Predicting prostate cancer behavior using transcript profiles

Abstract

Purpose: Prostate cancer represents a disease with diverse clinical outcomes. Treatment strategies that optimize benefit and minimize morbidities depend on accurate estimates of disease status and likelihood of progression. Emerging technologies capable of qualitatively and quantitatively profiling genes expressed by neoplastic tissues may provide insights into tumor behavior. This review discusses the use of microarray based transcript expression profiling to stratify human cancers into risk categories.

Materials and methods: MEDLINE was used to perform a comprehensive literature review of reports describing the assessment of gene expression profiles in malignant diseases. Particular emphasis was placed on studies developing models using individual genes or gene cohorts as predictors of prostate cancer outcome.

Results: Alterations in the expression of individual genes identified by microarray analyses have been used in studies of outcome in cancers of the prostate and other tissue types. Profiles of expressed genes have been used to develop prediction models that stratify cancers into prognostic categories based on relapse rates or responses to therapy.

Conclusions: Gene expression profiles offer an opportunity for acquiring molecular determinants correlating with clinical outcome. With rare exceptions these profiles have yet to be validated or used in prospective studies. Future research will benefit from assessments of intratumor heterogeneity and host factors such as the immune response and hormonal milieu. The prospective validation of predictive models will serve to prove usefulness in the clinical arena.