ProDiet: A Phase II Randomized Placebo-controlled Trial of Green Tea Catechins and Lycopene in Men at Increased Risk of Prostate Cancer

Prostate cancer is the commonest noncutaneous male malignancy worldwide with higher incidence in developed countries, in part due to screening with PSA (1). Around three quarters of men with an elevated PSA will not have cancer diagnosed immediately, so safe and effective chemoprevention would be beneficial as they are at the risk of future diagnosis (2). Randomized controlled trials (RCT) of prostate cancer prevention have included pharmaceutical and nutritional agents but with limited success. Finasteride [a 5-alpha-reductase inhibitor (ARI)] reduced prostate cancer risk in an randomized controlled trial (RCT) but the increase of high-grade tumors in the finasteride group and other adverse effects prevented licensing (3). Dutasteride (another ARI) also showed a lower incidence of prostate cancer but with a small excess of cardiac failure events (14 extra cases in 4,105 men, P = 0.03; ref. 4) More recently, the risk of higher grade cancers on 5-ARIs was not confirmed although there is an increased sexual side-effect profile (5). The lower prostate cancer incidence in a lung cancer prevention trial for smokers with vitamin E (6) and in a selenium trial for patients with melanoma (7) informed the design of the selenium and vitamin E phase III SELECT trial. However, despite being a well-conducted RCT with an intensive recruitment strategy and high enrolment of ethnic minority participants, there were no benefits at a 5-year interim analysis of SELECT. The trial was stopped prematurely and subsequently a higher prostate cancer incidence was shown in the vitamin E group (8).

Promising dietary chemoprevention agents based on preclinical and observational studies include lycopene, the major carotenoid in tomatoes. Lycopene is an active singlet oxygen quencher that assists DNA repair mechanisms (9). The World Cancer Research Fund/American Institute for Cancer Research (WCRF/AICR) systematic review categorized lycopene as “probable for decreased prostate cancer risk” (10) which was revised to “limited - no conclusion” in 2014 (11). One of the few RCTs of lycopene supplementation for prostate cancer prevention in Afro-Caribbean men with high-grade prostatic intraepithelial neoplasia (HGPIN, a potential prostate cancer precursor) did not alter PSA levels (12). However, participants were not blinded, follow-up was only 4 months, and the acceptability of lycopene was not assessed in this trial.

Prostate cancer incidence in Asian countries with high consumption of green tea such as Japan is lower than in western countries (13). Epigallocatechin-3-gallate (EGCG) is the most abundant and potent catechin in green tea with antitumor activities on inflammatory, insulin growth factor, and androgen signaling pathways (14). The WCRF/AICR review found “limited- no conclusions” possible for green tea for prostate cancer although a more recent review found it preventive (15). A recent small trial of EGCG capsules in men with HGPIN concluded that there was no difference in prostate cancer rates on rebiopsy at 1 year, although PSA levels were reduced by around 1 ng/mL and the capsules were well tolerated (16).

Evidence is now required from RCTs of dietary interventions for prostate cancer prevention to detect clinical benefits and harms. However, given the paucity of randomized trials, we firstly aimed to investigate the acceptability and feasibility of dietary modification and supplementation for green tea and lycopene in men at increased risk of prostate cancer in a placebo-controlled trial. Here we report the ProDiet (Prostate Diet) feasibility trial primary and secondary endpoints.

This was a placebo-controlled phase II randomized trial that compared green tea and lycopene dietary modification and capsules in men with PSA results between 2.0 and 2.974 ng/mL, or between 2.975 and 19.95 ng/mL with a negative biopsy. The study was approved by the UK Healthcare Research Authority Trent Multicentre Research Ethics Committee (08/H0405/61), conducted according to Declaration of Helsinki, 1964 and the trial number is ISCTRN 95931417.

Men who had participated in the Prostate testing for cancer and Treatment (ProtecT) trial at nine family practices in a UK city were invited to the ProDiet study between 2009 and 2010. The ProtecT trial (ISCTRN 20141297) is a population-based RCT of treatments for localized prostate cancer (17) In brief, 82,849 men aged 50–69 years were invited for PSA testing in nine centers and those with a raised PSA result were invited for standardized prostate biopsies. No ProtecT tests were offered if the PSA result was below 3.0 ng/mL or for negative biopsy results.

