Pathology of combined hepatocellular-cholangiocarcinoma


Matthew M Yeh, Department of Pathology, University of Washington School of Medicine, 1959 NE Pacific Street, NE140D, Box 356100, Seattle, WA 98195, USA. Email:


Combined hepatocellular-cholangiocarcinoma is a rare primary neoplasm in the liver. It has gained increasing recognition recently, which in part may be due to more extensive sampling of the explants and surgical resection specimens, the diagnostic challenges encountered in the clinical practice, and the yet to be determined clinical outcome, but partly may be attributed to its intriguing histogenesis/cells of origin. This review aims to update combined hepatocellular-cholangiocarcinoma with an emphasis on the pathological diagnosis, including the differential diagnosis and its diagnostic pitfalls, the possible cell of origin of this neoplasm, and its clinical outcome.


Combined hepatocellular-cholangiocarcinoma (HCC-CC), also known as mixed HCC-CC, is a rare (incidence among primary liver cancer ranges from 1.0% to 4.7%) but an increasingly recognized primary malignant neoplasm in the liver.1–4 It shares unequivocal features of both hepatocellular carcinoma (HCC) and cholangiocarcinoma (CC) as defined by the World Health Organization (WHO) classification,5 which also emphasizes that the diagnosis should not be used for neoplasms in which either form of growth is insufficiently differentiated for positive identification.5 Although several sporadic reports existed as early as the turn of the 20th century, this neoplasm was first described and reviewed in detail by Allen and Lisa in 1949.6 Popper and Schaffner in 1957 stated that with careful examination, most primary hepatic carcinomas could be found to have hepatocellular and ductal elements,7 but Edmondson in the following year pointed out that in the majority of cases, these ductal elements were from hepatocyte-like tumor cells and that such tumors are really a variant of HCC.3 It is now generally recognized that most of these ductal elements mentioned may reflect what we see as pseudoglands in classic HCC, but not the true glandular structures with mucin production observed in the rare combined HCC-CC.8 Goodman and colleagues subsequently examined 24 cases in the mid-1980s. This was the largest series studying combined HCC-CC using immunohistochemistry.9 Both Allen's and Goodman's studies attempted to classify these neoplasms into subtypes. It is worth mentioning that the subtypes under each classification scheme are arranged as described by the authors and are not necessarily equivalent to one another. Type 1 tumor by Allen and Lisa and type I tumor designated by Goodman et al. appear to be the collision type tumor. It appears that the subtype with an intimate intermingling of hepatocellular and glandular elements, classified as type 3 tumor by Allen and Lisa and subsequently defined by Goodman et al.9 as type II (transitional) neoplasms, draws most interest among investigators in exploring its histogenesis, establishing the pathological diagnosis and characterizing the clinical outcome, whereas cases of separate HCC and CC coincidentally found in the same liver are generally considered as collision tumors and are excluded by the WHO classification of combined HCC-CC. The concept of collision tumor (Fig. 1a) occurring as separate HCC and CC in the same liver has been further supported by the genetic findings that two independent neoplastic clones develop at close proximity10 and hence no histological transitions exist. It is interesting that in the study by Allen and Lisa, most of the intimately intermingled type and most of Goodman's type II tumors (80%) were cirrhotic.9,10 However, cirrhosis is not a prerequisite for combined HCC-CC and it may also arise in non-cirrhotic liver (Fig. 1b). In fact, prevalence of background cirrhosis (Fig. 1c) in combined HCC-CC varies among different studies and it may be attributed to different patient populations as well as diagnostic criteria for combined HCC-CC used in respective studies.4,11,12 Similarly, the underlying cause of liver disease, such as chronic hepatitis B or hepatitis C infection, also varies among different studies and this may be at least partially explained by the different geographic regions and diagnostic criteria used in each study.4,11,12

Figure 1.

