- Email: [email protected]
Glypican-3 expression in hepatocellular tumors: diagnostic value for preneoplastic lesions and hepatocellular carcinomas
Human Pathology (2006) 37, 1435 – 1441
www.elsevier.com/locate/humpath
Original contribution
Glypican-3 expression in hepatocellular tumors: diagnostic value for preneoplastic lesions and hepatocellular carcinomas Xiao Ying Wang MDa, Franc¸oise Degos MDb, Sylvie Dubois MDa, Sandrine Tessiore MDa, Mark Allegretta MDc,d, Ronald D. Guttmann MDc, Serge Jothy MDe, Jacques Belghiti MDf, Pierre Bedossa MDa,g,*, Vale´rie Paradis MDa,g a
Service d’Anatomie Pathologique Hoˆpital Beaujon, 92110 Clichy, France Service d’He´patologie Hoˆpital Beaujon, 92110 Clichy, France c BioMosaics Inc. Burlington, VT 05405, USA d Department of Pathology, University of Vermont, Burlington, VT 05405, USA e Department of Laboratory Medicine St Michael’s Hospital, Toronto, Canada M5B 1W8 f Service de Chirurgie He´patique Hoˆpital Beaujon, 92110 Clichy, France g CNRS UMR 8149 Faculte´ de Pharmacie, 75006 Paris V, France b
Received 18 February 2006; revised 2 May 2006; accepted 9 May 2006
Keywords: Hepatocellular carcinoma; Cirrhosis; Macronodules; Glypican-3
Summary Glypican-3 (GPC3), a member of heparan sulfate proteoglycans, plays a role in cell growth, differentiation, and migration. The objectives of this study were to assess the diagnostic value of GPC3 immunostaining in hepatocellular carcinomas (HCCs) and to analyze its expression profile in preneoplastic lesions. Tissue microarrays were built by sampling 54 HCCs and adjacent liver tissues (21 developing from cirrhosis and 33 from normal liver) and 94 cirrhotic macronodules. Fourteen typical liver cell adenomas and 5 with malignant foci were also included. Sections were assessed for GPC3 expression by immunohistochemistry. GPC3 staining was observed in 19 (90%) of 21 HCC cases with cirrhosis and in 18 (64%) of 28 HCC cases with normal liver ( P b .01). When staining was positive, it was both membranous and cytoplasmic. Positive staining was observed in 1 case of nonneoplastic adjacent liver. In cases of adenomas, only malignant foci were positive. Among the 94 macronodules, GPC3 immunostaining was noted in 48% (14/29) of high-grade dysplastic or early HCC and in 3% (2/65, P b .001) of benign or low-grade dysplastic macronodules. This study shows that GPC3 is an efficient diagnostic marker of HCC, potentially useful in the differential diagnosis of liver cell adenomas and well-differentiated HCC. Our results also suggest that GPC3 may be considered as an early marker of liver carcinogenesis because it is able to identify some cirrhotic macronodules with malignant potential. D 2006 Elsevier Inc. All rights reserved.
* Corresponding author. Service d’Anatomie Pathologique Hoˆpital Beaujon, Clichy, France. E-mail address: [email protected] (P. Bedossa). 0046-8177/$ – see front matter D 2006 Elsevier Inc. All rights reserved. doi:10.1016/j.humpath.2006.05.016
1436
X. Y. Wang et al.
1. Introduction Glypican-3 (GPC3) is a member of the glypican family, a group of heparan sulfate proteoglycans linked to the cell surface through a glycosylphosphatidylinositol anchor [1]. Glypicans play an important role in cell growth, differentiation, and migration [2,3]. Although down-regulation of this protein has been shown in several cancers and cell lines, including mesothelioma and ovarian and breast cancers, GPC3 is highly expressed, both at the mRNA and protein level, in hepatocellular carcinomas (HCCs) [4-9]. The timing of GPC3 expression in liver carcinogenesis is unknown. Indeed, most HCCs arise in liver cirrhosis through progressive malignant transformation of cirrhotic nodules. Among them, large parenchymal nodules, also called macronodules, represent an early step in the carcinogenesis pathways of HCC [10-13]. According to histopathology, macronodules are categorized as benign, dysplastic (low or high grade), and malignant [14]. Although grading of macronodules has been described in detail, classification of borderline macronodules is sometimes challenging for pathologists, supporting the search for additional features able to identify macronodules with high risk of malignant degeneration. To address this issue, we and others previously showed that benign and dysplastic macronodules could be differentiated from malignant nodules by gene expression analysis [15,16]. Although such a molecular approach can help to gain further insight into the biologic behavior of macronodules, it cannot be used as a diagnostic tool in routine practice. Therefore, reliable tissue markers able to identify cirrhotic macronodules with malignant potential would be potentially helpful, and GPC3 may be one of these. In addition to HCC arising in cirrhotic livers, some such tumors develop in normal livers by malignant progression of monoclonal benign tumors such as liver cell adenomas [17,18]. Diagnosis of malignancy within an adenoma is of significant importance in patient follow-up but is difficult to achieve. Imaging techniques are of little use in this context, and pathologists may have some difficulty in reaching a conclusion as to malignant transformation of adenomas because these malignant foci are usually well-differentiated HCC. In this context again, GPC3 might also be useful. Table 1
Finally, mixed liver tumors showing dual hepatocellular and biliary epithelial differentiations within the same tumor are referred to as hepatocholangiocarcinomas (HCCC) [19]. Although such a diagnosis is difficult to attempt, especially in the light of the hepatocellular component, we sought to determine whether GPC3 staining might help because cholangiocarcinomas (CCCs) appear to be GPC3-negative [20]. Therefore, the objective of the present study was to evaluate the diagnostic performance of GPC3 in epithelial tumors and preneoplastic conditions of the liver. To conduct this analysis, we used an immunohistochemical approach that encompasses a large number of tumors, including HCC, HCCC, cirrhotic macronodules, and typical and atypical liver cell adenomas. We confirmed the high sensitivity of GPC3 as a diagnostic marker not only in overt HCC but also in preneoplastic lesions and in the differential diagnosis of primary liver tumors. Finally, these data suggest that this marker may be very helpful in improving early diagnosis of HCC in routine practice.