Men attended a ProDiet appointment at their family practice with a study nurse who assessed eligibility, explained the potential risks and benefits of participation and obtained written consent. Men were excluded with a history of allergies to lycopene-containing foods or green tea; current or prior prostate cancer; major comorbidities or 5-ARI medication.

Participants were randomly allocated to one of three lycopene interventions and to one of three green tea interventions using a blocked random allocation [1:1:1 ratio; generated by the trial statistician (C. Metcalfe) using the Stata uniform() function] so that around 14 men were allocated to each of the nine lycopene and green tea intervention combinations. The intention had been to stratify the allocation by baseline PSA; however, this was impractical with opaque envelopes for allocation.

The allocation was concealed from the study nurse recruiting individuals until the participant's details were logged electronically with the research center, the nurse then opening the next numbered envelope containing the participant's allocation. If the participant was allocated to capsules, the bottle number would be indicated. Participants allocated to green tea drink were given a month's supply, and those to lycopene diet were given verbal and written advice (each intervention had a patient information leaflet).

To maintain blinding of participants and the study nurse, active and placebo capsules were very similar in appearance and provided to the study nurse in sealed packs. There was no blinding of participants allocated to dietary modification.

Men in the lycopene dietary group were advised to consume one to two portions of preferably cooked tomatoes daily (examples given included a bowl of soup, a heaped tablespoon of tomato puree or ketchup, two tinned tomatoes, one medium fresh tomato or a glass of tomato juice). Fruits such as watermelon or pink grapefruit were allowed, but men were advised that their lycopene content was lower than tomatoes. No foods were provided to this group. The other lycopene interventions comprised one daily gel capsule of 15 mg tomato-derived extract of lycopene (Solanum lycopersicon L. Solanaceae, Lyc-O-Mato, Lycored Ltd, authorized as dietary supplement in the United Kingdom) or a matched placebo capsule (Lycored Ltd) taken with a meal with water. The green tea (GT) interventions consisted of either drinking at least three cups (two U.K. mugs) of green tea daily (Camellia sinensis L, Theaceae, around 600 mL/day, green tea bags, e.g., PG Tips, Unilever Ltd, provided by the study), or two 300 mg green tea leaf–derived extract capsules (600 mg/d EGCG, Frutarom Ltd, authorized as a dietary supplement in the United Kingdom) or two matched placebo capsules (Frutarom Ltd). The dose was planned as 800 mg/day but was replaced by 600 mg/day as the supplier only manufactured 300 mg capsules.

Participants were provided with the first month's supply of capsules and/or green tea bags after randomization, then by post at 1 and 3 months (providing a 6-month supply). The research nurse telephoned men at one month to arrange supplies and give further advice regarding the interventions. Participants were given weekly study logs to aid compliance and to return unused supplies at 6 months. No foods or supplements were prohibited during follow-up. It was planned to advise all groups to consume five fruits or vegetables daily but this was removed as it may have hindered adherence and acceptability. The design and delivery of participant information was informed by the Theory of Planned Behaviour (18) and men received a study fridge magnet (19) and there was a study website and newsletter.

The study nurse recorded sociodemographic information, weight, blood pressure, and clinical characteristics at recruitment and took nonfasting blood samples for PSA and metabolite analyses. Participants completed a paper questionnaire following randomization (baseline), at one (postal) and at 6 months, including a food frequency questionnaire (FFQ), smoking status, and consumption of alcohol, the Hospital Anxiety and Depression Scale (20), Profile of Moods States (21) and International Continence Society Male Short Form (22). Additional questions on intervention compliance and satisfaction, adverse events in the preceding month and dietary changes were completed during follow-up. Participants were given an adverse event form with a freepost envelope. The nurse telephoned participants at one month to discuss PSA results, interventions and to arrange further supplies (planned as in-person so PSA and weight were not measured at one month). At the 6-month appointment the nurse recorded weight and blood pressure, asked about intervention adherence and acceptability, and took blood samples. Serum samples were separated by centrifugation and were stored at −80°C until analysis. Body mass index (BMI) used height measured in the ProMPT translational study linked to the ProtecT trial (60% uptake). Follow-up was completed between 2009 to 2010.

