‐control studies ‐ all leukaemia	2	785	Risk Ratio (IV, Random, 95% CI)	1.03 [0.50, 2.14]
2.25.3 Hospital‐based case‐control studies ‐ all leukaemia	2	1101	Risk Ratio (IV, Random, 95% CI)	0.64 [0.45, 0.91]
2.26 Lymphoma	1	96218	Risk Ratio (IV, Random, 95% CI)	0.89 [0.61, 1.30]
2.26.1 Cohort studies	1	96218	Risk Ratio (IV, Random, 95% CI)	0.89 [0.61, 1.30]
2.27 Multiple myeloma	2	96385	Risk Ratio (IV, Random, 95% CI)	0.50 [0.26, 0.95]
2.27.1 Cohort studies	1	95945	Risk Ratio (IV, Random, 95% CI)	0.74 [0.39, 1.41]
2.27.2 Hospital‐based case‐control studies	1	440	Risk Ratio (IV, Random, 95% CI)	0.38 [0.27, 0.53]
2.28 Non‐melanoma skin cancer	1	450	Risk Ratio (IV, Random, 95% CI)	0.82 [0.35, 1.90]
2.28.1 Population‐based case‐control studies	1	450	Risk Ratio (IV, Random, 95% CI)	0.82 [0.35, 1.90]
2.29 Thyroid cancer	1	100666	Risk Ratio (IV, Random, 95% CI)	0.88 [0.56, 1.37]
2.29.1 Cohort studies	1	100666	Risk Ratio (IV, Random, 95% CI)	0.88 [0.56, 1.37]
2.30 Brain cancer	1	106479	Risk Ratio (IV, Random, 95% CI)	1.07 [0.71, 1.62]
2.30.1 Cohort studies	1	106479	Risk Ratio (IV, Random, 95% CI)	1.07 [0.71, 1.62]

