Introduction

Cholangiolocellular carcinoma is characterized by tumor cells arranged in anastomosing cords and glands in an abundant fibrous stroma [1,2,3,4,5,6,7]. This pattern is reminiscent of cholangioles or canals of Hering. Despite its initial description more than 50 years ago [1], cholangiolocellular carcinoma remains an incompletely understood entity, largely due to variability in its interpretation regarding terminology, criteria for diagnosis and cell of origin.

In its initial description and in the 2000 World Health Organization classification [1, 8], cholangiolocellular carcinoma was regarded as a subtype of intrahepatic cholangiocarcinoma. In World Health Organization 2010 classification, the tumor was classified as subtype of combined hepatocellular-cholangiocarcinoma with stem cell features [9], which led to confusion about this entity. An international group of liver experts has proposed consensus terminology which no longer supports the World Health Organization 2010 designation of stem cell features, but states that cholangiolocellular carcinoma should be classified as a primary liver carcinoma as there is insufficient evidence that it is a subtype of cholangiocarcinoma [10]. In contrast to this recommendation, several recent studies note the overlap of immunohistochemical and molecular characteristics of cholangiolocellular carcinoma and intrahepatic cholangiocarcinoma, suggesting that cholangiolocellular carcinoma is a subtype of intrahepatic cholangiocarcinoma rather than a distinct entity [11,12,13].

In addition to terminology, variability in diagnostic criteria of cholangiolocellular carcinoma has likely contributed to the difference in results across published studies. In its initial description [1], branching cords and ductular configuration with or without admixed small glands were typical features attributed to cholangiolocellular carcinoma, and this criterion has been used in most studies and case reports [2, 3, 5, 6, 13,14,15,16,17,18,19,20]. In other studies, however, uniform small glands without a branching configuration was considered sufficient for the diagnosis of cholangiolocellular carcinoma [12, 21]. The initial description of cholangiolocellular carcinoma included cases with ‘anaplastic lining cells’ [1], and ‘poorly differentiated’ cholangiolocellular carcinoma cases have been described in some studies [20]. On the other hand, a vast majority of studies with cholangiolocellular carcinoma have included cases with low grade morphology that resembles ductular reaction [3,4,5, 7, 19]. A cutoff of 90% cholangiolocellular carcinoma component is used for diagnosis of pure cholangiolocellular carcinoma in some publications [3], while others have used a cutoff of 50% [6] and 80% [10, 22].

The origin of cholangiolocellular carcinoma has been variously proposed from stem cells, canals of Hering and interlobular bile ducts. Liver stem cells/progenitor cells have been postulated to reside in the regions of canals of Hering; cord-like and ductular-like features of cholangiolocellular carcinoma has led to the speculation that this tumor has stem cell characteristics [3]. Other studies have proposed origin from interlobular bile duct based on morphometric studies [16, 18]. Immunohistochemical staining for markers like CD133, CD117, NCAM, OCT4, and SALL4, as well as biliary epithelial markers like CK7 and CK19, has been cited as evidence to support a stem cell phenotype in cholangiolocellular carcinoma [3, 16]. This led to the characterization of cholangiolocellular carcinoma as a subtype of combined hepatocellular-cholangiocarcinoma in the World Health Organization 2010 classification [10]. Other differences in immunohistochemical staining such as membranous pattern with EMA in cholangiolocellular carcinoma and cytoplasmic pattern in intrahepatic cholangiocarcinoma have been reported [5, 16]. In contrast, other studies have reported that immunohistochemical staining profile of cholangiolocellular carcinoma using one or more of these putative stem cell markers is similar to small duct intrahepatic cholangiocarcinoma [5, 11].

The aim of this study is to examine the immunohistochemical and molecular features of well-characterized and strictly defined cholangiolocellular carcinoma cases, and compare them with well-differentiated intrahepatic cholangiocarcinoma.

