Associated measurement of fucosylated levels of AFP,DCP,and GPC3 for early diagnosis in hepatocellular carcinoma

Abstract

Background:

Hepatocellular carcinoma is a serious health problem worldwide, especially in Asian countries, such as China. However, there are difficulties in diagnosing and treating hepatocellular carcinoma. The alteration of fucosylated proteins was closely associated with carcinogenesis. This study is designed to evaluate the early diagnostic value of associated detection of fucosylated alpha-fetoprotein (fuc-AFP), fucosylated des-γ-carboxy prothrombin (fuc-DCP), and fucosylated glypican 3 (fuc-GPC3) in hepatocellular carcinoma.

Methods:

All serum specimens collected from patients were diagnosed by complete clinicopathological examination and then subjected to the associated detection of fuc-AFP, fuc-DCP, and fuc-GPC3 by protein microarray. Canonical discriminant analysis was adopted to discriminate between the hepatocellular carcinoma group and the benign liver disease group.

Results:

A total of 51 patients with hepatocellular carcinoma and 47 patients in the benign liver disease group were included in this study. Fuc-AFP, fuc-DCP, and fuc-GPC3 were significantly higher in the hepatocellular carcinoma group than in the benign liver disease group. The sensitivity, specificity, and accuracy of canonical discriminant analysis classification were 80.4%, 97.9%, and 88.8%, respectively.

Conclusions:

Fuc-AFP, fuc-DCP, and fuc-GPC3 are effective and useful tumor biomarkers. Associated measurement of these biomarkers with canonical discriminant analysis classification is a promising method for the early diagnosis of hepatocellular carcinoma.

Keywords

Fucosylation AFP DCP GPC3 hepatocellular carcinoma

Introduction

Hepatocellular carcinoma (HCC) is a leading cause of cancer-related global death and the fifth most common cancer around the world.^1-3 Compared with western countries, the incidence of HCC is much higher in Eastern and South-Eastern Asia, especially in china. The increased prevalence of chronic hepatitis B virus and hepatitis C virus (HCV) infection has resulted in approximately 70% of liver cirrhosis (LC).^3,4 HCC has the characteristics of high morbidity and mortality, poor prognosis, difficult early diagnosis and treatment, and a low 5-year survival rate. Most patients with HCC are in advanced stages, so the treatment is palliative and will not include radical liver resection and liver transplantation. Thus, HCC has seriously depressed patients’ quality of live and has aggravated the economic burden of society. Therefore, there is an urgent need to find new biomarkers or novel methods for improving the early diagnosis rate and the survival rate of HCC.

In order to diagnose HCC early, a lot of technology (e.g. magnetic resonance imaging (MRI) and computed tomography) is currently available and often used, including massive candidate tumor biomarkers, such as Golgi protein 73(GP73),^5,6 alpha-fetoprotein (AFP), glypican 3 (GPC3),⁷ vitamin K absence or antagonist-Ⅱ(PIVKA-Ⅱ, also known as des-γ-carboxy prothrombin (DCP)).⁸ The level of AFP especially, is currently widely applied to diagnose HCC in clinical practice. As an excellent protein research method, protein microarray has been extensively applied to measure AFP.^9-11 For instance, in our previous research, we have shown that the AFP serum level can be detected faithfully and correctly by protein microarray with a sensitivity of 91.67% and a specificity of 93.24%.¹² However, the serum level of AFP could also be increased in benign liver diseases, such as LC and hepatitis, and up to 40% of HCC patients’ AFP levels can be normal.¹³ Thus, the poor specificity of AFP is an inherent concern for its application in cancer diagnosis.

In Chinese and Japanese hospitals, DCP is also a commonly tested biomarker for HCC diagnosis and monitoring for tumor relapse. Studies have shown that the combined measurement of AFP and DCP was more useful than measuring AFP alone to diagnose HCC.^4,14,15 Libbrecht et al.⁷ have reported that GPC3 was expressed more in HCC tissues than in benign liver lesions, which correlated with the degree of dysplasia and was even positively associated with HCC-related genes. It has been estimated that GPC3 could play a more important role in HCC diagnosis than AFP. An alteration of the cellular glycosylation phenotype may affect the normal cellular biological processes, such as cell–cell interaction, and result in tumor development and progression. Abnormal protein glycosylations were closely related to carcinogenesis and metastasis formation, and provided new biomarkers for tumor diagnosis and specific targets for tumor therapeutic intervention.^16,17 One of the most commonly investigated abnormal glycosylations was fucosylation. Several researchers have estimated that the abnormal enrichments of fucosylated proteins had strongly promoted tumor dissemination and growth.^18-20 Some representative biomarkers, such as fuc-AFP, fuc-haptoglobin and α1-antitrypsin, have been screened for cancer diagnosis.^21-23

