Genomics
Our major effort is focused on defining prognostic gene expression signatures (i.e., a molecular classification) for prostate cancer.
Prostate cancer is a leading cause of cancer-related death in American men, second only to lung cancer. Gene expression profiles that molecularly characterize prostatic neoplasms may identify genes involved in prostate carcinogenesis, some of which may represent targets for therapeutic intervention. Molecular characterization of prostate cancer should also elucidate clinical biomarkers with diagnostic and prognostic utility, and may help direct clinicians to the most effective and least toxic existing therapies.
In our previous work, we learned that cDNA microarrays can be used to molecularly distinguish normal prostate tissue, benign prostatic hypertrophy (BPH), localized prostate cancer, and metastatic, hormone-refractory prostate cancer (Nature 412:822). The Genomics Laboratory is exploring in detail the molecular "fingerprint" of clinically-localized prostate cancer to determine whether global gene expression analyses can delineate classes of prostate cancer that correlate with clinical outcome and/or histopathologic grade.
In addition to using various clustering algorithms to analyze our prostate cancer data sets, we also will apply supervised learning strategies to predict prostate cancer prognosis and grade. Furthermore, we will characterize selected candidate biomarkers, using high-density tissue microarrays, which allow us to look at hundreds of clinically stratified patient specimens. Thus, based on our experience to date with these approaches, our overall hypothesis is that global gene expression analysis will elucidate a select set of genes that dictate whether clinically localized prostate cancer exhibits a progressive or non-progressive phenotype.
Molecular profiling of prostate cancer may allow for both an improved prostate cancer classification and lead to the identification of candidate biomarkers and regulatory genes.
We believe that we will continue to make rapid progress in this area because we have accumulated an extensive amount of experience with the use of high-throughput genomic and proteomic tools, an established bioinformatics and biostatistics infrastructure within the laboratory, and our assembly of an organized tissue resource set with linked clinical/pathology data.
