expression of glypican 3 enriches hepatocellular carcinoma

Received: October 8, 2014; Revised: November 6, 2014; Accepted: December 13, 2014

© The Author 2014. Published by Oxford University Press. All rights reserved. For Permissions, please email: [email protected]

Carcinogenesis, 2015, Vol. 36, No. 2, 232–242

doi:10.1093/carcin/bgu245Advance Access publication December 25, 2014Original Manuscript

232

original manuscript

Expression of glypican 3 enriches hepatocellular carcinoma development-related genes and associates with carcinogenesis in cirrhotic liversXiyong Liu1, Sean K. Wang2, Keqiang Zhang1, Hang Zhang1, Qin Pan3, Zhiwei Liu3, Hongming Pan3, Lijun Xue2, Yun Yen1,4,* and Peiguo G. Chu2,*1Department of Molecular Pharmacology, Beckman Research Institute, 2Department of Pathology, City of Hope National Medical Center, Duarte, CA 91010, USA, 3Department of Medical Oncology, Sir Run Run Shaw Hospital, Zhejiang University School of Medicine, Hangzhou 310016, China and 4Taipei Medical University, Taipei, Taiwan (R.O.C)

*To whom correspondence should be addressed. Tel: +1 626 359 8111; Fax: +1 626 301 8145; Email: [email protected]

Correspondence may also be addressed to Yun Yen. Tel: +1 626 359 8111; Fax: +1 626 301 8233; Email: [email protected]

Abstract

Glypican-3 (GPC3) protein expression was determined by immunohistochemical analysis from 29 normal livers, 80 cirrhotic livers sample taken near hepatocellular carcinoma (HCC), and 87 cirrhotic livers without HCC. The levels for miR-657 and HCC-related gene mRNAs were determined by quantitative real-time polymerase chain reaction (qRT-PCR). Also, a published microarray dataset was used for gene set enrichment analysis (GSEA) to investigate the relationship between GPC3- and HCC-related gene signatures. Kaplan–Meier analysis was used to evaluate the relationship between GPC3 and HCC recurrence. GPC3 protein expression was not detected in any of the 29 (0%) normal livers, but was detected in 32 of 87 (37%) cirrhotic livers without HCC, and 51 of 80 (64%) cirrhotic liver samples taken near HCC sites (P < 0.001). The GPC3-positive rate in cirrhotic livers of viral origin was 68% (27/40), which was significantly higher than for non-viral cirrhotic livers (11%, 5/47) (P < 0.001). Also, GPC3 expression positively correlated with mRNA expression of HCC-related genes in the qRT-PCR and GSEA evaluations. Furthermore, HCC recurrence in cirrhotic liver samples taken near HCC sites was significantly higher in the GPC3-positive group than the GPC3-negative group (Log-rank P = 0.02, HR = 3.26; 95% CI = 1.20–10.29). This study demonstrated that highly expression of GPC3 could enrich HCC-related genes’ mRNA expression and positive associate with dysplasia in cirrhotic livers. Therefore, GPC3 may serve as a precancerous biomarker in cirrhotic livers.

IntroductionHepatocellular carcinoma (HCC) is the eighth most common malignancy worldwide and one of the most deadly cancers. Approximately 70–90% of HCCs are associated with liver cirrho-sis. Chronic infection with hepatitis B virus (HBV), hepatitis C virus (HCV) or both is the most common risk factor for cirrho-sis and HCC worldwide (1,2). In Western countries, HCV infec-tion and alcohol-induced hepatic injury are the leading causes of liver cirrhosis and constitute the most important HCC risk

factors. Other risk factors for liver cirrhosis include autoim-mune hepatitis (AIH), alpha-1-antitrypsin deficiency (A1ATD), chemotherapy for primary or metastatic liver malignancies, hemochromatosis, nonalcoholic steatohepatitis (NASH), pri-mary biliary cirrhosis (PBC) and Wilson’s disease. The frequency of malignant transformation of cirrhotic liver to HCC varies greatly, depending on the underlying cause of cirrhosis, but is highest in patients with HBV/HCV infections.

Downloaded from https://academic.oup.com/carcin/article-abstract/36/2/232/334548by gueston 13 February 2018

mailto:[email protected]?subject=

mailto:[email protected]?subject=

X.Liu et al. | 233

Carcinogenesis, 2015, Vol. 36, No. 2, 232–242

doi:10.1093/carcin/bgu245Advance Access publication December 25, 2014Original Manuscript

The precise cellular and molecular mechanisms involved in the cancerous transformation of cirrhotic liver have not yet been elucidated. However, it is generally thought that the development of HCC may be a multistep process similar to that described for epithelial tumors, such as breast carcinoma, where the disease progresses from hyperplasia to atypical hyperplasia, and then to carcinoma in situ and finally invasive carcinoma (3,4). Histologically, precancerous liver lesions are not well defined. Prior studies have indicated that hepatic precancerous lesions may include dysplastic nodule (DN) (5) and liver cell dysplasia (6–8). DN is characterized by marked enlargement of individual cirrhotic nodules with thick liver cell plates and a moderate increase in cell density. These nodules show variable atypia but lack features of definite malignancy (3,5). The term ‘liver cell dysplasia’ was first coined by Anthony et al. (9) and encompasses two types of changes: large and small cell. Large cell dysplasia is characterized by cellular enlargement, nuclear pleomorphism and multinucleation of liver cells occurring in groups or occupy-ing whole cirrhotic nodules. An increased nuclear/cytoplasmic ratio without multinucleation is the hallmark of small cell type dysplasia (10). However, despite these morphologic descriptions, there are no immunohistochemical (IHC) or molecular markers to evaluate whether these lesions are precancerous (11,12).

Gene expression profiling studies have shown that some genes are up-regulated in HCC relative to benign lesions and normal hepatic tissue. These genes are not only useful in tumor diagnosis (13–15), but also correlated to patient survivability (16). The protein products of these genes regulate cell growth, inter-action, differentiation, and intracellular signal transduction (14,15). Our previous study also demonstrated that microRNA (miR)-657 expression was correlated with HCC development. We therefore hypothesized that the expression levels of these genes in precancerous lesions of cirrhotic livers may shed light on the differentiation status of the lesions.