Independent hepatocellular carcinoma (HCC) (thin arrow, yellow and soft nodule) and cholangiocarcinoma (CC) (thick arrow, white, firm and infiltrative mass) occuring (a) separately in the same liver is generally considered as collision tumor rather than true combined HCC-CC. Combined HCC-CC can arise in the background of either (b) non-cirrhotic or (c) cirrhotic liver.

Image studies of combined HCC-CC

Image studies of combined HCC-CC may vary, depending on whether HCC or CC is the predominant component and their enhancement patterns reflect the distribution of HCC and CC elements.2,13 Overall, when the HCC component is predominant, the CT images show marked hyperenhancement throughout the tumor in the early phase, which subsequently attenuates to hypoenhancement in the late phase because of washout of the contrast medium. The classic CT image of combined HCC-CC is best characterized as early enhancement in the periphery of the mass and delayed enhancement at the center of the mass but with hypoenhancement in the periphery.2 These changes may be explained histologically by the circular layered zones of HCC and CC, and the transitional elements of these two components. When CC components are in the center of the tumor, they exhibit delayed enhancement.2

Pathological diagnosis of combined HCC-CC

The typical histology of HCC shows carcinoma with hepatocytic differentiation, that is, trabecular or pseudoglandular growth pattern, bile in the canaliculi, and carcinoma cells resembling hepatocytes, such as fat, Mallory-Denk bodies or α-1 antitrypsin globules in the cytoplasm, whereas the classic histopathology of CC is characterized by desmoplastic stroma, and carcinoma cells forming glandular structures and producing mucin. These features are not typically seen in HCC. The definitive diagnosis of combined HCC-CC can only be established by histopathology. Although a unified criteria for combined HCC-CC is still not available, it is generally accepted that a firm diagnosis of combined HCC-CC requires evidence of HCC differentiation (Fig. 2a), such as trabecular growth pattern, bile production, or bile canaliculi as well as clear evidence of CC (Fig. 2b), such as true glandular structures formed by biliary-type epithelium, mucin production or prominent desmoplastic stoma.4,14 In addition, ideally, an interface of these two components showing they intermingle intimately with each other should also be appreciated (Fig. 3). Practically, a definitive diagnosis always requires the use of immunohistochemical and special stains to demonstrate both hepatocytic and biliary phenotypes. Commonly used stains include Hep Par 1 (Fig. 4a), polyclonal carcinoembryonic antigen (Fig. 4b), or CD10 for the hepatocytic differentiation, and mucin (Fig. 4c), CK7, and CK19 (Fig. 4d) stains for the biliary differentiation. Combined HCC-CC often expresses both biliary cytokeratins and markers of HCC, and is an important diagnosis to consider when there is an apparently conflicting or overlapping immunophenotype.15

Figure 2.

In combined hepatocellular-cholangiocarcinoma (HCC-CC), the HCC component shows (a) the typical trabecular and pseudo-acinar pattern and increased unpaired arteries (arrows), whereas the CC component shows (b) the typical desmoplastic stroma surrounding the neoplastic glands.

Figure 3.

Combined hepatocellular-cholangiocarcinoma (HCC-CC): (a) At low power magnification, the histology of combined HCC-CC shows both HCC (right, thick black arrow) and CC (left, thin arrows) components. There is an interface between both components (thick white arrow). (b) Higher magnification shows HCC (right) and CC (left) components. (c) The interface of HCC and CC components intimately intermingle with each other. (d) At the interface, single neoplastic cells (arrows) morphologically between HCC and CC are present.

Figure 4.

In the HCC component of combined hepatocellular-cholangiocarcinoma (HCC-CC), immunohistochemical staining for hepatocytic differentiation show (a) the positive cytoplasmic staining for Hep Par 1, (b, arrows) canalicular staining pattern for polyclonal carcinoembryonic antigen, but not the CC component, whereas (c, arrows) mucin stain and (d, left) CK19 immunostain are positive for the CC component, but not the HCC component. Notice there are (arrows) single CK19-positive cells with scant cytoplasm and round or oval nuclei, morphologically reminiscent of the hepatic progenitor cell or intermediate hepatocyte.