2. Materials and methods 2.1. Tissue samples Liver tissue samples were retrieved from the files of the Department of Pathology, Beaujon Hospital, Clichy, France. Malignant tumor samples obtained from liver resections included 54 HCC, 9 HCCC, and 10 cases of CCC. Among HCC, 21 occurred in cirrhotic liver (16 related to viral hepatitis infection and 5 to chronic alcohol consumption), and 28 were present in normal liver. Five HCC were of a fibrolamellar type on a background of normal liver. Preneoplastic liver lesions consisted of 94 cirrhotic macronodules, which were selected from 27 additional cirrhosis cases (related to hepatitis infection in 10 cases, chronic alcohol consumption in 16 cases, and autoimmune disease in 1 case). Cirrhotic macronodules were diagnosed as benign, low-grade dysplastic, high-grade dysplastic, and early HCC according to the terminology of the international working party [21]. The histologic criteria were the following: clone-like populations, nuclear hyperchromasia, irregularity of nucleus, pseudogland formation, increase in
Clinical data on patients and liver tumors
Tumor type (n)
Sex (M/F)
Mean age (y) (range)
Mean tumor size (cm) (range)
Tumor differentiation (well/moderate/poor)
HCC in cirrhosis (21) HCC in normal liver (28) Fibrolamellar HCC (5) HCCC (9) Cholangiocarcinomas (10) Cirrhotic macronodules (94) Liver cell adenomas (14) Atypical adenomas (5)
19/2 23/5 2/3 8/1 5/5 22/5 0/14 4/1
56 54 27 61 59 56 35 48
4.9 10.3 15 5.6 8 – 9 12
0/14/7 17/11/0 – 0/5/4 2/8/0 – – 5/0/0
(33-74) (23-72) (21-38) (43-76) (48-75) (38-63) (23-52) (36-63)
(1.5-8) (2.3-22) (7-22) (2.7-12) (2-14) (3.3-16) (8-16)
Glypican-3 expression in hepatocellular tumors
1437
cell density, thickness of plates, reticulin framework, and stromal invasion. Of the 27 cirrhosis cases with macronodules, 8 also had an HCC. Fourteen typical liver cell adenomas and 5 with malignant foci (called batypical adenomasQ), which occurred in normal liver, were also studied. Ten cases of CCCs and 9 cases of HCCCs were retrieved from our pathologic files. All cases were obtained from surgical specimens. All specimens were formalin-fixed and embedded in paraffin. Tissue samples were handled for routine histologic diagnosis. The HCC cases (n = 54), cirrhotic macronodules (n = 94), liver cell adenomas (n = 14), and atypical adenomas (n = 5) were selected to build tissue microarrays (TMAs). The primary clinical data of patients and liver tumors are reported in Table 1.
2.2. TMA construction Representative blocks of tumors and adjacent nontumorous liver, except for HCCC and CCC, were selected for GPC3 immunostaining. Areas of interest were identified and marked on corresponding hematoxylin and eosin
Fig. 2 GPC3 expression in HCCCs. HCCC with areas of trabecular hepatocellular carcinoma (white arrow) and glandular cholangiocarcinoma (black arrows). Positive staining was restricted to the area of hepatocellular differentiation. There was no labeling in glandular carcinoma areas.
(H&E)–stained slides. TMAs were constructed by sampling 3 cores for tumor samples and 2 cores for adjacent liver. Tissue cores of 1 mm in diameter were arrayed into a recipient paraffin block using a tissue microarrayer (MTA-1, Beecher Instruments, Inc, Sun Prairie, WI). Sections 3-lm thick were cut from completed array blocks and transferred to silanized glass slides. Sections from these arrays were then stained with H&E to assess adequacy of the sampling. Microarrays and whole sections (for HCCC and CCC) were immunostained using the same technique.
2.3. Immunohistochemistry
A
B
Immunohistochemical study was performed using an automated immunohistochemical stainer according to the manufacturer’s guidelines (streptavidin-peroxidase protocol; BenchMark, Ventana, Tucson, AZ). Sections measuring 3-lm were stained with monoclonal anti-GPC3 (1G12, 1:10, BioMosaics Inc, Burlington, VT). Sections were then evaluated by the same pathologist who was blinded to clinical information on patients. A case was considered negative in TMA if all 3 spots were negative. Positive staining was defined as membranous and/or cytoplasmic expression. GPC3 labeling was considered as positive when at least 10% of cells were positive. Uninterpretable results (because of loss of the tissue) were eliminated from further consideration. To investigate the usefulness of GPC3 as an immunohistochemical marker of HCC, serial sections of TMAs were further stained with primary antibody to a fetoprotein (AFP) (1/2000, Dako, Glostrup, Denmark). Similar to GPC3, AFP was considered positive when at least 10% of cells were stained.
2.4. Data analysis and statistics Fig. 1 GPC3 expression in HCC. A, Membranous and cytoplasmic staining of liver tumor cells. B, Hepatocellular carcinoma displaying prominent canalicular immunostaining.
Distribution between groups was compared using v 2 or Fisher exact test for qualitative data and the Student t test for
1438
X. Y. Wang et al.
Fig. 3 GPC3 and AFP expression in cirrhotic macronodules: low-grade dysplastic macronodule (A, GPC3-negative; B, AFP-negative) and high-grade dysplastic macronodule (C, GPC3-positive; D, AFP-negative).
quantitative data. A P value of .05 was chosen as the significance level.
3. Results 3.1. GPC3 expression in primary liver malignant tumors In the first step, we studied the immunohistochemical expression of GPC3 in HCC. Among 54 cases of HCC, 38 (70%) expressed GPC3 with both membranous and diffuse cytoplasmic staining (Fig. 1A). In a few cases, GPC3 immunostaining was granular and prominent adjacent to canaliculi (Fig. 1B). GPC3 staining was significantly more often observed in HCC from cirrhotic livers (19 [90%] of 21 cases) compared with HCC in normal liver (18 [64%] of 28 cases, P b .01). Among the 5 fibrolamellar HCCs present in normal liver, only 1 case (20%) displayed significant positive staining. In the group of HCC in cirrhotic livers, GPC3 expression was expressed in 13 (93%) of 14 moderately differentiated and in 6 (86%) of 7 poorly
differentiated HCC ( P = nonsignificant [NS]). Size of the tumor and any other clinicopathologic factors (etiology of cirrhosis, age of the patient, presence of vascular invasion) were not significantly related to GPC3 expression. In the group of HCCs occurring in normal liver, GPC3 was expressed in 9 (53%) of 17 well-differentiated and in 9 (82%) of 11 moderately differentiated HCCs ( P = .05). Nontumor tissue, including 21 cases of cirrhosis and 31 cases of normal liver (loss of tissue in 2 cases) obtained at a distance from the HCC, was consistently negative, except in 1 case of cirrhosis. Immunostaining of HCCC revealed GPC3 expression in 7 (78%) of 9 cases, with positive staining restricted to the area of HCC differentiation. There was no labeling in glandular carcinoma areas (Fig. 2). Regarding CCC tumors, 9 of 10 cases displayed no significant GPC3 expression.
