[CANCER RESEARCH 62, 3939–3944, July 15, 2002] Advances in Brief Comparison of Gene Expression Profiles between Hepatitis B Virus- and Hepatitis C Virus-infected Hepatocellular Carcinoma by Oligonucleotide Microarray Data on the Basis of a Supervised Learning Method1 Norio Iizuka, Masaaki Oka,2 Hisafumi Yamada-Okabe, Naohide Mori, Takao Tamesa, Toshimasa Okada, Norikazu Takemoto, Akira Tangoku, Kenji Hamada, Hironobu Nakayama, Takanobu Miyamoto, Shunji Uchimura, and Yoshihiko Hamamoto Departments of Surgery II [N. I., M. O., N. M., T. T., T. O., N. T., A. T.] and Bioregulatory Function [N. I.], Yamaguchi University School of Medicine, Yamaguchi 755-8505; Department of Computer Science and Systems Engineering, Faculty of Engineering, Yamaguchi University, Yamaguchi 755-8611 [T. M., S. U., Y. H.]; and Department of Oncology, Nippon Roche Research Center, Kanagawa 247-8530 [H. Y-O., K. H., H. N.], Japan Abstract nisms responsible for the pathogenesis of HCC differ between HBV and HCV infections. Several studies compared gene expression be- Gene expression profiles of hepatocellular carcinomas (HCCs) associ- tween nontumorous liver and HCC and revealed gene expression ated with hepatitis B virus (HBV) and hepatitis C virus (HCV) were patterns that are rather specific to HCC (10–14). However, there is analyzed and compared. Oligonucleotide microarrays containing >6000 genes and subsequent gene selection by a supervised learning method only one study that compared gene expression patterns between HCC yielded 83 genes for which expression differed between the two types of with HBV infection (B-type HCC) and HCC with HCV infection HCCs. Expression levels of 31 of these 83 genes were increased in HBV- (C-type HCC; 14), and only a limited number of specimens were associated HCCs, and expression levels of the remaining 52 genes were analyzed. Therefore, additional studies are needed to understand mo- increased in HCV-associated HCCs. The 31 genes up-regulated in HBV- lecular mechanisms involved in the development and progression of associated HCC included imprinted genes (H19 and IGF2) and genes virus-induced HCCs. In this study, we investigated gene expression relating to signal transduction, transcription, and metastasis. The 52 genes patterns of 45 HCC samples using high-density oligonucleotide mi- up-regulated in HCV-associated HCC included a number of genes respon- croarrays and the supervised learning method to gain additional in- sible for detoxification and immune response. These results suggest that sight into hepatocarcinogenesis or cancer progression related to HBV HBV and HCV cause hepatocarcinogenesis by different mechanisms and provide novel tools for diagnosis and treatment of HBV- and HCV- or HCV infection. The results of this study provide additional markers associated HCCs. and molecular targets for the diagnosis and treatment of B- and C-type HCCs. Introduction Materials and Methods HCC3 is one of the most common fatal cancers worldwide (1). The most clearly established risk factor for HCC is chronic infection with Tumor Samples. Surgical specimens were obtained from 45 patients who HBV or HCV (2). More than 350 million people worldwide are underwent surgical treatment for HCC at Yamaguchi University Hospital known to be chronic carriers of HBV (3). It is reported that the between May 1997 and August 2000. Written informed consent was obtained incidence of HCC is increasing in many countries in parallel to an from all patients before surgery. The study protocol was approved by the increase in chronic HCV infection (1, 2). Therefore, clarification of Institutional Review Board for Human Use at the Yamaguchi University the genetic portraits of hepatocarcinogenesis caused by HBV or HCV School of Medicine. Histopathological diagnosis of HCC was made after infection might provide clues toward effecting a decrease in the surgery in each case. The clinicopathological characteristics of the 45 patients incidence of HCC and establishing effective treatments for each type based on the International Union against Cancer TNM classification (15) are shown in Table 1. Of the 45 patients, 14 were positive for serum HBs Ag and of HCC. 31 were positive for HCV Ab; none were positive for both HBs Ag and HCV Recent development of DNA microarray technology, a type of Ab. Thus, the patients were classified into two groups, those positive for HBs high-throughput analysis for gene expression, has opened a new era in Ag (B-type HCC, n ϭ 14) and those positive for HCV Ab (C-type HCC, medical sciences (4–6). Supervised learning and unsupervised learn- n ϭ 31). ing methods have been introduced into gene expression analysis of Control Liver Samples. Six nontumorous liver samples were obtained DNA microarray data (7, 8). Using hierarchical cluster analysis, an from patients who underwent hepatic resection for benign