While fibrolamellar carcinoma, a variant of HCC, very rarely occurs in association with CC, a single case of combined fibrolamellar HCC-CC has been reported previously.16 Studies with a series of combined fibrolamellar HCC-CC are needed to further characterize this rare neoplasm.

Glypican-3 (GPC3) is a novel serological and immunohistochemical marker of hepatocellular carcinoma.17–19 A recent study to examine GPC3 immunoreactivity in combined HCC-CC shows the expression is sensitive and specific to the HCC component of combined HCC-CC but few cases also show weak immunoreactivity in the cholangiocarcinoma component of combined HCC-CC.20 While the positivity of GPC3 in the cholangiocarcinoma component may be a drawback, this antibody may offer as an additional immunohistochemical stain in diagnosing combined HCC-CC, if used along with other antibodies and in careful correlation with morphology.

Histogenesis of combined HCC-CC

The cell of origin of combined HCC-CC has been a matter of dispute. Overall three possibilities may be postulated regarding its cell of origin: (i) collision (double) tumor of HCC and CC that incidentally coexist in the same liver; (ii) subsequent differentiation of HCC or CC into the other component; and (iii) the cancer derives from the hepatic progenitor cells (Fig. 5). The fact that the HCC and CC elements intermingle with each other in a transitional area in most combined HCC-CC makes the first hypothesis less likely. Depending on various investigations, patients with combined HCC-CC share similar clinical and pathological features with patients with HCC4,21 or CC12,15,22 or the tumors are clinicopathologically different from those of CC11 or HCC.15 Caution needs to be practiced when interpreting these data according to how the diagnosis of combined HCC-CC was defined in these studies, such that the results may have been derived, and the results would have also differed given the difference in population or causes of underlying liver diseases. Apparently some studies used looser criteria. For example, it is well known that a significant number of HCC may stain positive for immunomarkers that are often used to indicate biliary differentiation, such as CK7 and CK19,23–25 which are by no means absolutely specific. Therefore, HCC with the so-called ‘pseudoglandular structure’ or reactivity with CK7 and/or CK19 by immunohistochemistry, but without the confirmation using a mucicarmine stain, in these studies may have been classified as combined HCC-CC and hence may have led to the conclusion that combined HCC-CC resembles HCC clinically and pathologically. Taken together, it appears that combined HCC-CC lacks consistent clinical characteristics and it may be explained by various studies across different geographic regions, causes, and populations.

Figure 5.

Postulated origin of combined hepatocellular-cholangiocarcinoma (HCC-CC). In general, three possibilities are postulated regarding the cell of origin of combined HCC-CC: (1) collision (double) tumor of HCC and CC that incidentally coexist in the same liver; (2) subsequent differentiation of HCC or CC into the other component; (3) the cancer derives from the hepatic progenitor cells.

A previous study using genome-wide allelotyping analysis has demonstrated that recurrent loss of heterozygosity at 3 p and 14 q are common in both CC and combined HCC-CC, whereas no beta-catenin was observed in combined HCC-CC and CC, a common mutation in HCC, suggesting common carcinogenic pathways shared by combined HCC-CC and CC,11 however, the clinical and demographic features of the patients with combined HCC-CC in this study were intermediate to those patients with HCC and CC, yet with a disproportionate male predilection (M : F = 14:1), closer to patients with HCC than CC.

