The article by Kelloff et al is a useful, comprehensive review of the current strategy underlying the development of clinically useful chemoprevention agents. One important topic that is not addressed in the article is the failure of micronutrients (selected on the basis of favorable epidemiologic finding) when tested as chemopreventive agents in clinical trials. Two examples of this are particularly noteworthy: In two large randomized trials involving heavy current or former smokers, b-carotene supplementation resulted in an increase in lung cancers compared to placebo.[1,2] Also, in two large randomized trials, folic acid(Drug information on folic acid) supplementation had no effect on the natural history of cervical intraepithelial neoplasia.[3,4]
Neither of these agents had undergone the types of preclinical and phase I/II evaluation or the development process outlined by Kelloff and colleagues. These findings thus argue strongly for following the strategy presented in the article before a large, multimillion dollar, multiyear phase III or IV trial is launched. Simply put, favorable epidemiologic observations are not sufficient to launch an agent into a definitive phase III or IV clinical chemoprevention trial.
A New View of Carcinogenesis
Although the authors' division of chemoprevention and chemotherapy development into approaches that they feel are divergent has heuristic value, and was an approach with which I once agreed, my thinking on this issue has evolved. I now strongly believe that carcinogenesis is a continuous, albeit nonlinear, process and that cancer, even at its most advanced stage, is suppressible and/or reversible by "chemoprevention agents." In animal tumors, many nonchemotherapeutic agents can shut down malignancy without exerting a cytotoxic effect. In humans, the sole example of such an effect is the response of acute promyelocytic leukemia (APL) to retinoic acid, which, one can argue, represents the equivalent of prevention (preventing progression).
To my mind, every tumor, even at a late stage, has its "retinoid equivalent." We just have to find the relevant ligand and receptor, or perhaps, like in APL, stumble upon the observation.
The systematic development of a marker that can be modulated and is valid as a measure of chemopreventive effect is an approach that we have followed in the development of retinoic acid for cervical intraepithelial neoplasia and difluromethylornithine (DFMO [Eflornithine]) for colon polyps. However, an equally valid approach would be to develop appropriate agents (eg, inhibitors or analogs, as the case may be) directed to products of the gene alterations that are cumulatively acquired during carcinogenesis. The ability to reverse the entire phenotype, even late in the disease, using chromosome or gene transfer suggests that protein products are provided that can "correct" the carcinogenesis progression process. Maybe the first priority marker (ie, differentiation and proliferation) outlined in the authors' Table 3 is secondary and derivative, while the second priority marker should be viewed as causal. Perhaps a specific molecular chemoprevention approach would be more fruitful.
Issues Related to Toxicity
Finally, two areas related to toxicity require further comment: the evaluation of long-term toxicity and of toxicity in women of child-bearing age. The assessment of any agent for toxicity over the long term is a hard task. Placebo-controlled studies are clearly needed, as the "noise level" of common complaints may be as high as 20% to 30%, and false-positive abandonment of a compound due to "toxicity" is a real danger without a control group.
The special issue of the use of chemoprevention agents in young women also needs to be approached thoughtfully and risk-benefit considered closely. For example, although DFMO is an excellent chemopreventive agent to try in cervical intraepithelial neoplasia III, it is highly teratogenic and embryotoxic during early gestation. Therefore, its widespread usage, even if a phase III trial confirms the early favorable phase II results, would be difficult.
These comments are meant to extend the thinking of the material presented in the review by Kelloff et al, which represents an excellent presentation of an important new approach to cancer management.