Lycopene, green tea metabolites and PSA were measured by laboratory staff blinded to the allocation (samples were identified by study number). PSA was measured at the local hospital which used the UK National Health Service External Quality Assessment for PSA. Plasma lycopene was measured by reverse-phase high-performance liquid chromatography (HPLC) with diode array detection following extraction into heptane (23). Plasma catechins, including ()-epicatechin-O-glucuronide, () epicatechin-sulfate, EGCG and 5-(dihyroxyphenyl)-ϒ-valerolactone above 1.0 nmol/L were measured using HPLC–mass spectrometry (24, 25).

Dietary intake was assessed using a validated 117-item food frequency questionnaire (FFQ) adapted from the UK EPIC study (26) with frequency reported across nine categories from “never/less than once per month” to “six or more times per day” for the previous 6 months (green tea was added in the same format). Analyses were conducted as described previously (27) and men were excluded from dietary analyses if they were assessed as misreporting energy intake [<800 kcal/day or >4,000 kcal/day; ref. 28). Alcohol consumption was categorized as above or below national recommendations (20 units/week).

Men were interviewed at around 6 and 30 weeks after randomization to assess intervention acceptability and men's experience of the interventions (21 men as one participant declined after the baseline interview because he felt he had nothing further to contribute). Interviews were fully transcribed with baseline results published previously (27, 18).

Primary outcomes were recruitment (randomization rate) and adherence to interventions (metabolite levels). It was, therefore, planned to invite around 250 men until 126 men were recruited, allowing an anticipated 50% recruitment rate to be estimated with 95% confidence interval from 44% to 56%. If 126 men were enrolled, the study would have 90% power at the 5% significance level to detect a true 67% increase in circulating lycopene and green tea metabolites between the placebo and intervention groups (including accommodation for skewed distribution of measures) at 6 months (29). There were no planned interim analyses or stopping guidelines.

Analyses were conducted on an intention-to-treat basis using Stata 14 (StataCorp). Analysis of covariance (ANCOVA) compared the log-transformed lycopene levels across lycopene groups at 6 months adjusted for transformed baseline lycopene levels. The exponentiated coefficients are the ratio of the geometric mean lycopene levels for active lycopene groups versus placebo capsules. Many green tea values were zero so the nonparametric generalized Hodges–Lehmannn (30) median difference (plasma values) and two-proportion z test (dietary reports) were used to compare active and placebo distributions and calculate 95% CIs (P value calculated using the Mann–Whitney test).

As this was a factorial design, we assessed whether there was an interaction between lycopene and green tea interventions on 6-month lycopene levels. To maximize statistical power, an “active” group (dietary advice and lycopene capsule groups) was compared with the placebo group. Baseline lycopene was included as a covariate, and a dummy variable identifying participants receiving both active lycopene and green tea interventions allowed the evidence for an interaction to be evaluated. There was no evidence of an interaction between lycopene and green tea on serum lycopene levels (P_interaction = 0.4), allowing the two interventions to be examined separately.

ANCOVA was used to compare mean PSA levels, systolic blood pressure, and weight between intervention groups and placebo for green tea and lycopene at 6 months (corresponding baseline measure of the outcome included as a covariate). The distribution of PSA results was highly skewed, so was log-transformed before analysis.

Of 469 men approached between December 2009 and May 2010, 133 were randomized (28.4%; 95% CI, 24.3%–32.7%) giving around 45 participants in each lycopene and green tea intervention (consort diagram shown in Fig. 1). Of these, 132 men attended 6-month follow-up and 131 (98.5%) gave blood. Table 1 displays the baseline characteristics that were well matched by allocated intervention. In addition, 132 men were of white ethnicity and around half had a managerial occupation.

Plasma levels at 6 months were higher in the lycopene capsule and dietary advice groups than in the placebo group, although the prespecified target difference of 67% was not met (Table 2). Plasma metabolites (EGCG) were higher in the green tea drink and capsule groups than the prespecified target difference and were undetectable in the placebo group (Table 2).

Consumption of lycopene-containing foods was highest in the dietary advice group with no increase reported in capsule groups (Table 2; Supplementary Fig. S1). The proportion of men who reported that they drank green tea daily increased to 79% in the tea drinking group at 6 months but was low and unaltered in both capsule groups (Table 2; Supplementary Fig. S1).