References	Detailed summary results	The Newcastle‐Ottawa Scale (NOS)
Bandera 2010	Reference category: lowest exposure Endometrial cancer Intermediate exposure: OR 1.04 (95% CI 0.72‐1.50) Highest exposure: OR 0.76 (95% CI 0.48‐1.21)	Low risk: Selection: 4/4 stars, low risk Comparability: 2/2 stars, low risk Exposure: 2/3 stars, difference in response rate between cases and controls or no designation Total score: 8/9 stars, moderate quality.
Berroukche 2012	Reference category: lowest exposure Prostate cancer Intermediate exposure 1: OR 1.4 (95% CI 0.8‐2.2) Intermediate exposure 2: OR 0.9 (95% CI 0.5‐1.4) Highest exposure: OR 0.6 (95% CI 0.3‐1.1)	Low risk: Selection: 3/4 stars, hospital controls Comparability: 2/2 stars, low risk Exposure: 3/3 stars, low risk Total score: 8/9 stars, moderate quality.
Bonner 2005	Reference category: lowest exposure Lung cancer Intermediate exposure: OR 0.84 (95% CI 0.38‐1.85) Highest exposure: OR 0.59 (95% CI 0.26‐1.37)	Low risk: Selection: 3/4 stars, no description of source of controls Comparability: 2/2 stars, low risk Exposure: 2/3 stars, low risk Total score: 7/9 stars, moderate quality.
Chen 2011	Oesophageal cancer Reference category: lowest exposure Intermediate exposure 1: OR 1.27 (95% CI 0.72‐1.89) Intermediate exposure 2: OR 0.97 (95% CI 0.59‐2.56) Highest exposure: OR 0.92 (95% CI 0.49‐2.32)	Low risk: Selection: 3/4 stars, hospital controls Comparability: 2/2 stars, low risk Exposure: 3/3 stars, low risk Total score: 8/9 stars, moderate quality.
Chen 2017a	Oral cancer Reference category: lowest exposure Chen 2015 Highest exposure: OR 0.48 (95% CI 0.28‐0.82) Chen 2016 Highest exposure: OR 0.501 (95% CI 0.284‐0.883) Chen 2017a non‐smokers Highest exposure: OR 0.515 (95% CI 0.323‐0.821) Former/current smokers Highest exposure: OR 0.849 (95% CI 0.556‐1.298) Non‐alcohol drinkers Highest exposure: OR 0.551 (95% CI 0.372‐0.817) Former/current alcohol drinkers Highest exposure: OR 0.934 (95% CI 0.934‐1.563)	Low risk: Chen 2015 Selection: 4/4 stars, low risk Comparability: 2/2 stars, low risk Exposure: 3/3 stars, low risk Total score: 9/9 stars, high quality. Chen 2016 Selection: 3/4 stars, hospital controls Comparability: 2/2 stars, low risk Exposure: 3/3 stars, low risk Total score: 8/9 stars, moderate quality. Chen 2017a Selection: 3/4 stars, hospital controls Comparability: 2/2 stars, low risk Exposure: 3/3 stars, low risk Total score: 8/9 stars, moderate quality.
Fu 2013	Oral cancer Reference category: lowest exposure Men Intermediate exposure: OR 0.82 (95% CI 0.69‐1.07) Highest exposure: OR 0.72 (95% CI 0.54‐0.93) Women Intermediate exposure: OR 1.00 (95% CI 0.80‐1.25) Highest exposure: OR 0.93 (95% CI 0.74‐1.26)	Low risk: Selection: 3/4 stars, hospital controls Comparability: 2/2 stars, low risk Exposure: 3/3 stars, low risk Total score: 8/9 stars, moderate quality.
Gao 1994	Oesophageal cancer Reference category: lowest exposure Exposure assessment A: intake of green tea Men Highest exposure: OR 0.80 (95% CI 0.58‐1.09) Women Highest exposure: OR 0.50 (95% CI 0.30‐0.83) Exposure assessment B: green tea consumption Men Intermediate exposure: OR 0.79 (95% CI 0.53‐1.17) Highest exposure: OR 0.79 (95% CI 0.56‐1.13) Women Intermediate exposure: OR 0.77 (95% CI 0.39‐1.53) Highest exposure: OR 0.34 (95% CI 0.17‐0.69)	Low risk: Selection: 3/4 stars, hospital controls Comparability: 2/2 stars, low risk Exposure: 2/3 stars, non‐response rate difference between cases and controls or NR Total score: 7/9 stars, moderate quality.
Gao 2005	Reference category: lowest exposure Endometrial cancer Highest exposure: OR 0.81 (95% CI 0.65‐1.00)	Low risk: Selection: 4/4 stars, low risk Comparability: 1/2 stars, the study did not control for smoking. Exposure: 2/3 stars, nonresponse rate different between cases and controls or NR Total score: 7/9 stars, moderate quality.
Gavrilas 2018	Colorectal cancer Reference category: lowest exposure Intermedaite exposure 1: OR 0.34 (95% CI 0.15‐0.76) Intermediate exposure 2: OR 0.03 (95% CI 0.01‐0.10) Highest exposure: OR 0.14 (95% CI 0.05‐0.34)	Low risk: Selection: 3/4 stars, no description of controls Comparability: 2/2 stars, low risk Exposure: 2/3 stars, non‐response rate NR Total score: 7/9 stars, moderate quality.
Goodman 2003	Ovarian cancer Reference category: lowest exposure Highest exposure: OR 0.9 (95% CI 0.6‐1.5) Highest exposure A: OR 1.0 (95% CI 0.6‐1.9) Highest exposure B: OR 0.9 (95% CI 0.5‐1.6)	Low risk: Selection: 4/4 stars, low risk Comparability: 1/2 stars, the study did not control for smoking Exposure: 3/3 stars, low risk Total score: 8/9 stars, moderate quality.
Goto 1990	Pancreatic cancer Reference category: lowest exposure Highest exposure: OR 0.34 (95% CI 0.17‐0.67)	Low risk: Selection: 3/4 stars, no description of source of controls Comparability: 2/2 stars, low risk Exposure: 2/3 stars, non‐response rate different between cases and controls or NR Total score: 7/9 stars, moderate quality.
Green 2014	Reference category: lowest exposure Colorectal cancer Intermediate exposure 1: OR 0.99 (95% CI 0.64‐1.52) Intermediate exposure 2: OR 1.15 (95% CI 0.62‐2.13) Highest exposure: OR 0.99 (95% CI 0.62‐1.58) Proximal colon cancer Intermediate exposure 1: OR 1.36 (95% CI 0.76‐2.41) Intermediate exposure 2: OR 0.61 (95% CI 0.20‐1.83) Highest exposure: OR 0.95 (95% CI 0.50‐1.78) Distal colon cancer Intermediate exposure 1: OR 0.75 (95% CI 0.38‐1.46) Intermediate exposure 2: OR 1.31 (95% CI 0.55‐3.10) Highest exposure: OR 0.97 (95% CI 0.47‐2.03) Rectal cancer Intermediate exposure 1: OR 0.91 (95% CI 0.48‐1.71) Intermediate exposure 2: OR 1.49 (95% CI 0.67‐3.32) Highest exposure: OR 1.05 (95% CI 0.53‐2.10)	Low risk: Selection: 4/4 stars, low risk Comparability: 2/2 stars, low risk Exposure: 2/3 stars, non‐response rate different between cases and controls or NR Total score: 8/9 stars, moderate quality.
Hakim 2000	Skin cancer Reference category: lowest exposure Highest exposure: OR 0.82 (95% CI 0.35‐1.90)	Low risk: Selection: 4/4 stars, low risk Comparability: 1/2 stars, the study did not control for smoking. Exposure: 3/3 stars, low risk Total score: 8/9 stars, moderate quality.
Han 2008	Reference category: lowest exposure Lung cancer Exposure assessment 1: Highest exposure: OR 0.520 (95% CI 0.392‐0.691) Exposure assessment 2: Intermediate exposure 1: OR 0.222 (95% CI 0.127‐0.390) Intermediate exposure 2: OR 0.687 (95% CI 0.448‐1.052) Highest exposure: OR 0.734 (95% CI 0.476‐1.132)	Low risk: Selection: 4/4 stars, low risk Comparability: 2/2 stars, low risk Exposure: 2/3 stars, non‐response rate different between cases and controls or NR Total score: 8/9 stars, moderate quality.
Hemelt 2010	Bladder cancer Reference category: lowest exposure Exposure assessment 1: Highest exposure: OR 0.95 (95% CI 0.68‐1.32) Exposure assessment 2: Intermediate exposure: OR 0.83 (95% CI 0.54‐1.27) Highest exposure: OR 1.02 (95% CI 0.71‐1.48) further divided by cups/d: < 4 cups/d OR 1.23 (95% CI 0.76‐1.97); > 4 cups/d OR 0.83 (95% CI 0.53‐1.28)	Low risk: Selection: 3/4 stars, hospital controls Comparability: 2/2 stars, low risk Exposure: 2/3 stars, non‐response rate different between cases and controls or NR Total score: 7/9 stars, moderate quality.
Hoshiyama 1992	Reference category: lowest exposure Single stomach cancer Intermediate exposure: OR 1.2 (95% CI 0.8‐1.7) Highest exposure: OR 0.9 (95% CI 0.6‐1.3) Multiple stomach cancer Intermediate exposure: OR 1.5 (95% CI 0.6‐3.5) Highest exposure: OR 1.6 (95% CI 0.7‐3.9)	Low risk: Selection: 4/4 stars, low risk Comparability: 2/2 stars, low risk Exposure: 3/3 stars, non‐response rate different between cases and controls or NR Total score: 8/9 stars, moderate quality.
Hsu 2012	Reference category: lowest exposure Nasopharyngeal carcinoma Intermediate exposure: OR 0.58 (95% CI 0.35‐0.98) Highest exposure: OR 0.61 (95% CI 0.40‐0.91)	Low risk: Selection: 4/4 stars, low risk Comparability: 2/2 stars, low risk Exposure: 2/3 stars, non‐response rate different between cases and controls or NR Total score: 8/9 stars, moderate quality.
Huang 1999	Huang 1999 Reference category: lowest exposure Stomach cancer Intermediate exposure 1: OR 0.88 (95% CI 0.73‐1.05) Intermediate exposure 2: OR 1.08 (95% CI 0.90‐1.24) Highest exposure: OR 0.90 (95% CI 0.73‐1.11) Inoue 1994 Reference category: lowest exposure Stomach cancer (total) Highest exposure: OR 1.09 (95% CI 0.83‐1.43) Subsite Cardia Highest exposure: OR 1.12 (95% CI 0.70‐1.79) Middle Highest exposure: OR 1.06 (95% CI 0.73‐1.54) Antrum Highest exposure: OR 1.10 (95% CI 0.78‐1.57) Kato 1990a Reference category: lowest exposure Stomach cancer Men Intermediate exposure: RR 1.14 (95% CI 0.82‐1.60) Highest exposure: RR 1.01 (95% CI 0.70‐1.47) Women Intermediate exposure: RR 0.71 (95% CI 0.45‐1.14) Highest exposure: RR 0.81 (95% CI 0.51‐1.27)	High risk: Huang 1999 Selection: 3/4 stars, hospital controls Comparability: 2/2 stars, low risk Exposure: 2/3 stars, non‐respondent described Total score: 7/9 stars, moderate quality. Kato 1990a Selection: 2/4 stars, potential bias in selection of cases and hospital controls Comparability: 1/2 stars, the study did not control for smoking. Exposure: 2/3 stars, self‐reported exposure not within a structured interview or questionnaire. Total score: 5/9 stars, low quality. Inoue 1994 Selection: 3/4 stars, hospital controls Comparability: 1/2 stars, the study did not control for smoking. Exposure: 2/3 stars, self‐reported exposure not within a structured interview or questionnaire. Total score: 6/9 stars, low quality.
Inoue 1998	Reference category: lowest exposure Oesophageal cancer Intermediate exposure 1: OR 1.02 (95% CI 0.50‐2.10) Intermediate exposure 2: OR 1.07 (95% CI 0.58‐2.00) Intermediate exposure 3: OR 0.96 (95% CI 0.50‐1.83) Highest exposure: OR 1.14 (95% CI 0.55‐2.34) Stomach cancer (also reported in Inoue 1994 in: Huang 1999 and Inoue 2009a) Intermediate exposure 1: OR 1.00 (95% CI 0.77‐1.44) Intermediate exposure 2: OR 0.96 (95% CI 0.70‐1.32) Intermediate exposure 3: OR 1.01 (95% CI 0.74‐1.39) Highest exposure: OR 0.69 (95% CI 0.48‐1.00) Colon cancer Intermediate exposure 1: OR 0.62 (95% CI 0.36‐1.05) Intermediate exposure 2: OR 0.64 (95% CI 0.42‐1.00) Intermediate exposure 3: OR 0.76 (95% CI 0.49‐1.17) Highest exposure: OR 0.77 (95% CI 0.47‐1.26) Rectal cancer Intermediate exposure 1: OR 1.41 (95% CI 0.70‐2.83) Intermediate exposure 2: OR 1.04 (95% CI 0.55‐1.98) Intermediate exposure 3: OR 1.42 (95% CI 0.75‐2.69) Highest exposure: OR 1.25 (95% CI 0.62‐2.51)	Low risk: Selection: 3/4 stars, hospital controls Comparability: 2/2 stars, low risk Exposure: 2/3 stars, self‐reported exposure not within a structured interview or questionnaire. Total score: 7/9 stars, moderate quality.
Inoue 2008	Reference category: lowest exposure Breast cancer Intermediate exposure: OR 0.65 (95% CI 0.45‐0.94) Highest exposure: OR 1.00 (95% CI 0.82‐1.22)	Low risk: Selection: 4/4 stars, low risk Comparability: 1/2 stars, the study did not control for smoking Exposure: 2/3 stars, non‐response rate different between cases and controls Total score: 7/9 stars, moderate quality.
Islami 2009	Reference category: lowest exposure Oesophageal cancer Highest exposure: OR 0.89 (95% CI 0.38‐2.09)	Low risk: Selection: 4/4 stars, low risk Comparability: 2/2 stars, low risk Exposure: 2/3 stars, non‐respondents described Total score: 8/9 stars, moderate quality.
Iwasaki 2014	Reference category: lowest exposure Breast cancer Intermediate exposure: OR 0.86 (95% CI 0.53‐1.41) Highest exposure: OR 1.27 (95% CI 0.75‐2.14)	Low risk: Selection: 3/4 stars, hospital controls Comparability: 2/2 stars, low risk Exposure: 2/3 stars, non‐respondents described Total score: 7/9 stars, moderate quality.
Ji 1996	Reference category: lowest exposure Stomach cancer Men Exposure assessment A: green tea drinking status Highest exposure: OR 0.96 (95% CI 0.77‐1.21) Exposure assessment B: consumption of green tea leaves Intermediate exposure 1: OR 1.06 (95% CI 0.76‐1.49) Intermediate exposure 2: OR 1.15 (95% CI 0.82‐1.61) Intermediate exposure 3: OR 0.88 (95% CI 0.55‐1.24) Highest exposure: OR 0.76 (95% CI 0.55‐1.27) Women Exposure assessment A: green tea drinking status Highest exposure: OR 0.77 (95% CI 0.52‐1.13) Exposure assessment B: consumption of green tea leaves Intermediate exposure: OR 0.74 (95% CI 0.45‐1.21) Highest exposure: OR 0.81 (95% CI 0.46‐1.43)	Low risk: Selection: 4/4 stars, low risk Comparability: 2/2 stars, low risk Exposure: 2/3 stars, non‐response rate different between cases and controls or NR Total score: 8/9 stars, moderate quality.