3.2. GPC3 expression in cirrhotic preneoplastic liver lesions Because GPC3 appears to be a relevant sensitive marker of fully developed HCC, we wondered whether it could be
Glypican-3 expression in hepatocellular tumors
1439
3.3. GPC3 expression in liver cell adenomas To evaluate the specificity of GPC3 expression as a marker of malignant transformation, we collected a series of liver cell adenomas. None of the 14 typical adenomas expressed GPC3, nor did the adjacent normal liver tissue. Interestingly, among the 5 adenomas with histopathological features suggestive of malignancy, 3 (60%) of them displayed significant GPC3 staining in the malignant areas (Fig. 5). Adjacent adenomas were always negative.
3.4. Comparison of GPC3 with AFP expression in hepatocellular tumors and preneoplastic lesions HCC tissues from cirrhotic livers were less frequently stained with AFP than with GPC3 (AFP: 12 (57%) of 21 cases; GPC3: 19 (90%) 21 cases; P b .05). HCC in normal liver tissues were also less frequently stained with AFP than GPC3, but the difference was not significant (AFP: 12 [42%] of 28 cases; GPC3: 18 [64%] of 28 cases, P = NS). No cases of adjacent nontumor liver tissue, typical adenomas, or HCC arising from preexisting adenomas were positive. For the cirrhotic macronodules, 5 cases of benign or low-grade dysplastic macronodules (7% [5/65]) and 5 (17%) of the 29 high-grade dysplastic and early HCC were AFP-positive. This was significantly less than GPC3 (14 [48%] of 29 cases; P b .001) Results are shown in Table 2.
4. Discussion Fig. 4 Early hepatocellular carcinoma. A, Presence of a malignant area (arrow) in a large regenerative macronodule (H&E staining). B, The malignant area displayed positive GPC3 imunostaining.
viewed also as an early marker of liver carcinogenesis. To address this issue, we analyzed 94 cirrhotic macronodules with various grades of dysplasia from 27 cirrhotic specimens, 8 of them with overt HCC. Sixty-five nodules were benign or had low-grade dysplasia, whereas 9 had highgrade dysplasia and 20 were early HCC. In the group of benign and low-grade dysplastic macronodules, only 2 lowgrade dysplastic macronodules (3%) displayed significant GPC3 expression. Among the high-grade dysplastic and early HCC, 14 (48%) of 29 of them were immunostained with GPC3 ( P b .001). In addition, GPC3 staining was significantly more often observed in early HCC (60% [12/ 20]) than in high-grade dysplastic macronodules (22% [2/9], P b .01). Adjacent cirrhotic tissue obtained from the same specimen was consistently GPC3-negative. Representative GPC3 and AFP immunostaining are illustrated in low(Fig. 3A and B) and in high-grade (Fig. 3C and D) dysplastic macronodules. Fig. 4 illustrates a GPC3 positive malignant foci in a regenerative macronodule.
Diagnosis of malignant hepatocellular liver tumors can be a challenge for pathologists and points out the need to develop relevant sensitive markers of malignant nodules. In this study, we showed that GPC3, a member of the heparan sulfate proteoglycan family, is a sensitive and specific immunohistochemical marker of HCC. More interestingly, we demonstrated that GPC3 is also an early marker of HCC because a significant proportion of highgrade dysplastic macronodules and early HCC already displayed GPC3 expression. Similar to AFP, GPC3 potentially behaves as an oncofetal liver protein; overexpression of GPC3 is detected in fetal liver, whereas it is absent in normal adult liver [9]. Reactivation of the fetal phenotype, which is common in malignant tumors, may explain the expression of GPC3 in malignant hepatocellular nodules. Up until now, AFP was regarded as the most useful marker of HCC, although its sensitivity is limited. By comparing AFP and GPC3 immunostaining in the different groups of lesions, our data confirm that GPC3 is more sensitive, especially in the group of HCC, which developed from cirrhosis. It is of note that GPC3 immunostaining appeared to be less frequent in HCC arising in normal liver, compared with HCC associated with cirrhosis. However, when the degree of differentiation was taken into account, the difference was not significant. Moreover, although no AFP immunostaining was
1440
X. Y. Wang et al.
Fig. 5 Liver cell adenoma. A, Microscopic view of a typical liver cell adenoma (H&E staining). B, No staining for GPC3. C, Adenoma with histopathological features suggestive of malignancy with enlarged hepatocellular plates and pseudoglandular formations (H&E staining). D, GPC3-positive staining in the malignant area.
observed in high-grade dysplastic macronodules and early HCC, a higher proportion of such macronodules displayed significant GPC3 positivity. These results are consistent with those reported by Yamauchi et al [9] and confirmed the performance of GPC3 as an early marker of malignant transformation to HCC. Regarding the multistep process of carcinogenesis in the setting of chronic liver diseases, it is noteworthy that almost all cases of benign and low-grade dysplastic macronodules were GPC3-negative. This observation supports previous molecular studies demonstrating that benign and low-grade dysplastic macronodules undoubtedly share similar molecular behavior [15]. As far as the GPC3 expression is concerned, we observed that early HCC are significantly more often positive than high-grade dysplastic macronodules (60% versus 20%). Classification of high-grade dysplastic macronodules relies on several elementary morphological lesions, including presence of unpaired arteries, pseudoglandular formations, and cytologic atypias [14,21]. Because clear delineation of these events are difficult for interpretation by the pathologist, GPC3
immunostaining will provide additional criteria for characterizing those macronodules with malignant foci in cirrhotic liver tissue. Table 2 tumors
Expression of GPC3 and AFP in hepatocellular
Tumor type (n)
GPC3
AFP
P
HCC in cirrhosis (21) HCC in normal liver (28) Fibrolamellar HCC (5) HCCC (9) Cholangiocarcinomas (10) Cirrhotic macronodules (94) Regenerative benign (42) Low-grade dysplastic (23) High-grade dysplastic (9) Early HCC (20) Liver cell adenomas (14) Atypical adenomas (5)
19 18 1 7 1 16 0 2 2 12 0 3
12 12 1 0 0 10 3 2 1 4 0 0
.05 NS NS .001 NS .05 NS NS .001 .001 NS .001
(90%) (64%) (20%) (78%) (10%) (17%) (8%) (22%) (60%) (60%)
(57%) (42%) (20%)
(10%) (7%) (8%) (1%) (20%)
Glypican-3 expression in hepatocellular tumors It has already been demonstrated that benign hepatocellular nodules, including cases of focal nodular hyperplasia and adenoma, are GPC3-negative [9]. Although this marker has no pathological relevance in differentiation between such benign disorders, it could be helpful for identifying cases of adenomas with malignant progression. In previous studies, the risk of malignant progression from adenoma is seen in approximately 10% of cases, with a higher rate in men and in patients with large tumors [17]. However, the accurate diagnosis of such well-differentiated malignant foci is quite difficult when based solely on morphological analysis. Interestingly, immunohistochemical study using GPC3 clearly demonstrated the GPC3 immunostaining of malignant foci within liver cell adenomas. GPC3 is down-regulated in CCC, a pattern that may help to differentiate HCC from CCC [20]. Such a differential expression pattern might also allow the accurate diagnosis of combined HCCC by showing both positive immunostained areas with hepatocellular differentiation and negative foci with biliary differentiation. Interestingly, we observed that among 10 cases of HCCC carefully selected based on morphological analysis, 9 of them displayed GPC3 staining in restricted areas with trabecular hepatocellular differentiation. These foci were not stained with AFP. Our immunohistochemical data provide further evidence of the biologic behavior of such combined tumors, which are clinically and genetically much closer to CCC than to HCC [22]. Finally, there have been many attempts to develop noninvasive markers of HCC to detect early HCC. Because GPC3 is secreted into blood, it is a potentially interesting serum marker of HCC [23-25]. A previous study showed that a GPC3 immunoassay had potential as a diagnostic marker of HCC, with serological sensitivity and specificity of 53% and 95%, respectively [23]. Our study, demonstrating the efficiency of the 1G12 GPC3 monoclonal antibody for immunohistochemistry, is promising and supports the need for further studies to test for circulating GPC3 with an immunoassay of the sera of cirrhotic patients with and without HCC.