However, the diagnostic value and the presence of fuc-DCP and fuc-GPC3 in HCC have rarely been studied. Thus, the purpose of this study is to investigate the possible diagnostic utility of fuc-AFP, fuc-DCP, and fuc-GPC3 in HCC using high through-put protein microarray and canonical discriminant analysis (CDA) method.

Materials and methods

Serum specimen collection

The serum specimens included in this study were collected from patients at the Beijing YouAn Hospital and were classified as the HCC group and the benign liver disease (BLD) group. The BLD group was set as the control. Patients in the HCC group had their diagnoses confirmed through histological findings or imaging technological characteristics, and then classified with the American Joint Committee on Cancer malignant tumors staging system (tumor–node–metastasis). Moreover, to exclude the potential HCC, the control subjects were screened by MRI. Patients who had been diagnosed with HCV, alcoholic liver disease, biliary cirrhosis, and intrahepatic cholangiocarcinoma were excluded from this study. A blood sample of 5 mL was obtained from each patient. All experimental procedures were approved by the Institutional Ethics Committee of the Beijing YouAn Hospital and informed consent was obtained from each patient before collection. Blood specimens were centrifuged at 3500 rpm/min for 30 min after clotting. The specimens were then sub-packed, coded, and stored at −80℃. (To avoid the degradation of proteins, each aliquot should not be thawed for five times before use.)

Antibodies and reagents

Rabbit horseradish peroxidase (HRP)-labeled polyclonal AFP antibody and mouse monoclonal AFP antibody were purchased from Fapon Biotech Inc. (Shenzhen, China). HRP conjugation kits were purchased from Abcam Trading Company Ltd. (Shanghai, China). Mouse monoclonal DCP antibody was purchased from Fujirebio Inc. (Tokyo, Japan). Prothrombin factor II antibody was purchased from Fitzgerald Industries International, Inc. (North Acton, MA, USA). Mouse monoclonal to human glypican 3 antibody was obtained from R&D Bio-techne China Co. Ltd. (Shanghai, China). Bovine serum albumin (BSA) and normal goat serum were purchased from Abcam Company Ltd (Cambridge, MA, USA). The Hotgen Biotech glycosyl capture spin column was supplied by Beijing Hotgen Biological Technology Co., Ltd. (Beijing, China). Chemiluminescent HRP substrate RapidStep™ ECL was purchased from Merck KgaA Ltd. (Darmstadt, Germany). Aldehyde-coated microscope slides were purchased from Shanghai BaiO Technology Co., Ltd. (Shanghai, China).

Protein microarray fabrication

Approximately 100 pL of mouse monoclonal AFP antibody solution (1.25 mg/mL), mouse monoclonal prothrombin factor II antibody (2.5 mg/mL), mouse monoclonal glypican 3 antibody (1 mg/mL), and 10% BSA were spotted on the surfaces of slides in triplicates at 8℃ with relative humidity of 50% by using the GeSiM Nano-Plotter™ Micropipetting System (Radeberg, Germany), as shown in Figure 1. BSA was spotted as a negative control. The detailed process of protein microarray making was carried out according to the manufacturer’s instructions and settings. The spotted slides were placed at 4℃ for at least 24h and then blocked with a mixed buffer (10% normal goat serum with 0.1% sodium azide) at 37℃ for 1.5h. Then all slides were briefly rinsed with 1 × PBS with 0.1% Tween-20 (PBST0.1) three times for 3 min each. The slides were dried by centrifugation in a plastic tube and then placed at 4℃ ready for use.

Figure 1.

The workflow of serum alpha1,6-fucosylated protein analysis using protein microarray.

