0 array, 41,751 probe sets corresponding to the known human genes were selected. Among them, the numbers of probe sets marked as “present” in >70% of samples derived from cancer tissue and adjacent noncancerous tissue were 17,303 and 16,431, respectively. Using cancer tissue samples, 11 probe sets satisfied the criteria by the Cox likelihood ratio test for recurrence free (P < 0.01) and more than 2-fold change between the recurrence and nonrecurrence groups (Supporting Table 1). Among them, the criteria by Wilcoxon signed-rank test (P < SAHA HDAC 0.01) and more than 2-fold change between the paired cancer and noncancerous tissues

were satisfied by two probe sets: immunoglobulin kappa locus and family with sequence similarity 134, member B (Table 1). Using noncancerous tissue samples, five probe sets satisfied the criteria by Cox regression analysis and more than 2-fold change (Supporting Table 1). Among them, three probe sets, cytochrome P450 1A2 (CYP1A2), carnosine dipeptidase 1 (CNDP1), and ornithine aminotransferase (OAT), showed significant down-regulation in paired cancer tissue. It is of interest that both carnosine and ornithine are known as amino acids closely related

to oxidative stress in the liver.15, 16 The multivariate Cox regression analysis with the forward variable-selection procedure revealed that a MLN0128 model containing only the gene-expression pattern of CYP1A2 (HR, 0.447; 95% CI, 0.249-0.808; P = 0.010) was the best predictive model for the recurrence of HCC (Table 1). We examined the association between clinicopathological selleck chemicals factors and recurrence-free survivals of 78 patients with early-stage HCC meeting the Milan criteria (Table 1; lower). Median recurrence-free survival time was 23.7 months. Univariate Cox regression analysis demonstrated that HCV infection (HR, 2.095; 95% CI, 1.048-4.188; P = 0.031) and low serum albumin level (HR, 0.447; 95% CI, 0.248-0.808; P = 0.010) correlated with the cumulative recurrence-free rate. Other clinicopathological

factors were not statistically significant. We further performed the forward variable-selection procedure using the two factors, and the serum albumin level was identified as the only risk factor for recurrence after curative resection for early-stage HCC (Table 1). To evaluate the best predictive model for patients with HCC meeting the Milan criteria, we considered the combined model using the serum albumin level and the gene-expression pattern of CYP1A2 (Table 2). A multiple logistic regression analysis demonstrated that CYP1A2 (OR, 0.534; 95% CI, 0.276-0.916; P = 0.036) was the only factor that contributed to the prediction of the recurrence. Moreover, the variable-selection procedure by logistic regression analysis identified that CYP1A2 was the best predictor for the recurrence of HCC meeting the Milan criteria (OR, 0.493; 95% CI, 0.273-0.787; P = 0.008).