References [1] Filmus J, Selleck SB. Glypicans: proteoglycans with a surprise. J Clin Invest 2001;108:497 - 501. [2] Li M, Choo B, Wong ZM, Filmus J, Buick RN. Expression of OCI-5/ glypican3 during intestinal morphogenesis: regulation by cell shape in intestinal epithelial cells. Exp Cell Res 1997;235:3 - 12. [3] Farooq M, Hwang SY, Park MK, Kim JC, Kim MK, Sung YK. Blocking endogenous glypican-3 expression releases Hep 3B from G1 arrest. Mol Cells 2003;3:356 - 60. [4] Lin H, Huber R, Schlessinger D, Morin JP. Frequent silencing of the GPC3 gene in ovarian cancer cell lines. Cancer Res 1999;59:807 - 10.
1441 [5] Murthy SS, Shen T, De Rienzo A, et al. Expression of GPC3, an Xlinked recessive overgrowth gene, is silenced in malignant mesothelioma. Oncogene 2000;19:410 - 6. [6] Xiang YY, Ladeda V, Filmus J. Glypican-3 expression is silenced in human breast cancer. Oncogene 2001;20:7408 - 12. [7] Hsu HC, Cheng W, Lai PL. Cloning and expression of a developmentally regulated transcript MXR7 in hepatocellular carcinoma. Cancer Res 1997;57:5179 - 84. [8] Zhu Z-W, Friess H, Wang L, et al. Enhanced glypican-3 expression differentiates the majority of hepatocellular carcinomas from benign hepatic disorders. Gut 2001;48:558 - 64. [9] Yamauchi N, Watanabe A, Hishinuma M, et al. The glypican 3 oncofetal protein is a promising diagnostic marker for hepatocellular carcinoma. Mod Pathol 2005;18:1591 - 8. [10] Furuya K, Nakamura M, Yamamoto Y, Togei K, Otsuka H. Macroregenerative nodule of the liver: a clinicopathologic study of 345 autopsy cases of chronic liver disease. Cancer 1988;61:99 - 105. [11] Kondo F, Ebera M, Siguira N, et al. Histological features and clinical course of large regenerative nodules: evaluation of their precancerous potentiality. Hepatology 1990;12:592 - 8. [12] Okuda K. Hepatocellular carcinoma: recent progress. J Hepatol 2002;32:225 - 37. [13] Paradis V, Laurendeau I, Vidaud M, Bedossa P. Clonal analysis of macronodules in cirrhosis. Hepatology 1998;28:953 - 8. [14] Ferrell LD, Crawford JM, Dhillon AP, Scheuer PJ, Nakanuma Y. Proposal for standardized criteria for the diagnosis of benign, borderline and malignant hepatocellular lesions arising in chronic advanced liver disease. Am J Surg Pathol 1993;17:1113 - 23. [15] Paradis V, Bieche I, Dargere D, et al. Molecular profiling of hepatocellular carcinomas (HCC) using a large-scale real-time RTPCR approach: determination of a molecular diagnostic index. Am J Pathol 2003;163:733 - 41. [16] Anders RA, Yerian LM, Tretiakova M, et al. cDNA microarray analysis of macroregenerative and dysplastic nodules in end-stage hepatitis C virus–induced cirrhosis. Am J Pathol 2003;162:991 - 1000. [17] Foster JH, Berman MM. The malignant transformation of liver cell adenomas. Arch Surg 1994;129:712 - 7. [18] Ito M, Sasaki M, Wen CY, et al. Liver cell adenoma with malignant transformation: a case report. World J Gastroenterol 2003;9: 2379 - 81. [19] Ng IO, Shek TW, Nicholls J, Ma LT. Combined hepatocellularcholangiocarcinoma: a clinicopathological study. J Gastroenterol Hepatol 1998;13:34 - 40. [20] Man X-B, Tang L, Zhang B-H, Qiu X-H, Wu M-C, Wang H-Y. Upregulation of glypican-3 expression in hepatocellular carcinoma but downregulation in cholangiocarcinoma indicates its differential diagnosis value in primary liver cancers. Liver Int 2005;25:962 - 6. [21] Wanless I, International Working Party. Terminology of nodular hepatocellular lesions. Hepatology 1995;22:983 - 92. [22] Cazals-Hatem D, Rebouissou S, Bioulac-Sage P, et al. Clinical and molecular analysis of combined hepatocellular-cholangiocarcinomas. J Hepatol 2004;41:292 - 8. [23] Capurro M, Wanless I, Sherman M, et al. Glypican-3: a novel serum and histochemical marker for hepatocellular carcinoma. Gastroenterology 2003;125:89 - 97. [24] Nakatsura T, Yoshitake Y, Senju S, et al. Glypican-3, overexpressed specifically in human hepatocellular carcinoma, is a novel tumor marker. Biochem Biophys Res Commun 2003;306:16 - 25. [25] Hippo Y, Watanabe K, Watanabe A. Identification of soluble NH2terminal fragment of glypican-3 as a serological marker for early-stage hepatocellular carcinoma. Cancer Res 2004;64:2418 - 23.