Using microdissection, Fujii et al. found single clonal tumor with homogeneous genetic background in both HCC and CC components of combined HCC-CC, suggesting that histological diversity is a phenotypic expression of divergent differentiation potential of a single clone, or single clonal process in which genetic heterogeneity in the process of clonal evolution within the tumor parallel histological phenotype and the neoplasm is composed of mosaics of closely related subclones.10 Fujii's observations support the hypothesis the tumor is derived from a single clone, which shows bidirectional phenotypic diversity and the tumor phenotype may also be altered by the divergent genetic changes in the process of tumor progression, given the mosaics of closely related subclones with varied phenotypic potential. An earlier experiment showing a primary cell line derived from resected combined HCC-CC differentiated to not only the characteristics of HCC but also those of CC, also demonstrated that single clonally expanded tumor cell can give rise to both hepatocytic and biliary components.26 So much so that whether combined HCC-CC arises from malignant transformation of the hepatic progenitor cells or from de-differentiation of the malignant hepatocytes or cholangiocytes has remained a debatable issue.

Although it has been long speculated that the cancer cells of combined HCC-CC derive from the regenerative compartment of the liver, that is, the canal of Herring, where intermediate (transitional) cells reside,27,28 it is not until recently that some advanced investigations shed light to support the hypothesis that at least a subset of the combined HCC-CC may derive from the hepatic progenitor cells.29–31 For instance, Theise et al.31 demonstrated that cells morphologically and immunohistochemically resembling hepatic progenitor cells merged with the HCC and CC components and with mature-appearing hepatocytes within some combined HCC-CC, supporting the notions that carcinogenesis of this unique neoplasm may be explained by the malignant transformation of the hepatic progenitor cells. Furthermore, a recent study32 demonstrated the cell of origin of cholangiolocellular carcinoma (Fig. 6), a very rare neoplasm accounting for less than 1% of primary liver cancer,33,34 may also be the hepatic progenitor cells. Because the HCC (Fig. 6a) and CC (Fig. 6b) components altogether comprised less than 10% of the neoplasm and the cholangiolocellular carcinoma area (mixture of small monotonous glands, antler-like anastomosing pattern, Fig 6c) occupied more than 90% of the neoplasm in this study, although these three histological components showed transitions between each other, the exact relationship between this unique neoplasm and the typical combined HCC-CC remains to be clarified. It is possible they may overlap to some degree and belong to a spectrum of the primary liver neoplasm arising from the hepatic progenitor cells.

Figure 6.

In cholangiolocellular carcinoma, besides the classic (a) HCC and (b) CC components, a significant portion of neoplasm shows (arrows, c) mixtures of small monotonous glands in antler-like anastomosing patterns. Marked hyalinized and fibrous stroma is present. There is also lymphocytic infiltration.

Differential diagnosis and diagnostic pitfalls

For the purpose of diagnosis, combined HCC-CC needs to be distinguished from conventional HCC or CC. Pseudoglands (Fig. 7) reflecting rapid and active neoplastic replication are very common in HCC and they should not be confused with the true glandular formation in CC. In fact, Popper and Schaffner in 1957 stated that with careful examination most primary hepatic carcinomas could be found to have both hepatocellular and ductal elements,7 but Edmondson in the following year pointed out that in the majority of cases these ductal elements were from hepatocyte-like tumor cells and that such tumors are in fact a variant of HCC.3 Retrospectively, most of these ductal elements likely represent pseudoglands in HCC. In this regard, detection of mucin by a mucin stain in the CC component or identification of bile in the HCC component can be very helpful. As mentioned previously, it has been recognized that the expression of CK7 and CK19 in HCC is not uncommon from several series and therefore a diagnosis solely based on immunohistochemistry may not be fully reliable.23–25 In fact, a recent study using a comparative functional genomics approach has demonstrated that the CK19-associated gene expression signature may predict poor patient survival.35 Whether this poorer outcome can be largely attributed to the progenitor cell lineage of the carcinoma awaits further investigation. Although rarely used nowadays, electron microscopy demonstrating dual differentiation in different cells within the same acini, such as bile canaliculi and abundant organelles (mitochondria, rough endoplasmic reticulum, lipid droplets, and abundant glycogen—characteristic of hepatocellular differentiation), and microvillus projections at the luminal surface or junctional complexes between the neoplastic cells, features of biliary differentiation can be useful on some occasions when immunohistochemical and mucin stains are equivocal.21

Figure 7.

