Perihilar Cholangiocarcinoma Originating in Peribiliary Glands: Insights from a Case without Precancerous Lesions

Introduction

Insight into embryogenesis and the use of genetic lineage tracing systems in experimental models have allowed us to identify the cells of origin of cholangiocarcinoma (CCA), and genomic evidence in support of the distinct cell of origin hypotheses is growing. For example, peribiliary glands (PBGs) can represent the potential cell origin of CCA as a stem cell niche [1–3]. In fact, precancerous lesions in CCA, such as biliary intraepithelial neoplasm (BilIN) and intraductal papillary neoplasm of the bile duct (IPNB), have been reported as contained at the base of the PBG in humans [4–12]. However, whether these precancerous lesions that originate from PBGs progress to CCA is controversial [13], and only a few reports have suggested that CCA originates from PBGs [14–16].

Herein, the case of an adult patient with obstructive jaundice in which no carcinoma cells were detected histologically before autopsy is presented. She was treated for a diagnosis of sclerosing cholangitis and died about 11 months after visiting our institute. The unique distribution of carcinoma cells and detailed microscopic investigation on autopsy suggested that the carcinoma arose from the PBG and grew locally within the PBG compartment without invading the bile duct mucosa. Interestingly, no abnormalities, such as stenosis or tumorous lesions, had been observed in the biliary tract 1 year earlier, when she had been treated for choledocholithiasis.

This case report may inform our understanding of the natural history of CCA, providing clues about cholangiocarcinogenesis and progression.

Case Report

A 75-year-old woman with concerns of jaundice consulted a gastroenterologist of our institute. Contrast-enhanced computed tomography (CT) showed stenosis, with wall thickening at the bifurcations of the right posterior and anterior sectoral ducts from the left hepatic duct. The wall of the common hepatic duct was also thickened, but it was locally dilated (Figure 1A, 1B). Magnetic resonance cholangiopancreatography visualized these stenoses and dilatation stereoscopically (Figure 1C). She had a history of a confluence stone packed at the orifice of the cystic duct that had been treated by electrohydraulic lithotripsy under direct cholangioscopic control, followed by laparoscopic cholecystectomy about 1 year earlier in our institute. At that time, endoscopic retrograde cholangiopancreatography (ERCP; Figure 2A) and drip infusion cholecystocholangiography CT before cholecystectomy (Figure 2B) showed no abnormality of the bile duct, such as stenosis. In addition, no malignancy was detected histologically in the specimen of the resected gallbladder and cystic duct. Liver function test results showed that total bilirubin, alanine transaminase, aspartate transaminase, and gamma-glutamyl transpeptidase levels were increased, at 8.25 mg/dL (reference range 0.30–1.20 mg/dL), 530 U/L (reference range 5–45 U/L), 357 U/L (reference range 10–40 U/L), and 1555 U/L (reference range <48 U/L), respectively. In contrast, inflammatory indicators such as the white blood cell count and C-reactive protein level were within almost normal ranges, at 5980/μL (reference range 3500–9700/μL) and 0.43 mg/dL (reference range <0.30 mg/dL) respectively. Serological studies showed a normal immunoglobulin subclass 4 (50 mg/dL, range 11–121 mg/dL), normal carcinoembryonic antigen (1.7 ng/mL, range <5.0 ng/mL), and increased carbohydrate antigen 19-9 (CA19-9) (292.0 U/mL, range <37 U/mL), elevated from the value of 210.2 U/mL at the time of cholelithiasis 1 year earlier, and that of 20.1 U/mL after treatment of cholelithiasis 7 months earlier, but not extremely. ERCP, direct cholangioscopy (SpyGlass DS; Boston Scientific Japan, Tokyo, Japan), and intraductal ultrasonography (IDUS; UMQ240; Olympus, Tokyo, Japan) were performed to make a diagnosis, followed by drainage. ERCP showed multiple non-continuous, smooth stenoses at the right posterior sectoral duct root, common hepatic duct, and upper common bile duct, and dilation of intrahepatic bile ducts (Figure 3). Direct cholangioscopy showed the whitish smooth surface mucosa suspected to be fibrosis throughout the bile duct, including around the stenoses (Figure 4A, 4B, 4F). In the small parts of the tightest stenoses of the common hepatic duct and common bile duct, relatively well-ordered, small blood vessel distension was observed (Figure 4E), but none of the findings characteristic of malignancy, such as villous mass, papillary projection, granular mucosa, or irregular mucosal nodularity [17,18] was observed. IDUS showed continuous thickening of the bile duct wall from around the bifurcation of the right posterior sectoral duct from the left hepatic duct (Figure 4B, 4C) to the lower common bile duct (Figure 4F), including at the bifurcation of the right anterior sectoral duct from the common hepatic duct (Figure 4D) and at the common hepatic duct stenosis (Figure 4E). The bile duct wall thickening extended beyond the site of stenoses indicated on ERCP. Biopsy was performed under direct cholangioscopy using biopsy forceps with a 1.0-mm jaw outer diameter (SpyBite; Boston Scientific Japan) at the stenosis of the common bile duct. Transpapillary biopsy was also performed at the common bile duct, where wall thickening was detected by IDUS using side-opening biopsy forceps with a 2.2-mm jaw outer diameter (FB-45Q-1; Olympus). Biopsy at the sites of small vessel distension was not attempted, because of the risk of bleeding. Bile was also collected during cholangioscopy for cell block cytology. Finally, endoscopic nasal biliary drainage (ENBD) was performed to complete the procedure. No malignant finding was detected in the 3 specimens obtained by SpyBite, 4 specimens of transpapillary biopsy, and a bile cell block. ERCP performed 3 weeks after ENBD for replacement of inside stents instead of ENBD showed that the stenoses extended to the anterior and posterior sectoral duct roots more clearly (Figure 5A). After that, ursodeoxycholic acid was given for sclerosing cholangitis.

