Histopathological classification of hepatocellular carcinoma

Digestive and Liver Disease 42S (2010) S228–S234

Histopathological classification of hepatocellular carcinoma Massimo Roncalli a, *, Young Nyun Park b , Luca Di Tommaso a a Department

of Pathology, University of Milan School of Medicine and IRCCS Istituto Clinico Humanitas, Rozzano, Milan, Italy of Pathology, Center for Chronic Metabolic Disease, Brain Korea 21 Project for Medical Science, Yonsei University College of Medicine, 134 Shinchon-dong, Seodaemoon-gu, Seoul 120-752, Korea

b Department

Abstract Recent advances in the understanding of HCC heterogeneity have shown not only distinct molecular classes with potential therapeutical implications but also have suggested that subclasses might be mirrored by tumour morphology and phenotype. The classification of HCC is therefore shifting from a more traditional, morphology-based approach to a more functional approach with clinical implications. The classical histological patterns (trabecular, pseudoglandular, solid, etc.) are still taken into account but in combination with other parameters such as cell grading, clonal changes, type and extent of vascularisation, tumour immunophenotype etc. This array of morpho-phenotypical features is expected to ultimately provide information about cell of origin, tumour behaviour and, hopefully, treatment sensitivity. In this chapter we will give a brief overview on modern principles of hepatic carcinogenesis by outlining the events/lesions bridging the microscopic dysplasia to advanced HCC. In this context special emphasis will be given to novel concepts and diagnostics and differential diagnosis of small HCC (of early and progressed type), which is the main target of the surveillance in the cirrhotic population. Novel and emerging subpopulations of HCC will also be considered, for example HCC with stem/progenitor cell phenotype and mixed hepatobiliary forms which are filling the traditional separation between two entities (HCC and cholangiocarcinoma), likely more inter-related than previously thought. © 2010 Editrice Gastroenterologica Italiana S.r.l. Published by Elsevier Ltd. All rights reserved. Keywords: Dysplasia; Early/progressed HCC; Hepatocarcinogenesis; Stem/progenitor cell

1. Introduction In tumour classification, pathology is moving fast from a mere descriptive and purely morphological approach to a more functional and clinically relevant one. Today the extraordinary amount of information originated from molecular studies is matched with clinical data, including epidemiology, tumour phenotype, outcome, and treatment sensitivity. As a consequence, some morphological features which had been thought to mirror the tumour biology have not proved as clinically significant while novel histological features and biomarkers are emerging as potentially useful in the clinical * Correspondence to: Massimo Roncalli, MD, PhD, University of Milan Medical School, Department of Pathology, Istituto Clinico Humanitas, Via Manzoni 56, 20089 Rozzano (Milan), Italy. Tel: +39-02-8224-4714, fax: +39-02-8224-4791. E-mail address: [email protected] (M. Roncalli).

practice. This understanding is rapidly changing the morphological approach to tumour classification which is now more oriented towards providing information on cell of origin, tumour heterogeneity, behaviour, and treatment sensitivity. It is known that certain HCCs, despite an apparently similar morphology and size, can show an unpredictable distinct outcome and that an indolent or aggressive tumour signature is likely to be embedded in most of tumours ab initio. The modern role of pathologists is to interrogate the tumour and the adjacent background and to be able to interpret this signature for a clinical use. Thus, a major task of any HCC classification should be to provide morpho-phenotypic tools or surrogate markers to predict aggressive/indolent forms or features associated to sensitivity/resistance to novel therapies. Pathologists classifying multinodular HCC will be increasingly requested to establish whether the tumour is multicentric and not metastatic as opposed to unicentric but metastatic (Fig. 1). This information is important as the tumour grading and staging

1590-8658/$ – see front matter © 2010 Editrice Gastroenterologica Italiana S.r.l. Published by Elsevier Ltd. All rights reserved.

