Gene Expression-Based Recurrence Prediction of Hepatitis B Virus-Related Human Hepatocellular Carcinoma

GSK 후원 해외연수지원 기금 연구보고

Gene expression-based recurrence prediction of hepatitis B virus-related human hepatocellular carcinoma

Yoon Jun Kim, M.D.

Department of Internal Medicine and Liver Research Institute, Seoul National University College of Medicine, Seoul, Korea

ABSTRACT Hepatocellular carcinoma (HCC) is one of poor prognostic malignancies because of the high rate of recurrence even after curative resection of tumors. To predict and classify the molecular signatures associated with early recurrence, we profiled the gene expression of 65 HCC samples with hepatitis B infection using genome-scale oligonucleotide microarray. We identified 348 unique gene set well reflecting early recurrence (ER) of HCC, which revealed to be enriched by GTPase signaling related genes, transcription, immune response, cell adhesion and motility related genes. We also generated a signature responding to recurrence time by using Cox proportional hazard model (HR genes). Hierarchical clustering showed that HR genes are more accurate classifier than ER genes. In addition, we applied a meta-analysis to integrate earlier expression data (Iizuka et al, 2003), and obtained 232 genes consistently expressed in both the independent data. This signature was validated in an independent study indicating its robustness for the prediction of HCC recurrence. In conclusion, the gene signatures retrieved from different but complementary methods may provide clues to predict patients with increased risk of developing early recurrence, and to identify novel therapeutic targets for HCC.

Key Words: Hepatitis B Virus; Hepatocellular carcinoma; Recurrence; Microarray; Gene Expression Profile

Corresponding Author: Yoon Jun Kim, Department of Internal Medicine, Seoul National University Hospital, 28 Yongon-dong, Chongno-gu, Seoul 110-744, Korea. Phone: 82-2-2072-3081, 82-2-740-8112; Fax: 82-2-743-6701; E-mail: [email protected] * 본 원고는 대한간학회-GSK 해외연수지원 기금 연구보고를 목적으로 2007년 보스턴에서 열린 미국간학회와 Clinical Cancer Research 2008년 4월호에 발표된 내용과 같은 연구를 토대로 작성되었음

- S3 - 2008년 대한간학회 추계학술대회

Introduction

Hepatocellular carcinoma (HCC) is one of the major causes of death from malignancy throughout the world. Unfortunately, HCC has a poor prognosis of overall five-year survival rate and it is mainly ascribed by the high intrahepatic recurrence rate even after the curative resection. Therefore, it is important to understand the molecular basis of the early recurrence of HCC and to develop a prediction of likelihood of recurrence after treatment, which may be helpful to plan a therapeutic strategy and to improve the outcome survival of HCC patients. Recently, DNA microarray technology offers a genome-wide view of transcription profile that enables us to understand the systematic changes of gene expression patterns. However, although many studies using microarray has been challenged to predict the survival time and prognosis of cancers, few of them were focused on the recurrence rate of cancer progression. In present study, we used DNA microarrays to examine the gene expression profiles to uncover the biological mechanisms that affect recurrence rate of HCC, and predicted the recurrence time of HCCs after curative resection. Several gene signatures for HCC recurrence were identified by applying different analyzing methods. First, we obtained gene sets reflecting early recurrence of tumors by a traditional two-sample t-test method. Second, we applied a Cox regression analysis method to the recurrence time and obtained a classifier gene set according to the hazard rate of each gene. Finally, it is critical to validate the signature can predict the new case of independent dataset, since the comparison of different microarray datasets dealing with the prediction of patient’s outcome or identification of differentially expressed genes often shows poor congruence with independent datasets. To overcome the limited success of the cross-comparison of microarray data, we applied a meta-analysis by combining the previous data of Iizuka et al.1 that had addressed the same sample label (eg. early recurrence vs. non-early recurrence) of HCC. The meta-analysis was accomplished by effect size method previously introduced by Choi et al2, which is known to select the genes small but consistent expression profiles from independent datasets regardless of their differences of experimental or computational methods used. By comparing and characterizing these different but complementary signatures, we tried to figure out the biological mechanisms involved in the progression of recurrence, and to develop a prediction model for the recurrence rate of HCC.

Methods

Patients

All 65 HCC samples were obtained from the hepatitis B-positive Korean patients who underwent curative resection at the Seoul National University hospital (Seoul, Korea).

- S4 - 김윤준 Gene expression-based recurrence prediction of hepatitis B virus-related human hepatocellular carcinoma

Microarray Experiments and data analysis

Total RNA of HCC samples were isolated and hybridized to Affymetrix HGU133a2 chips according to the manufacturer’s instruction (Affymetrix, Santa Clara, CA). The raw data for 65 samples were normalized using Robust Multi-array Average (RMA) method3 available at Bioconductor (www.bioconductor.org). Unsupervised and supervised hierarchical clustering of expression profile were performed using Cluster and TreeView software.4 A multidimensional scaling (MDS) plot was also carried out using Bioconductor’s package. In addition, we performed a meta-analysis with Iizuka’s data which previously proposed early recurrence related genes of HCC.1 The dataset of ER (n=20) and non-ER samples (n=40) was downloaded from the author’s website. In order to avoid the effect of sample size, we randomly selected 15 ER samples and 25 non-ER samples. The meta-analysis was accomplished by effect size method previously introduced by Choi et al.2

Results

1. Unsupervised clustering

We attempted an unsupervised clustering to find similarities in gene expression patterns in 65 HCC samples. Genes were further filtered by the criteria of more than 20 % of un-logged expression values have at least 100, and the unsupervised clustering was carried out on the retained 4,775 genes. Ten out of 15 ER samples were co-clustered, and five samples were clustered to the other cluster (accuracy rate 75 %). These results suggest that ER samples had an overall similarity of expression patterns and were readily distinguished from the non-ER samples.