Materials and methods

Case selection

The study group comprised 17 cases including cholangiolocellular carcinomas (n = 5), well-differentiated intrahepatic cholangiocarcinomas (n = 7), and combined cholangiolocellular and well-differentiated intrahepatic cholangiocarcinomas (n = 5). The cases were retrieved from the case files of University of California San Francisco, Southern California Kaiser Permanente and Yale University hospitals. The study was approved by the institutional review boards of participating institutions. Hematoxylin and eosin-stained sections were reviewed by two liver pathologists to confirm the diagnosis (SK, LF). Cholangiolocellular carcinoma component was defined based on extent of nuclear atypia and architectural growth pattern. Low grade nuclear atypia and a branching cord-like anastomosing appearance resembling ductular reaction embedded in a fibrous stroma were considered as defining features of cholangiolocellular carcinoma (Fig. 1). Tumors composed predominantly of small glands without an anastomosing cord-like pattern or tumors with branching configuration but high grade nuclear atypia were not considered as cholangiolocellular carcinoma. If the cholangiolocellular carcinoma component comprised 90% or more of the tumor, it was designated as cholangiolocellular carcinoma. Well-differentiated intrahepatic cholangiocarcinomas showed >95% glandular differentiation; moderately and poorly differentiated intrahepatic cholangiocarcinomas were not included in the study. Cholangiolocellular carcinoma component was absent or comprised <10% of the tumor in all selected intrahepatic cholangiocarcinoma cases. All intrahepatic cholangiocarcinomas were peripheral tumors with predominant small duct morphology. Tumors with at least 10% of each component (but less than 90% cholangiolocellular carcinoma component) were designated as mixed cholangiolocellular carcinoma-intrahepatic cholangiocarcinoma (Fig. 2). Tumors with a component of hepatocellular carcinoma were excluded even if cholangiolocellular carcinoma component was present.

Fig. 1
figure 1

Cholangiolocellular carcinoma characterized by branching cord-like configuration in a fibrous stroma (a, H&E stain, ×100). The tumor cells show mild nuclear atypia (b, H&E stain, ×200)

Fig. 2
figure 2

Mixed cholangiolocellular-intrahepatic cholangiocarcinoma with a characteristic cholangiolocellular carcinoma component characterized by branching configuration (long arrow) and intrahepatic cholangiocarcinoma component composed of discrete well-formed glands (short arrow) (a, H&E stain, ×40). A closer view of the cholangiolocellular component (b, H&E stain, ×200) and intrahepatic cholangiocarcinoma (c, H&E stain, ×200) are illustrated. Both these components were seen in the same tumor

Immunohistochemistry

Immunohistochemistry was performed on whole tissue sections in 16 cases with an automated staining process (Leica BOND 3 and the Ventana Ultra automated stainers) at the University of California San Francisco Immunohistochemistry Laboratory. Immunohistochemistry could not be performed in 1 intrahepatic cholangiocarcinoma as the tissue in the block was exhausted after sequencing analysis. The details of antibodies and dilutions used are summarized in Table 1. Moderate to strong staining in 10–50% of tumor cells was considered focal positive and >50% of tumor cells as diffuse positive. All other results were considered negative. For EMA, cytoplasmic and luminal patterns of staining were recorded for each tumor. Staining for hepatocellular markers performed during the diagnostic work-up of the cases was reviewed.