Ji 1997	Reference category: lowest exposure Men Colon cancer Exposure assessment A: green tea drinking habit Highest exposure: OR 0.99 (95% CI 0.74‐1.33) Exposure assessment B: green tea consumption Intermediate exposure 1: OR 1.13 (95% CI 0.80‐1.61) Intermediate exposure 2: OR 0.92 (95% CI 0.62‐1.37) Highest exposure: OR 0.82 (95% CI 0.52‐1.28) Exposure assessment C: lifetime green tea consumption Intermediate exposure 1: OR 0.93 (95% CI 0.63‐1.35) Intermediate exposure 2: OR 1.06 (95% CI 0.72‐1.56) Highest exposure: OR 1.01 (95% CI 0.67‐1.51) Rectal cancer Exposure assessment A: green tea drinking habit Highest exposure: OR 0.82 (95% CI 0.61‐1.10) Exposure assessment B: green tea consumption Intermediate exposure 1: OR 0.99 (95% CI 0.69‐1.41) Intermediate exposure 2: OR 0.66 (95% CI 0.43‐0.99) Highest exposure: OR 0.72 (95% CI 0.46‐1.13) Exposure assessment C: lifetime green tea consumption Intermediate exposure 1: OR 0.88 (95% CI 0.60‐1.27) Intermediate exposure 2: OR 0.75 (95% CI 0.50‐1.13) Highest exposure: OR 0.80 (95% CI 0.53‐1.23) Pancreatic cancer Exposure assessment A: green tea drinking habit Highest exposure: OR 0.88 (95% CI 0.60‐1.31) Exposure assessment B: green tea consumption Intermediate exposure 1: OR 1.23 (95% CI 0.79‐1.92) Intermediate exposure 2: OR 0.57 (95% CI 0.32‐1.03) Highest exposure: OR 0.63 (95% CI 0.34‐1.17) Exposure assessment C: lifetime green tea consumption Intermediate exposure 1: OR 1.11 (95% CI 0.69‐1.79) Intermediate exposure 2: OR 0.86 (95% CI 0.51‐1.45) Highest exposure: OR 0.59 (95% CI 0.32‐1.07) Women Colon cancer Exposure assessment A: green tea drinking habit Highest exposure: OR 0.77 (95% CI 0.56‐1.06) Exposure assessment B: green tea consumption Intermediate exposure: OR 0.83 (95% CI 0.57‐1.21) Highest exposure: OR 0.67 (95% CI 0.41‐1.10) Exposure assessment C: lifetime green tea consumption Intermediate exposure: OR 0.87 (95% CI 0.60‐1.27) Highest exposure: OR 0.62 (95% CI 0.38‐1.02) Rectal cancer Exposure assessment A: green tea drinking habit Highest exposure: OR 0.51 (95% CI 0.36‐0.73) Exposure assessment B: green tea consumption Intermediate exposure: OR 0.51 (95% CI 0.33‐0.79) Highest exposure: OR 0.57 (95% CI 0.34‐0.97) Exposure assessment C: lifetime green tea consumption Intermediate exposure: OR 0.54 (95% CI 0.35‐0.84) Highest exposure: OR 0.52 (95% CI 0.30‐0.89) Pancreatic cancer: Exposure assessment A: green tea drinking habit Highest exposure: OR 0.47 (95% CI 0.29‐0.77) Exposure assessment B: green tea consumption Intermediate exposure: OR 0.47 (95% CI 0.25‐0.89) Highest exposure: OR 0.53 (95% CI 0.25‐1.09) Exposure assessment C: lifetime green tea consumption Intermediate exposure: OR 0.59 (95% CI 0.32‐1.10) Highest exposure: OR 0.38 (95% CI 0.18‐0.82)	Low risk: Selection: 4/4 stars, low risk Comparability: 2/2 stars, low risk Exposure: 2/3 stars, non‐respondents described Total score: 8/9 stars, moderate quality.
Jia 2016	Reference category: lowest exposure Lung cancer + mesothelioma mortality Intermediate exposure 1: OR 0.88 (95% CI 0.66‐0.87) Intermediate exposure 2: OR 2.18 (95% CI 0.96‐3.50) Highest exposure: OR 0.70 (95% CI 0.57‐0.86)	Low risk: Selection: 4/4 stars, low risk Comparability: 1/2 stars, the study did not control for smoking Exposure: 2/3 stars, interview not blinded to case/control status Total score: 7/9 stars, moderate quality.
Jian 2004	Prostate cancer Reference category: lowest exposure Exposure assessment A: green tea drinking habits Highest exposure: OR 0.28 (95% CI 0.17‐0.47) Exposure assessment B: intake of green tea Intermediate exposure: OR 0.53 (95% CI 0.30‐0.94) Highest exposure: OR 0.27 (95% CI 0.15‐0.48)	Low risk: Selection: 3/4 stars, hospital controls Comparability: 2/2 stars, low risk Exposure: 3/3 stars, low risk Total score: 8/9 stars, moderate quality.
Jin 2013	Reference category: lowest exposure Lung cancer Exposure assessment A: green tea drinking habits Highest exposure: OR 0.78 (95% CI 0.65‐0.95) Exposure assessment B: consumption of green tea Intermediate exposure 1: OR 0.76 (95% CI 0.54‐1.07) Intermediate exposure 2: OR 0.84 (95% CI 0.66‐1.08) Highest exposure: OR 0.74 (95% CI 0.56‐0.99)	Low risk: Selection: 4/4 stars, low risk Comparability: 2/2 stars, low risk Exposure: 3/3 stars, low risk Total score: 9/9 stars, high quality.
Kakuta 2009	Reference category: lowest exposure Endometrial cancer Intermediate exposure 1: OR 0.77 (95% CI 0.37‐1.58) Intermediate exposure 2: OR 0.61 (95% CI 0.30‐1.23) Highest exposure: OR 0.33 (95% CI 0.15‐0.75)	Low risk: Selection: 3/4 stars, hospital controls Comparability: 2/2 stars, low risk Exposure: 3/3 stars, low risk Total score: 8/9 stars, moderate quality.
Kato 1990	Reference category: lowest exposure Colon cancer Highest exposure: RR 0.61 (95% CI 0.41‐0.91) Rectal cancer Highest exposure: RR 1.32 (95% CI 0.78‐2.23)	High risk: Selection: 3/4 stars, potential for selection bias in cases Comparability: 1/2 stars, the study did not control for smoking Exposure: 2/3 stars, self‐reported exposure not within a structured interview or questionnaire. Total score: 6/9 stars, low quality.
Kono 1988	Stomach cancer Reference category: lowest exposure Population controls Highest exposure: OR 0.3 (95% CI 0.1‐0.7) Hospital controls Highest exposure: OR 0.5 (95% CI 0.3‐1.1)	High risk: Selection: 4/4 stars when using population controls and 3/4 when using hospital controls Comparability: 2/2 stars, low risk Exposure: 1/3 stars, different methods of exposure ascertainment in cases and controls and self‐reported exposure not within a structured interview or questionnaire. Total score: 7/9 stars, moderate quality and 6/9 stars low quality.
Kubik 2008	Kubik 2008 Reference category: lowest exposure (OR for trend only reported) Lung cancer Men nonsmokers Highest exposure: OR 1.08 (95% CI 0.56‐2.08) Men smokers Highest exposure: OR 0.93 (95% CI 0.63‐1.38) Women nonsmokers Highest exposure: OR 0.88 (95% CI 0.61‐1.27) Women smokers Highest exposure: OR 1.09 (95% CI 0.80‐1.49) Data on former smokers who quit from 10 to < 20 years were excluded from the reported analysis. Previous report on fewer female cases reported in Kubik 2004. Kubik 2004 Reference category: lowest exposure Lung cancer Intermediate exposure: OR 0.97 (95% CI 0.69‐1.38) Highest exposure: OR 1.02 (95% CI 0.74‐1.40) Also analysis available stratified by smoking habits Subsequent report in Kubik 2008 with longer recruitment of women from the same area	Low risk: Kubik 2004 Selection: 3/4 stars, hospital controls Comparability: 2/2 stars, low risk Exposure: 2/3 stars, non‐response rate different between cases and controls or NR Total score: 7/9 stars, moderate quality. Kubik 2008 Selection: 3/4 stars, hospital controls Comparability: 2/2 stars, low risk Exposure: 2/3 stars, non‐response rate different between cases and controls or NR Total score: 7/9 stars, moderate quality.
Kuo 2009	Leukaemia Green‐tea‐only consumption compared to no‐tea consumption Cases exposed = 9, controls exposed = 24, cases not exposed = 63, controls not exposed = 123 OR 0.73 (95% CI 0.32‐1.67) Consumption of green tea with other types of tea compared to no‐tea consumption Cases exposed = 17, controls exposed = 61, cases not exposued = 63, controls not exposed = 123 OR 0.54 (95% CI 0.29‐1.01)	High risk: Selection: 4/4 stars, low risk Comparability: 0/2 stars, only crude analysis without adjustment for confounding factors Exposure: 2/3 stars, non‐response rate different between cases and controls or NR Total score: 6/9 stars, low quality.
Lassed 2016	Reference category: lowest exposure Postate cancer Intermediate exposure: OR 0.64 (95% CI 0.36‐1.15) Highest exposure: OR 0.40 (95% CI 0.08‐1.92)	High risk: Selection: 3/4 stars, hospital controls Comparability: 0/2 stars, crude analysis with no adjustment for confounding factors Exposure: 3/3 stars, low risk Total score: 6/9 stars, low quality.
Le Marchand 2000	Reference category: lowest exposure Lung cancer Intermediate exposure 1: OR 1.0 (95% CI 0.6‐1.7) Intermediate exposure 2: OR 0.7 (95% CI 0.4‐1.3) Highest exposure: OR 0.9 (95% CI 0.5‐1.6)	Low risk: Selection: 4/4 stars, low risk Comparability: 2/2 stars, low risk Exposure: 3/3 stars, low risk Total score: 9/9 stars, high quality.
Lee 2017	Reference category: lowest exposure Prostate cancer Exposure assessment A Highest exposure: OR 0.60 (95% CI 0.37‐0.98) Exposure assessment B Intermediate exposure 1: OR 1.53 (95% CI 0.39‐5.93) Intermediate exposure 2: OR 0.16 (95% CI 0.01‐2.22) Highest exposure: OR 0.34 (95% CI 0.02‐6.05)	Low risk: Selection: 3/4 stars, hospital controls Comparability: 2/2 stars, low risk Exposure: 2/3 stars, non‐response rate different between cases and controls or NR Total score: 7/9 stars, moderate quality.
Lei 1994	Reference category: lowest exposure Lung cancer Highest exposure: OR 0.71 (95% CI 0.43‐1.18)	High risk: Selection: 1/4 stars, no description of identification of cases with potential selection bias and hospital controls Comparability: 1/2 stars, the study did not control for smoking. Exposure: 1/3 star, no description of exposure ascertainment and non‐response rate between cases and controls NR Total score: 3/9 stars, low quality.
Leung 2016	Reference category: lowest exposure Ovarian cancer Exposure assessment A Highest exposure: OR 0.84 (95% CI 0.54‐1.30) Exposure assessment B Intermediate exposure 1: OR 1.03 (95% CI 0.50‐2.00) Intermediate exposure 2: OR 0.78 (95% CI 0.37‐1.55) Highest exposure: OR 0.77 (95% CI 0.39‐1.46)	Low risk: Selection: 4/4 stars, low risk Comparability: 2/2 stars, low risk Exposure: 3/3 stars, low risk Total score: 9/9 stars, high quality
Li 2011a	Reference category: lowest exposure Results using population controls All cancer Exposure assessment A Highest exposure: OR 0.78 (95% CI 0.60‐1.01) Exposure assessment B Intermediate exposure: OR 0.71 (95% CI 0.47‐1.08) Highest exposure: OR 0.51 (95% CI 0.27‐0.74) Breast cancer Exposure assessment A Highest exposure: OR 0.61 (95% CI 0.32‐1.18) Exposure assessment B Intermediate exposure: OR 0.60 (95% CI 0.24‐1.50) Highest exposure: OR 0.07 (95% CI 0.01‐0.47) Colorectal cancer Exposure assessment A Highest exposure: OR 0.62 (95% CI 0.42‐0.92) Exposure assessment B Intermediate exposure: OR 0.55 (95% CI 0.28‐1.06) Highest exposure: OR 0.45 (95% CI 0.25‐0.82) Leukaemia Exposure assessment A Highest exposure: OR 1.55 (95% CI 0.61‐3.96) Exposure assessment B Intermediate exposure: OR not estimable Highest exposure: OR 0.57 (95% CI 0.07‐4.79) Results using hospital controls All cancer Exposure assessment A Highest exposure: OR 0.77 (95% CI 0.59‐1.00) Exposure assessment B: Intermediate exposure: OR 0.60 (95% CI 0.40‐0.92) Highest exposure: OR 0.51 (95% CI 0.31‐0.83) Breast cancer Exposure assessment A Highest exposure: OR 0.55 (95% CI 0.28‐1.06) Exposure assessment B Intermediate exposure: OR 0.54 (95% CI 0.22‐1.33) Highest exposure: OR 0.06 (95% CI 0.01‐0.61) Colorectal cancer Exposure assessment A Highest exposure: OR 0.67 (95% CI 0.45‐0.99) Exposure assessment B Intermediate exposure: OR 0.51 (95% CI 0.26‐0.99) Highest exposure: OR 0.52 (95% CI 0.29‐0.94) Leukaemia Exposure assessment A Highest exposure: OR 1.46 (95% CI 0.55‐3.84) Exposure assessment B Intermediate exposure: OR not estimable Highest exposure: OR 0.65 (95% CI 0.08‐5.63)	Low risk: Selection: 4/4 stars when using population controls and 3/4 when using hospital controls, low risk Comparability: 2/2 stars, low risk Exposure: 2/3 stars, non‐response rate different between cases and controls or NR Total score: 8/9 stars and 7/9 stars, moderate quality.
Li 2014	Reference category: lowest exposure Prostate cancer Highest exposure: OR 0.59 (95% CI 0.40‐0.87)	Low risk: Selection: 4/4 stars, low risk Comparability: 1/2 stars, the study did not control for smoking Exposure: 2/3 stars, self‐reported exposure not within a structured interview or questionnaire. Total score: 7/9 stars, moderate quality.

