The human hepatoma cell lines PLC/PRF/5, HuH7, and HpeG2 were obtained from the Japanese Cancer Research Resources Bank (Tokyo, Japan), and the Hep3B cell line was obtained from the Institute of Development, Aging and Cancer, Tohoku University (Sendai, Japan). All cell lines were maintained in Dulbecco’s modified Eagle’s medium (DMEM) plus 10% fetal bovine serum, 100 U/ml penicillin, and 100 μg/ml streptomycin at 37°C in a humidified incubator with 5% CO₂. For hypoxic conditions, cells were maintained in a continuously monitored atmosphere of 1% O₂, 5% CO₂, and 94% N₂ in a multigas incubator (model 9200; Wakenyaku Company, Kyoto, Japan).

A total of 139 HCC patients who underwent hepatectomy at Osaka University Hospital and its associated hospitals were enrolled in this study. All aspects of our study protocol were approved by the ethics committee of the Graduate School of Medicine, Osaka University. All patients provided written informed consent to use their surgical specimens and clinicopathological data for research purposes. Clinical staging was based on the TNM classification of the Union for International Cancer Control (UICC), and histological grading was based on World Health Organization classification.

Immediately after surgical resection, a tissue sample was collected from the fresh specimens and stored in RNA Stabilization Reagent (RNA Later; Ambion, Inc., Austin, TX, USA) at −80°C until RNA extraction.

Total RNA was extracted by a single-step method with TRIzol reagent (Life Technologies, Inc., Gaithersburg, MD, USA) at Osaka University. Complementary DNA (cDNA) was generated by using avian myeloblastosis virus reverse transcriptase (Promega, Madison, WI, USA), as described previously (3). Real-time monitoring of PCR reactions was performed with the LightCycler system (Roche Applied Science, Indianapolis, IN, USA) for quantification of mRNA expression, as described previously (29). The housekeeping gene glyceraldehyde 3-phosphate dehydrogenase (GAPDH) was used as an internal standard. The sequences of the GAPDH primers were as follows: sense primer, 5′-CAACTACATGGTTTACATGTTC-3′ and antisense primer, 5′-GCCAGTGGACTCCACGAC-3′. EFNA1 primer sets were designed to flank one intron and were tested to ensure amplification of only cDNA to avoid amplification of possible contaminating genomic DNA. The sequences of these PCR primers were as follows: EFNA1 sense primer, 5′-TGCC GTCCGGACGAGACAGGC-3′ and antisense primer, 5′-CTG GAGCCAGGACCGGGACTG-3′.

Microarray results were evaluated in accordance with previously described methods (30). Briefly, total RNA was extracted with TRIzol reagent (Invitrogen, Carlsbad, CA, USA) according to the instructions supplied by the manufacturer. The integrity of RNA was assessed with Agilent 2100 Bioanalyzer and RNA 6000 LabChip kits (Yokokawa Analytical Systems, Tokyo, Japan). Only high-quality RNA was used for analysis. Seven RNA extractions from different normal liver tissue samples were mixed and used as the control reference. Next, 2 μg of total RNA was used to synthesize double-stranded cDNA that contained a promoter for T7 RNA polymerase. Amplified antisense RNA was synthesized by in vitro transcription of the cDNA templates using the Amino Allyl MessageAmp aRNA kit (Ambion, Austin, TX, USA). The reference and test samples were labeled with Cy3 and Cy5, mixed, and hybridized on a microarray covering 30,336 human probes (AceGene Human 30K; DNA Chip Research Inc. and Hitachi Software Engineering Company, Yokohama, Japan). The microarrays were scanned using ScanArray Lite, and signal values were calculated using DNASIS array software (Hitachi Software Engineering Company). The local background was subtracted from each spot, and the ratio of the intensity of fluorescence from the Cy5 channel to the intensity of fluorescence from the Cy3 channel was calculated for each spot. The ratio of expression levels of each gene was converted to a logarithmic scale (base 2), and the data matrix was normalized.

For clinicopathological analyses, study samples were divided into high- and low-expression groups based on the median EFNA1 mRNA expression levels in tumor tissue. All statistical analyses were carried out using the StatView J-5.0 program (Abacus Concepts, Inc., Berkeley, CA), USA. The post-operative period was measured from the date of surgery to the date of the last follow-up or death. Differences were estimated using Fisher’s exact probability test. Survival curves were calculated by the Kaplan-Meier method and compared statistically using the log-rank test. To estimate relative risk (RR) and 95% confidence intervals (95% CI), univariate and multivariate analyses were performed using the Cox proportional hazards regression model. Data are reported as mean ± standard deviation. Mean values were compared using the Mann-Whitney test. A probability value of <0.05 was deemed to be statistically significant.

First, we evaluated expression of EFNA1 under hypoxic conditions. EFNA1 was expressed in all four hepatoma cell lines and gradually increased under hypoxia in HuH7, HepG2 and Hep3B cell lines, but not in PLC/PRF/5 cells (Fig. 1). This result suggests that hypoxic conditions are associated with increased EFNA1 expression in HCC.

Next, we evaluated the expression of EFNA1 in clinical samples by using microarray analysis. The patients selected for microarray analysis included 113 (81.3%) men and 26 (18.7%) women. Twenty-six patients had hepatitis B virus infection, and 85 patients were positive for hepatitis C virus antibody. A total of 102 patients had a single tumor in the liver, and 65 patients had a tumor <3 cm in diameter. Macroscopic vascular invasion was seen in 15 patients. With regard to TNM staging, 96 patients (69.1%) were stage I, 31 patients (22.3%) were stage II, and 12 patients (8.6%) were stage III. The characteristics of the 139 patients are summarized in Table I.

We examined the correlation between expression levels of EFNA1 and EPHA2 and the clinicopathological factors of the 139 HCC patients who had undergone hepatic resection. The 139 patients were divided into two groups, a high-expression group (n=70) and a low-expression group (n=69), based on median expression levels from the microarray data for each gene in Table II. There was no correlation between EFNA1 expression and clinicopathological factors including tumor size, vascular invasion and number of tumors. EPHA2 expression was not significantly correlated with any clinicopathological factors, except for microscopic portal invasion. Tumors with high expression of EPHA2 had a tendency to have microscopic vascular invasion, although this result was not statistically significant (p=0.0786) (Tables I and II).

We next evaluated the correlation between EFNA1 and EphA2 expression levels using microarray data. We found that EFNA1 expression levels were significantly correlated with those of EPHA2 (Fig. 2).

We next examined the correlation between expression data from the microarray and quantitative RT-PCR (qRT-PCR) analysis of EFNA1 to validate the microarray data. qRT-PCR analysis was performed on 53 HCC tissue samples that were randomly selected from among the 139 HCC tissue specimens. Individual mRNA levels were normalized to GAPDH. In the 53 samples, qRT-PCR data for EFNA1 were significantly correlated with the results obtained from the microarray data (Fig. 3).

Kaplan-Meier survival curves demonstrated that patients with high EFNA1 expression had significantly shorter disease-free survival (DFS) than those with low EFNA1 expression based on both microarray (Fig. 4A) and qRT-PCR (Fig. 4B) data. EPHA2 expression was not correlated with the prognosis for HCC after curative resection (Fig. 4C). Univariate analysis for survival revealed that tumor number, microscopic vascular invasion, microscopic intrahepatic metastasis, and EFNA1 expression were significantly associated with DFS based on microarray data (Table III). Multivariate Cox regression analysis clarified that only EFNA1 expression remained an independent prognostic factor (Table IV).