2. Supervised clustering of recurrence related expression profiles

Next, we selected classifier genes from expression profiles of 15 of ER and 25 of non-ER samples. Permutation P-values of significant genes were computed based on 10,000 random permutations, which yielded 348 unique ER genes (193 up-regulated and 155 down-regulated genes). Hierarchical clustering of the dataset showed that ER samples were considerably well classified into a same cluster (accuracy rate, 89.4%) (Figure 1A). Similar to this, multidimensional scaling (MDS) analysis also showed that the ER samples were well separated, although the distribution of non-ER samples were not separated from the ER samples in Euclidean space. In order to validate the class prediction, we applied cross-validation approach with different algorithms including CCP, LDA, 1-NN, 3-NN, NC, SVM, which showed 75 to 80% of prediction accuracy rate. The statistical significances of misclassification rate were determined by leave-one-out cross validation test (P=0.043, 0.04, 0.013, 0.019, 0.027, and 0.0012, respectively). To identify key functional elements that govern the gene expression, we applied gene identification methods on

- S5 - 2008년 대한간학회 추계학술대회

A B

Figure 1. Comparison of ER and HR signatures (A) Heatmap view of hierachical clustering result of 40 ER and non-ER samples. (B) Hierarchical clustering of all 65 samples with HR gene set. (C) Kaplan-Meier analysis of recurrence time for the subclasses classified by ER and HR genes, respectively. The color bar under the hierarchical tree showed the sample information. Red: ER samples (recurred within 1 year after curative resection). Dark blue: non-ER samples (not recurred within 1 year and followed up more than 1 year). Grey: unclassified samples with censored less than 1 year of follow up time. Dichotomic classification of samples was indicated with dark cyan and dark red color. the basis of functional enrichment analysis. Recently, several methods for analyzing the gene set enrichment in the context of ontology information has been proposed, and they were thought to provide more robust and interpretable information.5,6 Therefore, we focused on the genes within the significantly enriched function categories, rather than those genes identified by single-gene analysis. This approach may be more suitable and helpful to understand the

- S6 - 김윤준 Gene expression-based recurrence prediction of hepatitis B virus-related human hepatocellular carcinoma

pattern changes and complex systems that govern the underlying alteration of molecular pathogenesis. Of the 348 ER genes, functional enrichment analysis revealed that the genes related with transcription, immune or inflammatory response, GTPase activity, blood coagulation, and cell motility were significantly enriched (Table 1). Remarkably, GTPase related functions including small GTPase mediated signal transduction (P=0.004) and regulation of GTPase activity (P=0.007) were enriched. There are a number of families functioning GTPase, which have key regulating roles for Ras signaling. Ras related proteins are very important effecter molecules for a wide variety of signal pathways that regulate such processes as cytoskeletal integrity, proliferation, cell adhesion, apoptosis, and cell migration. Also, the overexpression of Rho GTPase is well established in numerous cancers including breast cancer,7,8 lung cancer,9 and gastric cancer,10,11 and was proposed to lead invasion and metastasis of tumors. In line with this, we identified the significant enrichment of Rho GTPase family members in the early-recurred HCCs including RGL1, RAB31, RAB20, RHOQ, RIT1, IQGAP1, RRAGC, and RAP2B. In HCC, small GTPase Rho controls cell adhesion and motility through reorganization of the actin cytoskeleton and regulation of actomyosin contractility in rat MM1 hepatoma cells.12 These results support our finding that the Ras

Table 1. Functional enrichment of ER genes Probe ID Gene ID Symbol Gene Name GO:0006355 (P=0.001) regulation of transcription, DNA-dependent 221530_s_at 79365 BHLHB3 basic helix-loop-helix domain containing, class B, 3 207233_s_at 4286 MITF microphthalmia-associated transcription factor 31845_at 2000 ELF4 E74-like factor 4 (ets domain transcription factor) 206074_s_at 3159 HMGA1 high mobility group AT-hook 1 200989_at 3091 HIF1A hypoxia-inducible factor 1, alpha subunit 203278_s_at 51317 PHF21A PHD finger protein 21A 211105_s_at 4772 NFATC1 nuclear factor of activated T-cells, calcineurin-dependent 1 203348_s_at 2119 ETV5 ets variant gene 5 (ets-related molecule) 209204_at 8543 LMO4 LIM domain only 4 GO:0006955 (P=0.002) immune response 202988_s_at 5996 RGS1 regulator of G-protein signalling 1 205419_at 1880 EBI2 Epstein-Barr virus induced gene 2 201487_at 1075 CTSC cathepsin C 214511_x_at 440607 LOC440607 NA 216950_s_at 2209 FCGR1A Fc fragment of IgG, high affinity Ia, receptor (CD64) 203561_at 2212 FCGR2A Fc fragment of IgG, low affinity IIa, receptor (CD32) 201422_at 10437 IFI30 interferon, gamma-inducible protein 30 204896_s_at 5734 PTGER4 prostaglandin E receptor 4 (subtype EP4) 209499_x_at 8741 TNFSF13 tumor necrosis factor (ligand) superfamily, member 13 lymphocyte cytosolic protein 2 (SH2 domain containing leukocyte 205269_at 3937 LCP2 protein of 76kDa)

- S7 - 2008년 대한간학회 추계학술대회

sema domain, immunoglobulin domain (Ig), transmembrane domain 203528_at 10507 SEMA4D (TM) and short cytoplasmic domain, (semaphorin) 4D GO:0007264 (P=0.003) small GTPase mediated signal transduction 209568_s_at 23179 RGL1 ral guanine nucleotide dissociation stimulator-like 1 217762_s_at 11031 RAB31 RAB31, member RAS oncogene family 219622_at 55647 RAB20 RAB20, member RAS oncogene family 212119_at 23433 RHOQ ras homolog gene family, member Q 209882_at 6016 RIT1 Ras-like without CAAX 1 200791_s_at 8826 IQGAP1 IQ motif containing GTPase activating protein 1 218088_s_at 64121 RRAGC Ras-related GTP binding C 213923_at 5912 RAP2B RAP2B, member of RAS oncogene family GO:0007242 (P=0.003) intracellular signaling cascade 219257_s_at 8877 SPHK1 sphingosine kinase 1 201294_s_at 26118 WSB1 WD repeat and SOCS box-containing 1 207629_s_at 9181 ARHGEF2 rho/rac guanine nucleotide exchange factor (GEF) 2 lymphocyte cytosolic protein 2 (SH2 domain containing leukocyte 205269_at 3937 LCP2 protein of 76kDa) 205147_x_at 4689 NCF4 neutrophil cytosolic factor 4, 40kDa 209321_s_at 109 ADCY3 adenylate cyclase 3 212873_at 23526 HA-1 NA 207540_s_at 6850 SYK spleen tyrosine kinase 203760_s_at 6503 SLA Src-like-adaptor 203741_s_at 113 ADCY7 adenylate cyclase 7 221581_s_at 7462 WBSCR5 Williams-Beuren syndrome chromosome region 5 GO:0009968 (P=0.004) negative regulation of signal transduction 202388_at 5997 RGS2 regulator of G-protein signalling 2, 24kDa 202988_s_at 5996 RGS1 regulator of G-protein signalling 1 204319_s_at 6001 RGS10 regulator of G-protein signalling 10 GO:0006350 (P=0.005) transcription 31845_at 2000 ELF4 E74-like factor 4 (ets domain transcription factor) 206074_s_at 3159 HMGA1 high mobility group AT-hook 1 209204_at 8543 LMO4 LIM domain only 4 214512_s_at 10923 PC4 NA 218088_s_at 64121 RRAGC Ras-related GTP binding C GO:0043087 (P=0.007) regulation of GTPase activity 215435_at 50807 DDEF1 development and differentiation enhancing factor 1 219358_s_at 55803 CENTA2 centaurin, alpha 2 204982_at 9815 GIT2 G protein-coupled receptor kinase interactor 2 GO:0007186 (P=0.014) G-protein coupled receptor protein signaling pathway 205419_at 1880 EBI2 Epstein-Barr virus induced gene 2