On hematoxylin–eosin, the pseudoglandular formation in hepatocellular carcinoma (HCC) due to rapid proliferation of the carcinoma cells may mimic true glandular structure in cholangiocarcinoma (CC) component of combined HCC-CCC.

Importantly, marked bile ductular reaction (Fig. 8) with atypia adjacent to HCC due to mass effect is not an uncommon finding and one should be cautious not to overinterpret these atypical ductules as the CC component of the combined HCC-CC. Unfortunately, immunohistochemistry is of little value in distinguishing malignant biliary epithelium from reacting ductules. In general, ductular reaction is often accompanied by inflammatory cell infiltrate, whereas the cholangiocarcinoma component is typically surrounded by a desmoplastic stroma that lacks inflammatory cells. Despite intensive preoperative imaging studies, many combined HCC-CC may be misdiagnosed either as HCC or CC before surgery. Tissue sampling is always an issue and may preclude an accurate diagnosis in a core needle biopsy specimen.

Figure 8.

Marked bile ductular reaction (thick arrows) next to hepatocellular carcinoma (HCC), especially within the cirrhotic bands, may mimic the cholangiocarcinoma (CC) component of combined HCC-CC. The accompanied inflammatory infiltrates (thin arrows) may help to distinguish from true adenocarcinoma component.

Clinical outcome

Accurate preoperative diagnosis is important because the decision on the most appropriate treatment may depend on the predominant component of the tumor (HCC or CC); however most patients with combined HCC-CC are seldom diagnosed before surgery. This may largely be attributed to the specimen sampled and unless the interface area is biopsied, a confident diagnosis of combined HCC-CC may not be reached. Detection and treatment option may be optimized with advanced imaging studies, high index of suspicion, serum tumor markers (alpha-fetoprotein, carbohydrate antigen 19-9), and histopathology with appropriate use of immunohistochemistry.36 Recent studies reported that the survival rate of patients with combined HCC-CC after liver resection was poorer than that of patients with HCC or CC37 and pathologically combined HCC-CC showed more significantly vascular invasion and lymph node metastasis than HCC, with a similarity to CC.22 These results are similar to those that were previously reported,38,39 suggesting that a more aggressive treatment modality, such as postoperative adjuvant therapy and multimodality treatment for recurrent disease, may have to be explored to improve the survival rate of these patients. There are very few outcome data on liver transplantation and the role of liver transplantation in combined HCC-CC needs to be defined.40 This is largely hindered by the lack of accurate preoperative diagnosis of combined HCC-CC. Further studies are also warranted to seek optimal therapeutic options in treating combined HCC-CC.41

In addition, in the recently published American Joint Committee on Cancer manual,42 combined HCC-CC is included in the section on Carcinoma of the Intrahepatic Bile Ducts. In the previous edition,43 it was categorized within primary liver cancer, which included both HCC and cholangiocarcinoma. Whether this recent categorization truly reflects the histogenesis of this unique neoplasm and the similarity of its biological behavior to cholangiocarcinoma is an issue yet to be resolved.


Although a rare cancer, combined HCC-CC is gaining recognition and histopathology remains the gold standard for its diagnosis. The clinical outcome of combined HCC-CC may differ from HCC and CC and a specific treatment modality towards this unique cancer may be required. It has become clearer that hepatic progenitor cells are present in not only HCC23,25,44 and CC45 but also in combined HCC-CC.29–32 The role of hepatic progenitor cells as the cell of origin in combined HCC-CC is an interesting subject but it has been limited by the current lack of an animal model. Ongoing studies may elucidate the pathways for the development of novel targeted therapy.


The author thanks Virginia Lore and Cynthia Long for their excellent assistance in preparing this manuscript.