*Study characteristics*
Methods	Cohort study in Japan
Participants	Participants: more than 119,500 men from Hiroshima and Nagasaki. 18,115 men included in the present study Inclusion criteria: resident in Hiroshima and Nagasaki included in the Adult Health Study (a subcohort of the Life Span study cohort) carried out among atomic‐bomb survivors and in residents in the cities in either city in the period between 1950 and 1953 and who completed the questionnaire surveys (in 1963, 1965 and 1979 to 1981), free from prostate cancer before the survey Parent cohort: Adult Health Study, a subcohort of the Life Span Study Recruitment: from 1963‐1996
Interventions	N/A
Outcomes	Number of cases prostate cancer: 193 cases (out of total 196 identified)
Green tea in exposure categories	Exposure assessment: intake of green tea Lowest exposure: < 1 time/d Intermediate exposure: 2‐4 times/d Highest exposure: > 5 times/d
Notes	Funding: RERF Research Protocols (RP) no. 26‐63 and 14‐78 Statistical methods: Poisson regression Variables controlled in analysis: age, calendar period, city of residence, radiation dose and education level Variables controlled by matching: ‐

*Study characteristics*
Methods	PCC in USA
Participants	Participants: 397 cases and 373 controls (all women) Inclusion criteria: women aged ≥ 21 years, newly diagnosed and histologically confirmed from the Estrogen, Diet, Genetic and Endometrial Cancer study, from 6 counties, New Jersey, USA Recruitment: from 1 July 2001‐30 June 2005
Interventions	N/A
Outcomes	Number of cases Endometrial cancer: 397 cases from the 469 eligible (all women)
Green tea in exposure categories	Exposure assessment: green tea intake Lowest exposure: 0 cup/week Intermediate exposure: < 1 cup/week Highest exposure: ≥ 1 cups/week
Notes	Funding: NIH‐K07 CA095666 and R01CA83918 Statistical methods: unconditional logistic regression Variables controlled in analysis: age, education, race, age at menarche, menopausal status and age at menopause for postmenopausal women, parity, oral contraceptive use, HRT use, BMI, smoking (pack‐years), smoking status, added sugar/honey, milk, cream, or nondairy creamer in tea Variables controlled by matching ‐

*Study characteristics*
Methods	HCC in Algeria
Participants	Participants: 160 cases and 160 controls (all men) Inclusion criteria: free from other prostatic diseases or malignant tumours, not being under dietary restrictions or patients in critical conditions from Department of Urology of Sidibel‐Abbes UHC and of Saida Hospital, Algeria Recruitment: from January 2007‐March 2011
Interventions	N/A
Outcomes	Number of cases Prostate cancer: 160 cases (all men)
Green tea in exposure categories	Exposure assessment: intake of green tea Lowest exposure: ≤ 1 cup/d Intermediate exposure 1: 2‐3 cups/d Intermediate exposure 2: 4‐5 cups/d Highest exposure: > 6 cups/d
Notes	Funding: not declared Statistical methods: conditional logistic regression Variables controlled in analysis: tobacco smoking, total energy intake and family history of prostate cancer Variables controlled by matching: age (± 5 years)

*Study characteristics*
Methods	RCT, parallel, double‐blind in Italy
Participants	Participants: 60 men with HG‐PIN, 30 in each group Inclusion criteria: age 45‐75 years, with HG‐PIN diagnosed needle biopsies collection, not consuming green tea or taking antioxidants, not vegetarians and not under antiandrogenic therapy Recruitment: NR
Interventions	Treatment group: 3 capsules of GTEs, containing green tea catechins (200 mg each) capsules/d = total 600 mg/d, corresponding to approximately 300 mg/d of EGCG Control group: placebo Duration: 1 year
Outcomes	Primary outcome Prostate cancer incidence Secondary outcomes LUTS using IPSS PSA levels QoL data Safety data
Green tea in exposure categories	N/A
Notes	Grant support: PRIN 2004 (MIUR, Italy). Dr. Rizzi was supported by Genprofiler Srl (Bolzano, Italy).
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "Volunteers were randomly assessed to a placebo‐ or GTCs [green tea catechins]‐arm by simple randomisation" Comment: it is unclear how sequence was generated, however the baseline characteristics reported in Table 4 are mainly equally distributed.
Allocation concealment (selection bias)	Unclear risk	Quote: "That same day [when they signed the informed consent, NDR], they were alternatively assigned to the placebo‐ or GTCs [green tea catechins]‐arm and given the appropriate treatment. To all subjects, capsules were given by the urologist according to the double blind method" Comment: insufficient information to answer
Blinding of participants and personnel (performance bias) Prostate cancer incidence	Low risk	Quote: "capsules were given by the urologist according to the double blind method" Comment: probably done
Blinding of participants and personnel (performance bias) Lower urinary tract symptoms	Low risk	Quote: "In the second arm, men received placebo (three identical capsules per day). To all subjects, capsules were given by the urologist according to the double blind method". Comment: probably done
Blinding of participants and personnel (performance bias) PSA levels	Low risk	Review authors do not believe this will introduce bias since this measurement is independent from individual evaluation
Blinding of outcome assessment (detection bias) Prostate cancer incidence	Unclear risk	No explicit statement on blinded outcome assessment
Blinding of outcome assessment (detection bias) Lower urinary tract symptoms	Unclear risk	No explicit statement on blinded outcome assessment
Blinding of outcome assessment (detection bias) PSA levels	Low risk	Review authors do not believe this will introduce bias since this measurement is independent from individual evaluation
Incomplete outcome data (attrition bias) Prostate cancer incidence	Low risk	All randomised participants were included in the analysis
Incomplete outcome data (attrition bias) Lower urinary tract symptoms	Unclear risk	Quote: "patients, diagnosed with prostate cancer at the 6 months biopsy check, left the study" Comment: number of participants included in analysis not stated
Incomplete outcome data (attrition bias) PSA levels	Low risk	All randomised participants were included in the analysis
Selective reporting (reporting bias)	Unclear risk	The study protocol is not available and it is not clear if the published reports include all expected outcomes. Insufficient information to answer
Other bias	Low risk	Study controlled for total serum PSA at the time of enrolment, prostate volume at the time of enrolment, prostate volume at the end of study, total number of HG‐PIN scores vs total scores taken at the time of enrolment, total number of HG‐PIN scores taken at the end of study; total number of mono‐focal or plurifocal HG‐PIN lesions by means of a multivariate analysis

*Study characteristics*
Methods	PCC in China
Participants	Participants: 122 (male/female: 79/43) cases and 121 (male/female: 78/43) controls Inclusion criteria: newly diagnosed lung cancer, mean age 54.71 (SD 11.45) in cases and 54.44 (SD 11.97) in controls from residents of Xuan Weu, China Recruitment: from March 1995‐March 1996
Interventions	N/A
Outcomes	Number of cases Lung cancer: 122 (male/female: 79/43) cases
Green tea in exposure categories	Exposure assessment: intake of green tea Lowest exposure: never Intermediate exposure: 2‐3 times/week Highest exposure: ≥ 1 time/d
Notes	Funding: not declared Statistical methods: unconditional logistic regression Variables controlled in analysis: smoking (pack‐years) Variables controlled by matching: age, sex, village of residence, type of heating and cooking fuel

*Study characteristics*
Methods	HCC in China
Participants	Participants: 150 (male/female: 102/48) cases and 300 (male/female: 204/96) controls Inclusion criteria: histologically confirmed squamous cell oesophageal carcinoma, mean age 54.5 (SD 6) in cases and 54.0 (SD 7) in control from First Affiliated Hospital of Ji'nan University, China Recruitment: from June 2004‐May 2010
Interventions	N/A
Outcomes	Number of cases Oesophageal cancer: 150 (male/female: 102/48) cases
Green tea in exposure categories	Exposure assessment: green tea intake Lowest exposure: never Intermediate exposure 1: < 100 g/month Intermediate exposure 2: 100‐250 g/month Highest exposure: > 250 g/month
Notes	Funding: Medical Science fund of Guangong Province (B2008094) Statistical methods: conditional logistic regression Variables controlled in analysis: age, sex, education level, annual income, cancer family history, smoking status and alcohol drinking status Variables controlled by matching: age (± 3 years) and sex

