Cholangiocarcinoma (CCA) is a relatively rare malignancy with a generally poor prognosis, often attributed to advanced stage at the time of diagnosis. The only potential cure is radical surgery. According to the World Health Organization (WHO) classification of gastrointestinal tumors, cholangiocarcinoma is regarded as a primary tumor arising from any part of the biliary tree including the intrahepatic bile ducts, gallbladder and extrahepatic bile ducts [1–3]. The left and the right hepatic bile ducts and their first to third branches are called hilar or perihilar bile ducts. These are followed proximally by the intrahepatic bile ducts. Intrahepatic CCA is a liver tumor that may arise in any part of an intrahepatic biliary tree. Intrahepatic carcinoma may be further subdivided into a peripheral subtype, which arises from the small intrahepatic bile ducts, and a hilar subtype, which involves the left or right bile duct or their junction. Combined hepatocelullar-cholangiocarcinoma (HCC-CCA) is a rare variant that closely resembles peripheral CCA [4].

Hilar CCA shares morphological, molecular and prognostic features with extrahepatic CCA [5] and pancreatic ductal adenocarcinoma (PDAC) [6, 7]. Peripheral intrahepatic CCA, on the other hand, has morphological, molecular and clinical similarities to combined HCC-CCA [4, 7]. Hilar and extrahepatic CCA and PDAC usually affect older people, are frequently recognized in advanced stages when they are unresectable and have a dismal prognosis. Histologically, these tumors display tubular, cribriform and tubulo-papillary arranged cylindrical epithelium. They express the immunohistochemical markers S100P and AGR2 [7, 8]. Peripheral CCA occurs in the peripheral parts of the liver as a large mass and microscopically consists of cuboid epithelium with anastomosing or trabecular architecture. Reliable makers for detecting of peripheral CCA include CRP immunohistochemistry [9] and in situ hybridization of albumin messenger RNA [10].

Distinction between peripheral CCA, hilar CCA and metastatic PDAC is of great clinical relevance because they are treated differently and have different prognoses. In many cases, both CCA subtypes may be distinguished radiologically by contrast-enhanced computed tomography (CE-CT) and magnetic resonance (MRI). Generally, CT and MRI have high sensitivity in detecting mass forming liver and pancreatic tumors but suffer from relatively low specificity in regard to predicting the histological type of the tumor. Differentiating between hilar and peripheral CCA can also be difficult, especially for large tumors [11]. Hilar and extrahepatic CCA and PDAC have a worse prognosis than intrahepatic peripheral CCA [6]. Resectability of CCA varies among institutions between 10 and 75% [12]. The extent and type of surgery differs substantially depending on the anatomical subtypes of CCA. Patients with metastatic PDAC do not benefit from surgical resection.

A histological or immunohistochemical marker of both peripheral and hilar type of CCA or metastatic PDAC may lead to accurate diagnosis in the case of unclear radiological finding. The forkhead box f1 (FOXF1, previously known as HFH-8 or Freac-1) is a homeobox gene that encodes a transcription factor expressed in the extraembryonic mesoderm, allantois, splanchnic mesoderm, and septum transversum mesenchyme [13]. It is involved in mesenchymal-epithelial signaling required for development of structures that arise from the foregut endoderm, such as the lung, gallbladder, and pancreas. FOXF1 haploinsufficiency has been described to cause malformations of pulmonary vessels and the colon in mice [14, 15]. Several recent works have linked FOXF1 deletions to lethal alveolar capillary dysplasia with misalignment of pulmonary veins (ACDMPV) in human infants [16–20]. Kalinichenko et al. detected expression of FOXF1 in the septum transversum and gall bladder in mice. In their study, mice with heterozygous expression of FOXF1 had severe structural abnormalities of the gallbladder and extrahepatic bile ducts involving both mesenchymal and epithelial parts of the organs [21]. FOXF1 expression has also been associated with high grade and advanced stage colorectal cancer [22].

All facts mentioned above concerning importance of FOXF1 in extrahepatic bile ducts and pancreas development lead us to the idea to test expression of FOXF1 in CCAs and metastatic pancreatic carcinomas by immunohistochemistry. The aim of this study is to discover if there is any difference in FOXF1 expression in the peripheral subtype of CCA vs. hilar and extrahepatic CCA and liver metastasis of PDAC. We hypothesize that FOXF1 may be expressed in the latter group of tumors due to its involvement embryonic development of distal bile ducts and pancreatic ducts.

Concerning age of patients, the mean was 67.1 years in Group 1 and 63.7 years in Group 2, the difference is not significant (p = 0.3956).

Morphological, histochemical and clinical similarities between hilar and extrahepatic CCA and PDAC have been described in recent studies [7]. Macroscopically, these tumors are frequently seen infiltrating the peribiliary soft tissue and growing along the bile ducts, whereas intrahepatic peripheral CCA more frequently manifests as a mass lesion [1]. Microscopically, hilar and extrahepatic CCAs have cylindrical epithelium and tubular architecture in contrast to the cuboidal epithelium and trabecular arrangements observed in intrahepatic peripheral CCAs. On the other hand, intrahepatic peripheral CCA and combined HCC-CCA tend to show anastomosing and trabecular histology and expression of hepatocellular markers such as CRP [9].