Enrichment of serum alpha1,6-fucosylated proteins

Serum alpha1,6-fucosylated proteins, including fuc-AFP, fuc-DCP, and fuc-GPC3, were enriched from serum specimens by using the Hotgen Biotech glycosyl capture spin column. This has been commonly used to detect glycosyl proteins in Chinese hospitals. The fundamental principle of the Hotgen Biotech glycosyl capture spin column is that, Lens culinaris agglutinin (LCA) is preassembled in the kit to catch the fucosylated proteins, and then monosaccharide is used to compete with the captured fucosylated proteins for binding to LCA. Thus, the serum fucosylated proteins, including fuc-AFP, fuc-DCP, and fuc-GPC3, are finally eluted out of the spin column.

Processing of antibody arrays

All serum elutions were measured by prepared protein microarrays. For the protein microarray measuring, 15 uL serum elution was added to each microscope slide and incubated at 37 for 30 min. The slides were rinsed with PBST0.1 three times for 3 min each. Then 15 uL of the mixed HRP-labeled detection antibodies, including 1 ug/mL HRP-labeled polyclonal AFP antibody, 20 ug/mL HRP-labeled polyclonal DCP antibody, and 20 ug/mL HRP-labeled monoclonal GPC3 antibody, was added to each microscope, and incubated at 37°C for 30 min again, as summarized in Figure 1. The slides were rinsed again as above. Chemiluminescent HRP substrate was added to each microscope and then the slides were scanned by a chemiluminescent scanner, Chemi Doc^TM MP System (Bio-Rad, CA, USA). Image Lab software Version 5 was employed to quantify the data.

Statistical analysis

GraphPad Software version 7 (GraphPad Software, San Diego, CA, USA) and SPSS version 22.0 (IBM, Armonk, NY, USA) were used to analysis the data. The test of normal distribution was conducted for all data. The Mann–Whitney test for non-normally distributed data, and the student t test for normally distributed data were applied to compare the difference between the HCC group and the BLD group. CDA was performed to distinguish the HCC group and the BLD group, and the classification results were compared with the clinicopathological diagnosis. A leave-one-out cross validation was carried out to evaluate the classificatory power of CDA. The sensitivity, specificity, and accuracy were also calculated. A P-value <0.05 was considered to be statistically significant in all instances.

Results

In this study, 51 patients in the HCC group and 47 patients in the BLD group were included. The BLD group contained 18 patients with liver cirrhosis and 29 patients with chronic hepatitis. The clinical features of all patients are summarized in Table 1. The HCC group samples show higher levels of fuc-AFP, fuc-DCP, and fuc-GPC3 than the BLD group samples (Figure 2A). The levels of fuc-AFP and fuc-DCP were significantly higher in the HCC group than in the BLD group. However, there was no statistical difference in the value of fuc-GPC3 between the HCC group and the BLD group (Figure 2B). Furthermore, the association relationship of fuc-AFP, fuc-DCP, and fuc-GPC3 in all patients was analyzed (Figure 3). The Pearson correlation coefficient of fuc-AFP and fuc-DCP was 0.528 (P <0.001). In addition, the Pearson correlation coefficient of fuc-AFP and fucGPC3 was 0.332 (P =0.001). However, the Pearson correlation coefficient of fuc-DCP and fuc-GPC3 was 0.659 (P <0.001), which showed a relatively high correlation.

Table 1.

Summary of patients’ demographics and cancer characteristics.

Clinical features	HCC (No. of patients)	BLD (No. of patients)
Total number	51	47
Male	24	25
Female	27	22
Age, year	53±9	50±12
Stage
Pre-cancer	0	47
Ⅰ	24	0
Ⅱ	8	0
Ⅲ	12	0
Ⅳ	7	0
Metavir score
F0 F1	9	8
F2	7	2
F3	6	3
F4	29	20
Not available	0	14

HCC: hepatocellular carcinoma; BLD: benign liver disease.

Figure 2.

The variation of fuc-AFP, fuc-DCP and fuc-GPC3 in HCC group and BLD group sera. (a) Representative array images of the fuc-AFP, fuc-DCP and fuc-GPC3 in HCC group and BLD group. (b) The levels of fuc-AFP, fuc-DCP and fuc-GPC3 in HCC group and BLD group. HCC: hepatocellular carcinoma; BLD: benign liver disease; fuc-AFP: fucosylated alpha-fetoprotein; fuc-DCP: fucosylated des-γ-carboxy prothrombin; fuc-GPC3: fucosylated glypican 3.