- S8 - 김윤준 Gene expression-based recurrence prediction of hepatitis B virus-related human hepatocellular carcinoma

209201_x_at 7852 CXCR4 chemokine (C-X-C motif) receptor 4 204896_s_at 5734 PTGER4 prostaglandin E receptor 4 (subtype EP4) 206171_at 140 ADORA3 adenosine A3 receptor GO:0005975 (P=0.017) carbohydrate metabolism 219508_at 9245 GCNT3 glucosaminyl (N-acetyl) transferase 3, mucin type 212510_at 23171 GPD1L glycerol-3-phosphate dehydrogenase 1-like 202275_at 2539 G6PD glucose-6-phosphate dehydrogenase 219634_at 50515 CHST11 carbohydrate (chondroitin 4) sulfotransferase 11 202497_x_at 6515 SLC2A3 solute carrier family 2 (facilitated glucose transporter), member 3 203217_s_at 8869 ST3GAL5 ST3 beta-galactoside alpha-2,3-sialyltransferase 5 201481_s_at 5834 PYGB phosphorylase, glycogen, brain GO:0006096 (P=0.018) glycolysis 201037_at 5214 PFKP phosphofructokinase, platelet 201251_at 5315 PKM2 pyruvate kinase, muscle 200966_x_at 226 ALDOA aldolase A, fructose-bisphosphate GO:0006916 (P=0.021) anti-apoptosis secreted phosphoprotein 1 (osteopontin, bone sialoprotein I, early 209875_s_at 6696 SPP1 T-lymphocyte activation 1) 208296_x_at 25816 TNFAIP8 tumor necrosis factor, alpha-induced protein 8 sema domain, immunoglobulin domain (Ig), transmembrane domain 203528_at 10507 SEMA4D (TM) and short cytoplasmic domain, (semaphorin) 4D 200787_s_at 8682 PEA15 phosphoprotein enriched in astrocytes 15 GO:0007229 (P=0.028) integrin-mediated signaling pathway integrin, alpha M (complement component receptor 3, alpha, also 205786_s_at 3684 ITGAM known as CD11b (p170), macrophage antigen alpha polypeptide) integrin, alpha V (vitronectin receptor, alpha polypeptide, antigen 202351_at 3685 ITGAV CD51) 207540_s_at 6850 SYK spleen tyrosine kinase GO:0015031 (P=0.036) protein transport 218094_s_at 55861 C20orf35 chromosome 20 open reading frame 35 202260_s_at 6812 STXBP1 syntaxin binding protein 1 219622_at 55647 RAB20 RAB20, member RAS oncogene family 212119_at 23433 RHOQ ras homolog gene family, member Q 74694_s_at 79874 RABEP2 rabaptin, RAB GTPase binding effector protein 2 213923_at 5912 RAP2B RAP2B, member of RAS oncogene family GO:0006928 (P=0.036) cell motility 209083_at 11151 CORO1A coronin, actin binding protein, 1A 210845_s_at 5329 PLAUR plasminogen activator, urokinase receptor 202910_s_at 976 CD97 CD97 antigen 201954_at 10095 ARPC1B actin related protein 2/3 complex, subunit 1B, 41kDa GO:0006461 (P=0.038) protein complex assembly

- S9 - 2008년 대한간학회 추계학술대회

solute carrier family 7 (cationic amino acid transporter, y+ 204588_s_at 9056 SLC7A7 system), member 7 206074_s_at 3159 HMGA1 high mobility group AT-hook 1 membrane protein, palmitoylated 6 (MAGUK p55 subfamily 205429_s_at 51678 MPP6 member 6) 207540_s_at 6850 SYK spleen tyrosine kinase GO:0007596 (P=0.040) blood coagulation 203305_at 2162 F13A1 coagulation factor XIII, A1 polypeptide 210845_s_at 5329 PLAUR plasminogen activator, urokinase receptor 205479_s_at 5328 PLAU plasminogen activator, urokinase GO:0006954 (P=0.045) inflammatory response 204445_s_at 240 ALOX5 arachidonate 5-lipoxygenase 204924_at 7097 TLR2 toll-like receptor 2 202910_s_at 976 CD97 CD97 antigen 204150_at 23166 STAB1 stabilin 1 206171_at 140 ADORA3 adenosine A3 receptor and its related signaling are related with the recurrence rate of HCC. G-protein related signaling such as G-protein coupled receptor protein signaling pathway (EBI2, CXCR4, PTGER4, and ADORA3) and G-protein signaling regulator genes (RGS1, RGS2, and RGS10) were also overexpressed in early recurrence HCC (P=0.015, P=0.005, respectively). Among the ER genes, adenosine A3 receptor (ADORA3) was proved to be overexpressed in colon and breast cancers,13 and was proposed that adenosine signals may lead to the increase in HIF1-mediated effects in cancer cells.14 In particular interest, HIF1 was identified as one of overexpressed ER genes, and previously suggested as a lead gene in the signature associated with the low survival of HCC.15 The hypoxic response is manifested in many pathophysiological processes such as tumor growth and metastasis. During hypoxia, it was known that many glycolytic genes were up-regulated by HIF1-dependent manner.16 This was also supported by our findings that glycolysis and carbohydrate metabolism were enriched (P= 0.018, P=0.017, respectively). Taken together, these results suggest that the overexpression of HIF1 and its related genes may be one of the leading features of molecular alteration in the early-recurred HCC. In the perspective of liver function, we found that blood coagulation associated genes were overexpressed, and of which PLAU and PLAUR have also known to play an important role in the aggressiveness of HCC.17,18 Addi- tionally, liver-specific cytochrome p450 genes including CYP4F3, CYP27A1, CYP2J2, CYP4F12 were down- regulated showing good correspondence with previous data.15 Not surprisingly, metastasis associated functions such as anti-apoptosis (P=0.022), integrin-mediated signaling pathway (P=0.029), and cell motility (P=0.037) were also enriched in ER genes. One of remarkable findings in ER genes, SSP1 (osteopontin) had the highest fold difference of geometric mean (6.5 fold), which was previously noticed as a therapeutic target for metastatic HCC.19,20 Besides, there is a line of evidences supporting the role of ER genes in relation to cancer metastasis or recurrence. Calcium-binding proteins such as S100A4, S100A6, and S100A11 were also significantly overexpressed in ER samples. Corro-