M. Roncalli et al. / Digestive and Liver Disease 42S (2010) S228–S234

S229

Fig. 1. Multinodular HCC: main features of aid in the distinction between multicentric vs. metastatic disease. Legend: Multicentric vs. metastatic disease can be reasonably addressed by gross (radiological) features only in the conditions depicted on the left side of the figure. In all the other conditions (right side of the figure) only a microscopic examination can address the issue by proving the multistep carcinogenesis. The possibility of a multicentric disease, followed by a metastatic one, is not illustrated in the figure. As reported in the text the accuracy of this evaluation is not absolute and tumour allelotyping should be performed.

are clinically more significant than the mere description of the tumour pattern (pseudoglandular, solid, sarcomatoid, etc.). In HCC, heterogeneity is a well known biological feature reflecting an array of oncological agents and complex pathways to tumourigenesis. A clinico-functional classification of HCC has rapidly gained wide acceptance worldwide (BCLC staging system), based not only on HCC extent (size and number of nodules) but also on clinical features of patients, with tumour stratification according to the different available surgical and medical therapies. Notably, this classification still does not mention molecular or morphological features related to the biology of individual tumours in different categories (very early/early stage, intermediate and advanced stage). Recent advances in the understanding of HCC complexity have shown not only distinct molecular classes of HCC with potential therapeutical implications [1–3], but also suggest that subclasses might be mirrored by tumour morphology and phenotype. The present chapter will, therefore, survey the actual morphological classification of HCC with emphasis on clinically useful (or potentially useful) subtypes and related biomarkers.

2. Principles of hepatocarcinogenesis and HCC classification In humans, 90% of HCC cases arise as complication of chronic liver disease/cirrhosis with fibrosis playing a major predisposing role. HCC usually takes place after a multistep sequence of morphologically recognizable lesions as shown in Fig. 2. The very early onset of events is still poorly understood, and it is characterized by hepatocytes acquiring genetic/epigenetic changes and telomere shortening, allowing the cells to survive and to expand clonally [4].

Both differentiated hepatocytes and the hepatic progenitor cells can be the source of these abnormal cells, being the above cell types characterized by a lower turnover and longer survival, essential features of the hepatic longlasting process of carcinogenesis. Progressed HCCs have two main phenotypes: differentiation toward a single cell lineage (hepatocyte phenotype) and a bi-linear shift toward a mixed hepatobiliary phenotype with variable proportion of an immature component (so-called progenitor/stem cell). The hepatocarcinogenetic process is mostly characterized by the appearance of dysplastic lesions (dysplastic phase) in the form of microscopic foci (large and small cell changes) and macroscopic dysplastic nodules (low and high grade) [5,6]. This phase is very important because patients harbouring these lesions are at increased risk of developing HCC, allowing early tumour detection and consequently improving survival. The possibility of de novo development of HCC without premalignant antecedents is rare, but can not be excluded. The surveillance of cirrhotic patients with early detection of HCC [7] and an accurate study of hepatic resection/explants [8] have shown that high grade dysplastic nodules are the most advanced precursors of HCC and that small and well differentiated HCCs with indistinct margins (early HCC) are the earliest and smallest recognizable forms of HCC (usually less than 2 cm) [6,9]. Conversely, small HCCs with round pushing borders are progressed HCC, and such form can take place in two different settings: in a high grade dysplastic nodule and within an early-stage small HCC. In both cases, HCCs may show the radiologic and macroscopic appearance of so-called “nodule in nodule” [6]. The vascular changes occurring in the hepatocarcinogenetic process are very important, and consist of progressive disappearance of the native portal and arterial supply provided by portal tracts

S230

M. Roncalli et al. / Digestive and Liver Disease 42S (2010) S228–S234

Fig. 2. Main microscopic and macroscopic lesions occurring in hepatocarcinogenesis in chronic liver disease.

and by the progressive appearance of a neoarterial supply outside the portal tracts (unpaired arteries). As such, both dysplastic and early malignant lesions are characterized by a variable but incomplete vascular remodelling with an arterial shift. The study of these vascular changes may be useful to morphologically characterize the different steps of the carcinogenetic process, and sets both the basis and the limits for their radiologic detection.