Table 1 Details of antibodies used for immunohistochemistry

Capture-based next-generation DNA sequencing

Genomic DNA was extracted from formalin-fixed paraffin-embedded tumor tissue from each case, including separate cholangiolocellular and intrahepatic cholangiocarcinoma areas from the 5 mixed cholangiolocellular-intrahepatic cholangiocarcinoma cases. Sequencing libraries were prepared from genomic DNA and target enrichment was performed by hybrid capture using custom baits designed to target all coding regions of 479 cancer-related genes, select introns from 47 genes, and the TERT promoter, with a total sequencing footprint of ~2.9 Mb (Supplementary Table S1). The baits also capture 2000 unique sequences containing common single nucleotide polymorphisms within regions devoid of constitutional copy number variations to assist in genome-wide copy number and allelic imbalance analysis. Sequencing was performed on a HiSeq 2500 (Illumina, San Diego, CA). Duplicate sequencing reads were removed computationally to allow for accurate allele frequency determination and copy number calling. The analysis was based on the human reference sequence UCSC build hg19 (NCBI build 37), using the following software packages: BWA: 0.7.13, Samtools: 1.1 (using htslib 1.1), Picard tools: 1.97 (1504), GATK: Appistry v2015.1.1–3.4.46–0ga8e1d99, CNVkit: 0.7.2, Pindel: 0.2.5b8, SATK: Appistry v2015.1.1–1-gea45d62, Annovar: v2016 Feb01, Freebayes: 0.9.20 and Delly: 0.7.2 [23,24,25,26,27,28,29,30,31,32,33]. Pathogenic and likely pathogenic somatic mutations were identified by first excluding all but non-synonymous, splice-site, and TERT promoter mutations. Next, we excluded variants that had >0.005 frequency in the Exome Sequencing Project (esp6500) (http://evs.gs.washington.edu/EVS/), Exome Aggregation Consortium (ExAc) [34] or the 1000Genomes database [35]. The remaining somatic variants were manually reviewed and annotated using knowledge present in the following databases: Catalog of Somatic Mutations in Cancer (COSMIC, https://cancer.sanger.ac.uk/cosmic), cBioPortal for Cancer Genomics (www.cbioportal.org), ClinVar (https://www.ncbi.nlm.nih.gov/clinvar/), and PubMed (https://www.ncbi.nlm.nih.gov/pubmed/). Genome-wide copy number analysis based on on-target and off-target reads was performed using CNVkit [29] and Nexus Copy Number (Biodiscovery, Hawthorne, CA). Focal copy number changes were defined as those encompassing less than 10 megabases. Large scale chromosomal changes were defined as those involving entire chromosomes or chromosome arms.

Results

Clinicopathologic features

The average age at diagnosis was 60 years (range 27–73 years), and 31% of patients were male (Supplementary Table S2). The average tumor size was 5.7 cm (range 1.4–13.5 cm). Cirrhosis was present in 1 (20%) mixed cholangiolocellular-intrahepatic cholangiocarcinoma (alcoholic liver disease) and 4 (67%) intrahepatic cholangiocarcinoma cases (2 hepatitis C, 1 each non-alcoholic steatohepatitis and hepatitis B); none of the cholangiolocellular carcinoma cases had underlying cirrhosis. One mixed cholangiolocellular-intrahepatic cholangiocarcinoma developed in the setting of chronic hepatitis C with bridging fibrosis.

Immunohistochemistry

Immunohistochemistry was performed on 16 cases (5 cholangiolocellular carcinomas, 6 intrahepatic cholangiocarcinomas, 5 mixed cases; Table 2). CK19 was positive in all cases with diffuse staining in 4/5 cholangiolocellular carcinoma, 5/6 intrahepatic cholangiocarcinomas and 4/5 mixed cases. CD56 was positive in 2/5 cholangiolocellular carcinomas (1 focal, 1 diffuse), 4/6 intrahepatic cholangiocarcinomas (2 focal, 2 diffuse) and 2/5 mixed cases (1 focal, 1 diffuse) (Fig. 3, Table 2). EMA was positive in all cases; both luminal and cytoplasmic staining was present in 3/5 cholangiolocellular carcinomas (luminal only in 2 cases), 3/6 intrahepatic cholangiocarcinomas (cytoplasmic only in 2 cases; luminal only in 1 case) and 4/5 mixed cases (luminal only in 1 case) (Fig. 4). CD117 and SALL4 were negative in all cases. Hepatocellular markers (arginase-1, Hep Par 1 and/or glypican-3) had been performed during the diagnostic work-up in 10 cases (3 cholangiolocellular carcinomas, 3 intrahepatic cholangiocarcinomas, 4 mixed cases), and were negative in all tumors (Supplementary Table S2).