The differences in macroscopic, microscopic, histological, molecular and clinical properties of both CCA subtypes, combined HCC-CCA and PDAC may be explained by the different origins of these tumors. Hilar and extrahepatic CCA and PDAC are believed to arise from multipotent progenitor cells located in the peribiliary glands around the bile and pancreatic ducts [7]. These cells are of endodermal origin and are able to differentiate into hepatocytes, cholangiocytes and pancreatic cells [23]. Intrahepatic small bile ducts arise from hepatic progenitor cells via ductal plates at the terminal portal tracts [24]. The rare combined HCC-CCA may be proof of the continuous nature of liver tumor differentiation in the sense of hepatocellular-cholangiocellular phenotype. The biliary tract and pancreas are anatomically closely related. Embryologically, the extrahepatic bile ducts and the ventral pancreas arise from the diverticulum of the posterior ventral foregut almost at the same time and several transcription factors, such as Pdx1, Hes1, SOX9, and SOX17, are sequentially involved in the biliary and pancreatic differentiation [23]. There are studies documenting developmental defects of the pancreato-biliary system in SOX17-and Hes1-deficient mice [25, 26]. FOXF1 is a transcription factor involved in foregut and extrahepatic bile duct development. It is expressed mainly in fibroblasts, smooth muscle cells and endothelium [27, 28]. Several studies have described developmental defects of the extrahepatic bile ducts and gallbladder in FOXF1-deficient mice [21]. Moreover, FOXF1 has been described to play a significant role in cancerogenesis and cancer promoting molecular processes in colorectal cancer [22, 29, 30]. Expression of FOXF1 in lung cancer fibroblast promotes the cancer invasion and spread [31]. In HCC and breast cancer FOXF1 has been described as a tumor suppressor [32, 33]. On the other hand, FOXF1 has also been shown to be involved in tumor progression. In breast and lung cancer it has been shown to induce epithelial-mesenchymal transition, a crucial step in metastasis [27, 34]. All these findings suggest that the role of FOXF1 in tumorigenesis is extremely complex and tissue-specific. The association of FOXF1-deficiency with biliary tract malformations described in mice [21] and its importance in various carcinomas led us to use FOXF1 antibody as a potential immunohistochemical marker of CCA and metastatic PDAC. Moreover, we evaluated the histological pattern and CRP expression as a marker of intrahepatic peripheral CCA.

The results of our study display unreliability of pure histomorphology as there was no significant association in trabecular-anastomosing or tubular arrangement with the CCA subtype or metastatic PDAC. This finding does not corroborate findings by Gandou et al., who observed that columnar-tubular pattern was more frequently associated with perihilar CCA and PDAC and cuboidal-trabecular pattern was more frequently associated with intrahepatic peripheral CCA and combined HCC/CCA [6].

Although CRP immunohistochemistry was not the main aim of our study, we confirmed the association of CRP expression with intrahepatic peripheral CCA, which has been well described by Yeh et al. In their study, the sensitivity and specificity of CRP expression in the diagnosis of iCCA were 75.7% and 91.1% when using tissue microarrays and 93.3% and 88.2% when using whole tissue sections, respectively [9]. Our cohort displayed surprisingly similar sensitivity and specificity of 78.6% and 88.2%, respectively. Additionally to the corroboration of previous studies, this finding may signify accuracy of radiological diagnosis in our study.

To the best of our knowledge, this is the first study investigating FOXF1 as an immunohistochemical marker of CCA or metastatic PDAC. Our result suggests that FOXF1 is more frequently expressed in hilar and extrahepatic CCAs and metastatic PDACs in comparison to intrahepatic peripheral CCA and combined HCC/CCA. As an nuclear immunohistological marker, FOXF1 shows a fair specificity of 86.7% (only a few nuclear FOXF1 positive CCAs were peripheral) but low sensitivity of 64.7% (two thirds of hilar/extrahepatic CCAs and metastatic PDACs were FOXF1 negative). If we regard only nuclear positivity, FOXF1 may be labeled as a specific but insensitive marker of hilar and extrahepatic CCA or metastatic PDAC. Concerning cytoplasmic positivity of FOXF1, the cytoplasmic vs. nuclear localization of this transcription factor has been described to correlate with progression of colorectal cancer [22]. In the case of CCAs and PDACs, the significance of intracellular compartmentalization of FOXF1 remains unclear. However, our study showed lower statistical significance (p = 0.03) of cytoplasmic and nuclear FOXF1 to discern between CCA subtypes and metastatic PDAC in comparison with nuclear positivity only. The cytoplasmic and nuclear FOXF1 staining as a method to uncover hilar/extrahepatic CCA and metastatic PDAC showed higher sensitivity (76.5%) but poor specificity (66.7%) in comparison with solely nuclear staining. Like in many transcription factors detectable by immunohistochemistry used in the human pathology (i.e. CDX2, PAX2, PAX8, BOB1, SOX10, SOX11, SALL4, SATB2 etc.), we regard only nuclear FOXF1 positivity to have a diagnostic value; whereas the significance of cytoplasmic FOXF1 localization needs to be further clarified.

A surprising finding was the unreliability of histomorphological pattern in distinguishing the subtypes of CCA. From all the methods tested in the study, radiological finding, together with immunohistochemical CRP positivity may serve as hallmark features when making the diagnosis of intrahepatic peripheral CCA. In line with recent publications, there is no histopathological method to distinguish between hilar or extrahepatic CCA and metastatic PDAC. Our study is limited by its relatively small cohort size. However, there is a space for further research to identify diagnostic histopathological tests for both subtypes of CCA.