3. HCC classification We will review the following entities: 1. Early HCC (a) HCC of small size (≤2 cm), well differentiated (G1), vaguely nodular type. 2. Progressed HCC (a) HCC of small size (≤2 cm), usually moderately differentiated (G2), distinctly nodular type and of not small size (>2 cm); (b) HCC with stem/progenitor cell immunophenotype; (c) Mixed hepatobiliary carcinoma, classical type; (d) Mixed hepatobiliary carcinoma with stem/progenitor cell phenotype and immunophenotype. 3.1. Early HCC: HCC of small size (≤2 cm), well differentiated (G1), vaguely nodular type Early HCC is a starting form of hepatocellular malignant neoplasm because of the small size (usually less than 2 cm), incomplete arterial neovascularisation, and well differentiated

histology. It can be equated with so-called in situ carcinoma of other organs and sites, although early HCC has the potential to infiltrate portal tracts but not vessels (therefore it is not a metastatic tumour). It is an entity in between in situ carcinoma and invasive carcinoma, similar to microinvasive carcinoma. The definition of this entity was agreed on after a prolonged discussion with circulation of slides between a group of Eastern and Western pathologists led by Professor Masamichi Kojiro who first introduced early HCC as the very first modality of HCC growth [6]. The notion of early HCC is very important, and it has to be distinguished from both a high grade dysplastic nodule and progressed HCC with less differentiated histology. The frequency of early HCC is poorly understood because of the difference in hepatic background where HCC develops. Indeed, Western data are mainly based on the explanted liver, while Eastern data are obtained from surgical cases. Accordingly, explanted livers have more fibrosis so that an early well differentiated HCC is surrounded by cirrhotic septa with more distinct margins as compared to the earlier and less fibrotic setting of Eastern series. As shown in Table 1, at variance with progressed HCC, early HCC does not present a destructive and expansile growth, but it deceptively replaces the cirrhotic parenchyma, with a pre-existing dysplastic focus or nodule. It is not usually surrounded by a true capsule unless it occurs in an advanced stage of cirrhosis, and a few portal tracts can still be detected within the lesion. An important feature, particularly emphasized by Eastern authors and less frequently recognized by Western pathologists, is macrovesicular steatosis, reported in less than half of early HCCs (40%). Interestingly, steatosis is a feature which can be recognized by image

M. Roncalli et al. / Digestive and Liver Disease 42S (2010) S228–S234

S231

Table 1 Features of early vs. progressed HCC

Early HCC Progressed HCC

Size

Growth pattern

Steatosis

Grading

Neoarterial vascularization

Stromal invasion

Vascular invasion

5-year survival

≤2 cm Any

Replacing Expansile

Often (40%) Rare

G1 G2–G3

Incomplete Complete

+/– +

– +/–

89% 48%

analysis (US). As such, it should be carefully looked for by radiologists when small nodules arise in the cirrhotic liver. Steatosis is partly due to the incomplete neoarterialization of the neoplasm which is still partly supported by a venous supply. Most of the so-called hypovascular HCCs and HCC showing equivocal or controversial imaging are early HCCs [10,11]. Current international guidelines recommend liver biopsy to prove malignancy of 1–2 cm nodules that are equivocal at imaging. Histology of early HCC is that of a well differentiated hepatocellular neoplasm with increased cell density as compared to surroundings and a thin irregular trabecular and occasional pseudoglandular pattern. This histology poses, particularly on liver biopsy, problems of differential diagnosis mainly with immediate antecedents, i.e. high grade dysplastic nodules. This differential diagnosis can be addressed in both surgical specimens and in liver biopsy using morphology, histochemistry and immunocytochemistry. Currently three main tools are useful to prove a diagnosis of early HCC, namely the reduction/loss of reticulin framework, the stromal invasion and the immunoreactivity for a panel of biomarkers. Stromal invasion of neoplastic cells into residual portal tract and septa is an important criterion to separate malignancy from dysplastic precursors, but it is less suitable for use in liver biopsy where portal tracts are rarely found. The histological feature of stromal invasion can be surrogated by the use of Victoria blue staining outlining residual portal tracts or by the lack of ductular reaction, which usually takes place around non-malignant nodules, visualized by keratin (K) 19 immunostaining, next to the infiltrating edges of the tumour [12]. Interestingly, the tissue detection of at least two immunomarkers such as Glypican 3 (a membrane proteoglycane), Glutamyne synthetase (a target protein of β-catenin) and heat shock protein 70 (a chaperone stress protein) is very useful to prove malignancy of an equivocal hepatocellular nodule, particularly in liver biopsy [13,14]. The concept of early HCC is strictly pathological, referring to an incipient form of malignant transformation of the human liver. It should fit both dimensional (<2 cm) and morphological criteria. This designation should not be confused with the early or very early stage HCC of the BCLC scheme. Indeed in that scheme, both the very early and early stage HCCs can be, histologically, either early or progressed HCC according to pathological criteria, and can even be a combination of both in multinodular tumours. The outcome of early HCC is also very different from that occurring in progressed HCC of similar size (Table 1). For example, the 5-year survival rate of early HCC is twice as high as that of progressed HCC [15]. This is because early HCC is able to infiltrate the hepatic stroma but not the vessels, while progressed HCC, including small ones, can be angioinvasive in at least 25% of cases. Therefore, early

