OBJECTIVETo investigate whether gene expression profiles can be used to determine risk genes and predict HBV-related cirrhosis progression to liver carcinoma using Significance Analysis of Microarray (SAM) and Prediction Analysis of Microarray (PAM) methods.

METHODSThe Affymetrix GeneChip was used to establish the gene expression profiles of liver tissues from 15 patients with chronic hepatitis B and cirrhosis or hepatocellular carcinoma (HCC). Differentially expressed genes (fold-change more than 2; P value less than 0.01) were selected by GeneSpring GX software. Risk genes related to cirrhosis and liver carcinoma were generated by SAM and PAM methods. Real-time PCR was used to verify the expression of risk genes in the liver tissues.

RESULTSSamples were clustered into the cirrhosis subgroup (n =15) or the HCC subgroup (n =15). A total of 497 differentially expressed genes were identified, SAM identified 162 significant genes, including 18 up-regulated genes and 144 down-regulated genes (fold-change:-1.46 to 1.28). PAM identified 22 genes with a "poor risk signature" (defined with a threshold of 5.5), which were associated with classifying cirrhosis and liver carcinoma; of these risk genes, 4 were down-regulated and 18 were up-regulated in the HCC group compared to the cirrhosis group (fold-change: 2.038 to 7.897, P value less than 0.01). The correction of classification was more than 80% . FOXP1, SPINK1 and KCNJ16 were verified by real-time PCR as differently expressed in the two subgroups (P value =0.011, 0.002 and 0.004, respectively).

CONCLUSIONThe altered gene profiles of carcinogenesis in HBV-related cirrhosis involves hundreds of genes. The combination of three "poor risk genes" may represent potential targets for diagnosis and prediction of liver carcinoma progression.