Section 4 -

Molecular Diagnostics in the Development and the Use of Novel Cancer Therapeutics Gerald Batist Chair, McGill University Department of Oncology & Director, Segal Cancer Centre Jewish General Hospital

The identification and validation of molecular targets has been the focus of new treatment research in cancer. This spawned the development of pathology platforms that provide higher throughout capacity, especially tissue microarrays. Techniques continue to be developed to easily and quickly identify particular DNA sequence variants in small tissue samples, or protein biomarkers in tissue or even blood. More and more specific DNA polymorphisms or mutations and protein biomarkers identified in experimental research labs are being validated as effective predictors of response to a particular treatment. Important examples of molecular targets that are used in cancer therapeutics include nuclear hormone receptors such as estrogen (ER) and androgen (AR) in breast and prostate cancer. While the current standard for measurement of ER is immunohistochemistry, with pathologist rating of number of tumor cells staining and intensity of staining, a recent PCR-based panel of gene expression that predicts response to anti-estrogen therapy in breast cancer, appears to be heavily weighted by the ER. It is suggested that more quantitative measurement of ER represents a more reliable predictor of response. The androgen receptor is generally not assayed as a part of clinical pathology, since most, though not all prostate cancer cells are initially hormone dependent and therefore responsive to anti-androgen therapies. Furthermore, hormone-independence is not accompanied by the loss of expression of androgen receptor protein. The phenomenon of the "antagonist-agonist switch" provides some insights, since it appears that a specific mutation at a site remote from the AR ligand binding domain creates a change in the 3-D conformation of the receptor, such that a drug that is usually an antagonist has the opposite effect; it acts as an agonist. It may be that the molecular diagnosis of prostate cancer should include examination for some key DNA sequence variants. The ErbB family of receptors is being developed or are approved for use as targeted therapies in breast, lung, head and neck and possibly other tumor types. In each case, there is a continuing evolution of the understanding of the potentially critical features of the target molecule, and this requires ongoing research capacity, intimately tied to clinical diagnostics, in order to stay abreast of the optimal molecular context for the therapy in question. Level of gene expression, protein content, specific mutations are amongst the molecular diagnostic questions that have become clinical vital in diagnosis, prognosis and therapeutic options. For Erb-B2 (Her-2 neu), the amount of protein, as expressed either by immunohistochemistry, or where equivocal, by FISH analysis of gene copy number, is the benchmark in selecting Bevacizumab or other targeted agents. ErbB-2 (EGFr) represents a still-evolving target from a molecular diagnostics standpoint. Multiple studies failed to show a correlation between efficacy of EGFr inhibitors and levels of the receptor protein as measured by immunohistochemistry. This was followed by the revelation that specific mutations in the ATP binding site of the EGFr were associated with responses to these targeted agents. Although attractive, this has been challenged by large clinical correlative studies, which have instead indicated that gene copy number, as measured by FISH analysis, is the real predictor. Part of the solution to this puzzle is undoubtedly technologic: improved antibodies for IHC, improved sequencing techniques to identify mutations in small tissue samples, etc. This is an excellent example of why the pathologist's molecular diagnostic lab should be closely linked to the evolving world of molecular target research. Furthermore, as patients are treated with targeted agents, we have seen the emergence of novel mutations in the therapeutic molecular target. These too are recognized and studied in research laboratories. Some mutations in target genes have been unequivocally shown to be central to the tumor's response to a targeted agent. The best example is the c-kit tyrosine kinase recptor in Gastrointestinal stromal tumors (GIST). Only the tumors with particular activating mutations in c-kit are dramatically responsive to Imatinib (Gleevac). Computational biology and dynamic modeling of protein targets that incorporates these mutations can be used to predict response, or resistance, to a given therapeutic agent, and this data is fed back to drug development programs that can either alter the therapeutic model to better bind the altered target pharmacophore, or aim therapeutic targeting at another submolecular site. This talk will provide a few examples of targeted agents in use and in development where molecular diagnostics is as critical as morphology in the diagnosis, prognosis and guide for therapy. The importance of interfacing of the experimental research laboratory, with the molecular diagnostics lab and the pathology department will be highlighted. While it may seem that in many ways the weight of the molecular target is approaching that of the tumor morphology, the pathologist is the focal point of these activities, who can integrate these various components with the morphologic data to arrive at a more powerful diagnostic and prognostic determination.