HCC is not a metastatic disease which, conversely, may be a feature of progressed HCC. 3.2. Progressed HCC (a) HCC of small size (≤2 cm), usually moderately differentiated (G2), distinctly nodular type and of not small size (>2 cm) Progressed HCC may develop from a pre-existing dysplastic focus or nodule or from an early HCC (Fig. 2). In the latter two cases, it can take the gross and radiological appearance of the so-called “nodule in nodule” which is virtually diagnostic of malignancy in a cirrhotic setting. At the earliest stage, progressed HCC can be small, less than 2 cm, similar to early HCC. At variance with the latter, however, progressed HCC is rarely a diagnostic dilemma either radiologically and histologically, even in liver biopsy. It is morphologically characterized by a destructive and expansile growth pattern with complete neoarterialization and frequent vascular invasion. Portal tracts are no longer visualized, and the expansile round border of the tumour is generally rimmed by a condensed fibrosis showing up as a tumour capsule. Histologically small but progressed HCC is usually moderately differentiated (G2), rarely steatotic for more complete and advanced neoarterialization, but it can be poorly differentiated (G3) with size enlargement. During progression, it can develop different clonal populations and an array of morphological patterns that are well recognized in advanced HCC (pseudoglandular, solid, macrotrabecular, sarcomatoid, etc.). The most ominous histological features are satellite nodules, expression of a peritumoural (within 2 cm) metastatic growth as well as a micro- and/or macro-vascular invasion. In this setting or in recurrent tumours, liver cytology rather than liver biopsy is sufficient for diagnostic purposes, if clinically needed [6]. As already mentioned in Fig. 1, multiple tumours may be either progressed and metastatic HCC or multicentric HCC with obvious clinical implications. Multicentric tumours are expected to show a slow behaviour with eventual late recurrence while metastatic HCCs are more prone to develop early recurrences. To address this issue, which is still very challenging for both clinicians and pathologists, several gross and morphological criteria may be used (see Fig. 1), although the clinical history and the histopathological features, when available, may not be absolutely accurate [16]. Tumour allelotyping of the nodules has been proposed as the most accurate way to address the issue, but it has not been adopted in the clinical practice [16]. (b) HCC with stem/ progenitor cell immunophenotype There is a recently proposed subtype of progressed HCC (mostly G2/G3) in which a fraction of tumour cells (>5%)