- S10 - 김윤준 Gene expression-based recurrence prediction of hepatitis B virus-related human hepatocellular carcinoma

Table 2. Clinicopathological features of HCC patients Non-recur Early Cluster Cluster P-value P-value rence recurrecne P-value A B P-value HR (Wald) (log-rank) (n=25) (n=15) (n=39) (n=26) Sex (woman/man) 4/21 3/12 1.000 6/33 7/19 0.345 0.948 0.910 0.960 Age (<55/>=55) 12/13 13/2 0.019 23/16 16/10 1.000 0.420 0.047 0.039 AFP (<300/>=300) 19/6 12/3 1.000 27/12 20/6 0.579 0.608 0.360 0.363 PLT (<10/>=10) 8/17 5/10 1.000 14/25 8/18 0.791 0.992 0.990 0.997 Tumor Size (<5/>=5) 14/11 6/9 0.514 25/14 16/10 1.000 1.975 0.130 0.116 Gross Type 14/11 4/11 0.104 21/18 9/17 0.204 1.823 0.130 0.135 (single/multiple) Histologic grade 20/5 12/3 1.000 32/7 18/8 0.247 0.941 0.910 0.906 (I/II,III) T stage 10/15 6/9 1.000 20/19 11/15 0.613 0.804 0.570 0.538 (T0,T1/T2,T3,T4) Venous invasion 16/9 6/9 0.194 24/15 12/14 0.309 1.349 0.440 0.443 (no/yes) Extranodal invasion 15/10 4/11 0.055 22/17 8/18 0.047 1.864 0.120 0.121 (no/yes) TACE (no/yes) 19/6 11/4 1.000 32/7 20/6 0.754 0.708 0.490 0.477 HBeAg (-/+/non-C) 4/12 4/8 1.000 17/8 20/6 0.331 0.973 0.960 0.950 anti-hbe (-/+/non-C) 4/12 4/8 0.406 7/16 6/16 1.000 0.912 0.850 0.856 24.76 19.10 8.55 Recurrence (month,SE) 5.2 (0.8) (2.39) (2.26) (1.62) P-value was obtained from Fisher-exact t-test, and the statistical significance (P<0.05) was indicated by bold type. borating evidences has been demonstrated that S100A6 (calcyclin) is significantly associated with poor survival in pancreatic cancer21 and colorectal carcinoma.22 Also, S100A4 has been proposed as a metastasis mediator23 and has shown to be a prognostic marker in a number of cancers including esophageal-squamous cancers,24 non small lung cancers,25 and primary gastric cancers.26 S100A11 was observed to be overexpressed in numerous cancers and associated with metastasis of gastric cancers.27,28 Nonetheless, the molecular basis of these calcium-binding proteins in HCC recurrence is not fully understood yet, which thought to be good candidates for future studies. Furthermore, solute carrier families were suggested to be overexpressed and may provide essential nutrients needed to support the rapid growing tumors.29 In relation to this finding, we found that solute carrier families including SLC7A7, SLC4A7, SLC39A6, SLC38A1, SLC2A3, SLC1A5, SLC16A7 were overexpressed in the early recurrence HCC. Meanwhile, we also found that SLC6A12, SLC38A3, SLC25A15, SLC25A13, SLC22A7, and SLC17A2 were significantly down-regulated. These results indicate that the dysregulation of solute carrier family genes might be associated with the recurrence rate of HCC, although all of them were not appeared to be overexpressed contradicting to the previous findings.

- S11 - 2008년 대한간학회 추계학술대회

Figure 2. Network construction with ER genes. The genetic network was constructed by Cyto- scape as described in METHODS. The nodes re- present genes identified by natural language processing on PubMed literature. Up-regulated (red) and down-regulated (light blue) ER genes were presented with different colors.

3. Classification of HCC by Cox proportional hazard model

In another approach, we attempted to identify responding genes for recurrence time using Cox proportional hazard model. To determine the number of significant genes, we simulated the P-value of Wald test and determined the cut-off value that maximizes the significance of log-rank test of hazard rates between the two subclasses generated by hierarchical clustering. The significance of log-rank test between subclasses was highest at the cut- point of P=0.0124 (Wald test), which yielded 517 genes (HR genes). Compared to ER signature, HR signature well classified early recurrence tumors more accurately (92.5% accuracy rate, Figure 1B). Kaplan-Meier plot also showed that HR signature is better than ER signature to predict the recurrence rate of HCC (P=5.5×10-8, 6.6×10-4, respectively) (Figure 1C). Of the HR genes, 223 genes were common to ER genes and 294 genes were newly identified. To compare the HR genes to the ER genes, we selected same number of top-ranked 517 ER genes (P<0.0022), and observed that the only 292 ER genes were presented in HR gene set indicating the inconsistency of HR genes with ER genes. The functional enrichment of HR genes or the common 292 genes in both ER and HR signatures were similar to that of ER signature (data not shown). Therefore, we analyzed the enrichment in the difference set that belongs to HR signature but not to ER signature (HR-ER). This analysis found additional characteristics of HR signature enriched by recurrence/metastasis-related functions such as cell differentiation (P=0.011) and cell adhesion (P=0.022). In addition, we re-identified a series of following genes that were previously addressed to be associated with HCC. For example, TSPAN3 was identified to increase in cancerous tissue showing intrahepatic spreading compared with tumors without such spreading.30,31 Tissue factor (F3) is a transmembrane glycoprotein involved in initiating blood coagulation, and also related to tumor angiogenesis and invasiveness of HCC.32,33 Several genes related with prostanoids (PTGS1, PTGDS, and PTGER4) were identified as overexpressed ER genes. In concert with this, PTGS2 was newly identified in HR genes, which was previously well studied to be associated with invasiveness or proliferation of HCC.34,35 Taken together, our assessments to

- S12 - 김윤준 Gene expression-based recurrence prediction of hepatitis B virus-related human hepatocellular carcinoma

Figure 3. Comparison with Lee’s data (A) IDD genes were applied to 66 samples of Lee’s data that available recurrence time. Kaplan- Meier analysis of recurrence time for the subclasses, which were classified by hierarchical clustering of expression profiles of IDD genes in Lee’s data. (log-rank test P-values was indicated) (B) Dendrogram of hierarchical clustering B

characterize the HR genes suggest that the HR genes are biologically relevant and congruent with the previous knowledge of molecular features of HCC.