*Study characteristics*
Methods	Cohort study in USA
Participants	Participants: 7991 male participants out of 8006 recruited Inclusion criteria: American men of Japanese ancestry, born from 1990‐1919 and residing on the Hawaiian island of Oahu, Hawaii, USA Parent cohort: Honolulu Hearth Program Recruitment: from 1965‐1968
Interventions	N/A
Outcomes	Number of cases Outcome assessment: May 1991 Urinary tract cancer: 96 cases, including urinary bladder (N = 83), pelvis (N = 8) and ureter (N = 5)
Green tea in exposure categories	Exposure assessment: green tea intake Lowest exposure: almost never Highest exposure: not specified, probably merging 4 categories: < 2 times/week, 2‐4 times/week, almost daily, ≥ 1 time/d
Notes	Funding: grants provided by National Cancer Institute (R01 CA33644) Statistical methods: proportional hazard regression Variables controlled in analysis: age and smoking Variables controlled by matching: ‐

*Study characteristics*
Methods	Cohort study in China
Participants	Participants: 72,861 Chinese women Inclusion criteria: women aged 40‐70 years, no history of cancer at baseline. 381 women regularly drinking only black or oolong tea excluded Parent cohort: Shangai Women's Health Study Recruitment: from March 1997‐May 2000 Data on the same cohort also reported in Nechuta 2012
Interventions	N/A
Outcomes	Number of cases Outcome assessment: 31 December 2005 Breast cancer: 614 cases (all female)
Green tea in exposure categories	Exposure assessment: intake of green tea Lowest exposure: no Highest exposure: yes Exposure assessment: dosage of green tea Lowest exposure: no drinking Intermediate exposure 1: 0‐1.67 g/d Intermediate exposure 2: 1.68‐3.33 g/d Intermediate exposure 3: 3.34‐5.00 g/d Highest exposure: > 5.00 g/d
Notes	Funding: grants provided by National Institutes of Health (R01CA70867, R01CA106591, N02 CP1101066) Statistical methods: Cox proportional hazard regression Variables controlled in analysis: age, educational achievement, income, family history of breast cancer, history of fibroadenoma, BMI, waist‐to‐hip ratio, physically active, smoking status, alcohol consumption status, passive smoking status, ginseng intake, age at menarche, age at first live birth, menopausal status, age at menopause, use of HRT and dietary intake of total energy, fruits, vegetables, red meat, fish and isoflavones Variables controlled by matching: ‐

*Study characteristics*
Methods	RCT, parallel, double‐blind in USA ‐ Minnesota Green Tea Trial (MGTT)
Participants	Participants: 1075 post‐menopausal women (538 in the treatment group and 537 in the control group) Inclusion criteria: age 50‐70 years, classified as having high mammography density attending annual screening mammogram at 8 clinical centres in the Minneapolis‐St. Paul metropolitan area (Minnesota Green Tea Trial (MGTT), planning to reside in or near Minnesota for study duration. Exclusion criteria: tested positive for serological status of hepatitis B surface antigen or antibodies to hepatitis C virus; baseline ALT > 1.5 times the upper limit of 60 U/L; any history of cancer; any history of proliferative breast disease; history of breast augmentation; BMI < 18.5 or > 40 kg/m2; weight change > 4.6 kg during the previous 12 months; current or recent (within 6 months) use of HRT; current use of anti‐inflammatory agents including methotrexate or etanercept; current smoker; regular consumption of ≥ 7 alcoholic beverages/week; and regular consumption of ≥ 1 cups of green tea/week. Full details reported in Samavat et al. 2015. Recruitment: from August 2009‐April 2013
Interventions	Treatment group: 4 oral GTE capsules, i.e. 1315 (± 116) mg of total catechins/d, including 843 (± 44) mg of EGCG Control group: placebo Duration: 1 year
Outcomes	Primary outcome Effects on biomarkers of breast cancer risk: mammographic density, circulating reproductive hormones and circulating insulin‐like growth factor axis proteins Secondary outcome Circulating F2‐isoprostanes, urinary oestrogen metabolites, anthropometric variables and obesity‐associated hormone concentrations QoL Safety data
Green tea in exposure categories	N/A
Notes	Funding: from National Institutes of Health/National Cancer Institute grant R01 CA127236, Award Number T32CA132670 from the National Cancer Institute, the Department of Defense/US Army Medical Research and Materiel Command Award Number W81XWH‐11‐1‐0013, the University of Minnesota Agricultural Experiment Station Project Number MIN‐18‐103 and the National Center for Advancing Translational Sciences of the National Institutes of Health Award Number UL1TR000114
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "A detailed description of the Minnesota Green Tea Trial (MGTT) design, eligibility criteria, study conduct and patient flow through the trial will be published separately (Samavat et al., Cancer Causes and Control)." reporting that "Investigational Drug Services (IDS) pharmacy utilized a computer generated randomisation scheme using the permuted block method" Comment: probably low risk
Allocation concealment (selection bias)	Low risk	Quote: "A detailed description of the Minnesota Green Tea Trial (MGTT) design, eligibility criteria, study conduct and patient flow through the trial will be published separately (Samavat et al., Cancer Causes and Control)." reporting that "Randomization was performed by the Investigational Drug Services (IDS) pharmacy at University of Minnesota Medical Center ‐ Fairview". Central allocation. Comment: probably done
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Quote: from Samavat 2015 “In this double‐blinded study, study staff, participants, laboratory personnel and all parties involved with assessment of the study endpoints were blinded to treatment assignment. The treatment codes were only available to the IDS [Investigational Drug Services] pharmacy staff in charge of randomisation and a study biostatistician.” From Dostal 2015 : “Participants, investigators, laboratory staff and those monitoring clinical outcomes and adverse events were blinded to treatment assignment” and “Placebo capsules were identical in appearance to GTE". Study investigators were kept blinded to the assigned treatment of all participants experiencing an adverse effects Comment: probably done
Blinding of participants and personnel (performance bias) Non‐melanoma skin cancer incidence	Low risk	Quote: from Samavat 2015 “In this double‐blinded study, study staff, participants, laboratory personnel and all parties involved with assessment of the study endpoints were blinded to treatment assignment. The treatment codes were only available to the IDS [Investigational Drug Services] pharmacy staff in charge of randomisation and a study biostatistician.” From Dostal 2015 : “Participants, investigators, laboratory staff and those monitoring clinical outcomes and adverse events were blinded to treatment assignment” and “Placebo capsules were identical in appearance to GTE". Study investigators were kept blinded to the assigned treatment of all participants experiencing an adverse effects Comment: probably done
Blinding of participants and personnel (performance bias) Gynaecological cancer incidence	Low risk	Quote: from Samavat 2015 “In this double‐blinded study, study staff, participants, laboratory personnel and all parties involved with assessment of the study endpoints were blinded to treatment assignment. The treatment codes were only available to the IDS pharmacy staff in charge of randomisation and a study biostatistician.” From Dostal 2015 : “Participants, investigators, laboratory staff and those monitoring clinical outcomes and adverse events were blinded to treatment assignment” and “Placebo capsules were identical in appearance to GTE". Study investigators were kept blinded to the assigned treatment of all participants experiencing an adverse effects Comment: probably done
Blinding of outcome assessment (detection bias) All outcomes	Low risk	From Dostal 2015 "Participants, investigators, laboratory staff and those monitoring clinical outcomes and adverse events were blinded to treatment assignment" and "Placebo capsules were identical in appearance to GTE". Study investigators were kept blinded to the assigned treatment of all participants experiencing an adverse effects Comment: probably done
Blinding of outcome assessment (detection bias) Non‐melanoma skin cancer incidence	Low risk	From Dostal 2015 "Participants, investigators, laboratory staff and those monitoring clinical outcomes and adverse events were blinded to treatment assignment" and "Placebo capsules were identical in appearance to GTE". Study investigators were kept blinded to the assigned treatment of all participants experiencing an adverse effects Comment: probably done
Blinding of outcome assessment (detection bias) Gynaecological cancer incidence	Low risk	From Dostal 2015 "Participants, investigators, laboratory staff and those monitoring clinical outcomes and adverse events were blinded to treatment assignment" and "Placebo capsules were identical in appearance to GTE". Study investigators were kept blinded to the assigned treatment of all participants experiencing an adverse effects Comment: probably done
Incomplete outcome data (attrition bias) All outcomes	Low risk	Number and reason of participants' withdrawal from the study reported. Study authors performed an ITT analysis.
Incomplete outcome data (attrition bias) Non‐melanoma skin cancer incidence	Low risk	Number and reason of participants' withdrawal from the study reported. Study authors performed an ITT analysis.
Incomplete outcome data (attrition bias) Gynaecological cancer incidence	Low risk	Number and reason of participants withdrawal from the study reported. Authors performed an ITT analysis.
Selective reporting (reporting bias)	Low risk	Results to the Minnesota Green Tea Trial RCT (ClinicalTrials.gov identifier (NCT number): NCT00917735) reported in several publications all included in this review.
Other bias	Unclear risk	Elevated number of withdrawals

*Study characteristics*
Methods	RCT, parallel, double‐blind in USA
Participants	Participants: 20 people with ulcerative colitis. Ratio 4:1 randomisation treatment/placebo Inclusion criteria: age ≥ 18 years, with mildly to moderately active disease Recruitment: NR
Interventions	Treatment groups: green tea catechins (using Polyphenon E, Mitsu‐Norin, Fujieda) in low dose of catechins containing 200 mg of EGCG and high dose of catechins containing 400 mg of EGCG Control group: placebo Duration: 56 d
Outcomes	Primary outcome Safety data Secondary outcome: QoL
Green tea in exposure categories	N/A
Notes	Funding: National Institutes of Health grant (5K23DK073750), University of Louisville Research Foundation Project Initiation Grant, Polyphenon E supplied by Mitsui‐Norin (Fujieda, Japan) through the Chemoprevention Agent Development Group of the National Cancer Institute. Baseline characteristics were not equally distributed: both men and women included in the treatment groups, while only women were included in the placebo group. Treated participants were also slightly older and with higher mean weight and all used azathioprine. Tobacco use prevalence was 7% and 50% in treatment and control group, respectively.
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "Patients...were randomized in a double‐blinded fashion according to a random number generator compiled by a statistician not involved in the study" Comment: probably done
Allocation concealment (selection bias)	Low risk	Quote: "Patients...were randomized in a double‐blinded fashion according to a random number generator compiled by a statistician not involved in the study" Comment: probably done
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Quote: "Patients ….. were randomized in a double‐blinded fashion": Also UCDAI was assessed at day 0 and 56, while for other laboratory analyses at day 1, 14, 28, 56 and 70. Response: review authors do not believe this will introduce bias
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Quote: "Patients ….. were randomized in a double‐blinded fashion": Also UCDAI was assessed at day 0 and 56, while for other laboratory analyses at day 1, 14, 28, 56 and 70. Response: review authors do not believe this will introduce bias
Incomplete outcome data (attrition bias) All outcomes	Low risk	Number of participants included in analysis stated and reported
Selective reporting (reporting bias)	Low risk	The study protocol is available (ClinicalTrials.gov identifier (NCT number): NCT00718094) and the published reports include all expected outcomes
Other bias	High risk	Results from Inflammatory Bowel Disease Questionnaire, a validated, IBD‐specific indicator of QoL, was significantly different between groups at baseline. Some participants in the treatment group, but not in the placebo arm, took immunomodulatory drugs