S232

M. Roncalli et al. / Digestive and Liver Disease 42S (2010) S228–S234

expresses stem/progenitor cell markers such as K19, epithelial cell adhesion molecule (EpCAM), CD133, etc. [17–24] although this subtype is not otherwise recognizable by hematoxylin and eosin stain. It has been suggested to be a peculiar HCC linked to a stem/progenitor cell phenotype, and this subtype has also been suggested to be a biliary phenotype, as K19 is expressed in biliary epithelium as well as in stem/progenitor cell [22,23]. It is still unaddressed whether this phenotype reflects an origin from a stem/progenitor cells or dedifferentiation from mature hepatocytes [17,24,25]. Interestingly these tumours have been reported to behave more aggressively as compared to usual progressed HCC without stem/progenitor markers, with increased likelihood of recurrence following resection, resistance to chemotherapy, and radiotherapy, and increased likelihood of metastasis [17– 25]. It has, therefore, been suggested that the application of these specific immunomarkers on the biopsied or resected HCC tissue may be helpful to select patients with more aggressive HCC. It will provide useful information for optimal management and treatment of HCC patients. It is reasonable to assume that, in the near future, more tissue immunomarkers from both the tumour and the peritumour liver tissues will become available, allowing better characterization and prediction of HCC behaviour in each patient [25–28]. (c) Mixed hepatobiliary carcinoma, classical type Mixed hepatobiliary carcinoma, classical type, is not common, accounting for about 2.5% of primary liver cancers.

The most typical form is combined hepatocellular and cholangiocarcinoma, which has areas of typical HCC and areas of typical cholangiocarcinoma. The term “mixed hepatobiliary carcinoma” is not recommended to be used for collision tumours in which HCC and cholangiocarcinoma arise separately. These tumours can be defined morphologically, and confirmatory immunophenotyping is helpful. Confirmation of hepatocellular differentiation can be supported by immunostaining for hepatocyte specific antigen (HePar1), polyclonal anti-carcinoembryonic antigen (pCEA), and CD10. Cholangiocarcinoma components are usually positive for K7 and K19. Mucin is a strong evidence in favor of mature biliary differentiation. Careful examination will usually reveal small areas of intermediate morphology at the interface between HCC and cholangiocarcinoma areas. They are composed of small cells with a high nuclear cytoplasmic ratio and hyperchromatic nuclei, and the expression of K7 and K19, nuclear cell adhesion molecule (NCAM, CD56), c-kit, or EpCAM. Therefore, these small tumour cells at the interface suggest a recapitulation of stem/progenitor cells. The prognosis of this subtype has been reported to be poorer as compared to the usual progressed HCC. (d) Mixed hepatobiliary carcinoma with stem/progenitor cell phenotype and immunophenotype Recently a rare subtype has been reported, in which the majority of tumour cells show stem/progenitor like cells or intermediate features between hepatocyte and cholangio-

Fig. 3. Illustration of the main categories of HCC according to hepatocytes and biliary differentiation depicted using both morphology and the immunophenotype.

M. Roncalli et al. / Digestive and Liver Disease 42S (2010) S228–S234

cyte. These tumours include “Hepatic stem cell carcinoma”, “Primary liver carcinoma of intermediate (hepatocyte-cholangiocyte) phenotype (Intermediate carcinoma)”, “Cholangiolocellular carcinomas” [29–33]. Areas of typical HCC and/or of typical cholangiocarcinoma are not or only focally detectable in these tumours. Considering these diverse terminologies and combinations, the International Consensus Panel sponsored, in part, by the Laennec Society for Hepatopathology suggested a category of “mixed hepatobiliary carcinoma” with “stem/progenitor cell features”. The prognosis of this subtype is uncertain given the absence of large cohorts of patients. The large sized “cholangiolocellular carcinomas” (>4.0 cm) had higher recurrence rates following treatment [33], and the overall survival of intermediate carcinoma was intermediate between those of typical HCCs and typical cholangiocarcinoma [31], whereas small sized mixed hepatobiliary carcinomas with stem/progenitor cell phenotype and immunophenotype in explanted livers showed no recurrence [30]. Further expanded study with long term follow-up will properly determine the clinical significance of this subset of primary liver carcinomas.