4. Clinicopathological features of the data

Previously, pathologic features such as tumor size, venous invasion, presence of satellite nodules, and advanced TNM stage, are known as risk factors for recurrence and important aspects affecting the prognosis of patients with HCC.1 We compared the clinicopathological features of patients between ER and non-ER samples. Unexpectedly, we observed that only the patient age was significant (P=0.0004) indicating the younger patients are more sus- ceptible to early recurrence of HCC (Table 2). In addition, we compared the clinicopathological features between the category of subclass A and B that were classified by hierarchical clustering of HR genes. Univariate Cox regression analysis showed identical evidence that the higher rate of early recurrence in younger patients. Multi- variate Cox regression analysis also identified patient’s age as an independent risk factor for early recurrence (hazard rate=0.251, 95 % CI=0.082-0.765, P-value<0.015). Meanwhile, when the difference of clinical features between subclass A and B was tested, only the status of extra-nodal invasion was revealed to be significant with marginal P-value (P=0.048). Based on this finding, we assumed that there would be a dominant signature for extra nodal invasion. This might be expected to provide a clue to uncover the molecular mechanism of the HCC recurrence. Using a supervised comparison with P<0.01, we obtained a total of 264 significant genes. However, only 16 of the 264 genes were belong to the ER genes indicating that the genes related to extra nodal invasion

- S13 - 2008년 대한간학회 추계학술대회

Table 3. Functional enrichment of IDD genes Probe ID Gene ID Symbol Gene Name GO:0008284 (P=0.001) positive regulation of cell proliferation 211839_s_at 1435 CSF1 colony stimulating factor 1 (macrophage) signal sequence receptor, alpha (translocon-associated protein 200891_s_at 6745 SSR1 alpha) 210895_s_at 942 CD86 CD86 antigen (CD28 antigen ligand 2, B7-2 antigen) 203175_at 391 RHOG ras homolog gene family, member G (rho G) 213864_s_at 4673 NAP1L1 nucleosome assembly protein 1-like 1 GO:0009117 (P=0.001) nucleotide metabolism 201268_at 4831 NME2 non-metastatic cells 2, protein (NM23B) expressed in 201577_at 4830 NME1 non-metastatic cells 1, protein (NM23A) expressed in 216705_s_at 100 ADA adenosine deaminase GO:0000074 (P=0.004) regulation of progression through cell cycle 217620_s_at 5291 PIK3CB phosphoinositide-3-kinase, catalytic, beta polypeptide 203693_s_at 1871 E2F3 E2F transcription factor 3 protein phosphatase 3 (formerly 2B), catalytic subunit, beta 215586_at 5532 PPP3CB isoform (calcineurin A beta) 204682_at 4053 LTBP2 latent transforming growth factor beta binding protein 2 203175_at 391 RHOG ras homolog gene family, member G (rho G) 210021_s_at 10309 UNG2 uracil-DNA glycosylase 2 GO:0006457 (P=0.022) protein folding 208696_at 22948 CCT5 chaperonin containing TCP1, subunit 5 (epsilon) 200825_s_at 10525 HYOU1 hypoxia up-regulated 1 211538_s_at 3306 HSPA2 heat shock 70 kDa protein 2 201327_s_at 908 CCT6A chaperonin containing TCP1, subunit 6A (zeta 1) 210436_at 10694 CCT8 chaperonin containing TCP1, subunit 8 (theta) are distinct from the ER genes. These 264 genes were characterized by the enrichment of many transcription related genes (eg. transcription, RNA processing, and chromatin modification), whereas immune and apoptosis related genes were significantly down-regulated. Remarkably, the expression of MHC class I genes such as HLA-C, HLA-E, HLA-G, and HLA-J were down-regulated in the tumors with extra nodal invasion indicating the suppression of apoptosis and immune related signaling might have crucial roles in the extra nodal invasion of HCC. Although we were not able to evaluate the relationships between extra nodal invasion and the early recurrence of HCC, it might be possible to suppose that those dysregulation of RNA processing, chromatin modification, apoptosis or immune related genes may have functionally significant relationships, even though indirectly, with the molecular pathophysiology of the early recurrence of HCC.

- S14 - 김윤준 Gene expression-based recurrence prediction of hepatitis B virus-related human hepatocellular carcinoma

5. Construction of interaction network with ER genes

Previously, biological networks, such as protein-protein interaction and metabolic networks have node degree distribution that follows a power law distribution showing scale-free architectures. It has been suggested that the scale-free network is characterized by the existence of hubs with multiple interaction partners, which thought to have a functional importance in the maintaining cell functions.36,37 Dysregulation of the hub genes may lead to deleterious changes in cellular function, and more likely to responsible for pathogenesis of disease. For example, Jeong et al. have shown a simulation study that the deletion of hub proteins with more than 15 links in the protein-protein interaction network of yeast proves to more likely to be lethal.37 Considering the above observations, we intended to search hub genes by constructing an interaction network of ER gene set. The interaction network of ER genes and their related genes were constructed by using Agilent literature search software implemented in Cytoscape (http://www.cytoscape.org),38 which provide interaction information retrieved from PubMed literature database and protein-protein interactions. As shown in Figure 2, 1,323 genes were linked with 4,946 interactions. From the network, we found 138 genes with greater interaction partners (≥10 links) which thought to be hub genes. These genes can be classified into 16 functional groups including SRC, Rho GTPase, TNF, CHN, NFKB, CDKN, BCL2, PLAU, SRGAP, RGS, IL, chemokine, RAB, IFN, MAPK, and DLG related genes and 56 solitary non-clustered genes. Those hub genes, as expected, were previously studied extensively supporting the association with advanced cancer or carcinogenesis.39-44 Of the ER genes, 14 genes were highlighted as hub genes in ER network (CHN2, PLAU, SYK, RGS1, MGMT, ANG, PLAUR, PMAIP1, DDIT4, EBI2, MITF, TRIP10, RGS2, and SPHK1). Although our interpretation to highlighting hub genes is somewhat speculative, we suggest that the hub genes provided here are eligible for the putative diagnostic or therapeutic candidates of HCC.

