HOME
Return to Table of Contents
VALIDATION OF A GENE SIGNATURE PREDICTING RESPONSE TO TREATMENT IN CHRONIC HEPATITIS C VIRUS INFECTIONS
L Chen1,2, I Borozan1, J Sun1, N Anand3, J Heathcote3, AM Edwards1,2,5, ID McGilvray4*
1Banting and Best Department of Medical Research, 2Department of Medical Genetics and Microbiology, 3Departments of Medicine, 4Surgery, and 5Medical Biophysics, University of Toronto, Toronto, Ontario
INTRODUCTION: We previously identified an 18-gene signature that predicts treatment response in chronic hepatitis C (CHC) in a population of 31 patients. In this study we asked whether its predictive capacity held true for a larger cohort.
METHODS: Liver biopsies were taken from 78 CHC patients prior to initiating treatment with pegInterferon/ribavirin. RNA was extracted, amplified, and studied using a 19K cDNA microarray. CHC liver biopsies were compared to biopsies from 20 normal, uninfected livers. The levels of 10 of the genes from our previous set of 18 were assessed by real-time PCR. Classification accuracy was assessed using 4 different methods (KNN, DQDA, DLDA, CART).
RESULTS: All four classification methods yielded similar results. For predicting “responders”, the 18 gene signature had a positive predictive value (PPV) of 96% (sensitivity 80%, specificity 86%). There was higher variability in the 10 gene real-time PCR dataset, producing a lower PPV of 88% (sensitivity 75%, specificity 59%). In all, 123 genes were differentially expressed in CHC (p <0.001, fold<or> 1.5). As in our preliminary study, pre-treatment expression of a number of interferon-stimulated genes (ISGs) was higher in nonresponder liver tissue. These included the ISGs in our predictive subset of 18 genes. When the 123 genes were analyzed by multivariate analysis, responder status and genotype were the two most important variables contributing to levels of gene expression.
CONCLUSION: The pre-treatment difference between responder and nonresponder liver tissue is robust over large numbers of patients, a finding that has implications for diagnosis and disease pathogenesis.