Men stated in interviews that they were confident about adhering to their allocated options by quickly and easily establishing a routine to prompt them regarding their interventions (Table 3). Routines were assisted by taking capsules at meals or with prescribed medicines, while green tea drinkers either swapped some or all the normal black tea or added green tea drinks, e.g., at meals. Men randomized to dietary lycopene often added tomatoes to recipes, lunch plates, or sandwiches, and tomato juice was very popular. Men used the ProDiet log but typically discontinued it once a routine became established, often in the early days/weeks. Changes to their established routine (e.g. holidays or eating out) were the most commonly reported barriers to adherence because men forgot or chose not to take interventions with them (Table 3). However, many men reported always being able to adhere to the interventions. Some men were surprised that they found green tea palatable and few reported significantly disliking it. Men randomized to the lycopene-rich diet often reported liking tomatoes and so not minding or finding it enjoyable to increase their intake. Most men regarded capsules as “supplements” rather than medication, which may have assisted those men who usually disliked taking medication.

In future, men reported they would prefer capsules to dietary options for green tea (102/132, 77.3% preferred capsules; 26/132, 19.7% preferred drink) and lycopene (88/132, 66.7% preferred capsules, 41/132, 31.1% preferred dietary changes).

Around half of the participants intended to continue with a tomato-rich diet (64/132 48.5%) after the trial and around one third drinking green tea (45/132, 34.1%).

The frequency was generally low except for nocturia (night time urinary frequency), insomnia, and hypertension which occurred in similar frequencies across all groups (Table 4).

PSA levels did not differ between lycopene or green tea groups at 6 months (Table 5). Systolic blood pressure and weight were also comparable between all green tea and lycopene groups (Table 5).

We report a randomized feasibility trial of green tea and lycopene supplementation in men at increased risk of prostate cancer. Nearly one third of men agreed to be randomized to lycopene and green tea capsules or dietary options for 6 months. Lycopene and green tea (EGCG) plasma concentrations were higher in participants randomized to active capsule or dietary options compared with those randomized to placebo capsules. Men's accounts from interviews revealed that they quickly established routines to enhance adherence by incorporating interventions into daily life. There were also no major differences in adverse events between groups. PSA, systolic blood pressure, and weight were comparable between groups, although the trial was not designed to identify differences in these endpoints.

To our knowledge, this is the first placebo-controlled randomized trial of lycopene and green tea to include dietary options in men at increased risk of prostate cancer. The factorial design maximized the options tested and the standardized ProtecT trial prostate cancer detection process ensured a well-characterized ProDiet population. Several features helped increase trial quality such as measures to conceal random allocation, blinding of participants to capsule options, and outcome assessors. Self-reported consumption of green tea and lycopene-rich foods remained stable in men randomized to placebo capsules so contamination was low, which implies that participant blinding had been successful. Adherence and acceptability were assessed in multiple ways, including interviews which evaluated men's attitudes towards interventions. The trial design was pragmatic with no run-in period to remove noncompliant individuals, dietary restrictions, or provision of multivitamins (8).

Recruitment to cancer chemoprevention phase II trials is difficult (2) and was lower than planned in this trial, whereas an intensive process was required in the SELECT trial to ensure success (8). Adherence was high in ProDiet at 6 months but might have been lower beforehand as there were no interim assessments, although interviews indicated that men established routines to enhance adherence. Metabolites were not measured from fasted participants as this would have restricted participants attending afternoon appointments. Nonfasted measurements may have lowered values reported here although they were comparable with lycopene results from several studies (31, 32). The dose of green tea could have also been increased above 600 mg/L but was comparable with other prostate cancer prevention studies (15) and the equivalent of at least six green tea cups daily (as consumed in Japan) was unlikely to acceptable in the United Kingdom. There are some other limitations as the trial recruited previous trial participants whom might have been more disposed to adhere to dietary interventions than the general population. Participants were also predominantly white (as in the ProtecT trial), thus reducing generalization to other ethnicities with different prostate cancer risks. UK Afro-Caribbean men revealed in interviews that they were quite heavily involved in food preparation with tomatoes being central to their diet in contrast to most ProDiet and ProtecT participants (18, 27, 33).

