For translation of preclinical chemoprevention data to human testing in randomized clinical trials, several critical issues need to be considered, such as dose, form, and bioavailability of the agent, timing of treatment, clinical trial design, and predictive value of the preclinical models. The agent dose in preclinical studies is usually 40 – 80% of the maximally tolerated dose, but in humans lower doses may be required by regulatory agencies or deemed prudent by investigators. Naturally occurring chemoprevention agents are also often tested in animal models at doses higher than those physiologically relevant in humans, which may have different, sometimes potentially harmful effects. A recent report on such biphasic effects of genistein in preclinical models suggests the potential for harm at low, physiologically relevant doses.[34
] The form of the agent used in animal studies, often mixed into the diet, may not be feasible in humans, for whom administration in tablets or capsules is typically used. The bioavailability of some agents with considerable in vitro
cancer-inhibitory activity is poor, such as is the case for curcumin.[66
] In addition, there is very little information about the bioavailability of any agent to the relevant target, prostate tissue. Only for lycopene are there data in this respect, but the metabolism of this and other compounds may be complex and result in the presence of metabolites with unknown activity in prostate tissue.[113
] Thus, preclinical studies and subsequent RCTs must be coordinated, such that both form and bioavailability of the agents tested are considered and agent metabolism is addressed as well. Importantly, the timing of agent administration in animal studies and human clinical trials typically differs considerably; chemopreventive treatment of humans is typically not considered until middle age. Therefore, delayed administration of the agent under test must be included in the preclinical research phase, which has been proven feasible.[82
] For some agents, such as green tea polyphenols and genistein, preventive activity in preclinical models has only been identified when treatment occurred early in life,[7
] which is obviously not feasible in humans.
Other important issues in translating preclinical data pertain to the study design of clinical prevention trials.[3
] Following Phase I safety studies, short-term Phase II studies, with agent administration before radical prostatectomy, are needed, to further establish safety and generate data on efficacy using relevant intermediate end-points in the prostate tissue. Beyond the Phase II studies, trials of intermediate duration are needed prior to embarking on large Phase III studies. Several study designs have been proposed, most of which are applied to populations at high risk for prostate cancer, such as men with elevated PSA, but negative biopsies, men with high-grade PIN, but no cancer on biopsy, or men with a family history of prostate cancer.[3
] One such type of intermediate trial involves treatment of men with recurring prostate cancer, with reduction in the rise of PSA in these men as the end-point.[3
] The problem with this study design is that one cannot differentiate between the effects of the test agent on prostate cancer cell growth and effects on PSA expression, which are not necessarily linked and can even occur in opposite directions. One other intermediate trial design involves treatment of men at high risk of recurrence after radical prostatectomy.[3
] This design not only includes a relatively low sample size (250 – 300 subjects) and short duration (two to three years of treatment), but focuses on prostate cancer that is clinically significant and potentially lethal. This is important because many prostate cancers currently detected in the USA have questionable or low clinical significance and may not need to be prevented. Thus, clinical trials that focus on clinically significant prostate cancer are crucial in developing chemopreventive agents that are active against aggressive, potentially lethal forms of prostate cancer. However, no such studies have been completed to date. Of note, two of the currently completed Phase III RCTs for the chemoprevention of prostate cancer, SELECT and PCPT, involve average risk men, whereas participants of the REDUCE trial had elevated risk of prostate cancer associated with elevated PSA levels. In one other RCT with men at increased risk of prostate cancer because of the presence of high grade PIN on biopsy, but no cancer[127
], the antiestrogen toremifene did not significantly reduce prostate cancer development in three years of follow-up [http://prostatecancerinfolink.net/risk-prevention/prevention-prostatecancer/other-trials/
]. In all four studies, the majority of detected cancers were likely of low clinical significance but not distinguishable from potentially lethal cancers.
Predictive ability of preclinical models
The ability of preclinical models to predict the outcome of subsequent clinical trials is one of the most important issues in the translation of preclinical chemoprevention data. Preclinical studies of selenium and vitamin E, with rat models, have been uniformly negative, as indicated earlier, and were thus fully predictive of the negative outcome of SELECT. Similar preclinical model studies with antiandrogens[73
] were also predictive of the reduction in prostate cancer development in the PCPT and REDUCE trials with 5α-reductase inhibitors.[77
] Tamoxifen was not active in preventing prostate cancer in a rat model study,[74
] predictive of the lack of significant efficacy of the antiestrogen toremifene. The predictive value of animal models for other chemoprevention agents is not clear because of the lack of definitive clinical trials. The limited clinical trials with silibinin, green tea polyphenols, lycopene, soy, and 4-HPR are insufficient to allow a formal definitive evaluation of the predictive ability of preclinical animal models for these agents.