Table 2 Immunohistochemical results in cholangiolocellular carcinoma (CLC), intrahepatic cholangiocarcinoma (ICC), and mixed CLC-ICC
Fig. 3
figure 3

Variable immunohistochemical staining for CD56 in cholangiolocellular carcinoma: strong membranous staining (a, ×200) and largely negative in another case (b, ×200)

Fig. 4
figure 4

Immunohistochemical staining for EMA showing luminal staining (right, ×100) and cytoplasmic staining (left, ×100) in different areas of the same cholangiolocellular carcinoma

Genomic alterations

The mean target sequencing coverages for cholangiolocellular carcinoma, mixed cholangiolocellular-intrahepatic cholangiocarcinoma, and intrahepatic cholangiocarcinoma were 415 ± 207, 597 ± 104, and 558 ± 128 unique reads per target interval, respectively (Supplementary Table S3). Recurrent pathogenic and likely pathogenic alterations are summarized in Fig. 5. Variant details, including genomic coordinates and mutant allele frequencies, are provided in Supplementary Table S4. Recurrent pathogenic alterations in IDH1/2 and chromatin regulators (PBRM1, ARID1A, BAP1) were present in all three tumor categories. IDH1/IDH2 mutations were seen in 2 (40%) cholangiolocellular carcinomas, 1 (14%) well-differentiated intrahepatic cholangiocarcinoma, and 3 (60%) mixed cholangiolocellular-intrahepatic cholangiocarcinomas (Table 3, Supplementary Table S4). Truncating mutations or deep deletions in PBRM1, ARID1A, or BAP1 were present in 60%, 43%, and 80% of cholangiolocellular carcinomas, intrahepatic cholangiocarcinomas and mixed cholangiolocellular-intrahepatic cholangiocarcinomas, respectively. When all cases with cholangiolocellular carcinoma component were considered (both cholangiolocellular carcinomas and cholangiolocellular component of mixed tumors), IDH1/IDH2 were observed in 50%, and PBRM1, ARID1A, or BAP1 mutations in 70% of cases. In all mixed cholangiolocellular-intrahepatic cholangiocarcinomas, both components showed an identical mutational profile. Other changes that have been commonly described in the literature in intrahepatic cholangiocarcinoma were also observed in cases with cholangiolocellular carcinoma component: FGFR2 alteration (10%), ATM mutation (10%), and one case each of amplification involving CCND1 and MDM2 (Fig. 5, Supplementary Table S5).

Fig. 5
figure 5

Genomic changes in cholangiolocellular carcinoma, mixed cholangiolocellular-intrahepatic cholangiocarcinoma, and well-differentiated intrahepatic cholangiocarcinoma

Table 3 Genomic changes in cholangiolocellular carcinoma (CLC), intrahepatic cholangiocarcinoma (ICC), and mixed CLC-ICC

Copy number alterations

Copy number alterations were evaluated in 14 cases (Table 4); the analysis could not be performed in the remaining 3 cases due to low (less than 30%) tumor content. All three groups of cases showed frequent losses in 1p, 3p, 6q, 12q, and 14q, and frequent gains in 1q. The only notable differences between the groups was frequent gains in chromosome 7 in mixed cholangiolocellular-intrahepatic cholangiocarcinomas and intrahepatic cholangiocarcinomas, whereas this change was not seen in cholangiolocellular carcinomas.