4. Conclusions and perspectives In HCC classification, a shift is taking place from morphological to morpho-phenotypical categories with pathogenetic and clinical implications (Fig. 3). The aim is to help in selecting an optimal therapeutic strategy for individual HCC (personalized medicine) in the current clinical practice. Novel concepts in HCC pathogenesis, in which the earliest and less invasive forms of HCC (early HCC) are recognized, have clarified that even small (less than 2 cm) HCCs can be quite heterogeneous, ranging from an indolent to an already metastatic disease (progressed HCC of small size). Thus, incorporation of the above information into the current HCC classification for clinical use (BCLC staging system) might improve the stratification of tumours according to available therapies and oncological protocols. Another challenge for pathologists is to convincingly address whether an individual multinodular HCC has a multicentric or a metastatic origin. To resolve this, requires the co-operation of several specialists, including molecular pathologists. More information is also expected to come from HCC immunophenotyping and novel biomarkers, which are under scrutiny to be validated as effective diagnostic tools or predictors of behaviour. Finally, a rapidly evolving field is that devoted to testing the sensitivity of tumours to biological therapies. An array of molecular tools (tissue immunomarkers, serological markers, genetic and epigenetic markers, microRNA, etc.) is under investigation to select the most effective predictors of drug sensitivity/resistance [3]. It is likely that all of the above information will help to establish a more robust and clinically useful HCC classification system. It is the authors’ opinion that a well targeted characterization of HCC from liver biopsy or even cytological specimens will be, in the near future, the major task of liver pathologists, to meet the growing expectancy of hepatologists, surgeons and oncologists.

S233

Conflict of interest The authors have received a fee from Bayer HealthCare for their contribution to this supplement. Bayer HealthCare played no role in the preparation, review, or approval of the manuscript. The authors have no other conflict of interest to report.

References [1] Boyault S, Rickman DS, de Reyniès A, et al. Transcriptome classification of HCC is related to gene alterations and to new therapeutic targets. Hepatology 2007;45:42–52. [2] Katoh H, Ojima H, Kokubu A, et al. Genetically distinct and clinically relevant classification of hepatocellular carcinoma: putative therapeutic targets. Gastroenterology 2007;133:1475–86. [3] Llovet JM, Bruix J. Molecular targeted therapies in hepatocellular carcinoma. Hepatology 2008;48:1312–27. [4] Farazi PA, DePinho RA. Hepatocellular carcinoma pathogenesis: from genes to environment. Nat Rev Cancer 2006;6:674–87. [5] International Working Party. Terminology of nodular hepatocellular lesions. Hepatology 1995;22:983–93. [6] International Consensus Group for Hepatocellular Neoplasia The International Consensus Group for Hepatocellular Neoplasia. Pathologic diagnosis of early hepatocellular carcinoma: a report of the international consensus group for hepatocellular neoplasia. Hepatology 2009;49:658–64. [7] Borzio M, Fargion S, Borzio F, et al. Impact of large regenerative, low grade and high grade dysplastic nodules in hepatocellular carcinoma development. J Hepatol 2003; 39:208–14 [8] Libbrecht L, Bielen D, Verslype C, et al. Focal lesions in cirrhotic explant livers: pathological evaluation and accuracy of pretransplantation imaging examinations. Liver Transpl. 2002;8:749–61. [9] Kojiro M, Roskams T. Early hepatocellular carcinoma and dysplastic nodules. Semin Liver Dis 2005;25:133–42. [10] Bolondi L, Gaiani S, Celli N, et al. Characterization of small nodules in cirrhosis by assessment of vascularity: the problem of hypovascular hepatocellular carcinoma. Hepatology 2005;42:27–34. [11] Forner A, Vilana R, Ayuso C, et al. Diagnosis of hepatic nodules 20 mm or smaller in cirrhosis: prospective validation of the noninvasive diagnostic criteria for hepatocellular carcinoma. Hepatology 2008; 47:97–104. [12] Park YN, Kojiro M, Di Tommaso L, et al. Ductular reaction is helpful in defining early stromal invasion, small hepatocellular carcinomas, and dysplastic nodules. Cancer 2007;109:915–23. [13] Di Tommaso L, Franchi G, Park YN, et al. Diagnostic value of HSP70, glypican 3, and glutamine synthetase in hepatocellular nodules in cirrhosis. Hepatology 2007;45:725–34. [14] Di Tommaso L, Destro A, Seok JY, et al. The application of markers (HSP70 GPC3 and GS) in liver biopsies is useful for detection of hepatocellular carcinoma. J Hepatol 2009;50:746–54. [15] Takayama T, Makuuchi M, Hirohashi S, et al. Early hepatocellular carcinoma as an entity with a high rate of surgical cure. Hepatology 1998;28:1241–6. [16] Ng IO, Guan XY, Poon RT, et al. Determination of the molecular relationship between multiple tumour nodules in hepatocellular carcinoma differentiates multicentric origin from intrahepatic metastasis. J Pathol 2003;199:345–53. [17] Durnez A, Verslype C, Nevens F, et al. The clinicopathological and prognostic relevance of cytokeratin 7 and 19 expression in hepatocellular carcinoma. A possible progenitor cell origin. Histopathology 2006;49:138–51. [18] Uenishi T, Kubo S, Yamamoto T, et al. Cytokeratin 19 expression in hepatocellular carcinoma predicts early postoperative recurrence. Cancer Sci 2003;94:851–7.