6. Meta-analysis by combining independent microarray data

To validate our classifiers, we attempted to compare our data to earlier data. Previously, Iizuka et al. have found 12 recurrence genes of HCC from microarray expression profiling.1 Iizuka’s dataset was underwent with the same method which was previously applied to our dataset. We inspected the expression pattern of HR and ER genes in the Iizuka’s dataset. 112 out of 348 ER genes and 144 out of 517 HR genes were overlapped with Iizuka’s dataset, respectively, and which were well clustered together when the hierarchical clustering were carried out. These results imply that the overall expression pattern of the ER or HR genes in Iizuka’s dataset are similar to that of our dataset. Nonetheless, both ER and HR signatures were not able to stratify the recurrence time of Iizuka’s data. Previously, Lee et al.15 had shown the similar results in comparing independent datasets, and suggested that the inconsistency of datasets may be due to the different conditions of experimental or computational methods used in each dataset. At this point, in order to overcome the heterogeneity between datasets, we applied a meta-analysis using effect size model as described in METHODS, which able to generate a common signature expressed small but consistent in multiple datasets with statistical significance. As a result, we identified a total of 232 genes with the threshold of z score, |zth|>2.33 (P<0.01), integration-driven discovery genes (IDD genes). The fraction of newly identified genes from meta-analysis were estimated by calculating integration-driven

- S15 - 2008년 대한간학회 추계학술대회

discovery rate (IDR), which yield the IDR of 46.1% for up- regulated genes and 42.5% for down-regulated genes. Those IDD genes were characterized by functional enrichment of cell proliferation (CSF1, SSR1, CD86, RHOG, and NAP1L1), nucleotide metabolism (NME2, NME1, and ADA), regulation of progression through cell cycle (PIK3CB, E2F3, PPP3CB, LTBP2, RHOG, and UNG2), and protein folding (CCT5, HYOU1, HSPA2, CCT6A, CCT8) (P<0.05 for each category) (Table 3). Assuming that the dataset-driven noises can be eliminated by meta-analysis, we tested the reproducibility of the IDD signatures independently of the dataset. We applied IDD signatures to the independent gene expression profile of Lee’s data.15 192 of 232 IDD genes were annotated to Lee’s dataset, and their expression profiles of 66 patients which available recurrence time information were collected and clustered into two groups by hierarchical clustering method. As expected, IDD signature successfully stratified the recurrence time into two groups (Figure 3B). In spite of the difference that Lee’s dataset was came from the platform of oligonucleotide microarray, our inter- studies validation reveals that the IDD gene signature is robust and independent of experimental conditions, indicating the potential clinical utility as a predictor of HCC progression.

Discussion

This work established the early recurrence associated genes and their functional changes by using three different methods of supervised classification, Cox-regression model, and meta-analysis. At first, we attempted to profile the gene expression by supervised learning method. Previously, Iizuka et al. have noted that the ER vs. non-ER classification could not present accurate prediction of early recurrence of HCC in a long-term follow-up study.45 As mentioned earlier, the classification of early recurrence by the recurrence time is much complicated because there are many confounding factors affecting recurrence time besides the genetic changes in HCC. However, as shown in our results, the sample labeling of ER vs. non-ER is readily classifiable in unsupervised clustering, and the enriched functions in ER gene set were biologically relevant to carcinogenesis or cancer metastasis. In addition, many of the ER genes could be further validated by previous literatures. For example, it is noteworthy that one of the distinguishing features is the overexpression of SPP1 (Osteopontin) in concordance with the previous independent studies. Moreover, Ras-related signaling was emphasized to have a crucial role in HCC recurrence. Collectively, our study overall reveals that the classification of ER vs. non-ER is biologically meaningful and interpretable. We think that the ER signature in our study is more powerful predictor than that of Iizuka’s dataset, since our data samples were collected from more homogenous patient population of same viral background, ethnicity, and hospital treatment, which may allow the dataset to be clearer and more interpretable. Secondly, the HR signature was superior to the ER signature in predicting the recurrence rate of HCC. Moreover, additional biological insight into the underlying biological pathways can be further identified from HR signature. However, the HR gene signature was generated from the dataset including the unclassified samples and has relatively weaker strength of statistical significance than ER signature. Therefore, they may not be good

- S16 - 김윤준 Gene expression-based recurrence prediction of hepatitis B virus-related human hepatocellular carcinoma

candidates for target selection to get biological insights for the mechanism of recurrence progression. These results suggest that both the ER and HR signatures are useful in a complementary purpose with adequate understanding of their characteristics. In meta-analysis, we identified the IDD signature that showed a consistent expression in both independent datasets. Gene expression profile studies must be cross-compared by independent multiple studies, since there are a lot of confounding factors and systematic noises attributable to the signature selection. To remove the dataset– driven biases, we applied a meta-analysis by integrating the dataset with the expression profile of Iizuka. From this, we observed that the independent datasets are quite different each other. The primary reason for the discrepancy may be due to the patient’s different background and the difference of the experimental conditions. Indeed, the data samples were collected from heterogeneous populations because the Iizuka’s data included the samples with different viral infection (eg. HBV and HCV). Ethnic difference may also contribute to the difference between the datasets. Moreover, Iizuka’s data were normalized by MAS4 (Microarray Analysis Suite version 4, Affymetrix) while our dataset by RMA. In addition, different platform types were used, although both studies were performed in Affymetrix chips. It is clear that different platforms or normalization methods may lead to different results. Therefore, data integration from independent studies has important benefit potential to identifying signatures without dataset–driven biases. Here, we suggest that meta-analysis can yield a robust prognostic signature with the assurance that they are less prone to false findings independently of experimental conditions of individual studies. Then, we evaluated the IDD signature whether it can predict the status of new cases and showed its prediction power for recurrence rate of HCC. Although the IDD signature was derived from the differential expression of the clinical classification of ER vs. non-ER, it was able to predict the recurrence rate of the new case of independent study. A key point of our analysis is the capacity to predict recurrence rate indicating its usefulness in a clinical setting. The ability to accurately predict the recurrence rate on the basis of expression profile of HCC may help to plan a therapeutic strategy after tumor resection. Of course, much more data are needed to evaluate the benefit of our approach. In summary, we figured out the early recurrence related gene expression profile of HCCs by different gene profiling methods, and suggest that the gene signatures retrieved from different but complementary methods may provide clues predicting patient’s outcome and identifying novel diagnostic or therapeutic targets for HCC progression.