Glypican-3 (GPC3) is a heparan sulfate proteoglycan that is present in a secreted form, or is linked to the cell surface through a glycosylphosphatidylinositol-anchored membrane protein (17). Recently, GPC3 was identified as a sensitive and specific marker for HCC and hepatoblastoma. In patients with HCC, GPC3 is overexpressed in neoplastic liver tissue (18–20) and its levels are elevated in the serum. However, GPC3 is unde-tectable in normal and benign liver lesions and in the serum of healthy adults (21). GPC3 expression was also greater in HCC tissue than in tissue from cirrhotic liver (18,19,22–26). There has been controversy regarding the incidence of GPC3 expression in

DNs in cirrhotic livers. Wang et al. (22) reported that the expres-sion of GPC3 in DNs was correlated with the degree of dysplasia, with positive rates of 3%, 48% and 60% for lower-grade, interme-diate-grade and high-grade dysplasia, respectively. In contrast, Libbrecht et al. (24) found none of 14 (0%) lower-grade dysplasia and only 2 of 29 (7%) high-grade dysplasia samples expressed GPC3. Hepatic adenomas, focal nodular hyperplasias and large cell dysplasias, once thought to be precancerous lesions, usu-ally do not express GPC3 (24,27). To date, IHC analysis of GPC3 expression in cirrhotic livers without dysplasia has been limited to very small studies, which have limited statistical power to detect inter-group differences and are more susceptible to acci-dental bias than larger studies.

In the current study, we comprehensively assessed GPC3 IHC expression in cirrhotic livers without HCC caused by different factors, and in cirrhotic liver samples near HCC lesions. We also determined GPC3 levels and expression of HCC-related miR-657 and HCC-related genes in cirrhotic livers. The expression level of GPC3 and relative risk of HCC recurrence were also evaluated in cirrhotic liver samples taken from tissue near HCC.

Materials and methods

Case selectionCollection of human tissue samples was conducted on a protocol approved by the Institutional Review Board (IRB). All patients provided informed consent. Histologic diagnostic criteria for cirrhotic livers, DN, large cell changes (LCC), small cell changes (SCC), and HCC were as fol-lows: Cirrhosis, evidence of a diffuse process in which the normal acini were replaced by architecturally abnormal nodules separated by fibrous and inflamed tissue (28); DN, marked enlargement of individual cirrhotic nodules that showed thick liver cell plates and a moderate increase in cell density with a slightly irregular trabecular pattern (5); LCC, cellu-lar enlargement, nuclear pleomorphisms and multinucleation of liver cells occurring in groups or occupying whole cirrhotic nodules (28); SCC, increased nuclear/cytoplasmic ratio (10); HCC, tumor cells with evidence of hepatocellular differentiation and features of malignancy (sheets of tumor cells with nuclear atypia and mitoses). Tissue blocks were obtained from the 1991–2008 surgical pathology files from 1991–2008 at City of Hope National Medical Center (Duarte, CA). The tissues had been routinely fixed in 10% neutral buffered formalin and embedded in paraffin. One paraf-fin block was selected from each case as follows: cirrhotic livers without HCC (87 cases: 69 resections, 17 biopsies, Supplementary Table 1, available at Carcinogenesis Online), cirrhotic liver samples taken near HCC sites (80 cases: 78 resections, 2 biopsies), and normal livers (29 cases: all resection, Supplementary Table 2, available at Carcinogenesis Online).

Clinical and serologic information from patients with cirrhotic livers with or without HCC were obtained to determine the cause of disease. For patients with a cryptogenic cause, no related clinical and serologic information could be found. Of 87 cirrhotic liver samples without HCC (Supplementary Table 1, available at Carcinogenesis Online), 40 were asso-ciated with viral hepatitis (6 HBV, 31 HCV and 3 HBV/HCV), and 47 were associated with either metabolic abnormalities (2 hemochromatosis, 2 A1ATD and 2 Wilson’s disease) or other causes (13 chemotherapy, 11 alco-holic, 8 PBC, 5 NASH, 2 cryptogenic and 2 AIH). Fourteen of 87 cirrhotic liver samples without HCC showed dysplasia (16%), including 9 with LCC, 2 with SCC and 3 with DN. Of the 80 samples from cirrhotic livers taken near HCC sites (Supplementary Table 2, available at Carcinogenesis Online), 58 were associated with viral hepatitis (19 HBV, 36 HCV and 3 HBV/HCV), and 22 were associated with either metabolic (1 hemochromatosis) or other causes (7 NASH, 13 cryptogenic and 1 AIH). Twenty-six of 80 cirrhotic livers near HCC showed dysplasia (33%), including 15 with LCC, 5 with SCC and 6 with DN.

ImmunohistochemistryAll tissue sections were stained with a commercially available monoclo-nal antibody to GPC3 (Clone 1G12, 1:200, Cell Marque, Rocklin, CA). The

Abbreviations

DN dysplastic nodule FFPE formalin-fixed and paraffin-embedded; GPC3 glypican 3; GSEA gene set enrichment analysis; HBV hepatitis B virus; HCC hepatocellular carcinoma; HCV hepatitis C virus; IGFBP1 insulin-like growth factor binding protein 1; LCC large cell changes; NASH non-alcoholic steatohepatitis; NES normalized enrichment score; PBC primary biliary cirrhosis; RIGE retinoic acid-induced gene E; SCC small cell changes; UGT2B4 UDP-glucuronosyl transferase.


http://carcin.oxfordjournals.org/lookup/suppl/doi:10.1093/carcin/bgu245/-/DC1




234 | Carcinogenesis, 2015, Vol. 36, No. 2

sections were deparaffinized and rehydrated in graded alcohol. Heat-induced epitope retrieval of sections was by incubation in 10 mM citrate buffer (pH 6.0) at 105°C for 30 min in a pressure cooker (Biocare Medical, Concord, CA). Sections were stained on a Dako Autostainer Plus system (Dako, Carpinteria, CA) according to the manufacturer’s instructions. EnVision™ Plus and peroxidase were used to detect the primary antibody. Sections were counterstained in 5% Mayer’s Hematoxylin for 3 min.