www.elsevier.com/locate/humpath
Original contribution
Glypican-3 expression in hepatocellular tumors: diagnostic value for preneoplastic lesions and hepatocellular carcinomas Xiao Ying Wang MDa, Franc¸oise Degos MDb, Sylvie Dubois MDa, Sandrine Tessiore MDa, Mark Allegretta MDc,d, Ronald D. Guttmann MDc, Serge Jothy MDe, Jacques Belghiti MDf, Pierre Bedossa MDa,g,*, Vale´rie Paradis MDa,g a
Service d’Anatomie Pathologique Hoˆpital Beaujon, 92110 Clichy, France Service d’He´patologie Hoˆpital Beaujon, 92110 Clichy, France c BioMosaics Inc. Burlington, VT 05405, USA d Department of Pathology, University of Vermont, Burlington, VT 05405, USA e Department of Laboratory Medicine St Michael’s Hospital, Toronto, Canada M5B 1W8 f Service de Chirurgie He´patique Hoˆpital Beaujon, 92110 Clichy, France g CNRS UMR 8149 Faculte´ de Pharmacie, 75006 Paris V, France b
Received 18 February 2006; revised 2 May 2006; accepted 9 May 2006
Keywords: Hepatocellular carcinoma; Cirrhosis; Macronodules; Glypican-3
Summary Glypican-3 (GPC3), a member of heparan sulfate proteoglycans, plays a role in cell growth, differentiation, and migration. The objectives of this study were to assess the diagnostic value of GPC3 immunostaining in hepatocellular carcinomas (HCCs) and to analyze its expression profile in preneoplastic lesions. Tissue microarrays were built by sampling 54 HCCs and adjacent liver tissues (21 developing from cirrhosis and 33 from normal liver) and 94 cirrhotic macronodules. Fourteen typical liver cell adenomas and 5 with malignant foci were also included. Sections were assessed for GPC3 expression by immunohistochemistry. GPC3 staining was observed in 19 (90%) of 21 HCC cases with cirrhosis and in 18 (64%) of 28 HCC cases with normal liver ( P b .01). When staining was positive, it was both membranous and cytoplasmic. Positive staining was observed in 1 case of nonneoplastic adjacent liver. In cases of adenomas, only malignant foci were positive. Among the 94 macronodules, GPC3 immunostaining was noted in 48% (14/29) of high-grade dysplastic or early HCC and in 3% (2/65, P b .001) of benign or low-grade dysplastic macronodules. This study shows that GPC3 is an efficient diagnostic marker of HCC, potentially useful in the differential diagnosis of liver cell adenomas and well-differentiated HCC. Our results also suggest that GPC3 may be considered as an early marker of liver carcinogenesis because it is able to identify some cirrhotic macronodules with malignant potential. D 2006 Elsevier Inc. All rights reserved.
* Corresponding author. Service d’Anatomie Pathologique Hoˆpital Beaujon, Clichy, France. E-mail address: [email protected] (P. Bedossa). 0046-8177/$ – see front matter D 2006 Elsevier Inc. All rights reserved. doi:10.1016/j.humpath.2006.05.016
1436
X. Y. Wang et al.
1. Introduction Glypican-3 (GPC3) is a member of the glypican family, a group of heparan sulfate proteoglycans linked to the cell surface through a glycosylphosphatidylinositol anchor [1]. Glypicans play an important role in cell growth, differentiation, and migration [2,3]. Although down-regulation of this protein has been shown in several cancers and cell lines, including mesothelioma and ovarian and breast cancers, GPC3 is highly expressed, both at the mRNA and protein level, in hepatocellular carcinomas (HCCs) [4-9]. The timing of GPC3 expression in liver carcinogenesis is unknown. Indeed, most HCCs arise in liver cirrhosis through progressive malignant transformation of cirrhotic nodules. Among them, large parenchymal nodules, also called macronodules, represent an early step in the carcinogenesis pathways of HCC [10-13]. According to histopathology, macronodules are categorized as benign, dysplastic (low or high grade), and malignant [14]. Although grading of macronodules has been described in detail, classification of borderline macronodules is sometimes challenging for pathologists, supporting the search for additional features able to identify macronodules with high risk of malignant degeneration. To address this issue, we and others previously showed that benign and dysplastic macronodules could be differentiated from malignant nodules by gene expression analysis [15,16]. Although such a molecular approach can help to gain further insight into the biologic behavior of macronodules, it cannot be used as a diagnostic tool in routine practice. Therefore, reliable tissue markers able to identify cirrhotic macronodules with malignant potential would be potentially helpful, and GPC3 may be one of these. In addition to HCC arising in cirrhotic livers, some such tumors develop in normal livers by malignant progression of monoclonal benign tumors such as liver cell adenomas [17,18]. Diagnosis of malignancy within an adenoma is of significant importance in patient follow-up but is difficult to achieve. Imaging techniques are of little use in this context, and pathologists may have some difficulty in reaching a conclusion as to malignant transformation of adenomas because these malignant foci are usually well-differentiated HCC. In this context again, GPC3 might also be useful. Table 1
Finally, mixed liver tumors showing dual hepatocellular and biliary epithelial differentiations within the same tumor are referred to as hepatocholangiocarcinomas (HCCC) [19]. Although such a diagnosis is difficult to attempt, especially in the light of the hepatocellular component, we sought to determine whether GPC3 staining might help because cholangiocarcinomas (CCCs) appear to be GPC3-negative [20]. Therefore, the objective of the present study was to evaluate the diagnostic performance of GPC3 in epithelial tumors and preneoplastic conditions of the liver. To conduct this analysis, we used an immunohistochemical approach that encompasses a large number of tumors, including HCC, HCCC, cirrhotic macronodules, and typical and atypical liver cell adenomas. We confirmed the high sensitivity of GPC3 as a diagnostic marker not only in overt HCC but also in preneoplastic lesions and in the differential diagnosis of primary liver tumors. Finally, these data suggest that this marker may be very helpful in improving early diagnosis of HCC in routine practice.