Five months after the replacement of the inside stents, jaundice recurred due to stent dysfunction. Contrast-enhanced CT at this time showed dilatation of intrahepatic bile ducts upstream of inside stents, while it did not show findings suggesting extramural infiltration, such as bile ducts wall severe thickening and mass forming. Serological studies showed an increased CA19-9 (1573.3 U/mL). ERCP after treatment by ENBD showed that the stenosis extended from the left hepatic duct and right anterior sectoral duct branches to the lower end of the common bile duct continuously, and the right posterior sectoral duct was not visualized (Figure 5B). Five specimens acquired from the stenoses of the common hepatic duct and common bile duct by transpapillary biopsy showed no malignant findings. The inside stents were again replaced. Serological studies 1 month later still showed an increased CA19-9 (1376.8 U/mL), but it was slightly decreased from the previous value. Three months later, the patient developed acute obstructive suppurative cholangitis and a liver abscess. ENBD was performed, but was ineffective, and she died after an overall course of 11 months.

At autopsy, the common bile duct wall was irregularly thickened macroscopically, and the lumen was partially severely constricted, but the surface was smooth (Figure 6A). Severe adhesion was observed between around the common hepatic duct and the duodenal bulb (Figure 6B). Microscopically, with low magnification, periductal fibrous thickening was observed from the left hepatic duct to the lower common bile duct (Figure 6C–6E). With high magnification, well-differentiated adenocarcinoma with desmoplastic stroma was found in a periductal infiltrating pattern (Figure 6F), and carcinoma cells in the bile duct lumen were hardly seen in any of the bile duct excisions. The carcinoma cells invaded the hepatoduodenal ligament with severe perineural invasion, and they infiltrated not only the retroperitoneum around the artery and portal vein, but also the perihilar hepatic tissue and head of the pancreas. Immunohistochemical study using anti-laminin antibody (Novocastra, Newcastle upon Tyne, United Kingdom) delineated the basement membrane of the pre-existing intramural PBG and glandular canal (Figure 7A, 7D). The hematoxylin and eosin-stained specimens showed that in situ-like carcinoma cells spread, replacing the epithelium of the intramural PBG (Figure 7C, 7F). The epithelium of the most of the PGBs and glandular canals was replaced by carcinoma cells (Figure 7B, 7E), while some intact PBGs were also seen. The carcinoma cells in the bile duct lumen were hardly seen. The observed distribution of the carcinoma cells is shown schematically in Figure 7G and 7H. These findings were widely observed in many portions from the hilar bile duct to the lower common bile duct regardless of the presence or absence of extramural invasion. Immunohistochemical study using the antibody to E-cadherin (Leica Biosystems, Newcastle upon Tyne, United Kingdom), which has been thought to be one of the key molecules of cholangiocarcinogenesis and progression, showed that the expression of E-cadherin decreased in the in situ-like carcinoma cells in the PBG (Figure 8A–8D). E-cadherin was lost in the adenocarcinoma cells at the invasive portion to outside the bile duct wall (Figure 8E, 8F).

Discussion

The PBG can be subdivided in intra- and extramural glands, and these are located in the fibrous bile duct wall and in the loose connective tissue surrounding bile duct wall, respectively. PBG have a direct connection with the lumen via glandular canals. The PBG network is a set of anatomical complexes consisting of PBGs, the peribiliary vascular plexus, and the lymphatic and nervous systems [15,19,20–22]. It has been proven that PBGs elongate and form intricate intramural epithelial networks that communicate between different segments of the bile duct mucosa via glandular canals that run parallel to the central bile duct lumen in a mouse model [23]. Figure 7G and 7H show schematically the PBG network structure based on previous reports [13,23–26]. The observed distribution of the carcinoma cells examined with serial sections of the present case is shown by the red lines in these figures. The in situ-like carcinoma cells spread, replacing the epithelium of the intramural PBGs and the glandular canals without invading the bile duct mucosa in many portions throughout the extrahepatic bile duct regardless of the presence or absence of extramural invasion. It needs to be noted that because of the extramural invasion, it was difficult to isolate the distribution of in situ-like carcinoma cells in extramural PBGs.

