A high probability of benefit is desirable to justify the choice of anti-angiogenic therapy from an ever-expanding list of expensive new anticancer agents. However, biomarkers of response to cytotoxic agents are not optimal for predicting benefit from anti-angiogenic drugs. This discussion will focus on both preclinical and clinical research to identify biomarkers for anti-angiogenic therapies that can inform dosing, early clinical benefit, initial drug choice, emerging resistance and second-line treatments.

Breast cancer is currently the most common form of serious malignancy in women, affecting approximately 1 in 9 of the female population in the Western World [l]. The etiological background of breast cancer is currently largely unclm, however, diet, reproductive lifestyle and heredity are thought to be important risk factors [2]. The most life-heatening aspect of many forms of malignancies, including breast cancer, is the ability of cells from the primary tumour to disseminate to distant sites of the body. This dissemination, or metastasis as it is known, is a complex multistep process requiring a number of genetic interactions and a variety of gene products [3].In order to study the process of metastasis at the molecular level, a rodent mammary tumour model has been developed in our laboratory. This completely syngeneic system centres around the use of the highly-inbred Furth-Wistar rat [4]. A cell line, designated Rama 37 was isolated from a benign mammary tumour that developed following Uearment of the Furth-Wistar rat with dimethylbenz[a]antene (DMBA). This cell line, when injected into the mammary fat pads of the syngeneic host produces benign, nonmetastatic, encapsulated tumours. A metastatic derivative of the Rama 37 cell line has previously been produced by transfecting this cell line with genomic DNA fragments from a metastasizing human breast carcinoma derived cell line [5]. The resultant transfected cells, when injected into the mammary fat pads of the Furth-Wistar rat, developed metastases at a high incidence. From one such lung metastases, a cell line was isolated designated Ca2-5-LT1 which when introduced into the host also metastasized [5].In an aftempt to identify those gene products that are associated with the progression from the benign tumour producing cell line, Rama 37 to the metastatic derivative, Ca2-5-LTl a variation of the subtractive hybridisation technique has been used [6-71. This technique was performed in a manner that permitted the identification of those mRNAs expressed more highly in the Ca2-5-LT1 cell line when compared to the Rama 37 cell line. Following the implementation of the subtractive hybridisation procedure and the production of a subuacted cDNA library, Northern blot analysis was performed using individual subtracted cDNAs as hybridisation probes. From these studies a number of subtracted cDNAs were identified which corresponded to mRNAs that were expressed more highly in the metastatic cell line when compared to the non-metastatic counterpart.

Loss of heterozygosity (LOH) at the mannose 6-phosphate/insulin-like growth factor 2 receptor gene locus (M6P/IGF2R) on 6q26-27 has recently been demonstrated in approximately 30% of both invasive and in situ breast cancers. LOH was coupled with somatic point mutations in the remaining allele in several instances, leading to the proposition that M6P/IGF2R is a tumor suppressor gene. Somatic mutations in M6P/IGF2R have also been described in hepatoma and gastrointestinal cancers with the replication error positive (RER+) phenotype. These data indicate that M6P/IGF2R loss of function mutations may be involved in the pathogenesis of a wide spectrum of malignancies. Extensive data on the normal function of the M6P/IGF2R suggest that loss of M6P/IGF2R activity may contribute to multiple aspects of tumor pathophysiology, including deregulated growth, apoptosis, angiogenesis and invasion.

Benign rat mammary epithelial cells transfected with restriction enzyme-fragmented DNA from a human malignant metastatic cell line (Ca2-83) produces transfectants that yield metastatic tumours in syngeneic rats. The six metastasis-inducing DNAs (Met-DNAs) that have been isolated from such transfectants are subgene in size and do not code for any expressed mRNAs, but correspond to potential regulatory regions of human DNA from malignant, metastatic cells. In pilot studies the one Met-DNA tested is detectable in some human breast tumours but not in normal tissue. Transfection of all six Met-DNAs singly into the benign mammary epithelial cells causes enhanced expression of osteopontin, whilst transfection of cDNA for osteopontin also induces the metastatic state. These results show that short regulatory DNAs exist in human cancer cells that can induce metastatic spread via a common eector gene, osteopontin, in model rat mammary cell lines.