2. Materials and methods 2.1. Tissue samples Liver tissue samples were retrieved from the files of the Department of Pathology, Beaujon Hospital, Clichy, France. Malignant tumor samples obtained from liver resections included 54 HCC, 9 HCCC, and 10 cases of CCC. Among HCC, 21 occurred in cirrhotic liver (16 related to viral hepatitis infection and 5 to chronic alcohol consumption), and 28 were present in normal liver. Five HCC were of a fibrolamellar type on a background of normal liver. Preneoplastic liver lesions consisted of 94 cirrhotic macronodules, which were selected from 27 additional cirrhosis cases (related to hepatitis infection in 10 cases, chronic alcohol consumption in 16 cases, and autoimmune disease in 1 case). Cirrhotic macronodules were diagnosed as benign, low-grade dysplastic, high-grade dysplastic, and early HCC according to the terminology of the international working party [21]. The histologic criteria were the following: clone-like populations, nuclear hyperchromasia, irregularity of nucleus, pseudogland formation, increase in
Clinical data on patients and liver tumors
Tumor type (n)
Sex (M/F)
Mean age (y) (range)
Mean tumor size (cm) (range)
Tumor differentiation (well/moderate/poor)
HCC in cirrhosis (21) HCC in normal liver (28) Fibrolamellar HCC (5) HCCC (9) Cholangiocarcinomas (10) Cirrhotic macronodules (94) Liver cell adenomas (14) Atypical adenomas (5)
19/2 23/5 2/3 8/1 5/5 22/5 0/14 4/1
56 54 27 61 59 56 35 48
4.9 10.3 15 5.6 8 – 9 12
0/14/7 17/11/0 – 0/5/4 2/8/0 – – 5/0/0
(33-74) (23-72) (21-38) (43-76) (48-75) (38-63) (23-52) (36-63)
(1.5-8) (2.3-22) (7-22) (2.7-12) (2-14) (3.3-16) (8-16)
Glypican-3 expression in hepatocellular tumors
1437
cell density, thickness of plates, reticulin framework, and stromal invasion. Of the 27 cirrhosis cases with macronodules, 8 also had an HCC. Fourteen typical liver cell adenomas and 5 with malignant foci (called batypical adenomasQ), which occurred in normal liver, were also studied. Ten cases of CCCs and 9 cases of HCCCs were retrieved from our pathologic files. All cases were obtained from surgical specimens. All specimens were formalin-fixed and embedded in paraffin. Tissue samples were handled for routine histologic diagnosis. The HCC cases (n = 54), cirrhotic macronodules (n = 94), liver cell adenomas (n = 14), and atypical adenomas (n = 5) were selected to build tissue microarrays (TMAs). The primary clinical data of patients and liver tumors are reported in Table 1.
2.2. TMA construction Representative blocks of tumors and adjacent nontumorous liver, except for HCCC and CCC, were selected for GPC3 immunostaining. Areas of interest were identified and marked on corresponding hematoxylin and eosin
Fig. 2 GPC3 expression in HCCCs. HCCC with areas of trabecular hepatocellular carcinoma (white arrow) and glandular cholangiocarcinoma (black arrows). Positive staining was restricted to the area of hepatocellular differentiation. There was no labeling in glandular carcinoma areas.
(H&E)–stained slides. TMAs were constructed by sampling 3 cores for tumor samples and 2 cores for adjacent liver. Tissue cores of 1 mm in diameter were arrayed into a recipient paraffin block using a tissue microarrayer (MTA-1, Beecher Instruments, Inc, Sun Prairie, WI). Sections 3-lm thick were cut from completed array blocks and transferred to silanized glass slides. Sections from these arrays were then stained with H&E to assess adequacy of the sampling. Microarrays and whole sections (for HCCC and CCC) were immunostained using the same technique.
2.3. Immunohistochemistry
A
B
Immunohistochemical study was performed using an automated immunohistochemical stainer according to the manufacturer’s guidelines (streptavidin-peroxidase protocol; BenchMark, Ventana, Tucson, AZ). Sections measuring 3-lm were stained with monoclonal anti-GPC3 (1G12, 1:10, BioMosaics Inc, Burlington, VT). Sections were then evaluated by the same pathologist who was blinded to clinical information on patients. A case was considered negative in TMA if all 3 spots were negative. Positive staining was defined as membranous and/or cytoplasmic expression. GPC3 labeling was considered as positive when at least 10% of cells were positive. Uninterpretable results (because of loss of the tissue) were eliminated from further consideration. To investigate the usefulness of GPC3 as an immunohistochemical marker of HCC, serial sections of TMAs were further stained with primary antibody to a fetoprotein (AFP) (1/2000, Dako, Glostrup, Denmark). Similar to GPC3, AFP was considered positive when at least 10% of cells were stained.
2.4. Data analysis and statistics Fig. 1 GPC3 expression in HCC. A, Membranous and cytoplasmic staining of liver tumor cells. B, Hepatocellular carcinoma displaying prominent canalicular immunostaining.
Distribution between groups was compared using v 2 or Fisher exact test for qualitative data and the Student t test for
1438
X. Y. Wang et al.
Fig. 3 GPC3 and AFP expression in cirrhotic macronodules: low-grade dysplastic macronodule (A, GPC3-negative; B, AFP-negative) and high-grade dysplastic macronodule (C, GPC3-positive; D, AFP-negative).
quantitative data. A P value of .05 was chosen as the significance level.
3. Results 3.1. GPC3 expression in primary liver malignant tumors In the first step, we studied the immunohistochemical expression of GPC3 in HCC. Among 54 cases of HCC, 38 (70%) expressed GPC3 with both membranous and diffuse cytoplasmic staining (Fig. 1A). In a few cases, GPC3 immunostaining was granular and prominent adjacent to canaliculi (Fig. 1B). GPC3 staining was significantly more often observed in HCC from cirrhotic livers (19 [90%] of 21 cases) compared with HCC in normal liver (18 [64%] of 28 cases, P b .01). Among the 5 fibrolamellar HCCs present in normal liver, only 1 case (20%) displayed significant positive staining. In the group of HCC in cirrhotic livers, GPC3 expression was expressed in 13 (93%) of 14 moderately differentiated and in 6 (86%) of 7 poorly
differentiated HCC ( P = nonsignificant [NS]). Size of the tumor and any other clinicopathologic factors (etiology of cirrhosis, age of the patient, presence of vascular invasion) were not significantly related to GPC3 expression. In the group of HCCs occurring in normal liver, GPC3 was expressed in 9 (53%) of 17 well-differentiated and in 9 (82%) of 11 moderately differentiated HCCs ( P = .05). Nontumor tissue, including 21 cases of cirrhosis and 31 cases of normal liver (loss of tissue in 2 cases) obtained at a distance from the HCC, was consistently negative, except in 1 case of cirrhosis. Immunostaining of HCCC revealed GPC3 expression in 7 (78%) of 9 cases, with positive staining restricted to the area of HCC differentiation. There was no labeling in glandular carcinoma areas (Fig. 2). Regarding CCC tumors, 9 of 10 cases displayed no significant GPC3 expression.