Cytoplasmic and membrane-bound GPC3 immunoreactivity (brown granules) were assessed. A section was scored as positive if at least 1 hepatocyte stained positive (29) and as >5% positive if >5% of the cirrhotic hepatocytes were positive.

Quantitative reverse transcriptase PCR for miR-657 and HCC-related genesFormalin-fixed and paraffin-embedded (FFPE) tissue slides of the liver samples were evaluated by pathologists, and then the lesional tissues were carefully transferred into microfuge tubes. RNA was extracted from each sample with the FFPE kit (Qiagen). The quality and quantity of RNA samples were determined using a NanoDrop ND 1000 spectrophotom-eter (Thermo Scientific, Wilmington, DE). Expression levels of miR-657 and other miRNAs were measured with the TaqMan MicroRNA assay kit (Applied Biosystems, Foster City, CA) according to the manufacturer’s instructions. Patient RNA samples were used as templates for reverse transcription. Reverse transcription reactions were carried out in a 96-well plate format using the high-capacity complementary DNA (cDNA) archive kit, RNase inhibitors and specific reverse transcription primers (Applied Biosystems). Quantitative real-time polymerase chain reactions (qRT-PCR) were then performed with the reverse transcription products, TaqMan 2×Universal PCR Master Mix without UNG Amperase (Applied Biosystems), miRNA-specific TaqMan probes, and primers (Applied Biosystems) on an Applied Biosystems 7500 Fast Real Time PCR system with an initial dena-turation at 95°C, followed by 40 cycles at 95°C for 15 s and 60°C for 1 min. The threshold cycle (CT) was then determined and defined as the frac-tional cycle number at which the fluorescence detected passed a fixed threshold. CT values of different miRNAs were normalized to an endog-enous control (RNU6B).

Primers used were specific for the HCC-related genes galectin 4 (Gal-4), UDP-glucuronosyl transferase (UGT2B4), insulin-like growth factor binding protein 1 (IGFBP1), retinoic acid-induced gene E (RIGE) and the control glyceraldehyde 3-phosphate dehydrogenase (GAPDH) (sequences provided in Supplementary Table 3, available at Carcinogenesis Online). The mRNA expression detection protocol was based on a previous report (15). qRT-PCR was carried out using an ABI Prism 7900 HT Sequence Detection System (Applied Biosystems, Carlsbad, CA). The reaction mixture con-sisted of 1× ABI SYBR Green PCR Master Mix, 0.25 μl cDNA, and 0.2 μmol/l of each primer in a total volume of 20 μl. Relative gene expression was quantified using the comparative CT method. GAPDH was used as an endogenous control, and data were expressed as fold change relative to GAPDH for each sample.

Gene set enrichment analysisMicroarray data from a prospective study (E-GEOD-40873) were down-loaded from the ArrayExpress public database (http://www.ebi.ac.uk/arrayexpress/). The phenotypes GPC3-high and GPC-low were stratified by mRNA expression level. Gene set enrichment analysis (GSEA) was car-ried out by computing overlaps with canonical pathways (CP) and gene ontology (GO) gene sets (C2), obtained from the Broad Institute (30). We compared the gene expression levels from the GPC3-high versus GPC3-low groups and identified the genes that had significantly different expres-sion in the GSEA by using the Molecular Signatures Database (V4.0). The permutation number was set as 1000. The enrichment statistic was based on classic method; and metric for ranking genes was according to signal to noise.

Statistical analysisStatistical analyses for differences between viral cirrhotic livers and non-viral cirrhotic livers with or without HCC and with and without dysplasia were performed. JMP 8.0 software (SAS Institute, Cary, NC) and GraphPad Prism 5.0 were used for statistical analysis and figure drawing. Group comparisons for continuous data were performed using t-tests

for independent means or one-way analyses of variance. For categorical data, we used chi-square analyses, Fisher’s exact tests or binomial tests of proportions. Logistic regression models were used to determine the odds ratios (OR) and 95% confidential intervals (95% CI). Kaplan–Meier analysis and Cox proportional hazard model were applied for recurrence analysis. A P value <0.05 was considered statistically significant.

Results

GPC3 is expressed in cirrhotic livers

Of the 29 normal livers that did not have cirrhosis or HCC, none showed evidence of GPC3-positive staining. Of the 87 cirrhotic liver samples without HCC, 32 (37%) were positive for GPC3, including 2 samples from patients with alcoholic cirrhosis (Table 1; Figure 1A), 1 chemotherapy-associated cirrhosis case (Figure 1B), 1 NASH-associated cirrhosis case (Figure 1C), 1 PBC case (Figure 1D) and 27 HBV- and/or HCV-associated cirrhosis cases (Figure 1D–G). In 11 (34%) of the 32 cirrhotic livers without HCC that scored positive for GPC3 expression, >5% of the hepat-ocytes stained for GPC3. Ten of these 11 samples came from viral cirrhotic livers and one from a liver with NASH-associated cirrhosis. For most GPC3-positive cases, we did not see mor-phologic differences in hematoxylin and eosin (H&E)-stained sections between GPC3-positive hepatocytes and nearby GPC3 negative hepatocytes (Figure 1F and G).

For the majority of GPC3-positive samples, the GPC3-positive hepatocytes were present at the periphery of cirrhotic nod-ules immediately adjacent to the inflamed and fibrous tissue (Figure 2A). Occasionally, GPC3-positive hepatocytes were seen at the center of cirrhotic nodules (Figure 2B). In rare cases, all of the hepatocytes in the nodule were positive for GPC3 (Figure 2C). The GPC3 staining pattern in cirrhotic livers was predominantly granular cytoplasmic, with frequent Golgi and canalicular pat-terns (Figure 2D). GPC3 staining in cirrhotic liver samples was rarely in a membranous pattern (Figure 2E and F), which is the most common staining pattern seen in HCC samples (Figure 2G and H).

Fifty-one (64%) of the 80 samples from cirrhotic livers near HCC (Table 1) were positive for GPC3, including samples from 1 autoimmune, 13 cryptogenic, 2 hemochromatosis-associated, 7 NASH-associated and 40 HBV- and/or HCV-associated livers. The staining patterns were similar to those seen in cirrhotic livers without HCC. Similar to what we saw for cirrhotic livers without HCC, one-third (17/51, 33%) of the cirrhotic liver samples from near an HCC site had more than 5% GPC3-positive hepatocytes. Fifteen of these 17 samples were associated with HBV and/or HCV.