Lycopene intake and circulating lycopene were associated with a reduced risk of prostate cancer (34, 35) in some meta-analyses, although not universally (11). A recent meta-analysis suggested that there was a 3% reduction in prostate cancer incidence per mg/day increase in dietary lycopene intake (95% CI, 0.94–0.99; ref. 35), which matches the increase seen in lycopene dietary consumption reported in this trial. The ProDiet dietary lycopene intervention equates to tomato consumption previously associated with a reduced prostate cancer risk in ProtecT trial participants (36). However, a small placebo-controlled trial of lycopene supplementation in men with HGPIN showed no differences in expression of tissue markers for proliferation or cell-cycle inhibition (MCM-2 and p27, primary endpoints), PSA or cancer rates on rebiopsy at 6 months (37). Lycopene has cardiovascular benefits although not through lowering blood pressure so different endpoints would be needed in a definitive trial to measure its broader impacts (38).

Recent systematic reviews suggest a possible role for green tea in prostate cancer prevention for HGPIN (14, 39, 40) but evidence is conflicting for overall prostate cancer incidence (41). Polyphenols were detected in prostate tissue of men with prostate cancer who had consumed green tea (but not black tea or water) prior to radical prostatectomy (42) and showed systemic antioxidant effects. The second trial of green tea capsules for 3–6 weeks before prostatectomy showed no changes in PSA nor prostate tissue biomarkers of cell proliferation, apoptosis, or angiogenesis, which the authors hypothesized may have been due to rapid clearance or poor bioaccumulation (43).

The mode of action of most chemoprevention agents remains largely unknown and the concept for prostate cancer has been deemed a failure following the SELECT and finasteride trials (44). However, preclinical evidence should be used to identify biologically active agents to enhance their likelihood of success in clinical trials (2, 45). In one example, aspirin has recently been recommended for colorectal cancer prevention in the United States (46). The UK Add-Aspirin secondary chemoprevention trial with five years of aspirin or placebo in men with high-risk localized prostate cancer has an survival endpoint including prostate cancer and a prostate-specific endpoint of biochemical failure-free survival in around 2,000 patients (47). A phase III trial of lycopene, green tea, or other chemoprevention agents would need to be of a similar size to the prostate cohort of the Add-Aspirin trial with a biological target, incidence endpoints, and a good safety profile (45).

In conclusion, men at increased risk of prostate cancer adhered successfully to lycopene and green tea dietary and capsule interventions for 6 months with few side effects. Therefore, although recruitment was moderate, dietary interventions can be evaluated in clinical effectiveness randomized trials.

No potential conflicts of interest were disclosed.

The views expressed in this publication are those of the author(s) and not necessarily those of the NHS, the National Institute for Health Research or the Department of Health.

Conception and design: J.A. Lane, J. Horwood, J.L. Donovan, D. Gillatt, J. Holly, R.M. Martin, C. Metcalfe

Development of methodology: J.A. Lane, J. Horwood, D. Gillatt, J. Holly, C. Metcalfe

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): J.A. Lane, K.N.L. Avery, J. Horwood, M. Cantwell, G.P. Caro, A. Crozier, F.C. Hamdy, H. Moody, D.E. Neal

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): J.A. Lane, V. Er, K.N.L. Avery, J. Horwood, M. Cantwell, G.P. Caro, A. Crozier, R. Macefield, E. Walsh, R.M. Martin, C. Metcalfe

Writing, review, and/or revision of the manuscript: J.A. Lane, V. Er, K.N.L. Avery, J. Horwood, M. Cantwell, G.P. Caro, A. Crozier, G.D. Smith, J.L. Donovan, J. Holly, R. Macefield, D.E. Neal, R.M. Martin, C. Metcalfe

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): L. Down, D. Gillatt, E. Walsh

Study supervision: J.A. Lane

Funding for the study was provided by Cancer Research UK (C11046/A10052). This study was supported by the NIHR Biomedical Research Centre at University Hospitals Bristol NHS Foundation Trust and the University of Bristol (to A. Lane, V. Er, and R. Martin). BRTC receives NIHR CTU Support Funding (to J. Horwood and A. Lane). G.D. Smith works in a unit supported by MRC R120971-102. The authors wish to thank all participants and Christine Croker for administrative support.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.