ED04-03

The chemicallyinduced model of mammary cancer in rats was developed more than 40 years ago by Huggins and coworkers at the University of Chicago. The resulting tumors which appear within 6-20 wqeeks following treatment with carcinogen {methylnitrosourea, DMBA] are adenocarcinomas and are overwhelmingly ER+. The tumors appear histologically and by gene array analysis to be similar to highly differentiated ER+ tumors in humans. In contrast to humans roughly 50% of these tumors have mutations in the Ha Ras oncogene. The model was used by Huggins and coworkers to examine the role of hormonal manipulation in tumor development including role of estrogen , progesterone, pregnancy etc. The carcinogen induced model was subsequently employed to determine the efficacy of the SERMS (tamoixfen, raloxifene){Jordan and co workers}. It has also been used to examine the efficacy of aromatase inhibitors [Lubet et al 1994]. Not surprisingly agents which have proven to be highly effective in prevention and therapy of ER+ tumors in humans have also proven to be highly effective in this specific model.
 Because of the ease with which tumors can be generated and the substantial numbers of agents which have been tested in this model it has often been employed to screen for potential agents which might prove relevant in the prevention of human breast cancer. As mentioned above it has proven susceptible to a wide variety of agents which alter the hormonal axis. Interestingly however it has proven susceptible to classes of agents which presumably do not function by altering the hormonal axis. One of the first class of non hormonal agents that was profoundly effective in this model were the RXR agonists [Gottardis et al 1996]. This inital observation has been confirmed by various investigators including Dr. Sporn. Thus a wide variety of RXR agonists , most of which increase triglyceride levels as their primary toxicity, have proven effective. We recently examined a number of sythetic RXR agonists , based on a 9cis retinoic acid backbone, and shown that these agents were effective in this model. The particular interest of this is that at least one of these effective agents [UAB 30] does not increase triglyceride levels {Muccio 2006}.
 In addition to results with the RXR agonists a number of other non hormonal agents appear to be effective in this model. Thus the farnesyltransferase inhibitor (tipifarnitib) was effective and preferntially inhibited development of tumors with an Ha Ras mutation[Lubet et al 2006]. Also we have observed that a number of EGFr inhibitors {Iressa and Tarceva) are effective in this model.
 In addition Dr. Sporn has observed that a number of triterpenoids have been highly effective in this model.
 In addition to the large number of effective agents we have examined the widest variety of ineffective agents including various statins, Vitamin C, Vitamin E, 1,2 dithiol3-thione, phenyl butyrate etc. Finally we have used this model to determine variouis agents with moderate efficacy e.g. 5,6 benzoflavone, 9 cis RA, Celecoxib, various NSAIDS etc.
 The wide variety of agents which are both structurally and mechanistically varied has allowed us to use this model to examine for potential biomarkers in this model which we hope will prove applicable to humans. We initially observed that the aromatase inhibitor vorozole profoundly decreased tumor cell proliferation and increased tumor cell apoptosis in small palpable lesions treated short term (Christov 2000). Human trials in a neoadjuvant setting have similarly shown that short term treatment with aromatase inhibitors in humans profoundly decreased tumor cell proliferation and was highly correlated with the overall efficacy of these agents {Dowsett and coworkers 2005,2006,2007}. We have used these limited exposure studies to look at a wide variety of agents to determine whether short term treatment of palpable lesions followed by measurements of proliferation and apoptosis could predict long term chemopreventive efficacy. Initially we followed a variety of RXR agonists and showed that the short term markers could readily predict and differentiate highly active agents from agents that were less active [Muccio 2006]. More recently we have expanded this to the widest range of agents in this model and again found that in this model that one could readily differentiate highly effective agents from less effective agents based on short term biomarker changes (Christov et al Clin. Cancer Res 2007). We feel that this approach bears some similarities to what has been and can be employed in humans in the neoadjuvant setting.
 Because of the wide and varied agents for which we have complete prevention data we have examined other potential biomarkers which might prove relevant to human trials e.g. gene array technologies and imaging techniques. Preliminary results with these alternative biomarkers will be discussed.

Sixth AACR International Conference on Frontiers in Cancer Prevention Research-- Dec 5-8, 2007; Philadelphia, PA