Figure 3.

Correlation analysis of the values of fuc-AFP, fuc-DCP and fuc-GPC3 in all patients. (a) The Pearson correlation coefficient was 0.528, P <0.001. (b) The Pearson correlation coefficient was 0.332, P =0.001. (c) The Pearson correlation coefficient was 0.659, P <0.001. fuc-AFP: fucosylated alpha-fetoprotein; fuc-DCP: fucosylated des-γ-carboxy prothrombin; fuc-GPC3: fucosylated glypican 3.

The chemiluminescent values of fuc-AFP and fuc-DCP, except fuc-GPC3, were automatically selected as independent factors for the linear discriminant function of CDA. The function was listed as:

f (x) = 1.338 χ_{1} - 0.655 χ_{2}

where χ₁ represents the chemiluminescent value of fuc-AFP, and χ₂ represents the chemiluminescent value of fuc-DCP.

The original classification and cross-validation results of CDA established by fuc-AFP and fuc-DCP are summarized in Table 2. Of the 98 patients, 42 cases were categorized as the HCC group and 56 cases were diagnosed as the BLD group. Compared with the clinicopathological diagnosis, the sensitivity and specificity of CDA were 80.4% and 97.9%, respectively, with an accuracy of 88.8%. The receiver operating characteristic curve (ROC) analysis of CDA established by fuc-AFP and fuc-DCP is presented in Figure 4, which shows the high value of the area under the curve (AUC) 0.95, with a 95% confidence interval (CI) (0.901, 0.998).

Table 2.

Protein microarray classification compared with clinicopathological diagnosis of HCC.

Protein microarray	Clinicopathological diagnosis		Sensitivity (%)	Specificity (%)	Accuracy (%)
Protein microarray	HCC	BLD	Sensitivity (%)	Specificity (%)	Accuracy (%)
Original classification
HCC	41	1	80.4	97.9	88.8
BLD	10	46
Cross-validation
HCC	40	1	78.4	97.9	87.8
BLD	11	46

HCC: hepatocellular carcinoma; BLD: benign liver disease.

Figure 4.

ROC analysis of CDA; the value of AUC is 0.95, 95% CI (0.901, 0.998), P<0.001. AUC: area under the curve; CDA: canonical discriminant analysis; CI: confidence interval; ROC: receiver operating characteristic.

Discussion

Protein glycosylation plays an important role in various biological processes, such as cellular differentiation, cell signaling, especially in cancer progression, tumor invasion, and metastatic potentials, which has aroused a great deal of interest in researching its alterations in various tumor transformations in the scientific community.^16,24 Crucial glycosyltransferases have been intensively investigated in HCC.¹⁶ The activity of these enzymes was closely related to the development of HCC and is very much increased in the HCC sera or tissues.

Fucosylation is one of the most commonly observed post-translation modifications in relation to types of cancer. Several fucosylated proteins have been identified by using mass spectrometry, such as fibulin-1C, GP73, hemopexin, haptoglobin, and AFP.^18,25,26 Among these, fuc-AFP is the best known marker and is widely used in HCC diagnosis. Related investigations show that the level of alpha1,6-fucosyltransferase is positively correlated with the degree of dedifferentiation in HCC tissues and then with a subsequent increment of fuc-AFP, which means that the fuc-AFP level is closely related to the cancer behavior and prognosis.²⁷ Thus, fuc-AFP is more specific to malignant tumors rather than benign diseases, and can be used to screen and monitor patients at a high risk of progression to HCC.