The changes in the spread of bile duct stenoses on cholangiography during the clinical course and the characteristic and unique distribution of the in situ-like carcinoma cells suggested that the CCA arose from the PBG in the hilum, without a detectable precancerous lesion on the luminal surface of bile duct, and then spread to the lower end of the common bile duct via the PBG network, and finally invaded to outside the bile duct wall. The findings support strongly the hypothesis that not only precancerous lesions such as BilIN and IPNB, but also invasive cancers originate from PBGs in humans. Only 2 human cases suggesting that CCAs spread via the PBG network have been reported [15]. Nakagawa et al developed a mouse model of injury-related extrahepatic CCA with invasive periductal infiltration [27]. They suggested that the epithelial cells lining the bile duct lumen detached due to E-cadherin loss, whereas the cells could survive in the PBG, considered a stem cell niche, leading eventually to the development of carcinoma and periductal infiltration. Immunohistochemical study using anti-E-cadherin antibody in the present case showed that the expression of E-cadherin was lost not only in the infiltrating carcinoma cells, but also in the in situ-like carcinoma cells at the cervical portion of the PBG, similar to the mouse model.

In the present case, in situ-like carcinoma cells spread, replacing the epithelium of the bile duct at the intramural PBGs and glandular canals throughout the extrahepatic bile duct. When considering cholangiocarcinogenesis and progression, it is important to discuss whether these in situ-like carcinoma cells are high-grade BilIN, as a precancerous lesion. In general, neoplastic epithelia of high-grade BilIN involve the non-neoplastic, surrounding biliary mucosa and non-neoplastic glands, including the PBGs located at various depths from the luminal surface of the biliary tract [12]. On the other hand, in the present case, in situ-like carcinoma cells were scarce in the luminal surface. Furthermore, no abnormalities had been observed in the biliary tract 1 year earlier, when she had been treated for choledocholithiasis at the orifice of the cystic duct and had undergone laparoscopic cholecystectomy in our institute. At that time, no stenosis of the biliary lumen had been observed on ERCP and 3D-CT. In addition, contrast-enhanced CT had not shown intra- and periductal tumor of suspected BilIN or IPNB, which are considered to be precancerous lesions. Whereas BilIN usually does not produce clinical symptoms and is not detectable on imaging studies [12], considering the progression to advanced cancer in 1 year, it is conceivable that the CCA might have developed at the PBG without going through a detectable precancerous lesion, such as BilIN or IPNB. Recently, Lin et al proposed a BilIN-independent, “de novo” development path of gallbladder cancer by examining the genetic relationships among low-grade and high-grade BilIN and adenocarcinoma [28]. It is possible that the CCA in the present case arose in the de novo development path. Furthermore, if cancer develops, even well-differentiated adenocarcinoma progresses rapidly with periductal invasion. It remains unclear whether this case represents a universal characteristic or only a specific type of CCA. It is necessary to accumulate more case studies that compare imaging evaluation and surgical or autopsy specimens.

Interestingly, repeated biopsies of the site of wall thickening confirmed by IDUS did not lead to a diagnosis of malignancy in this case. Since adenocarcinoma of the gastrointestinal tract arises from the mucosa, diagnosis is based on tissue sampling from the surface of the lumen. However, the sensitivity of diagnosis of CCA by transpapillary biopsy is 50% when localized wall thickening is detected by IDUS, which is much lower than the about 80% to 90% when polypoid or intraductal tumors are detected [29]. In a recent meta-analysis of 61 studies, the diagnostic sensitivity of tissue sampling obtained with ERCP biopsy, ERCP brushing plus biopsy, and endoscopic ultrasound with fine-needle aspiration in patients with perihilar and distal CCA was 67%, 70.7%, and 73.6%, respectively [30,31]. In CCA spreading via the PBG network, not in the lumen, as in the present case, endoscopic ultrasound–fine-needle aspiration/biopsy is theoretically considered to be the optimal method of tissue sampling, but as noted above, it is not sufficiently sensitive. Further technological innovation in the tissue diagnosis of biliary tract cancer is needed, assuming that there are CCAs that are not exposed to the surface of the lumen. At present, the most important issue might be to recognize this type of CCA and perform repeated tissue sampling for early diagnosis.

Conclusions

This case suggests that extrahepatic CCA originated from the PBG and grew locally within the PBG compartment without invading the bile duct mucosa. Based on the imaging findings from 1 year earlier and autopsy results, it is speculated that the extrahepatic CCA progressed without a detectable precancerous lesion, such as BilIN or IPNB. These findings are valuable for understanding the carcinogenesis and progression of CCA. Additionally, histological and cytological diagnosis could be theoretically difficult when sampling tissue from the bile duct lumen in cases like this.