Expression of GPC3 is associated with dysplasia, viral status and HCC in cirrhotic livers

The GPC3-positive rate was significantly different between cir-rhotic livers with and without HCC (Figure 2H). Fourteen of 87 (16%) cirrhotic livers without HCC showed at least focal dyspla-sia, and 9 of these 14 were positive for GPC3 (64%). Twenty-three of 73 (32%) cirrhotic livers without dysplasia were positive for GPC3. The GPC3-positive rate was significantly higher in cirrhotic livers with dysplasia than in those without dysplasia (P = 0.020) (Figure 2I, Table 2). The GPC3-positive rate was also significantly higher in viral cirrhotic livers without HCC than in non-viral cir-rhotic livers without HCC (P < 0.001) (Figure 2J, Table 1).

Further logistic analysis revealed that the overall OR of GPC3-positive cirrhotic livers was 2.5 (95% CI 1.3–4.9) (with ver-sus without HCC) in comparison with that of GPC3-negative cir-rhotic livers (P < 0.001, Table 2). In cirrhotic livers without HCC,



http://www.ebi.ac.uk/arrayexpress/

http://www.ebi.ac.uk/arrayexpress/

X.Liu et al. | 235

the adjusted OR of GPC3-positive cirrhotic livers with dyspla-sia was 3.5 (95% CI 1.0–13.3) in comparison with that of GPC3-positive cirrhotic livers without dysplasia (Table 2). A similar result was seen for cirrhotic liver samples taken near HCC sites, but the comparison was not statistically significant. The above findings suggest that GPC3 was significantly associated with HCC and liver dysplasia.

GPC3 is positively associated with miR-657 and HCC-related genes in cirrhotic livers

Our previous study demonstrated that miR-657 plays a pivotal role in the development of HCC by targeting TLE1 through the NK-κB signaling pathway (31). Therefore, we compared the RNA expression of miR-657 and the downstream target gene TLE1 among normal livers, GPC3-negative cirrhotic livers, GPC3-positive cirrhotic livers and HCC tissue samples (Figure 3A and B). The feasibility of expression profiling from FFPE liver tissue has been validated by Kondoh et al. (15). Overall, the miR-657/TLE1 levels steadily increased/decreased as expected in the normal livers, GPC3-negative cirrhotic livers, GPC3-positive cir-rhotic livers and HCC samples. In particular, the expression of miR-657 was significantly greater in GPC3-positive relative to GPC3-negative cirrhotic livers (P = 0.03).

Based on cDNA microarray analysis, UGT2B4, GAL-4, RIGE and IGFBP1 have been identified as HCC related genes (15). Here, we determined the mRNA expression levels of the above HCC-related genes in GPC3-positive (n = 26) and GPC3-negative (n = 29) cirrhotic livers (Figure 3C). After normalization, qRT-PCR analysis indicated that expression of GAL-4 and RIGE was signifi-cantly greater in GPC3-positive than in GPC3-negative cirrhotic

livers (Student’s t-test, P < 0.05). The expression levels of UGT2B4 and IGFBP1 were also greater in GPC3-positive versus GPC3-negative cirrhotic livers, but these results were not statistically significant (P > 0.05). Nevertheless, the above findings indicate that the expression of GPC3 is positively associated with some HCC-related genes in cirrhotic livers, which further suggests that GPC3 is a potential biomarker for precancerous lesions.

To further explore the above results, GSEA was used to evaluate if GPC3 was associated with HCC-related genes. Each sample was stratified based on GPC3 mRNA expression levels near HCC sites as GPC3-high (24 cases) or GPC3-low (25 cases). The analysis indicated that a set of up-regulated gene signa-ture from transgeneic mice (32) was enriched in the GPC3-high subgroup [normalized enrichment score (NES) = 2.13, P = 0.002] (Figure 3D). The heat map further indicated up-regulation of the HCC-related genes in the GPC3-high subgroup (Figure 3D, lower panel). Additionally, an HCC-downregulated gene set (33) was enriched in the GPC3-low subgroup (NES = 1.99, P = 0.004) (Figure 3E). The heat map also showed that expression of these HCC-negative-related genes was less in the GPC3-high subgroup (Figure 3E, lower panel). These results further suggest that GPC3 is co-expressed with HCC-related genes in cirrhotic livers, which implies that GPC3 could be a valuable biomarker to indicate the presence of precancerous lesions in cirrhotic liver samples.

GPC3 is positively associated with HCC recurrence in cirrhotic livers near HCC

We also evaluated the association between GPC3 and HCC recur-rence and proliferation, using existing gene sets and the GPC3-high and GPC3-low samples from previous stratification. The

Table 1. Expression of GPC3 in cirrhotic livers with or without HCC

Cirrhotic livers without HCC (n = 87) Cirrhotic livers near HCC (n = 80)

Cases GPC3 positive (%) P value Cases GPC3 positive (%) P value

Sex Male 53 24 (45.3) 56 35 (62.6) Female 34 8 (23.5) 0.040 24 16 (66.7) 0.722Age <50 30 8 (26.7) 12 4 (33.3) 50–59 33 12 (36.4) 23 19 (82.6) 60–69 18 9 (50.0) 30 20 (66.7) ≥70 6 3 (50.0) 0.373 15 8 (53.3) 0.025Dysplasia With 14 9 (64%) 26 20 (78%) Without 73 23 (32%) 0.020 53 31 (59%) 0.108Viral status Viral 40 27 (68%) 58 40 (69%) Non-viral 47 5 (11%) <0.001 22 11 (50%) 0.115Viral detail HBV 6 5 (83%) 19 9 (47%) HCV 31 19 (61%) 36 30 (83%) HBV/HCV 3 3 (100%) 3 1 (33%)Non-viral AIH 2 0 (0%) 1 1 (100%) Alcoholic 11 2 (18%) A1ATD 2 0 (0%) Chemotherapy 13 1 (8%) Cryptogenic 2 0 (0%) 13 3 (23%) Hemochromatosis 2 0 (0%) 1 1 (100%) NASH 5 1 (20%) 7 6 (86%) PBC 8 1 (12.5%) Wilson’s disease 2 0 (0%)