Currently, DCP, an abnormal prothrombin, has been detected together with AFP for the early diagnosis of HCC in East Asian countries, such as China, South Korea and Japan. DCP has been reported as a more useful biomarker for screening small HCC with a sensitivity of 51.7% and a specificity of 86.7%.⁸ However, some studies also found that the combined detection of AFP and DCP could achieve a higher sensitivity of 78.3% than testing AFP or DCP alone.^28,29 GPC3, a heparin sulfate glycoprotein linked to the cell surface through a glycosylphosphatidylinositol-anchored membrane protein, is absent in hepatitis but is overexpressed in HCC tissues.³⁰ Several studies have reported that GPC3 could be used as a sensitive and specific serological tumor marker for HCC diagnosis.^30,31 However, fuc-DCP and fuc-GPC3 have rarely been researched. Therefore, in the present study, the levels of serum fucosylated proteins, such as fuc-AFP, fuc-DCP, and fuc-GPC3, were tested together and were found to be significantly higher in the HCC group than in the BLD group, indicating their potential utility in the diagnosis of HCC. A single tumor biomarker is not sufficiently sensitive and specific for the diagnosis of HCC. Thus, in this study, fuc-AFP, fuc-DCP, and fuc-GPC3 were detected together with protein microarray and then assessed by the CDA method. Compared with the clinicopathological diagnosis results, the CDA analysis shows a strong discriminant power and the potential to distinguish HCC from benign liver diseases, with a high accuracy of 88.8%. However, the value of fuc-GPC3 was not selected as a factor to establish the above linear discriminant equation for optimal discriminant performance. The reason for this might be that there was no significantly statistical difference in fuc-GPC3 between the HCC and the BLD group. Some studies have investigated the combined diagnosis value of AFP, fuc-AFP, and DCP in HCC with a sensitivity of 40.51% and a specificity of 96.7%.²⁹ Our study shows a higher sensitivity and specificity than those studies, indicating that the level of fucosylated proteins may be more helpful in diagnosing HCC.

One of the limitations of this study was the relatively small number of serum specimens included. Thus, in subsequent research, our most urgent task is to include more patients from multicenter hospitals in the study. We believe that the protein microarray combined CDA method will then be more powerful. Another aspect in future research is that more emphasis should be placed on including more HCC patients at an early stage, with the aim of analyzing the early diagnostic value of fucosylated proteins.

Conclusion

This study has demonstrated that fucosylated proteins, including fuc-AFP, fuc-DCP, and fuc-GPC3, could be useful in the diagnosis of HCC. Our results also demonstrated that protein microarray combined with CDA could be an accurate and cost-effective method in HCC diagnosis with a high sensitivity and specificity.

Footnotes

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study is funded by Beijing Municipal Science & Technology Commission (D171100003117004).

References

Giunchi

Vasuri

Fiorentino

. Epidemiology of hepatocelluar carcinoma. In: Loda

Mucci

Mittelstadt

et al . (eds). Pathology and Epidemiology of Cancer. Cham, Switzerland: Springer, 2017, pp.447–454.

Levrero

Zucman-Rossi

Mechanisms of HBV-induced hepatocellular carcinoma. J Hepatol 2016; 64: S84.

Nagaratnam

Cheuk

Hepatocellular carcinoma. In: Nagaratnam

Nagaratnam

Cheuk

(eds) Geriatric Diseases. Cham, Switzerland: Springer, 2017, pp.1–4.

Peipei

Jianjun

Yoshinori

et al . Biomarkers: Evaluation of screening for and early diagnosis of hepatocellular carcinoma in Japan and China. Liver Cancer 2013; 2: 31–39.

Kai

Wei

Zhang

et al . GP73 N-glycosylation at Asn144 reduces hepatocellular carcinoma cell motility and invasiveness. Oncotarget 2016; 7: 23530.

Zhao

Wang

Cui

et al . Significance of combined tests of serum golgi glycoprotein 73 and other biomarkers in diagnosis of small primary hepatocellular carcinoma. Cancer Biomark 2015; 15: 677.

Libbrecht

Severi

Cassiman

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

Zakhary

Khodeer

Shafik

et al . Impact of PIVKA-II in diagnosis of hepatocellular carcinoma. J Adv Res 2013; 4: 539–546.

Haab

BB.

Antibody-lectin sandwich arrays for biomarker and glycobiology studies. Expert Rev Proteomics 2010; 7: 9.

10.

Haab

Partyka

Cao

Using antibody arrays to measure protein abundance and glycosylation: considerations for optimal performance. Curr Protoc Protein Sci 2013; 73: Unit 27.26.

11.

Huang

Zhu

Protein array-based approaches for biomarker discovery in cancer. Genomics Proteomics Bioinformatics 2017; 15: 73–81.

12.

Zhao

Zhang

Lin

et al . Protein microarray with horseradish peroxidase chemiluminescence for quantification of serum α-fetoprotein. J Int Med Res 2015; 43: 35–40.