Table 4 Copy number changes in cholangiolocellular carcinoma (CLC), mixed CLC-ICC and intrahepatic cholangiocarcinoma (ICC)

Discussion

Cholangiolocellular carcinoma has been variously classified as a subtype of intrahepatic cholangiocarcinoma, combined hepatocellular-cholangiocarcinoma with stem cell features and a unique primary liver carcinoma [5, 8,9,10]. Its origin from stem cells, cholangioles and interlobular bile ducts has been debated [3, 5, 16, 18]. Our results show that the immunohistochemical features, mutational spectrum and copy number alterations in cholangiolocellular carcinoma are similar to well-differentiated intrahepatic cholangiocarcinoma.

The mutational profile of intrahepatic cholangiocarcinomas has been characterized in numerous studies [36,37,38,39,40,41,42,43,44,45,46,47]. The most frequent genomic alterations in intrahepatic cholangiocarcinomas occur in IDH1/2, ARID1A, PBRM1, BAP1, CDKN2A, and FGFR2. Other recurrently altered genes include KRAS, TP53, CCND1, CDK4, and PIK3CA. Among these, many including KRAS, PIK3CA, TP53, ARID1A, CDKN2A, and CCND1 are also frequently altered in other gastrointestinal adenocarcinomas, but IDH1/2, PBRM1, BAP1, and FGFR2 are relatively specific for intrahepatic cholangiocarcinoma. In our study, 60% of cholangiolocellular carcinomas and 80% of mixed cholangiolocellular-intrahepatic cholangiocarcinomas had alterations in IDH1/2, PBRM1, or FGFR2; these genomic alterations are rare or absent in other gastrointestinal adenocarcinomas including extrahepatic biliary tree, pancreas, stomach and colon [42], emphasizing the genetic similarity of cholangiolocellular carcinoma and intrahepatic cholangiocarcinoma. Other genetic alterations seen in cholangiolocellular carcinoma and mixed cholangiolocellular-intrahepatic cholangiocarcinomas in our study that have been reported in intrahepatic cholangiocarcinomas in the literature include mutations in ARID1A (19–36%) and ATM (1.8–3%), and amplifications involving CCND1 (1–13%), and MDM2 (5–12%) [38, 40, 47]. If less specific but commonly observed changes in intrahepatic cholangiocarcinoma such as ARID1A mutation are included, genomic changes typical of intrahepatic cholangiocarcinoma were present in 90% of cases with cholangiolocellular component. In addition, all tumors with mixed cholangiolocellular-intrahepatic cholangiocarcinomas in our series showed the same genomic profile in both components. These results strongly support the conclusion that cholangiolocellular carcinoma is a histologic subtype of intrahepatic cholangiocarcinoma, and is not a distinct entity or a subtype of combined hepatocellular-cholangiocarcinoma.

In our study, copy number alterations in cholangiolocellular carcinoma cases were similar to intrahepatic cholangiocarcinomas. The most common changes in cholangiolocellular carcinoma were gains involving 1q, and deletions involving 1p, 3p, 6q, 12q, and 14q. All these changes have been reported as common copy number alterations in intrahepatic cholangiocarcinomas across multiple studies [48,49,50,51,52]. Losses involving 3p and 14q have been considered particularly characteristic of intrahepatic cholangiocarcinomas as these are rare in extrahepatic cholangiocarcinoma [52]; 3p deletion was present in all cholangiolocellular carcinomas in our series while both 3p and 14q loss was present in 75% of cases. Although based on a limited number of cases, the cytogenetic data also provides support for the similarity between cholangiolocellular carcinoma and intrahepatic cholangiocarcinoma.