GSEA analysis revealed that GPC3-high status was positively correlated with Hoshida et al. (16) liver cancer recurrence gene sets (NES = 2.00 and 2.41, respectively, Figure 4A and B). The heat map further showed these two gene sets were up-regulated in the GPC3-high subgroup. We further analyzed GPC3 expression

in cirrhotic liver samples taken near HCC sites and evaluated recurrence in this cohort. Kaplan–Meier analysis indicated that the recurrence rate in the GPC3-high subgroup was significantly higher than that of the GPC3-low group (Log-rank P = 0.02, Figure 4C). At 70 weeks post-surgery, half of the HCC recurrence

Figure 1. GPC3 expression in cirrhotic livers. Alcoholic cirrhosis (A, 400×), chemotherapy associated cirrhosis (B, 400×), non-alcoholic steatohepatitis-associated cir-

rhosis (C, 400×), and primary biliary cirrhosis (D, 400×). Note the granular cytoplasmic staining with a vague canalicular pattern in A–C, and a canalicular and Golgi

pattern in D. (E) H&E-stained viral cirrhotic liver showing large cell dysplasia and weak and diffuse GPC3 immunostaining of normal-appearing hepatocytes in the left

field (arrows). Hepatocytes with large cell changes are negative for GPC3 (arrows) (F). (G) H&E-stained cirrhotic nodule without dysplasia. (H) GPC3 immunostaining of

the nodule is shown in G.


X.Liu et al. | 237

had occurred in the GPC3-high subgroup. However, the recur-rence rate in the GPC3-low group was only about 20% at this time point. Cox proportional hazard analysis further revealed that the hazard ratio (GPC3-high versus GPC3-low) of recurrence was 3.26 (95% CI 1.20–10.29). Finally, we compared prognostic significance of GPC3 with existing biomarkers for HCC. Alpha-fetoprotein (AFP) is considered a good diagnostic biomarker for HCC. However, for the cohort evaluated in this study, the prognostic value of AFP mRNA in cirrhotic livers near HCC was poor in pre-dicting the recurrence of HCC (Log-rank P = 0.639) (Figure 4D). Thus, the above findings suggest that GPC3 is a potential bio-marker for identifying pre-cancerous cirrhotic livers.

DiscussionSeveral types of liver lesions have been described as being pre-cancerous over the years, including DN, large cell dysplasia,

small cell dysplasia, nodular regenerative hyperplasia, hepatic adenoma and focal nodular hyperplasia (5,9,10,34,35). The first four types of lesions are commonly seen in cirrhotic livers, regardless of the presence of HCC. For instance, large cell dyspla-sia has been described in 7% of patients with cirrhosis and 65% of patients with HCC (36). However, there is controversy regarding whether these lesions are truly precancerous or are just archi-tectural changes associated with cirrhosis. Large cell dysplasia, originally reported as a precancerous lesion, is now thought to be a non-specific morphologic change that has little value in predicting malignancy (10–12). Similarly, hepatic adenoma and focal nodular hyperplasia are now thought to be benign morpho-logic changes resulting from use of female hormones or from ischemia, respectively (27). Therefore, there is a strong need for useful biomarkers that can predict cancer in the cirrhotic livers.

Because GPC3 is a sensitive and specific marker for HCC (19–22), several studies have focused on its expression in putative

Figure 2. Expression of GPC3, viral status and dysplasia. Sample micrographs show GPC3-positive cells on the periphery of cirrhotic nodules near fibrous and inflamed

stroma (A, common staining pattern), in the center of cirrhotic nodules (B, rare staining pattern) or in whole cirrhotic nodules (arrows) (C). GPC3-positive staining in

cirrhotic livers is shown as weak cytoplasmic granules concentrated near canaliculi and the Golgi apparatus (D). The positive granules in cirrhotic livers are distributed

throughout the cytoplasm and alongside the cell membrane (cytoplasmic and membranous staining pattern) (E). GPC3-positive staining in HCC shows a diffuse cyto-

plasmic (F) and membrane-associated staining pattern, although cytoplasmic and canalicular and Golgi staining patterns were also seen (G). GPC3 expression among

normal livers and cirrhotic livers with and without HCC (H). Correlation of GPC3 expression and dysplasia in cirrhotic livers with and without HCC (I). GPC3 and viral

status among cirrhotic livers (J).The error bar represents the standard deviation.



precancerous lesions (22,24,25,27,29,37,38). Benign hepatic ade-nomas and focal nodular hyperplasias, which mimic HCC, were always reported as negative for GPC3 (27). The reported inci-dence of GPC3 expression in dysplastic liver lesions is variable, ranging from 0% (23,38) to 25% in low-grade DNs, and up to 75% in high-grade DNs (29). In the current study, most GPC3-positive hepatocytes in cirrhotic livers appeared morphologically normal and the nearby dysplastic hepatocytes were frequently GPC3-negative. Interestingly, Di Tommaso et al. (23) found that 20 of 22 cases of high-grade DNs and all 15 cases of liver-regenerative nodules and 15 cases of low grade DNs they evaluated were GPC3-negative. Similarly, Libbrecht et al. (24) showed that all 14 low-grade DNs they evaluated and 27 of 29 high-grade DNs were GPC3-negative. These results strongly suggest that low-grade DNs and most high-grade DNs are regenerative rather than dys-plastic/precancerous (23,39). Abdul-Al et al. (29) also observed similar GPC3-positive staining in non-dysplastic, non-neoplastic hepatocytes in six cases of HCV-associated cirrhosis.