13.

Yao

Zhang

Chen

et al . Diagnostic value of immunohistochemical staining of GP73, GPC3, DCP, CD34, CD31, and reticulin staining in hepatocellular carcinoma. J Histochem Cytochem 2013; 61: 639–648.

14.

Hiraoka

Ishimaru

Kawasaki

et al . Tumor markers AFP, AFP-L3, and DCP in hepatocellular carcinoma refractory to transcatheter arterial chemoembolization. Oncology 2015; 89. 10.1159/000381808.

15.

Yamamoto

Imamura

Matsuyama

et al . AFP, AFP-L3, DCP, and GP73 as markers for monitoring treatment response and recurrence and as surrogate markers of clinicopathological variables of HCC. J Gastroenterol 2010; 45: 1272–1282.

16.

Blomme

Van

Callewaert

et al . Alteration of protein glycosylation in liver diseases. J Hepatol 2009; 50: 592–603.

17.

Chen

Laroche

Hamelinck

et al . Multiplexed analysis of glycan variation on native proteins captured by antibody microarrays. Nat Methods 2007; 4: 437–444.

18.

Dai

Liu

Y-K

Dr Cui

et al . Identification and analysis of altered α1,6-fucosylated glycoproteins associated with hepatocellular carcinoma metastasis. Proteomics 2006; 6: 5857–5867.

19.

Miyoshi

Moriwaki

Terao

et al . Fucosylation is a promising target for cancer diagnosis and therapy. Biomolecules 2012; 2: 34–45.

20.

Muinelo-Romay

Villar-Portela

F Poceiro

et al . Identification of alpha(1,6)fucosylated proteins as biomarkers for human colorectal carcinoma. In: Congress of the Federation-Of-European-Biochemical-Societies 2009, pp.102–102.

21.

Asazawa

Kamada

Takeda

et al . Serum fucosylated haptoglobin in chronic liver diseases as a potential biomarker of hepatocellular carcinoma development. Clin Chem Lab Med 2015; 53: 95–102.

22.

Miyoshi

Nakano

Fucosylated haptoglobin is a novel marker for pancreatic cancer: Detailed analyses of oligosaccharide structures. Proteomics 2008; 8: 3257–3262.

23.

Shang

Qin

et al . Aided diagnosis of hepatocellular carcinoma using serum fucosylated haptoglobin ratios. J Cancer 2017; 8: 887.

24.

Mehta

Herrera

Block

Chapter Seven – glycosylation and liver cancer. Adv Cancer Res 2015; 126: 257–279.

25.

Dai

Fan

Liu

et al . Identification and analysis of alpha1,6-fucosylated proteins in human normal liver tissues by a target glycoproteomic approach. Electrophoresis 2007; 28: 4382.

26.

Kobayashi

Nouso

Kinugasa

et al . Clinical utility of serum fucosylated-hemopexin in Japanese patients with hepatocellular carcinoma. Hepatol Res 2012; 42: 1187–1195.

27.

Aoyagi

Mita

Suda

et al . The fucosylation index of serum alpha-fetoprotein as useful prognostic factor in patients with hepatocellular carcinoma in special reference to chronological changes. Hepatol Res 2002; 23: 287–295.

28.

Morota

Nakagawa

Sekiya

et al . A comparative evaluation of Golgi protein-73, fucosylated hemopexin, α-fetoprotein, and PIVKA-II in the serum of patients with chronic hepatitis, cirrhosis, and hepatocellular carcinoma. Clin Chem Lab Med 2011; 49: 711–718.

29.

Gao

Song

Combination of triple biomarkers AFP, AFP-L3, and PIVAKII for early detection of hepatocellular carcinoma in China: Expectation. Drug Discov Ther 2017; 11: 168–169.

30.

Liu

Wang

Zhang

et al . Expression of glypican 3 enriches hepatocellular carcinoma development-related genes and associates with carcinogenesis in cirrhotic livers. Carcinogenesis 2015; 36: 232–242.

31.

Shirakawa

Kuronuma

Nishimura

et al . Glypican-3 is a useful diagnostic marker for a component of hepatocellular carcinoma in human liver cancer. Int J Oncol 2009; 34: 649–656.