Predicting the Prognosis of Hepatocellular Carcinoma Using Gene Expression

BackgroundMany risk factors affect survival after liver resection for hepatocellular carcinoma (HCC), but these lack specificity for the prognosis. Studies of gene expression profiles successfully predicted the survival of HCC. To date, few studies have focused on HBV-associated HCC. Therefore, we investigated the genes involved in tumor prognosis in patients with HBV-associated HCC.Materials and MethodsWe analyzed clinical data and microarray test results in 51 patients who underwent liver resection for HBV-associated HCC between August 1998 and December 2002. We used Kaplan-Meier plots to analyze survival rates and some well-known clinical staging systems. In addition, we performed microarray survival analysis using Cox univariate regression. Then, we devised a scoring system that adds the survival-associated gene expression signature to one or two independent clinical risk factors for survival.ResultsThe mean follow-up period was 61.4 mo. Thirty-six patients had recurrence during this period, and 22 died. The 5-y survival rate was 58.0%. Multivariate analysis showed that tumor size was an independent risk factor for survival. We identified 194 spots on the microarray in survival analysis. These genes were clustered into two groups and showed a statistically significant difference in the survival rates. In the clinical analysis, the CLIP and Okuda staging systems showed statistically significant relationships. When we added the survival-associated gene expression signature to tumor size, our new method showed a more statistically significant relationship between stage and survival.ConclusionsWe propose that adding the results of microarray survival analysis to the staging system predicts survival more precisely.