liver tumor or unsupervised learning method, Honda et al. (9) showed different gene metastatic liver tumor, which derive from gastrointestinal cancer. Liver func- expression profiles in the hepatic lesions of chronic hepatitis associ- tion for these 6 patients was shown to be normal, and the liver was shown to ated with HBV and HCV and suggested that the molecular mecha- histopathologically be normal. All 6 patients were seronegative for both HBs Ag and HCV Ab. Samples and RNA Extraction. Total RNA was extracted with Sepasol- Received 3/18/02; accepted 5/23/02. The costs of publication of this article were defrayed in part by the payment of page RNAI (Nacalai Tesque, Tokyo, Japan) and purified with the RNeasy Mini kit charges. This article must therefore be hereby marked advertisement in accordance with (Qiagen, Tokyo, Japan) according to the manufacturer’s instructions. Quality 18 U.S.C. Section 1734 solely to indicate this fact. of the total RNA was judged from the ratio between 28S and 18S RNase after 1 Supported in part by a Grant-in-Aid from the Ministry of Education, Science, Sports agarose gel electrophoresis. and Culture of Japan (12671230). 2 To whom requests for reprints should be addressed, at Department of Surgery II, cDNA Synthesis and in Vitro Translation for Labeled cRNA Probe. Yamaguchi University School of Medicine, 1-1-1 Minami-Kogushi, Ube, Yamaguchi cDNA was synthesized with the reverse SuperScript Choice System (Invitro- 755-8505, Japan. Phone: 81-836-22-2262; Fax: 81-836-22-2262; E-mail: 2geka-1@po. gen Life Technologies, Carlsbad, CA) according to the manufacturer’s instruc- cc.yamaguchi-u.ac.jp. tions. cRNA was synthesized from the cDNA template by use of the MEGAs- 3 The abbreviations used are: HCC, hepatocellular carcinoma; HB, hepatitis B; HBV, hepatitis B virus; HCV, hepatitis C virus; Ag, antigen; Ab, antibody; GST, glutathione cript T7 kit (Ambion, Austin, TX) according to the manufacturer’s S-transferase; MAPK, mitogen-activated protein kinase. instructions. Mononucleotides and short oligonucleotides were removed by 3939 Downloaded from cancerres.aacrjournals.org on September 26, 2021. © 2002 American Association for Cancer Research. GENE EXPRESSION PATTERNS LINKED TO VIRUS TYPE IN HCC Table 1 Patient characteristics per study group stock of 9 B-type and 12 C-type HCC samples that were subjected to microar- BT group (n ϭ 14)a CT group (n ϭ 31) P ray study. Reverse transcriptase step was performed as described previously (17). Five ␮l of cDNA solution (equivalent to the cDNA from 100 ng of initial Sex N.S. ␮ Male/Female 8/6 22/9 RNA) were amplified in 45 l of PCR mixture (17) containing 25 pmol of each Age (yrs) 51.5 Ϯ 3.0 66.0 Ϯ 1.1 0.0001b primer for IGF-2 and ␤-actin genes. PCR was performed for 26 cycles for Primary lesion N.S. IGF-2 and 24 cycles for ␤-actin. Each cycle consisted of denaturation at 94°C Single tumor 6 12 for 1 min, annealing at 60°C for 45 s, and elongation at 72°C for 2 min. The Multiple tumors 8 19 Tumor size (cm) 5.9 Ϯ 1.4 5.5 Ϯ 0.8 N.S. primers used in this study were as follows: IGF-2,5Ј-ctggtggacaccctccagttc-3Ј Stagec N.S. (sense) and 5Ј-gcccacggggtatctggggaa-3Ј (antisense); and ␤-actin,5Ј-CCA- I/II 5 12 GAGCAAGAGAGGTAT-3Ј (sense) and 5Ј-CTGTGGTGGTGAAGCTG- IIIA/IVA/IVB 9 19 Ј Histological gradec N.S. TAG-3 (antisense). The expected sizes were 235 and 436 bp for IGF-2 and G1 0 2 ␤-actin genes, respectively. PCR products were separated by electrophoresis G2 9 25 on 1.5% agarose gels and visualized under UV light after ethidium bromide G3 5 4 c staining. We determined the mean band densities using NIH Image 1.62 Venous invasion N.S. ␤ (Ϫ)722software, and we calculated levels of IGF-2 relative to -actin gene. (ϩ)79Statistical Analysis. Clinicopathological factors pertaining to B- and C- Nontumorous liver N.S. type HCCs were compared, and differences were analyzed by ␹2 test, Fisher’s Nonspecific change 2 1 exact test, Student’s t test, or Mann-Whitney’s U test (Table 1). P Ͻ 0.05 was Chronic hepatitis 4 13 Liver cirrhosis 8 17 accepted for statistical significance. Pearson’s correlation coefficient (r) was calculated to examine the relation between microarray and reverse tran- a BT group, patients with HBV-associated HCC; CT group, patients with HCV- 2 Ͼ Ͻ associated HCC; NS, not significant. scriptase PCR results. r 0.16 and P 0.05 were considered significant. b P by Mann-Whitney U test. Calculations were done with Statview 5.0 (Abacus Concepts, Berkeley, CA) on c Assessment based on TNM classification by the International Union against cancer. a Macintosh computer (Apple Computers, Inc., Cupertino, CA). Results and Discussion column chromatography on a CHROMA SPIN ϩSTE-100 column (Clontech, Palo Alto, CA). Clinicopathological Characteristics Pertaining to B- and C-type Gene Expression Analysis by Means of High-density Oligonucleotide HCCs.