The staining pattern of GPC3 in cirrhotic livers depends on the expression level. The GPC3 staining in cirrhotic livers is usu-ally weak and focal, but more than one-third of viral cirrhotic livers showed >5% of hepatocytes as GPC3 positive. At very low-levels of GPC3 expression, positive GPC3 staining appears cytoplasmic and to have a vague canalicular and Golgi pattern. As the GPC3 expression level increases, GPC3 staining is usu-ally diffusely cytoplasmic with attenuation along the cell mem-branes, giving an impression of a membranous staining pattern. Histologically, these GPC3-positive hepatocytes are present in small groups and are often seen at the peripheral edges of cir-rhotic nodules that are adjacent to inflamed and fibrotic stroma. Similar findings were also observed by Wang et al. (37). It has been shown that GPC3 is highly expressed in fetal hepatocytes and in hepatic progenitor cells after liver injury in adults (40). The histologic hallmark of cirrhotic livers is hepatocellular necrosis and regeneration. After hepatic necrosis, progenitor hepatocytes, which are the targets of malignant transformation in the Solt–Farber model of hepatic carcinogenesis, proliferate and express GPC3 (40). The overexpression of GPC3 modulates cell proliferation by inhibiting fibroblast growth factor 2 and bone morphogenetic protein 7 signaling, which may help to explain its role in liver carcinogenesis (41).

Recently, it was also shown that GPC3 stimulates Wnt activity by facilitating interaction with Wnt signaling receptors (42,43). Our GSEA results also indicated that the Wnt3A up-regulated gene signature was enriched in GPC3-high subgroup (NES = 1.74, P = 0.033). The HCC-related genes GAL-4, IGFBP1, RIGE and UGT2B4 play roles in cell–cell and cell–substrate interaction, cell proliferation and drug resistance, respectively (15). In addition, they may also regulate Wnt signal transduction in human can-cers (44–47). Because GPC3 and the HCC-related genes GAL-4 and RIGE regulate and are regulated through the Wnt signal trans-duction pathway, their expression may be up-regulated simulta-neously. GPC3-positive cirrhotic livers only express some but not all HCC-related genes, suggesting additional mechanisms for cirrhotic liver carcinogenesis other than the GPC3/Wnt pathway because only ~80–85% of HCCs express GPC3 (27).

In Western countries, >80% of cases of HCC occur in patients with underlying cirrhosis, regardless of etiology. The major causes of cirrhosis include viral hepatitis (HBV and HCV), alco-holism, obesity, insulin resistance/metabolic syndrome, biliary disease, metabolic disorders and autoimmune disease. Despite the many causes of cirrhosis, the major risk factors for HCC are infection with HBV, HCV or both (48,49). HBV is responsible for at least 75% of HCC cases worldwide (50). In industrialized coun-tries, 60% of cases of HCC are related to HCV infection. Patients older than 50 years who have viral hepatitis and a history of alcohol abuse have the highest risk of developing HCC (51). In the current study, we found that 84% (27/32) of GPC3-positive cir-rhotic livers without HCC are associated with HBV and/or HCV infection, which further supports the above epidemiologic find-ings. Therefore, if a liver biopsy shows cirrhosis with viral hepa-titis and is positive for GPC3 expression, the patient may need to be closely monitored for the development of HCC. Studies have shown that, in addition to injuring hepatocytes, HBV and HCV have direct oncogenic effects on hepatocytes through integra-tion of viral DNA into the genome (52), or through the activity of viral oncogenic proteins (53). In comparison with other fac-tors, HBV and HCV are the most potent stimulants of hepatocyte transformation to precancerous lesions and HCC. Non-viral fac-tors may have weak oncogenic effects and thus take a longer time to transform normal hepatocytes to precancerous lesions and HCC.

Table 2. Stratification analysis for relevance between GPC3 and dysplasia and HCC

Tissue types Cases GPC3 positive ORa (95% CI) Adjusted ORb (95% CI)

Overall cirrhotic liver Without HCC 87 32 (36.8%) Reference Reference Near HCC 80 51 (63.8%) 3.0 (1.6–5.7)** 2.5 (1.3–4.9)**Cirrhotic livers without HCC Without dysplasia 73 23 (31.5%) Reference Reference With dysplasia 14 9 (64.3%) 3.9 (1.2–14.0)* 3.5 (1.0–13.3)*Cirrhotic livers near HCC Without dysplasia 53 31 (57.4%) Reference Reference With dysplasia 26 20 (76.9%) 2.5 (0.9–7.7) 2.5 (0.9–7.7)Viral cirrhotic livers Without HCC 40 27 (67.5%) Reference Reference Near HCC 58 40 (69.0%) 1.1 (0.4–2.5) 1.6 (0.6–4.4)Non-viral cirrhotic livers Without HCC 47 5 (10.6%) Reference Reference Near HCC 22 11 (50.0%) 8.4 (2.5–31.7)** 7.0 (2.0–28.7)**

aNon-conditional logistic regression analysis.bThe OR was adjusted by age and sex.

*P < 0.05; **P < 0.01.


X.Liu et al. | 239

Figure 3. GPC3 expression was positively associated with expression of HCC-related RNAs in cirrhotic livers. Expression levels of miR-657 (A) and its target TLE1 (B) in

normal livers, GPC3- cirrhotic livers, GPC+ cirrhotic livers and HCC. mRNA expression levels of Gal-4, UGT2B4, IGFBP1 and RIGE (C) were determined for GPC3− (n = 26)

and GPC3+ (n = 29) cirrhotic livers. The error bar represents the standard deviation. Using downloaded microarray data, a GSEA analysis was conducted to evaluate

association between GPC3− and HCC-related gene sets in cirrhotic liver samples taken near HCCs. Genes up-regulated in an HCC transgenic mouse model were also

up-regulated in GPC3-high samples relative to GPC3-low subgroup. (D). Genes down-regulated in HCC were also down-regulated in GPC3-high subgroup relative to

GPC3-low controls (E).



As is the case in HCC, the GPC3-positive cirrhotic livers also express high levels of miR-657. Our analysis also revealed that HCC-related genes and HCC recurrence-related genes were

enriched in GPC3-high subgroups in cirrhotic liver samples. Some gene changes were validated by qRT-PCR, which provides further support for the interpretation that GPC3 is a precancerous

Figure. 4. Expression of GPC3 was positively associated with HCC recurrence in cirrhotic liver samples taken near HCC sites. Genes up-regulated in liver cancer recur-

rence (A) and liver cancer proliferation (B) were up-regulated in GPC3-high cirrhotic livers near HCC. Kaplan–Meier analysis also showed that mRNA expression level

of GPC3 in cirrhotic liver samples taken near HCC sites could predict the recurrence of HCC (C), but AFP expression could not (D).