S234

M. Roncalli et al. / Digestive and Liver Disease 42S (2010) S228–S234

[19] Ding SJ, Li Y, Tan YX, et al. From proteomic analysis to clinical significance: overexpression of cytokeratin 19 correlates with hepatocellular carcinoma metastasis. Mol Cell Proteomics 2004;3:73–81. [20] Yamashita T, Forgues M, Wang W, et al. EpCAM and alpha-fetoprotein expression defines novel prognostic subtypes of hepatocellular carcinoma. Cancer Res 2008;68:1451–61. [21] Ma S, Chan KW, Hu L, Lee TK, et al. Identification and characterization of tumorigenic liver cancer stem/progenitor cells. Gastroenterology 2007;132:2542–56. [22] Wu P, Fang J, Lau V, et al. Classification of hepatocellular carcinoma according to hepatocellular and biliary differentiation markers: clinical and biological implications. Am J Pathol 1996;1167–75. [23] Yoon DS, Jeong J, Park YN, et al. Expression of biliary antigen and its clinical significance in hepatocellular carcinoma. Yonsei Med J 1999;40:472–7. [24] Yamashita T, Ji J, Budhu A, et al. EpCAM-positive hepatocellular carcinoma cells are tumor-initiating cells with stem/progenitor cell features. Gastroenterology 2009;136:1012–24. [25] Lee JS, Heo J, Libbrecht L, et al. A novel prognostic subtype of human hepatocellular carcinoma derived from hepatic progenitor cells. Nat Med 2006;12:410–6.

[26] Budhu A, Chen Y, Kim JW, et al. Induction of a unique gene expression profile in primary human hepatocytes by hepatitis C virus core, NS3 and NS5A proteins. Carcinogenesis 2007;28:1552–60. [27] Hoshida Y, Villanueva A, Kobayashi M, et al. Gene expression in fixed tissues and outcome in hepatocellular carcinoma. N Engl J Med 2008;359(19):1995–2004. [28] Hoshida Y, Nijman SM, Kobayashi M, et al. Integrative transcriptome analysis reveals common molecular subclasses of human hepatocellular carcinoma. Cancer Res 2009;69:7385–92. [29] Theise ND, Yao JL, Harada K, et al. Hepatic “stem cell” malignancies in adults: four cases. Histopathology 2003;43:263–71. [30] Fujii T, Zen Y, Harada K, et al. Participation of liver cancer stem/progenitor cells in tumorigenesis of scirrhous hepatocellular carcinoma-human and cell culture study. Hum Pathol 2008;39:1185–96. [31] Kim H, Park C, Han KH, et al. Primary liver carcinoma of intermediate (hepatocyte-cholangiocyte) phenotype. J Hepatol 2004; 40: 298–304. [32] Shiota K, Taguchi J, Nakashima O, et al. Clinicopathologic study on cholangiolocellular carcinoma. Oncol Rep 2001;8:263–8. [33] Komuta M, Spee B, Vander Borght S, et al. Clinicopathological study on cholangiolocellular carcinoma suggesting hepatic progenitor cell origin. Hepatology 2008;47:1544–56