*Study characteristics*
Methods	Cohort study of Japanese population living in Hawaii, USA
Participants	Participants: 40,575 eligible participants (of whom 12,789 were Japanese). Final population 11,907 (male/female: 5610/6297), 882 excluded for missing information on covariates Inclusion criteria: Japanese participants, complete information on covariates Parent cohort: Hawaii Health Surveillance Program Cohort (HHSPC) Recruitment: from 1975‐1980
Interventions	N/A
Outcomes	Number of cases Oucome assessment: 31 December 1994 Stomach cancer: 108 (male/female: 64/44) cases
Green tea in exposure categories	Exposure assessment: intake of green tea Lowest exposure: none Intermediate exposure: 1 cup/d Highest exposure: ≥ 2 cups/d
Notes	Funding: not declared Statistical methods: Cox proportional hazard regression Variables controlled in analysis: age, years of education, Japanese place of birth and sex (in combined analyses). Analyses among men were also adjusted for cigarette smoking and alcohol intake status. Variables controlled by matching: ‐ Galanis 1997 is a letter reporting preliminary findings subsequently published in Galanis 1998.

*Study characteristics*
Methods	RCT, quadruple‐blind in USA
Participants	178 (male/female: 89/89) participants: 42 (male/female: 21/21) in group A, 63 (male/female: 31/32) in group B and 73 (male/female: 37/36) in group C, 89 men
Interventions	Group A: green tea beverage and placebo capsules Group B: placebo beverage and Polyphenon E capsules (Mitsui‐Norin Co, Ltd, Shizuoka, Japan), corresponding to 800 mg/d of ECGC Group C: placebo beverage and placebo capsules Duration: 6 months
Outcomes	Primary outcome Change in levels of biomarkers of oxidative stress Secondary outcome Safety data
Green tea in exposure categories	N/A
Notes	No results on lung cancer prevention published or reported on ClinicalTrials.gov (NCT00363805)
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "We are conducting a 6‐month randomized, controlled, double‐blind trial..." Comment: probably done
Allocation concealment (selection bias)	Unclear risk	No statement
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Review authors do not believe this would introduce bias since all biological analyses were performed
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Review authors do not believe this would introduce bias since all biological analyses were performed
Incomplete outcome data (attrition bias) All outcomes	High risk	Data on lung cancer NR, only on biomarkers
Selective reporting (reporting bias)	Low risk	All outcomes reported on ClinicalTrials.gov (NCT00363805)
Other bias	Unclear risk	The number of withdrawals is low, however since no full publication is available, only an abstract and report on ClinicalTrials.gov, it is difficult to judge if other biases are present.

*Study characteristics*
Methods	HCC in Romania
Participants	Participants: 151 (male/female: 92/59) cases and 151 (male/female: 90/61) controls Inclusion criteria: recently diagnosed cases undergoing conventional treatment recruited from MEDISPROF Oncology Hospital, Cluj‐Napoca, Romania Recruitment: from April 2015‐October 2017
Interventions	N/A
Outcomes	Colorectal cancer: 151 (male/female: 92/59) cases
Green tea in exposure categories	Exposure assessment: drinking of green tea Lowest exposure: < 1 serving/week Intermediate exposure 1: 1‐2 servings/week Intermediate exposure 2: 3‐5 servings/week Highest exposure: > 5 servings/week
Notes	Funding: not declared Statistical methods: crude analysis with raw data Variables controlled in analysis: crude analysis Variables controlled by matching: age (± 5 years) Gavrilas 2018 (Bulletin UASVM Food Science and Technology) reports same results subsequently published in Gavrilas 2018

PERMALINK

Green tea (Camellia sinensis) for the prevention of cancer

Tommaso Filippini

Marcella Malavolti

Francesca Borrelli

Angelo A Izzo

Susan J Fairweather-Tait

Markus Horneber

Marco Vinceti

Abstract

Background

Objectives

Search methods

Selection criteria

Data collection and analysis

Main results

Authors' conclusions

Plain language summary

Summary of findings

Background

Description of the intervention

How the intervention might work

Pharmacology of Camellia sinensis

Possible mechanisms of action of Camellia sinensis polyphenols

Why it is important to do this review

Objectives

Methods

Criteria for considering studies for this review

Types of studies

Types of participants

Types of interventions

Types of outcome measures

Primary outcomes

Secondary outcomes

Search methods for identification of studies

Electronic searches

Searching other resources

Data collection and analysis

Selection of studies

Data extraction and management

Assessment of risk of bias in included studies

Experimental studies

Nonexperimental studies

Measures of treatment effect

Unit of analysis issues

Dealing with missing data

Assessment of heterogeneity

Assessment of reporting biases

Data synthesis

Subgroup analysis and investigation of heterogeneity

Sensitivity analysis

Summary of findings and assessment of the certainty of the evidence

Summary of findings 1. Green tea extract supplementation compared with placebo for preventing cancer: experimental studies.

Summary of findings 2. Highest compared with lowest green tea exposure for preventing cancer: primary outcomes in nonexperimental studies.

Summary of findings 3. Highest compared with lowest green tea exposure for preventing cancer: individual cancer types in nonexperimental studies.

Results

Description of studies

Results of the search

1.

Included studies

1. Summary characteristics of experimental studies.

2. Summary characteristics of cohort studies.

3. Summary characteristics of case‐control studies.

Outcomes

4. Detailed summary results of included experimental studies.

5. Detailed summary results of included nonexperimental cohort studies.

6. Detailed summary results of included nonexperimental case‐control studies.

Diagnoses

Any cancer type

Gastrointestinal tract cancer

Respiratory tract cancer

Breast cancer

Urogenital tract cancer

Haematopoietic cancer

Other types of cancer

Exposure

Sponsorship

Excluded studies

Risk of bias in included studies

Experimental studies

*Study characteristics*
Methods	PCC in Japan
Participants	Participants: 71 (male/female: NR) cases and 142 (male/female: NR) controls Inclusion criteria: permanent residents in Hokkaido Prefecture, Japan Recruitment: NR
Interventions	N/A
Outcomes	Number of cases Pancreatic cancer: 71 (male/female: NR) cases
Green tea in exposure categories	Exposure assessment: drinking of green tea Lowest exposure: no drinking Highest exposure: drinking green tea almost every day
Notes	Article in Japanese Funding: not declared Statistical methods: unconditional logistic regression Variables controlled in analysis: sex, age and area of residence Variables controlled by matching: sex and age (± 3 years)

*Study characteristics*
Methods	PCC in Australia
Participants	Participants: 854 (male/female: 525/329) and 948 (male/female: 556/392) controls Inclusion criteria: histologically confirmed incident cases aged 40‐79 years, from the Western Australian Bowel Health Study (WABOHS) Recruitment: 1 June 2005‐31 August 2007
Interventions	N/A
Outcomes	Number of cases Colorectal cancer: 854 (male/female: 525/329) cases Proximal colon cancer: 281 cases Distal colon cancer: 260 cases Rectal cancer: 323 cases
Green tea in exposure categories	Exposure assessment: intake of green tea Lowest exposure: no drinking Intermediate exposure 1: < 1 cup/month Intermediate exposure 2: < 1 cup/week Highest exposure: ≥ 1 cup/week
Notes	Funding: Australian National Health and Medical Research Council (Project Grant #353568 and Fellowship #37614900) and Dutch Cancer Society Statistical methods: conditional logistic regression Variables controlled in analysis: age group, sex, energy intake from food, alcohol intake, smoking status, use of multivitamins, diabetes, physical activity during the age period 19–34 years, BMI at age 40 years, socioeconomic status and country of birth Variables controlled by matching: sex and age

*Study characteristics*
Methods	HCCs in Japan
Participants	Huang 1999: Participants: 887 (male/female: 595/292) cases and 28,619 (male/female: 7892/20,727) controls Inclusion criteria: aged 20‐79 years, from the Aichi Cancer Center Hospital (ACCH) in Aichi Prefecture, Japan Recruitment: from 1988‐1995 Inoue 1994 Participants: 127,477 inhabitants of 45 areas of Japan, data retrieved from 26,426 participants: 668 (male/female: 420/248) cases and 668 (male/female: 420/248) controls Inclusion criteria: aged 20‐79 years, from the Aichi Cancer Center Hospital (ACCH) in Aichi Prefecture, Japan Recruitment: from January 1988‐June 1991 Kato 1990a Participants: 427 (male/female: 289/138) cases and 3014 (male/female: 1247/1767) controls Inclusion criteria: aged ≥ 18 years, who underwent endoscopic examination and with no other types of cancer attending at Aichi Cancer Center Hospital, Japan. Controls considered in the present analysis are those with normal gastric mucosa Recruitment: from April 1985‐March 1989
Interventions	N/A
Outcomes	Huang 1999 Number of cases Stomach cancer: 887 (male/female: 595/292) cases Inoue 1994: Number of cases Stomach cancer: 668 cases, 123 of cardia, 218 of middle stomach and 256 of antrum Kato 1990a Number of cases Stomach cancer: 427 (male/female: 289/138) cases
Green tea in exposure categories	Exposure assessment: intake of green tea Lowest exposure: never Intermediate exposure 1: 1‐2 cups/d Intermediate exposure 2: 3‐5 cups/d Highest exposure: > 6 cups/d Inoue 1994 Exposure assessment: green tea drinking Lowest exposure: < every day Highest exposure: every day Kato 1990a Exposure assessment: consumption of green tea Lowest exposure: nondrinker Intermediate exposure: 1‐4 cups/d Highest exposure: ≥ 5 cups/d
Notes	Huang 1999 Funding: National Institutes of Health, USAA Statistical methods: unconditional logistic regression Variables controlled in analysis: sex and age Variables controlled by matching: ‐ Inoue 1994 Funding: Grant‐in‐Aid for Cancer Research (4‐2) and for a comprehensive 10‐year Strategy for Cancer Control from the Ministry of Health and Welfare, Japan Statistical methods: conditional logistic regression Variables controlled in analysis: age and intake of fresh vegetables Variables controlled by matching: sex, age (± 2 years) and first time of hospital visit (± 2 months) Kato 1990a Funding: Grant‐in‐Aid for a Comprehensive 10‐year Strategy for Cancer Control. Japan, from the Ministry of Health and Welfare Statistical methods: Walker‐Duncan logistic regression Variables controlled in analysis: age and residence Variables controlled by matching: ‐