X.Liu et al. | 241

biomarker in cirrhotic livers. A recent study reported that a human heavy-chain variable domain antibody, HN3, with high affinity (Kd = 0.6 nM) for cell-surface-associated GPC3 mole-cules. The HN3 inhibited proliferation of GPC3-positive cells and exhibited significant inhibition of HCC xenograft tumor growth in nude mice (54). Therefore, we are optimistic that GPC3 might provide useful prognostic and therapeutic values in liver dis-ease, particularly when combined with morphological assess-ment of hepatic lesions. However, the limitation of this project was that all above evidences were based on retrospective and cross-section designed studies. A prospective cohort study will carry out to further validate these findings in the future.

Taken together, GPC3 expression is significantly higher in cirrhotic livers with dysplasia. Meanwhile, GPC3 level was posi-tively associated with HCC-related genes’ mRNA expression in cirrhotic livers. It may serve as a precancerous biomarker for cirrhotic livers and prognostic for HCC.

FundingCity of Hope National Medical Center.

AcknowledgementsThe authors thank Drs. Nancy Linford and Dr. Keely Walker for editing this manuscript.

X.L. and P.G.C. were involved in study design, statistical anal-ysis and manuscript writing. S.K.W. and H.Z. were involved in RNA extraction and qRT-PCR conducting. K.Z. was involved in qRT-PCR primer design and qRT-PCR validation. Q.P. involved in human tissue sample collection and patients’ following-up. Z.L. and H.P. were involved in collecting human tissue sample and clinical information. Lijun Xue: sample preparation and Immunohistochemistry staining. Y.Y. was involved in study design and data analysis.Conflict of interest statement: None declared.

References 1. Akagi, G. et al. (1984) Liver cell dysplasia and hepatitis B surface anti-

gen in liver cirrhosis and hepatocellular carcinoma. Cancer, 54, 315–318.

2. Okuda, K. et al. (1982) Hepatocellular carcinoma arising in noncirrhotic and highly cirrhotic livers: a comparative study of histopathology and frequency of hepatitis B markers. Cancer, 49, 450–455.

3. Le, B.B. et al. (1995) Adenomatous hyperplasia in cirrhotic livers: his-tological evaluation, cellular density, and proliferative activity of 35 macronodular lesions in the cirrhotic explants of 10 adult French patients. Hum. Pathol., 26, 897–906.

4. Sakamoto, M. et al. (1991) Early stages of multistep hepatocarcinogen-esis: adenomatous hyperplasia and early hepatocellular carcinoma. Hum. Pathol., 22, 172–178.

5. Takayama, T. et al. (1990) Malignant transformation of adenomatous hyperplasia to hepatocellular carcinoma. Lancet, 336, 1150–1153.

6. Borzio, M. et al. (1995) Liver cell dysplasia is a major risk factor for hepatocellular carcinoma in cirrhosis: a prospective study. Gastroen-terology, 108, 812–817.

7. Cohen, C. et al. (1979) Liver cell dysplasia: association with hepatocel-lular carcinoma, cirrhosis and hepatitis B antigen carrier status. Can-cer, 44, 1671–1676.

8. Le, B.B. et al. (1997) Prevalence of liver cell dysplasia and association with HCC in a series of 100 cirrhotic liver explants. J. Hepatol., 27, 835–842.

9. Anthony, P.P. et al. (1973) Liver cell dysplasia: a premalignant condition. J. Clin. Pathol., 26, 217–223.

10. Watanabe, S. et al. (1983) Morphologic studies of the liver cell dysplasia. Cancer, 51, 2197–2205.

11. Park, Y.N. et al. (2006) Large liver cell dysplasia: a controversial entity. J. Hepatol., 45, 734–743.

12. Ikeda, H. et al. (2009) Large cell change of hepatocytes in chronic viral hepatitis represents a senescent-related lesion. Hum. Pathol., 40, 1774–1782.

13. Jia, H.L. et al. (2007) Gene expression profiling reveals potential bio-markers of human hepatocellular carcinoma. Clin. Cancer Res., 13, 1133–1139.

14. Skawran, B. et al. (2008) Gene expression profiling in hepatocellular carcinoma: upregulation of genes in amplified chromosome regions. Mod. Pathol., 21, 505–516.

15. Kondoh, N. et al. (1999) Identification and characterization of genes associated with human hepatocellular carcinogenesis. Cancer Res., 59, 4990–4996.

16. Hoshida, Y. et al. (2008) Gene expression in fixed tissues and outcome in hepatocellular carcinoma. N. Engl. J. Med., 359, 1995–2004.

17. Filmus, J. et al. (2001) Glypicans: proteoglycans with a surprise. J. Clin. Invest., 108, 497–501.

18. Nakatsura, T. et al. (2003) Glypican-3, overexpressed specifically in human hepatocellular carcinoma, is a novel tumor marker. Biochem. Biophys. Res. Commun., 306, 16–25.

19. Yamauchi, N. et al. (2005) The glypican 3 oncofetal protein is a promis-ing diagnostic marker for hepatocellular carcinoma. Mod. Pathol., 18, 1591–1598.

20. Kandil, D.H. et al. (2009) Glypican-3: a novel diagnostic marker for hepatocellular carcinoma and more. Adv. Anat. Pathol., 16, 125–129.

21. Capurro, M. et al. (2003) Glypican-3: a novel serum and histochemical marker for hepatocellular carcinoma. Gastroenterology, 125, 89–97.

22. Wang, X.Y. et al. (2006) Glypican-3 expression in hepatocellular tumors: diagnostic value for preneoplastic lesions and hepatocellular carcino-mas. Hum. Pathol., 37, 1435–1441.

23. Di Tommaso, L. et al. (2007) Diagnostic value of HSP70, glypican 3, and glutamine synthetase in hepatocellular nodules in cirrhosis. Hepatol-ogy, 45, 725–734.

24. Libbrecht, L. et al. (2006) Glypican-3 expression distinguishes small hepatocellular carcinomas from cirrhosis, dysplastic nodules, and focal nodular hyperplasia-like nodules. Am. J. Surg. Pathol., 30, 1405–1411.