It has been argued that cholangiolocellular carcinoma shows stem/progenitor cells features and immunohistochemical expression of a variety of putative stem cell markers such as CD133, CD117, NCAM, OCT4, SALL4, CD44, CK19, and NCAM [3, 9, 10, 16, 53,54,55] has been reported in cholangiolocellular carcinomas. However, none of these markers are specific for stem cells and it has recently been argued by an international group of liver pathologists that these stains cannot be considered to define stem cell characteristics [10]. Our study did not find significant differences in immunohistochemical results in cholangiolocellular carcinomas and intrahepatic cholangiocarcinomas using CK19 or putative stem cell markers like CD56 and CD117, although only a small number of cases were examined. Our results are similar to the findings in another recent study [12]. Luminal staining with EMA has been reported in cholangiolocellular carcinoma compared with cytoplasmic staining in intrahepatic cholangiocarcinoma [5, 10, 16], but this difference was not observed in our series based on a limited number of cases, and both patterns were observed in different areas of the same cholangiolocellular carcinoma. Detailed morphometric studies have been used to argue against the cholangiolar origin of cholangiolocellular carcinoma, and have favored origin from interlobular bile duct [16, 18]. Hence the unique nature of cholangiolocellular carcinoma based on stem cell or cholangiolar origin remains speculative, and does not provide valid arguments against the overall evidence of cholangiolocellular carcinoma being a subtype of intrahepatic cholangiocarcinoma.

Several recent studies have suggested subtyping of intrahepatic cholangiocarcinoma into large duct and small duct types [7, 17, 19, 22], and it has been proposed that cholangiolocellular carcinoma may be the same as small duct intrahepatic cholangiocarcinoma [17]. The lack of a clear morphologic boundary between small duct intrahepatic cholangiocarcinoma and cholangiolocellular carcinoma has been noted [19]. IDH1 mutation is characteristic of small duct intrahepatic cholangiocarcinoma [40, 45], further supporting this interpretation. Based on sequencing results, one study concluded that cholangiolocellular carcinomas is a distinct entity even though the discussion section of this study states that cholangiolocellular carcinoma should be regarded as an intrahepatic biliary carcinoma [11]. IDH1 mutation was identified in 40% of cholangiolocellular carcinomas in this study, further emphasizing the similarity to intrahepatic cholangiocarcinoma rather than being a distinct entity. Based on the definition of cholangiolocellular carcinoma provided in the World Health Organization 2010 classification, cholangiolocellular carcinoma cases without a hepatocellular component were erroneously labeled as combined hepatocellular-cholangiocarcinoma in several studies [53,54,55]. Cholangiolocellular carcinoma does not share morphologic, immunohistochemical or genomic characteristics with hepatocellular carcinoma, and should not be regarded as a combined hepatocellular-cholangiocarcinoma unless a distinct hepatocellular carcinoma component is present.

Outcome data was not available in our study. A favorable outcome for cholangiolocellular carcinomas compared with intrahepatic cholangiocarcinoma has been suggested [12], but the lack of strict diagnostic criteria makes these results unreliable as small duct intrahepatic cholangiocarcinoma was considered as cholangiolocellular carcinoma in this study. Since cholangiolocellular carcinomas are mostly well-differentiated tumors, their prognosis should be compared with well-differentiated intrahepatic cholangiocarcinoma for a proper comparison; survival comparison with intrahepatic cholangiocarcinoma as a group is likely to skew the results based on moderately and poorly differentiated cases in the intrahepatic cholangiocarcinoma group. Hence the prognostic significance of pure cholangiolocellular carcinomas or cholangiolocellular component in intrahepatic cholangiocarcinoma remains to be elucidated. The presence of cholangiolocellular carcinoma component can be noted in pathology reports, but does not currently bear any clinical connotations. There is currently no therapeutic difference between cholangiolocellular carcinoma and intrahepatic cholangiocarcinoma.

In summary, the immunohistochemical results, mutational profile and copy number alterations show that cholangiolocellular carcinoma is similar to intrahepatic cholangiocarcinoma, and is best regarded as a histologic subtype of small duct intrahepatic cholangiocarcinoma. There is insufficient evidence to conclusively establish stem cell characteristics in cholangiolocellular carcinomas based on immunohistochemistry. It is important to use strict diagnostic criteria for cholangiolocellular carcinoma to enable meaningful comparisons across different studies.