Determining proper treatment for individuals diagnosed with prostate cancer suffers because of the current unreliability of methods to predict the clinical course of the disease. Existing procedures for patient stratification include a series of clinical, biochemical, and histopathological examinations, such as prostate specific antigen (PSA) levels, tumor stage, Gleason score, and genetic approaches that look for specific oncogenic alterations. Used together, these tools still have only limited classification capability. Recent studies utilizing cDNA microarray technology have indicated that specific genetic markers may be useful as potential prostate cancer markers. In the March 15 issue of the Journal of Clinical Investigation, Gennadi Glinsky and colleagues, from the Sydney Kimmel Cancer Center, San Diego, CA, present a highly accurate gene-expression-based prostate cancer recurrence predictor algorithm. The authors used a microarray-based gene-expression profiling method to determine molecular signatures that could distinguish subgroups of patients with different disease outcomes. The algorithm was developed utilizing an analysis that assessed expression profiles of over 12,000 genes from 100 different tumors. Five highly discriminating gene-expression clusters were assigned from this analysis and then tested on a new set of 79 patients. Of those patients who ultimately had recurrence of prostate cancer, the algorithm correctly classified 88% of these patients into the poor-prognosis group. This study provides a significant advance in our ability to properly classify the clinical outcome of prostate cancer patients for use in determining appropriate treatment.
In an accompanying commentary, Mitchell Benson and James McKiernan of Columbia University, place this study in context with other methods for predicting clinical outcome, and how combination of these methods with that presented here should enhance
Contact: Laurie Goodman
Journal of Clinical Investigation