25. Zhu, Z.W. et al. (2001) Enhanced glypican-3 expression differentiates the majority of hepatocellular carcinomas from benign hepatic disor-ders. Gut, 48, 558–564.

26. Zynger, D.L. et al. (2006) Glypican 3: a novel marker in testicular germ cell tumors. Am. J. Surg. Pathol., 30, 1570–1575.

27. Coston, W.M. et al. (2008) Distinction of hepatocellular carcinoma from benign hepatic mimickers using glypican-3 and CD34 immunohisto-chemistry. Am. J. Surg. Pathol., 32, 433–444.

28. Anthony, P.P. et al. (1977) The morphology of cirrhosis: definition, nomenclature, and classification. Bull. World Health Organ., 55, 521–540.

29. Abdul-Al, H.M. et al. (2008) Glypican-3 expression in benign liver tissue with active hepatitis C: implications for the diagnosis of hepatocellular carcinoma. Hum. Pathol., 39, 209–212.

30. Subramanian, A. et al. (2005) Gene set enrichment analysis: a knowl-edge-based approach for interpreting genome-wide expression pro-files. Proc. Natl. Acad. Sci. U. S. A., 102, 15545–15550.

31. Zhang, L. et al. (2013) MicroRNA-657 promotes tumorigenesis in hepa-tocellular carcinoma by targeting transducin-like enhancer protein 1 through nuclear factor kappa B pathways. Hepatology, 57, 1919–1930.

32. Lee, J.S. et al. (2004) Application of comparative functional genomics to identify best-fit mouse models to study human cancer. Nat. Genet., 36, 1306–1311.

33. Lee, D.C. et al. (2008) Functional and clinical evidence for NDRG2 as a candidate suppressor of liver cancer metastasis. Cancer Res., 68, 4210–4220.

34. Tsuji, M. et al. (1999) An immunohistochemical study of hepatic atypi-cal adenomatous hyperplasia, hepatocellular carcinoma, and chol-angiocarcinoma with alpha-fetoprotein, carcinoembryonic antigen, CA19-9, epithelial membrane antigen, and cytokeratins 18 and 19. Pathol. Int., 49, 310–317.

35. Koelma, I.A. et al. (1986) Hepatocellular carcinoma, adenoma, and focal nodular hyperplasia. Comparative histopathologic study with immu-nohistochemical parameters. Arch. Pathol. Lab. Med., 110, 1035–1040.

36. Ganne-Carrié, N. et al. (1996) Predictive score for the development of hepatocellular carcinoma and additional value of liver large cell dys-plasia in Western patients with cirrhosis. Hepatology, 23, 1112–1118.



37. Wang, H.L. et al. (2008) Glypican-3 as a useful diagnostic marker that distinguishes hepatocellular carcinoma from benign hepatocellular mass lesions. Arch. Pathol. Lab. Med., 132, 1723–1728.

38. Llovet, J.M. et al. (2006) A molecular signature to discriminate dysplas-tic nodules from early hepatocellular carcinoma in HCV cirrhosis. Gas-troenterology, 131, 1758–1767.

39. Tornillo, L. et al. (2002) Chromosomal alterations in hepatocellular nodules by comparative genomic hybridization: high-grade dysplas-tic nodules represent early stages of hepatocellular carcinoma. Lab. Invest., 82, 547–553.

40. Grozdanov, P.N. et al. (2006) The oncofetal protein glypican-3 is a novel marker of hepatic progenitor/oval cells. Lab. Invest., 86, 1272–1284.

41. Midorikawa, Y. et al. (2003) Glypican-3, overexpressed in hepatocellular carcinoma, modulates FGF2 and BMP-7 signaling. Int. J. Cancer, 103, 455–465.

42. Capurro, M.I. et al. (2005) Glypican-3 promotes the growth of hepatocel-lular carcinoma by stimulating canonical Wnt signaling. Cancer Res., 65, 6245–6254.

43. Filmus, J. et al. (2008) Glypicans. Genome Biol., 9, 224. 44. Ueno, K. et al. (2011) IGFBP-4 activates the Wnt/beta-catenin signaling

pathway and induces M-CAM expression in human renal cell carci-noma. Int. J. Cancer, 129, 2360–2369.

45. Zhu, W. et al. (2008) IGFBP-4 is an inhibitor of canonical Wnt signalling required for cardiogenesis. Nature, 454, 345–349.

46. Satelli, A. et al. (2011) Galectin-4 functions as a tumor suppressor of human colorectal cancer. Int. J. Cancer, 129, 799–809.

47. Abrahams, N.A. et al. (2004) Chromophobe renal cell carcinoma: a comparative study of histological, immunohistochemical and ultras-tructural features using high throughput tissue microarray. Histopa-thology, 45, 593–602.

48. Kao, J.H. et al. (2000) Overview of hepatitis B and C viruses. In Goedert, J.J. (ed) Infectious Causes of Cancer. Human Press, NJ, pp. 313–330.

49. Zaman, S.N. et al. (1985) Risk factors in development of hepatocellu-lar carcinoma in cirrhosis: prospective study of 613 patients. Lancet, 1, 1357–1360.

50. Fattovich, G. et al. (2004) Hepatocellular carcinoma in cirrhosis: inci-dence and risk factors. Gastroenterology, 127, S35–S50.

51. Ostapowicz, G. et al. (1998) Role of alcohol in the progression of liver disease caused by hepatitis C virus infection. Hepatology, 27, 1730–1735.

52. Chang, M.H. et al. (1997) Universal hepatitis B vaccination in Tai-wan and the incidence of hepatocellular carcinoma in children. Taiwan Childhood Hepatoma Study Group. N. Engl. J. Med., 336, 1855–1859.

53. Kasprzak, A. et al. (2008) Role of hepatitis C virus proteins (C, NS3, NS5A) in hepatic oncogenesis. Hepatol. Res., 38, 1–26.

54. Feng, M. et al. (2013) Therapeutically targeting glypican-3 via a confor-mation-specific single-domain antibody in hepatocellular carcinoma. Proc. Natl. Acad. Sci. U. S. A., 110, E1083–E1091.


expression of glypican 3 enriches hepatocellular carcinoma

Documents