Conclusions

A bstract : The “considerations” addressed in this section consist of a number of thought‐provoking issues and unresolved questions that emerge from the papers in this volume. The evidence for tamoxifen carcinogenicity in animal models and, to a more restricted extent, in humans has led some investigators to question whether SERMs are ready or appropriate for clinical testing—specifically, in a disease prevention setting involving healthy but high‐risk individuals. There is, however, inconsistency in both efficacy and toxicity—specifically, carcinogenicity—between animal models and humans, leading others to question the value of basing the decision to proceed with clinical studies on preclinical results in animals. Although the molecular basis for SERM action is rapidly being clarified, the cellular activity of these agents is still elusive. We discuss the view that the efficacy of tamoxifen in breast cancer is based on its treatment of “occult cancers,” or small collections of cancer cells that are not clinically apparent, not only in the context of prevention but also in the treatment setting. As part of our approach that assumes estrogen activity to be the foundation upon which SERM development is being modeled, we discuss the inconsistency between the epidemiologic data and prospective randomized data with respect to the relationship between estrogen use and cardiovascular disease. The need to validate surrogate markers of SERM action is discussed in relation to bone but is clearly relevant to all disease sites. The semantics used in describing SERM action as agonistic or antagonistic in relation to estrogen at various target sites has been inconsistent, especially in the clinical context. We attempt to dissect out some of the inconsistencies in semantics in the hope that this will contribute to improved communication of data resulting from SERM research. In the clinical arena, we begin with the premise that the large, simple randomized trial offers the optimal design for the testing of SERMs. In view of limited resources, however, we counter this position with alternative, if less desirable, approaches to the clinical format for SERM testing. Finally, we explore the process by which statistically meaningful results from clinical trials are extrapolated into the specific drug indications that apply to clinical practice.