*Study characteristics*
Methods	HCC in Japan
Participants	Participants: 127,477 inhabitants of 45 areas of Japan, data retrieved from 21,128 participants Inclusion criteria: aged 40‐79 years, from the Aichi Cancer Center Hospital (ACCH) in Aichi Prefecture, Japan Recruitment: from January 1988‐June 1991
Interventions	N/A
Outcomes	Number of cases Oesophageal cancer: 185 (male/female: 161/24) cases Stomach cancer: 893 (male/female: 613/280) cases Colon cancer: 362 (male/female: 213/149) cases Rectal cancer: 266 (male/female: 173/93) cases
Green tea in exposure categories	Exposure assessment: intake of green tea Lowest exposure: rarely Intermediate exposure 1: occasional Intermediate exposure 2: 1‐2 cups/d Intermediate exposure 3: 4‐6 cups/d Highest exposure: ≥ 7 cups/d
Notes	Funding: Grant‐in‐Aid for Cancer Research (9‐4) from the Ministry of Health and Welfare, Japan and a grant from the Foundation of All Japan Coffee Association Statistical methods: unconditional logistic regression Variables controlled in analysis: coffee intake, black tea intake, sex, age, year and season at 1st hospital visit, habitual smoking, habitual alcohol drinking, regular physical exercise, fruit intake, rice intake and beef intake Variables controlled by matching: ‐ Data on Aichi also reported in Inoue 2009a for stomach cancer

*Study characteristics*
Methods	Cohort studies in Japan
Participants	Participants: 219,080 (male/female: 100,479/118,601) participants Inclusion criteria: all studies were carried out on Japanese populations starting from mid‐1980s to mid‐1990s with information on diet (including green tea) Parent cohorts Japan Public Health Center‐based Prospective Study (JPHC)‐I Japan Public Health Center‐based Prospective Study (JPHC)‐II Japan Collaborative Cohort Study (JACC) (only from 24 of the 45 investigated areas) Miyagi Cohort Study (MIYAGI) Three Prefecture Study ‐ Miyagi portion (3‐pref MIYAGI) Three Prefecture Study ‐ Aichi portion (3‐pref AICHI) Recruitment JPHC‐I: 1990 JPHC‐II: from 1993‐1994 JACC: from 1988‐1990 MIYAGI: 1990 3‐pref MIYAGI: 1984 3‐pref AICHI: 1985 Previous reports Fujino 2002 Participants: 127,477 inhabitants of 45 areas of Japan, data retrieved from 18,746 men and 26,184 women Inclusion criteria: aged 40‐79 (≥ 18 years in 1949). Details reported in Ohno 2001 Parent cohort: Japan Collaborative Cohort Study of Evaluation of Cancer Risk (JACC Study) Recruitment: from 1988‐1990 Hoshiyama 2002 Participants: 127,477 inhabitants of 45 areas of Japan, data retrieved from 18,746 men and 26,184 women Inclusion criteria: aged 40‐79 (≥ 18 years in 1949). Details reported in Ohno 2001 Parent cohort: Japan Collaborative Cohort Study of Evaluation of Cancer Risk (JACC Study) Recruitment: from 1988‐1990 Hoshiyama 2004: (nested case‐cohort study due to availability of serum samples) Participants: 127,477 inhabitants of 45 areas of Japan, data retrieved from 30,370 men and 42,281 women including 151 cases and 265 controls Inclusion criteria: aged 40‐79 (≥ 18 years in 1949). Details reported in Ohno 2001 Parent cohort: Japan Collaborative Cohort Study of Evaluation of Cancer Risk (JACC Study) Recruitment: from 1988‐1990 Koizumi 2003 Participants: 31,345 (male/female: 13,992/17,353) from cohort 1 and 47,605 (male/female: 22,836/24,769). Final included subjects 26,311 (male/female: 11,902/14,409) from cohort 1 and 39,604 (male/female: NR) from cohort 2 Inclusion criteria: aged ≥ 40 years, residents in three municipalities of the Miyagi Prefecture, Northern Japan (cohort 1); aged 40‐64 years (cohort 2) in residents of Miyagi Prefecture, Japan Parent cohorts Cohort 1: Three Prefecture Study ‐ Miyagi portion (3‐pref MIYAGI) Cohort 2: Miyagi Cohort Study (MIYAGI) Recruitment: from 1984 for cohort 2 and from 1990 for cohort 2 Tsubono 2001 Participants: 31,345 (male/female: 13,992/17,353) with final included participants 26,311 (male/female: 11,902/14,409) Inclusion criteria: aged ≥ 40 years, residents in 3 municipalities of the Miyagi Prefecture, Northern Japan Parent cohort: Three Prefecture Study ‐ Miyagi portion (3‐pref MIYAGI) Recruitment: from 1984 Sasazuki 2004 Participants: 140,420 (male/female: 68,722/71,698), with 61,595 from cohort 1 and 78,825 from cohort 2 and 72,943 (male/female: 34,832/38,111) participants with complete data Inclusion criteria: aged 40‐59 years, from 5 Public Health Centre areas (Iwate, Akita, Nagano, Okinawa and Tokyo) for cohort 1 and aged 40‐69 years for cohort 2, from 6 public health centre areas (Ibaraki, Niigata, Kochi, Nagasaki, Okinawa and Osaka), respondent at 5‐year follow‐up. Katsushika Public Health Center was excluded due to missing cancer data. Parent cohorts Cohort 1: Japan Public Health Center‐based Prospective Study (JPHC)‐I Cohort 2: Japan Public Health Center‐based Prospective Study (JPHC)‐II Recruitment: from 1990 for cohort 1 and 1993/1994 for cohort 2
Interventions	N/A
Outcomes	Number of cases Inoue 2009b Outcome assessment JPHC‐I: 2001 JPHC‐II: 2003‐2004 JACC: 2001 MIYAGI: 2001 3‐pref MIYAGI: 1992 3‐pref AICHI: 2000 Stomach cancer: 3577 (male/female: 2495/1082) cases Previous reports Fujino 2002 Outcome assessment: 31 December 1997 Stomach cancer mortality: 379 (male/female: 261/118) cases Hoshiyama 2002 Outcome assessment: 31 December 1997 Stomach cancer mortality: 359 (male/female: 240/119) cases Hoshiyama 2004 Outcome assessment: 31 December 1997 Stomach cancer: 151 cases of the 804 eligible cases due to availability of serum samples Koizumi 2003 Outcome assessment: December 1992 for 3‐pref MIYAGI, NR for MIYAGI (probably 1997) Stomach cancer: 733 (male/female: NR) cases; 419 (male/female: 296/123) cases in 3‐pref‐MIYAGI I and 314 (male/female: NR) cases in MIYAGI‐ Tsubono 2001 Outcome assessment: 31 December 1992 for 3‐pref MIYAGI only Stomach cancer: 419 (male/female: 296/123) cases from 3‐pref MIYAGI only Sasazuki 2004 Outcome assessment: 31 December 2001 Stomach cancer: 892 (male/female: 665/227) cases, including 109 (male/female: 88/21) including upper‐third gastric cancers and 631 (male/female: 461/170) distal cancers and as histological categorisation, 471 (male/female: 386/85) cases of differentiated type and 312 (male/female: 197/115) cases of undifferentiated type
Green tea in exposure categories	Exposure assessment: green tea intake (same for Inoue 2009a, Koizumi 2003, Tsubono 2001 and Sasazuki 2004) Lowest exposure: < 1 cup/d Intermediate exposure 1: 1‐2 cups/d Intermediate exposure 2: 3‐4 cups/d Highest exposure: ≥ 5 cups/d Exposure assessment: intake of green tea (in Fujino 2002) Lowest exposure: ≤ 3 times/week Intermediate exposure: > 3 times/week Highest exposure: every day Exposure assessment: intake of green tea (In Hoshiyama 2002 and Hoshiyama 2004) Intermediate exposure 1: 1‐2 cups/d Intermediate exposure 2: 3‐4 cups/d Intermediate exposure 3: 5‐9 cups/d Highest exposure: ≥ 10 cups/d
Notes	Funding: in Tsubono: grants from the Japanese Ministry of Health and Welfare and the Japanese Ministry of Education, Science and Culture. In Koisumi 2003 not declared and this study was supported by a Grant for the Third Term Comprehensive Control Research for Cancer from the Ministry of Health, Labour and Welfare of Japan. Statistical methods: Cox proportional hazard regression Inoue 2009b Variables controlled in analysis: age, area (JPHC‐I, JPHC‐II and JACC only), smoking, ethanol intake, rice intake, soy bean paste soup, coffee intake, pickled vegetable intake and green–yellow vegetable intake Fujino 2002 Funding: Grants‐in‐Aid for Scientific Research from the Ministry of Education, Science, Sports and Culture of Japan (Monbusho) (N. 61010076, 63010074, 1010068, 2151065, 3151064, 4151063, 5151069, 6279102, 11181101, 12218237 and 12218216). Statistical methods: Cox proportional hazard regression Variables controlled in analysis: age Hoshiyama 2002 Funding: Ministry of Education, Science, Sports and Culture of Japan grants: 61010076, 62010074, 63010074, 1010068, 2151065, 3151064, 4151063, 5151069, 6279102, 11181101, 12218237 Statistical methods: Cox hazard proportional regression Variables controlled in analysis: age, smoking status, history of peptic ulcer, family history of stomach cancer, consumption of rice, miso soup, green‐yellow vegetables, white vegetables, fruits and preference for salty foods Hoshiyama 2004 Funding: Ministry of Education, Science, Sports and Culture of Japan grants: 61010076, 62010074, 63010074, 1010068, 2151065, 3151064, 4151063, 5151069, 6279102, 11181101, 12218237 Statistical methods: Cox hazard proportional regression Variables controlled in analysis: age, smoking status, HP infection, history of peptic ulcer, family history of stomach cancer, educational level, consumption of rice, miso soup, green‐yellow vegetables, white vegetables, fruits and preference for salty foods Koizumi 2003 Variables controlled in analysis: sex, age, type of health insurance, parental history of gastric cancer, history of peptic ulcer, cigarette smoking, alcohol consumption, consumption of rice, black tea, coffee, pickled vegetables, bean‐paste soup. Plus consumption of meat, green or yellow vegetables, other vegetables and fruits for cohort 1 and consumption of pork, ham, spinach, carrot, cabbage, Chinese cabbage, orange and other fruits for cohort 2 Tsubono 2001 Variables controlled in analysis: sex, age, type of health insurance, history of peptic ulcer, smoking status, alcohol consumption, daily consumption of rice, consumption of black tea and consumption of coffee, and consumption of meat, green or yellow vegetables, pickled vegetables, other vegetables, fruits and bean‐paste soup. Sasazuki 2004 Funding: Grant‐in‐Aid for Cancer Research and for the Second Term Comprehensive 10‐year Strategy for Cancer Control from the Ministry of Health, Labour and Welfare of Japan Statistical methods: Cox proportional hazard regression Variables controlled in analysis: age, area and cigarette smoking (stratified by sex) Variables controlled by matching: ‐ Note: Ohno 2001 study report rationale for initiating the Japan Collaborative Cohort (JACC) Study