요 약 간세포암은 근치적 절제술 후 높은 재발률로 인하여 매우 나쁜 예후를 가지는 암종의 하나이다. 따라서 수술 후 간암의 재발과 관련된 예측 모델을 개발하고 기존의 수술로 도움을 받을 환자군과 불량한 예후가 예상되므 로 새로운 치료법으로 치료하여야 할 환자군을 분류하는 것은 임상적으로 매우 중요하다. 이와 같은 간암의 조기 재발과 관련된 분자생물학적 지표를 발굴하고 이에 의한 간암의 분류를 위하여 만성 B형간염에 감염된 65명의 간암 조직을 이용하여 전체 유전체 규모의 마이크로어레이 칩을 이용하여 유전자 발현 양상을 본 연구 에서는 시행하였다. 연구 결과, 간암을 조기 재발의 고위험군과 저위험군으로 분류할 수 있었다(P=1.9×10-6, log rank test). 이 결과의 일관성과 유효성을 완전히 다른 독립적인 간암군의 데이터에서 재확인하여 검증하였다. CD24는 조기 재발을 예측할 수 있는 바이오마커로서의 가능성을 보였으며 유전자 네트워크분석에서는 SP1과

- S17 - 2008년 대한간학회 추계학술대회

peroxisome proliferate-activated receptor α가 간암의 조기 재발을 조절하는 유전자일 가능성을 시사하였다. 결론 적으로 저자들은 본 연구를 통하여 간암의 조기 재발을 효과적으로 예측할 수 있는 유전자 발현 양상을 확인하 였고 이는 다른 간암 코호트와 다른 마이크로어레이 기반에서도 확인되었다. 이와 같은 연구 결과를 통하여 간암의 재발에 관한 분자생물학적 기반에 관한 추가적인 유용한 정보를 주었다고 하겠다.

색인 단어: 만성 B형간염, 간세포암, 재발, 마이크로어레이, 유전자 발현 양상

참고문헌

1.Iizuka N, Oka M, Yamada-Okabe H, Nishida M, Maeda Y, Mori N, Takao T, et al. Oligonucleotide microarray for prediction of early intrahepatic recurrence of hepatocellular carcinoma after curative resection. Lancet 2003;361:923-929. 2.Choi JK, Yu U, Kim S, Yoo OJ. Combining multiple microarray studies and modeling interstudy variation. Bioinformatics 2003;19:84i-90. 3. Irizarry RA, Hobbs B, Collin F, Beazer-Barclay YD, Antonellis KJ, Scherf U, Speed TP. Exploration, normalization, and summaries of high density oligonucleotide array probe level data. Biostat 2003;4:249-264. 4.Eisen MB, Spellman PT, Brown PO, Botstein D. Cluster analysis and display of genome-wide expression patterns. Proc Natl Acad Sci U S A 1998;95:14863-14868. 5.Subramanian A, Tamayo P, Mootha VK, Mukherjee S, Ebert BL, Gillette MA, Paulovich A, et al. Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles. Proc Natl Acad Sci U S A 2005;102:15545-15550. 6. Doniger SW, Salomonis N, Dahlquist KD, Vranizan K, Lawlor SC, Conklin BR. MAPPFinder: using Gene Ontology and GenMAPP to create a global gene-expression profile from microarray data. Genome Biol 2003;4: R7. 7. Baugher PJ, Krishnamoorthy L, Price JE, Dharmawardhane SF. Rac1 and Rac3 isoform activation is involved in the invasive and metastatic phenotype of human breast cancer cells. Breast Cancer Res 2005;7:R965-R974. 8.Mira JP, Benard V, Groffen J, Sanders LC, Knaus UG. Endogenous, hyperactive Rac3 controls proliferation of breast cancer cells by a p21-activated kinase-dependent pathway. Proc Natl Acad Sci U S A 2000;97:185-189. 9. Shikada Y, Yoshino I, Okamoto T, Fukuyama S, Kameyama T, Maehara Y. Higher expression of RhoC is related to invasiveness in non-small cell lung carcinoma. Clin Cancer Res 2003;9:5282-5286. 10. Kondo T, Sentani K, Oue N, Yoshida K, Nakayama H, Yasui W. Expression of RHOC is associated with metastasis of gastric carcinomas. Pathobiology 2004;71:19-25. 11. Pan Y, Bi F, Liu N, Xue Y, Yao X, Zheng Y, Fan D. Expression of seven main Rho family members in gastric carcinoma. Biochem Biophys Res Commun 2004;315:686-691. 12. Itoh K, Yoshioka K, Akedo H, Uehata M, Ishizaki T, Narumiya S. An essential part for Rho-associated kinase in the transcellular invasion of tumor cells. Nat Med 1999;5:221-225. 13.Madi L, Ochaion A, Rath-Wolfson L, Bar-Yehuda S, Erlanger A, Ohana G, Harish A, et al. The A3 adenosine receptor is highly expressed in tumor versus normal cells: potential target for tumor growth inhibition. Clin Cancer Res 2004;10:4472-4479. 14. Merighi S, Benini A, Mirandola P, Gessi S, Varani K, Leung E, MacLennan S, et al. A3 adenosine receptors modulate hypoxia-inducible factor-1alpha expression in human A375 melanoma cells. Neoplasia 2005;7:894-903. 15.Lee JS, Chu IS, Heo J, Calvisi DF, Sun Z, Roskams T, Durnez A, et al. Classification and prediction of survival in hepatocellular carcinoma by gene expression profiling. Hepatology 2004;40:667-676. 16.Minchenko A, Leshchinsky I, Opentanova I, Sang N, Srinivas V, Armstead V, Caro J. Hypoxia-inducible Factor-1-mediated Expression of the 6-Phosphofructo-2-kinase/fructose-2,6-bisphosphatase-3 (PFKFB3) Gene. ITS POSSIBLE ROLE IN THE WARBURG EFFECT. J. Biol. Chem. 2002;277:6183-6187. 17.Zhou L, Hayashi Y, Itoh T, Wang W, Rui J, Itoh H. Expression of urokinase-type plasminogen activator, urokinase-type plasminogen activator receptor, and plasminogen activator inhibitor-1 and -2 in hepatocellular