3.2. GPC3 expression in cirrhotic preneoplastic liver lesions Because GPC3 appears to be a relevant sensitive marker of fully developed HCC, we wondered whether it could be
Glypican-3 expression in hepatocellular tumors
1439
3.3. GPC3 expression in liver cell adenomas To evaluate the specificity of GPC3 expression as a marker of malignant transformation, we collected a series of liver cell adenomas. None of the 14 typical adenomas expressed GPC3, nor did the adjacent normal liver tissue. Interestingly, among the 5 adenomas with histopathological features suggestive of malignancy, 3 (60%) of them displayed significant GPC3 staining in the malignant areas (Fig. 5). Adjacent adenomas were always negative.
3.4. Comparison of GPC3 with AFP expression in hepatocellular tumors and preneoplastic lesions HCC tissues from cirrhotic livers were less frequently stained with AFP than with GPC3 (AFP: 12 (57%) of 21 cases; GPC3: 19 (90%) 21 cases; P b .05). HCC in normal liver tissues were also less frequently stained with AFP than GPC3, but the difference was not significant (AFP: 12 [42%] of 28 cases; GPC3: 18 [64%] of 28 cases, P = NS). No cases of adjacent nontumor liver tissue, typical adenomas, or HCC arising from preexisting adenomas were positive. For the cirrhotic macronodules, 5 cases of benign or low-grade dysplastic macronodules (7% [5/65]) and 5 (17%) of the 29 high-grade dysplastic and early HCC were AFP-positive. This was significantly less than GPC3 (14 [48%] of 29 cases; P b .001) Results are shown in Table 2.
4. Discussion Fig. 4 Early hepatocellular carcinoma. A, Presence of a malignant area (arrow) in a large regenerative macronodule (H&E staining). B, The malignant area displayed positive GPC3 imunostaining.
viewed also as an early marker of liver carcinogenesis. To address this issue, we analyzed 94 cirrhotic macronodules with various grades of dysplasia from 27 cirrhotic specimens, 8 of them with overt HCC. Sixty-five nodules were benign or had low-grade dysplasia, whereas 9 had highgrade dysplasia and 20 were early HCC. In the group of benign and low-grade dysplastic macronodules, only 2 lowgrade dysplastic macronodules (3%) displayed significant GPC3 expression. Among the high-grade dysplastic and early HCC, 14 (48%) of 29 of them were immunostained with GPC3 ( P b .001). In addition, GPC3 staining was significantly more often observed in early HCC (60% [12/ 20]) than in high-grade dysplastic macronodules (22% [2/9], P b .01). Adjacent cirrhotic tissue obtained from the same specimen was consistently GPC3-negative. Representative GPC3 and AFP immunostaining are illustrated in low(Fig. 3A and B) and in high-grade (Fig. 3C and D) dysplastic macronodules. Fig. 4 illustrates a GPC3 positive malignant foci in a regenerative macronodule.
Diagnosis of malignant hepatocellular liver tumors can be a challenge for pathologists and points out the need to develop relevant sensitive markers of malignant nodules. In this study, we showed that GPC3, a member of the heparan sulfate proteoglycan family, is a sensitive and specific immunohistochemical marker of HCC. More interestingly, we demonstrated that GPC3 is also an early marker of HCC because a significant proportion of highgrade dysplastic macronodules and early HCC already displayed GPC3 expression. Similar to AFP, GPC3 potentially behaves as an oncofetal liver protein; overexpression of GPC3 is detected in fetal liver, whereas it is absent in normal adult liver [9]. Reactivation of the fetal phenotype, which is common in malignant tumors, may explain the expression of GPC3 in malignant hepatocellular nodules. Up until now, AFP was regarded as the most useful marker of HCC, although its sensitivity is limited. By comparing AFP and GPC3 immunostaining in the different groups of lesions, our data confirm that GPC3 is more sensitive, especially in the group of HCC, which developed from cirrhosis. It is of note that GPC3 immunostaining appeared to be less frequent in HCC arising in normal liver, compared with HCC associated with cirrhosis. However, when the degree of differentiation was taken into account, the difference was not significant. Moreover, although no AFP immunostaining was
1440
X. Y. Wang et al.
Fig. 5 Liver cell adenoma. A, Microscopic view of a typical liver cell adenoma (H&E staining). B, No staining for GPC3. C, Adenoma with histopathological features suggestive of malignancy with enlarged hepatocellular plates and pseudoglandular formations (H&E staining). D, GPC3-positive staining in the malignant area.
observed in high-grade dysplastic macronodules and early HCC, a higher proportion of such macronodules displayed significant GPC3 positivity. These results are consistent with those reported by Yamauchi et al [9] and confirmed the performance of GPC3 as an early marker of malignant transformation to HCC. Regarding the multistep process of carcinogenesis in the setting of chronic liver diseases, it is noteworthy that almost all cases of benign and low-grade dysplastic macronodules were GPC3-negative. This observation supports previous molecular studies demonstrating that benign and low-grade dysplastic macronodules undoubtedly share similar molecular behavior [15]. As far as the GPC3 expression is concerned, we observed that early HCC are significantly more often positive than high-grade dysplastic macronodules (60% versus 20%). Classification of high-grade dysplastic macronodules relies on several elementary morphological lesions, including presence of unpaired arteries, pseudoglandular formations, and cytologic atypias [14,21]. Because clear delineation of these events are difficult for interpretation by the pathologist, GPC3
immunostaining will provide additional criteria for characterizing those macronodules with malignant foci in cirrhotic liver tissue. Table 2 tumors
Expression of GPC3 and AFP in hepatocellular
Tumor type (n)
GPC3
AFP
P
HCC in cirrhosis (21) HCC in normal liver (28) Fibrolamellar HCC (5) HCCC (9) Cholangiocarcinomas (10) Cirrhotic macronodules (94) Regenerative benign (42) Low-grade dysplastic (23) High-grade dysplastic (9) Early HCC (20) Liver cell adenomas (14) Atypical adenomas (5)
19 18 1 7 1 16 0 2 2 12 0 3
12 12 1 0 0 10 3 2 1 4 0 0
.05 NS NS .001 NS .05 NS NS .001 .001 NS .001
(90%) (64%) (20%) (78%) (10%) (17%) (8%) (22%) (60%) (60%)
(57%) (42%) (20%)
(10%) (7%) (8%) (1%) (20%)
Glypican-3 expression in hepatocellular tumors It has already been demonstrated that benign hepatocellular nodules, including cases of focal nodular hyperplasia and adenoma, are GPC3-negative [9]. Although this marker has no pathological relevance in differentiation between such benign disorders, it could be helpful for identifying cases of adenomas with malignant progression. In previous studies, the risk of malignant progression from adenoma is seen in approximately 10% of cases, with a higher rate in men and in patients with large tumors [17]. However, the accurate diagnosis of such well-differentiated malignant foci is quite difficult when based solely on morphological analysis. Interestingly, immunohistochemical study using GPC3 clearly demonstrated the GPC3 immunostaining of malignant foci within liver cell adenomas. GPC3 is down-regulated in CCC, a pattern that may help to differentiate HCC from CCC [20]. Such a differential expression pattern might also allow the accurate diagnosis of combined HCCC by showing both positive immunostained areas with hepatocellular differentiation and negative foci with biliary differentiation. Interestingly, we observed that among 10 cases of HCCC carefully selected based on morphological analysis, 9 of them displayed GPC3 staining in restricted areas with trabecular hepatocellular differentiation. These foci were not stained with AFP. Our immunohistochemical data provide further evidence of the biologic behavior of such combined tumors, which are clinically and genetically much closer to CCC than to HCC [22]. Finally, there have been many attempts to develop noninvasive markers of HCC to detect early HCC. Because GPC3 is secreted into blood, it is a potentially interesting serum marker of HCC [23-25]. A previous study showed that a GPC3 immunoassay had potential as a diagnostic marker of HCC, with serological sensitivity and specificity of 53% and 95%, respectively [23]. Our study, demonstrating the efficiency of the 1G12 GPC3 monoclonal antibody for immunohistochemistry, is promising and supports the need for further studies to test for circulating GPC3 with an immunoassay of the sera of cirrhotic patients with and without HCC.
