PASCAL - Pattern Analysis, Statistical Modelling and Computational Learning

A Four-Gene Expression Signature for Prostate Cancer Cells Consisting of UAP1, PDLIM5, IMPDH2, and HSPD1
Isabelle Guyon, Herb Fritsche, Paul Choppa, Li-Ying Yang and Stephen Barnhill
UroToday Int J. Volume 2, Number 4, 2009. ISSN 1944-5784

Abstract

INTRODUCTION: The objective of the study was to develop a gene expression test that is highly associated with the presence of prostate cancer for use as an adjunct to the pathology examination of tissue. METHODS: A gene expression database (U133A Affymetrix) was produced from 87 preparations of laser microdissected cells obtained from cancer (G3 and G4) and noncancer prostate tissues. The database was analyzed using univariate feature ranking and recursive feature elimination algorithms (support vector machine) to identify overexpressed genes that were associated with prostate cancer. RT-PCR assays were developed for the unique 4-gene set that was found to be reflective of prostate cancer. The gene expression data were used to construct a mathematical equation to classify tissues as cancer vs noncancer. The RT-PCR tests and the calculated gene expression score were validated in an independently collected set of formalin-fixed and fresh-frozen prostate tissues. RESULTS: Analysis of the U133A gene expression database identified a group of 63 genes that were overexpressed in cancer and also gave an AUC (area under the curve) of > 0.84 for separating cancer vs noncancer. The gene discovery was validated with a database of 164 independently collected tissues reported in the Oncomine database. The 63 gene set was reduced to a subset of 4 complementary genes (UAP1, PDLIM5, IMPDH2, and HSPD1), using univariate feature ranking and recursive feature elimination (RFE) algorithms that gave an AUC = 0.94 for discrimination between cancer and noncancer prostate cells. Quantitative RT-PCR (reverse transcriptase polymerase chain reaction) assays were developed and validated. A mathematical formula based on the gene expression values of the 4 genes along with a housekeeping gene was developed for the classification of cancer vs noncancer tissues. In a blinded validation study of 71 independent prostate tissue samples that included both fresh prostate tissues and formalin fixed tissues, the 4-gene test gave a sensitivity of 90% with a specificity of 97% (the 95% confidence interval was 86% - 100%). CONCLUSION: The 4-gene RT-PCR test can be used to detect Gleason grade 3 and grade 4 cancer cells in prostate tissue and may be useful as an adjunct test to the pathology examination of prostate tissue taken at biopsy or prostatectomy.

PDF - Requires Adobe Acrobat Reader or other PDF viewer.
EPrint Type:Article
Project Keyword:Project Keyword UNSPECIFIED
Subjects:Learning/Statistics & Optimisation
ID Code:6036
Deposited By:Isabelle Guyon
Deposited On:08 March 2010