- S18 - 김윤준 Gene expression-based recurrence prediction of hepatitis B virus-related human hepatocellular carcinoma

carcinoma. Pathol Int 2000;50:392-397. 18.Tang ZY, Ye SL, Liu YK, Qin LX, Sun HC, Ye QH, Wang L, et al. A decade's studies on metastasis of hepatocellular carcinoma. J Cancer Res Clin Oncol 2004;130:187-196. 19.Ye QH, Qin LX, Forgues M, He P, Kim JW, Peng AC, Simon R, et al. Predicting hepatitis B virus-positive metastatic hepatocellular carcinomas using gene expression profiling and supervised machine learning. Nat Med 2003;9:416-423. 20. Hung-Wei Pan Y-HOS-YPS-HLP-LLP-HLJ-CSC-LCH-CH. Overexpression of osteopontin is associated with intrahepatic metastasis, early recurrence, and poorer prognosis of surgically resected hepatocellular carcinoma. Cancer 2003;98:119-127. 21. Vimalachandran D, Greenhalf W, Thompson C, Luttges J, Prime W, Campbell F, Dodson A, et al. High Nuclear S100A6 (Calcyclin) Is Significantly Associated with Poor Survival in Pancreatic Cancer Patients. Cancer Res 2005;65:3218-3225. 22.Komatsu K, Andoh A, Ishiguro S, Suzuki N, Hunai H, Kobune-Fujiwara Y, Kameyama M, et al. Increased Expression of S100A6 (Calcyclin), a Calcium-binding Protein of the S100 Family, in Human Colorectal Adeno- carcinomas. Clin Cancer Res 2000;6:172-177. 23.Garrett SC, Varney KM, Weber DJ, Bresnick AR. S100A4: A mediator of metastasis. J. Biol. Chem. 2005: R500017200. 24.Ninomiya I, Ohta T, Fushida S, Endo Y, Hashimoto T, Yagi M, Fujimura T, et al. Increased expression of S100A4 and its prognostic significance in esophageal squamous cell carcinoma. Int J Oncol 2001;18:715-720. 25.Kimura K, Endo Y, Yonemura Y, Heizmann CW, Schafer BW, Watanabe Y, Sasaki T. Clinical significance of S100A4 and E-cadherin-related adhesion molecules in non-small cell lung cancer. Int J Oncol 2000;16:1125- 1131. 26. Yonemura Y, Endou Y, Kimura K, Fushida S, Bandou E, Taniguchi K, Kinoshita K, et al. Inverse expression of S100A4 and E-cadherin is associated with metastatic potential in gastric cancer. Clin Cancer Res 2000;6: 4234-4242. 27. El-Rifai We, Moskaluk CA, Abdrabbo MK, Harper J, Yoshida C, Riggins GJ, Frierson HF, Jr., et al. Gastric Cancers Overexpress S100A Calcium-binding Proteins. Cancer Res 2002;62:6823-6826. 28.Cross SS, Hamdy FC, Deloulme JC, Rehman I. Expression of S100 proteins in normal human tissues and common cancers using tissue microarrays: S100A6, S100A8, S100A9 and S100A11 are all overexpressed in common cancers. Histopathology 2005;46:256-269. 29.Patil MA, Chua MS, Pan KH, Lin R, Lih CJ, Cheung ST, Ho C, et al. An integrated data analysis approach to characterize genes highly expressed in hepatocellular carcinoma. Oncogene 2005;24:3737-3747. 30. Kanetaka K, Sakamoto M, Yamamoto Y, Yamasaki S, Lanza F, Kanematsu T, Hirohashi S. Overexpression of tetraspanin CO-029 in hepatocellular carcinoma. J Hepatol 2001;35:637-642. 31. Kanetaka K, Sakamoto M, Yamamoto Y, Takamura M, Kanematsu T, Hirohashi S. Possible involvement of tetraspanin CO-029 in hematogenous intrahepatic metastasis of liver cancer cells. J Gastroenterol Hepatol 2003;18:1309-1314. 32.Poon RT-P, Lau CP-Y, Ho JW-Y, Yu W-C, Fan S-T, Wong J. Tissue Factor Expression Correlates with Tumor Angiogenesis and Invasiveness in Human Hepatocellular Carcinoma. Clin Cancer Res 2003;9:5339-5345. 33.Kaido T, Oe H, Yoshikawa A, Mori A, Arii S, Imamura M. Tissue factor is a useful prognostic factor of recurrence in hepatocellular carcinoma in 5-year survivors. Hepatogastroenterology 2005;52:1383-1387. 34.Tang TC, Poon RT, Lau CP, Xie D, Fan ST. Tumor cyclooxygenase-2 levels correlate with tumor invasiveness in human hepatocellular carcinoma. World J Gastroenterol 2005;11:1896-1902. 35.Leng J, Han C, Demetris AJ, Michalopoulos GK, Wu T. Cyclooxygenase-2 promotes hepatocellular carcinoma cell growth through Akt activation: evidence for Akt inhibition in celecoxib-induced apoptosis. Hepatology 2003; 38:756-768. 36. Wuchty S, Oltvai ZN, Barabasi AL. Evolutionary conservation of motif constituents in the yeast protein interaction network. Nat Genet 2003;35:176-179. 37. Jeong H, Mason SP, Barabasi AL, Oltvai ZN. Lethality and centrality in protein networks. Nature 2001;411: 41-42. 38. Shannon P, Markiel A, Ozier O, Baliga NS, Wang JT, Ramage D, Amin N, et al. Cytoscape: a software envi- ronment for integrated models of biomolecular interaction networks. Genome Res 2003;13:2498-2504. 39. Park SS, Eom YW, Kim EH, Lee JH, Min do S, Kim S, Kim SJ, et al. Involvement of c-Src kinase in the

- S19 - 2008년 대한간학회 추계학술대회

regulation of TGF-beta1-induced apoptosis. Oncogene 2004;23:6272-6281. 40.Nagao M, Nakajima Y, Hisanaga M, Kayagaki N, Kanehiro H, Aomatsu Y, Ko S, et al. The alteration of Fas receptor and ligand system in hepatocellular carcinomas: how do hepatoma cells escape from the host immune surveillance in vivo? Hepatology 1999;30:413-421. 41.Pikarsky E, Porat RM, Stein I, Abramovitch R, Amit S, Kasem S, Gutkovich-Pyest E, et al. NF-kappaB functions as a tumour promoter in inflammation-associated cancer. Nature 2004;431:461-466. 42.Wong C-M, Yam JW-P, Ching Y-P, Yau T-O, Leung TH-Y, Jin D-Y, Ng IO-L. Rho GTPase-Activating Protein Deleted in Liver Cancer Suppresses Cell Proliferation and Invasion in Hepatocellular Carcinoma. Cancer Res 2005;65:8861-8868. 43.Yang C, Liu Y, Leskow FC, Weaver VM, Kazanietz MG. Rac-GAP-dependent inhibition of breast cancer cell proliferation by {beta}2-chimerin. J Biol Chem 2005;280:24363-24370. 44.Guo XZ, Shao XD, Liu MP, Xu JH, Ren LN, Zhao JJ, Li HY, et al. Effect of bax, bcl-2 and bcl-xL on regulating apoptosis in tissues of normal liver and hepatocellular carcinoma. World J Gastroenterol 2002;8: 1059-1062. 45. Iizuka N, Hamamoto Y, Oka M. Predicting individual outcomes in hepatocellular carcinoma. Lancet 2004;364: 1837-1839.

- S20 -