References [1] Filmus J, Selleck SB. Glypicans: proteoglycans with a surprise. J Clin Invest 2001;108:497 - 501. [2] Li M, Choo B, Wong ZM, Filmus J, Buick RN. Expression of OCI-5/ glypican3 during intestinal morphogenesis: regulation by cell shape in intestinal epithelial cells. Exp Cell Res 1997;235:3 - 12. [3] Farooq M, Hwang SY, Park MK, Kim JC, Kim MK, Sung YK. Blocking endogenous glypican-3 expression releases Hep 3B from G1 arrest. Mol Cells 2003;3:356 - 60. [4] Lin H, Huber R, Schlessinger D, Morin JP. Frequent silencing of the GPC3 gene in ovarian cancer cell lines. Cancer Res 1999;59:807 - 10.
1441 [5] Murthy SS, Shen T, De Rienzo A, et al. Expression of GPC3, an Xlinked recessive overgrowth gene, is silenced in malignant mesothelioma. Oncogene 2000;19:410 - 6. [6] Xiang YY, Ladeda V, Filmus J. Glypican-3 expression is silenced in human breast cancer. Oncogene 2001;20:7408 - 12. [7] Hsu HC, Cheng W, Lai PL. Cloning and expression of a developmentally regulated transcript MXR7 in hepatocellular carcinoma. Cancer Res 1997;57:5179 - 84. [8] Zhu Z-W, Friess H, Wang L, et al. Enhanced glypican-3 expression differentiates the majority of hepatocellular carcinomas from benign hepatic disorders. Gut 2001;48:558 - 64. [9] Yamauchi N, Watanabe A, Hishinuma M, et al. The glypican 3 oncofetal protein is a promising diagnostic marker for hepatocellular carcinoma. Mod Pathol 2005;18:1591 - 8. [10] Furuya K, Nakamura M, Yamamoto Y, Togei K, Otsuka H. Macroregenerative nodule of the liver: a clinicopathologic study of 345 autopsy cases of chronic liver disease. Cancer 1988;61:99 - 105. [11] Kondo F, Ebera M, Siguira N, et al. Histological features and clinical course of large regenerative nodules: evaluation of their precancerous potentiality. Hepatology 1990;12:592 - 8. [12] Okuda K. Hepatocellular carcinoma: recent progress. J Hepatol 2002;32:225 - 37. [13] Paradis V, Laurendeau I, Vidaud M, Bedossa P. Clonal analysis of macronodules in cirrhosis. Hepatology 1998;28:953 - 8. [14] Ferrell LD, Crawford JM, Dhillon AP, Scheuer PJ, Nakanuma Y. Proposal for standardized criteria for the diagnosis of benign, borderline and malignant hepatocellular lesions arising in chronic advanced liver disease. Am J Surg Pathol 1993;17:1113 - 23. [15] Paradis V, Bieche I, Dargere D, et al. Molecular profiling of hepatocellular carcinomas (HCC) using a large-scale real-time RTPCR approach: determination of a molecular diagnostic index. Am J Pathol 2003;163:733 - 41. [16] Anders RA, Yerian LM, Tretiakova M, et al. cDNA microarray analysis of macroregenerative and dysplastic nodules in end-stage hepatitis C virus–induced cirrhosis. Am J Pathol 2003;162:991 - 1000. [17] Foster JH, Berman MM. The malignant transformation of liver cell adenomas. Arch Surg 1994;129:712 - 7. [18] Ito M, Sasaki M, Wen CY, et al. Liver cell adenoma with malignant transformation: a case report. World J Gastroenterol 2003;9: 2379 - 81. [19] Ng IO, Shek TW, Nicholls J, Ma LT. Combined hepatocellularcholangiocarcinoma: a clinicopathological study. J Gastroenterol Hepatol 1998;13:34 - 40. [20] Man X-B, Tang L, Zhang B-H, Qiu X-H, Wu M-C, Wang H-Y. Upregulation of glypican-3 expression in hepatocellular carcinoma but downregulation in cholangiocarcinoma indicates its differential diagnosis value in primary liver cancers. Liver Int 2005;25:962 - 6. [21] Wanless I, International Working Party. Terminology of nodular hepatocellular lesions. Hepatology 1995;22:983 - 92. [22] Cazals-Hatem D, Rebouissou S, Bioulac-Sage P, et al. Clinical and molecular analysis of combined hepatocellular-cholangiocarcinomas. J Hepatol 2004;41:292 - 8. [23] Capurro M, Wanless I, Sherman M, et al. Glypican-3: a novel serum and histochemical marker for hepatocellular carcinoma. Gastroenterology 2003;125:89 - 97. [24] Nakatsura T, Yoshitake Y, Senju S, et al. Glypican-3, overexpressed specifically in human hepatocellular carcinoma, is a novel tumor marker. Biochem Biophys Res Commun 2003;306:16 - 25. [25] Hippo Y, Watanabe K, Watanabe A. Identification of soluble NH2terminal fragment of glypican-3 as a serological marker for early-stage hepatocellular carcinoma. Cancer Res 2004;64:2418 - 23.