*Study characteristics*
Methods	PCC in Iran
Participants	Participants: 300 cases (male/female: 150/150) and 571 (male/female: 278/293) controls Inclusion criteria: newly diagnosed histologically confirmed oesophageal cancer in residents in the eastern part of Golestan Province, Iran, attending the Atrak Clinic. Recruitment: from December 2003 to March 2007
Interventions	N/A
Outcomes	Number of cases Oesophageal cancer: 266/300 eligible cases
Green tea in exposure categories	Exposure assessment: frequency of green tea consumption Lowest exposure: never, < weekly Highest exposure: daily, weekly
Notes	Funding: not declared Statistical methods: conditional logistic regression Variables controlled in analysis: ethnicity, daily vegetable intake, alcohol consumption, tobacco or opium ever used, duration of residence in rural areas, education level, car ownership, tea temperature and black tea intake Variables controlled by matching: place of residence, age (± 2 years) and sex

*Study characteristics*
Methods	Cohort studies in Japan (Iwasaki 2010a) and case‐cohort nested study (Iwasaki 2010b)
Participants	Participants: 140,420 (male/female: 68,722/71,698) from cohort 1 and cohort 2. From the available 67,422 women Inclusion criteria: incident breast cancer cases in women at baseline aged 40‐59 years in cohort 1 and 40‐69 years in cohort 2, with no history of breast cancer, respondent to baseline questionnaire, not moved out of study area In Iwasaki 2010b: included data with available plasma tea polyphenol levels. Final population of 144 cases and 288 controls Parent cohorts Japan Public Health Center‐based Prospective Study (JPHC)‐I Japan Public Health Center‐based Prospective Study (JPHC)‐II Recruitment JPHC‐I: 1990 JPHC‐II: from 1993‐1994
Interventions	N/A
Outcomes	Number of cases Outcome assessment: 31 December 2006 in Iwasaki 2010a and 31 December 2002 in Iwasaki 2010b Breast cancer: 581 cases, based on baseline questionnaire data
Green tea in exposure categories	Exposure assessment: green tea intake at baseline Iwasaki 2010a Lowest exposure: < 1 cup/week Intermediate exposure 1: 1‐2 cups/week Intermediate exposure 2: 3‐4 cups/week Intermediate exposure 3: 1‐2 cups/d Intermediate exposure 4: 3‐4 cups/d Highest exposure: ≥ 5 cups/d Iwasaki 2010b Lowest exposure: < 1 cup/week Highest exposure: ≥ 5 cups/d
Notes	Iwasaki 2010a Funding: Grant‐in‐Aid for Scientific Research from Monbusho, the Japanese Ministry of Education, Science and Culture (N.11181101) Statistical methods: Cox proportional hazard regression Variables controlled in analysis: age, area (10 Public Health Centres), age at menarche, menopausal status at baseline, number of births, age at first birth, height, BMI, alcohol intake, smoking status, leisure time physical activity, exogenous hormone use, family history of breast cancer, oolong tea intake, black tea intake and coffee intake Variables controlled by matching: ‐ Iwasaki 2010b Funding: Grants‐in‐Aid for Cancer Research and for the Third Term Comprehensive Ten‐Year Strategy for Cancer Control from the Ministry of Health, Labour and Welfare of Japan and Grants‐in‐Aid for Scientific Research on Priority Areas (17015049) from the Ministry of Education, Culture, Sports, Science and Technology of Japan. Statistical methods: crude data analysis Variables controlled in analysis: crude data Variables controlled by matching: ‐ Data of parent study Iwasaki 2010a were used for breast cancer analysis including a larger population.

*Study characteristics*
Methods	Case‐control study in China
Participants	Participants: 53 (male/female: 19/34) cases and 106 (male/female: 48/58) controls Inclusion criteria: participants of the cohort of Dayao in Yunnan, a rural area with naturally occurring asbestos, China Recruitment: NR
Interventions	N/A
Outcomes	Number of cases Lung cancer: 23 cases Pleural mesothelioma: 26 cases
Green tea in exposure categories	Exposure assessment: green tea intake Lowest exposure: never Intermediate exposure 1: 1‐3 times/week Intermediate exposure 2: 4‐6 times/week Highest exposure: ≥ 7 times/week
Notes	Article in Chinese Funding: National Natural Research Fundation of China (no. 41071064) Statistical methods: conditional logistic regression Variables controlled in analysis: sex, age, ethnicity, cultural level, BMI, alcohol consumption, history of lung cancer, family history of cancer Variables controlled by matching: sex and age (± 3 years)

*Study characteristics*
Methods	PCC and HCC in Japan
Participants	Participants: 139 (male/female: 74/65) cases, 278 (male/female: 148/130) population controls and 2575 (male/female: 1171/1404) hospital controls Inclusion criteria: aged 20‐75 years, newly diagnosed cases from Karatsu city and 9 neighbouring villages in Saga Prefecture, Japan. Hospital controls were aged 25‐75 years, free from cancer at clinical/radiological examination. Population controls were residents of Saga Prefecture, Japan Recruitment: from 1979‐1982
Interventions	N/A
Outcomes	Number of cases Stomach cancer: 139 (male/female: 74/65) cases
Green tea in exposure categories	Exposure assessment: green tea intake Lowest exposure: none or 1‐4 cups/d Intermediate exposure: 5‐9 cups/d Highest exposure: ≥ 10 cups/d
Notes	Funding: Grant‐in‐Aid, Ministry of Education, Science and Culture, Japan Statistical methods: unconditional logistic regression for hospital controls and conditional logistic regression for population controls Variables controlled in analysis: sex, age, class, smoking, consumption of mandarin oranges, fruits and others Variables controlled by matching: sex, age, class

*Study characteristics*
Methods	HCC in Czech Republic
Participants	Kubik 2008 Participants: 1096 (male/female: 509/587) cases and 2966 (male/female: 788/2178) controls Inclusion criteria: aged 25‐89 years, with microscopically confirmed lung cancer attending Prague University Hospital Na Bulovce, Departments of pneumology, thoracic surgery and internal medicine Recruitment: from April 2002‐August 2006 (men) and from April 1998‐November 2006 (women) Kubik 2004 Participants: 435 cases and 1710 controls (all women) Inclusion criteria: women aged 25‐89 years, with microscopically confirmed lung cancer attending Prague University Hospital Na Bulovce, Departments of pneumology, thoracic surgery and internal medicine Recruitment: from April 1998‐November 2002
Interventions	N/A
Outcomes	Kubik 2008 Number of cases Lung cancer: 1096 (male/female: 509/587) cases: 308 (male/female: 101/207) cases of adenocarcinoma, 398 (male/female: 249/149) squamous cell cancers and 213 (male/female: 81/132) small‐cell cancers Kubik 2004 Number of cases Lung cancer: 435 cases (all women)
Green tea in exposure categories	Kubik 2008 Exposure assessment: green tea intake Lowest exposure: never Highest exposure: monthly or less, weekly or less, daily or several times/week Kubik 2004 Exposure assessment: green tea intake Lowest exposure: never Intermediate exposure: monthly (≤ 1 month ) or weekly (≤ 1 week but > 1 monthly) Highest exposure: daily (daily or several times/week)
Notes	Kubik 2008 Funding: grant (N. NR/8411‐3) from the Internal Grant Agency of the Ministry of Health of the Czech Republic and by an institutional research project (N. MZO 00064211) from the Ministry of Health, Czech Republic Statistical methods: unconditional logistic regression Variables controlled in analysis: age, residence, education and pack‐years of smoking Variables controlled by matching: ‐ Kubik 2004 Funding: Internal Grant Agency of the Ministry of Health of the Czech Republic (NJ/6732–3) Statistical methods: unconditional logistic regression Variables controlled in analysis: age, residence, education and pack‐years of smoking Variables controlled by matching: ‐