Introduction

Carcinoma of the extrahepatic biliary tract accounts for less than 2 % of all cancers. Its commonest type is cholangiocarcinoma accounting for approximately 80 % of the cases, while other types include squamous and adeno-squamous carcinoma, colloid carcinoma, papillary carcinoma, oat cell carcinoma, pleomorphic giant cell tumors and carcinoid tumors [1]. The “carcinoid tumors” have been renamed by the World Health Organization (WHO) [2–4] into “Neuro-Endocrine Tumors (NETs)”, in order to designate all gastrointestinal lesions with evidence of endocrine differentiation.

Extrahepatic bile ducts are among the rarest primary sites of NETs, accounting from 0.2 % to 2 % of all such malignancies [3]. These lesions are difficult to be diagnosed preoperatively and almost impossible to distinguish from cholangiocarcinomas [3, 4].

The aim of the present systematic review article is to elucidate the clinical and pathological features of extrahepatic biliary NETs (EBNETs), their treatment and prognosis by analyzing all published cases.

Material – Methods

In 1959 the first EBNET was reported by Davies [5]. From 1959 up to 2012, 100 articles describing 150 cases were published [5–104]. As EBNET, we describe any NET located on left or right hepatic duct, common hepatic duct, cystic duct or common bile duct. Tumors of the intra-hepatic biliary ducts, liver parenchyma, gallbladder and Vater’s region are excluded from this review. Composite tumors including both NET and adenocarcinoma features are excluded as well. Each article was carefully studied and a database was created including the following parameters: age; gender; tumor size and location; presenting symptoms; presence of metastases; time of diagnosis; treatment; immunohistochemistry; pathology and follow up. Immunohistochemistry findings was considered indispensable criterion for inclusion in the database. Cases reporting at least eight characteristics were considered as “well documented”.

Results

The majority of the articles (75 %) were published after the 1990’s, probably because of the advanced pathological and immunohistochemical techniques which facilitated the diagnosis.

Since we consider by convention that “well documented” is every case that presented minimum 8 characteristics in the database created, 72 cases were excluded from the review as inadequately documented, leaving a total number of 78 “well-documented” cases. Table 1 summarizes all “well-documented” cases with EBNET, while table 2 presents the characteristics of these patients. The female to male ratio was1.6/1 with a mean age of 47.04 ± 17.62 years (ranging from 6 to 79 years). Remarkably, seven cases of EBNETs were reported in children and adolescents [12, 42, 55, 63, 71, 91, 99]. The tumors were symptomatic in 88.5 % of the patients whereas the rest of the cases were incidentally diagnosed either during radiological imaging for other pathology [93] or during operation; usually cholecystectomy [52, 61]. The symptoms were mostly related to the tumor mass growth, invasion of adjacent structures or metastases rather than the hormone and vasoactive peptide secretion. The most common symptom was jaundice (60.3 %), while other symptoms described were abdominal pain (43.6 %), pruritus (19.2 %), nausea-vomiting (12.8 %), weight loss (15.4 %), weakness (5.1 %), anorexia (2.6 %) and fever (2.6 %). Concomitant cholelithiasis was reported in 15 cases (19.2 %). EBNETs rarely induced symptoms associated with hormones or peptides secretion. Only in 7 cases (9 %) presenting symptoms were caused by hormone and vasoactive peptide hyper-secretion; including 4 gastrinomas and 1 somatostatinoma [20, 48, 52, 58, 75, 92]. Diarrhea (7.7 %) and gastrointestinal ulcers (5.1 %) were the most characteristic symptoms caused by hyper-secretion. Though, in only two cases urinary 5-hydroxyindoleacetic acid (5-HIAA) was measured preoperative and found to be slightly elevated [9, 75], gastrin was increased in 3 cases [58, 92] and serotonin in 2 cases [9, 75] respectively. None of the reported hyper-functional EBNETs had liver metastases at the time of diagnosis. Moreover, EBNETs seem to be related to Von Hippel-Lindau syndrome (VHL) in two cases [35, 86], a rare inherited familial cancer syndrome. None of EBNETs cases had clinical features related with carcinoid syndrome.

Table 1 Well-documented extrahepatic biliary neuroendocrine tumors (EBNETs)
Full size table
Table 2 Extrahepatic biliary neuroendocrine tumors (EBNETs) characteristics
Full size table

Regarding the tumor location, the most frequent sites were the common hepatic duct and the distal common bile duct (19.2 %) followed by the middle of the common bile duct (17.9 %), the cystic duct (16.7 %) and the proximal common bile duct (11.5 %). It is noteworthy that the EBNET described by Ueyama et al. [44] arose in a congenital bile duct cyst in common bile duct. Data concerning the tumor size were available in only 68 patients (87.2 %). The mean tumor diameter was 2.15 ± 1.2 cm (ranging from 0.2 to 5.5 cm). It should be noted that very small tumors were found to be more aggressive and having liver metastases [50, 72, 77]. 34.6 % of EBNETs were metastatic either to local lymph nodes (19.23 %) or the liver (16.7 %). In three patients, local invasion of the surrounding structures, such as pancreas or portal vein, was found.

Surgical excision, when feasible, was considered as the main and only curative treatment for EBNETs. The type of procedure depended on the tumor location. The most popular procedure was excision of extrahepatic biliary tree (62.82 %) with portal vein lymphadenectomy (43.6 %). Reconstruction was made mainly by Roux en Y hepaticojejunostomy. Pancreatoduodenectomy was performed for tumors located in the distal common bile duct. Various types of hepatectomies were performed in 11 patients for proximal tumors or liver metastasis. Liver transplantation was performed in 3 patients (3.85 %). In 5 patients the tumors locate on cystic duct and cholecystectomy was considered as adequate treatment. Finally, 4 cases were proved to be inoperable and only biopsies were taken.

EBNETs were difficult to be diagnosed preoperatively. In almost all cases, the final diagnosis was made postoperatively by pathology. In two cases with extensive biliary and liver disease the diagnosis was made in autopsy [12, 39]. Preoperative diagnosis was feasible in only 4 cases (5.12 %). In two patients with biliary mass, serum blood serotonin levels were elevated [9, 75] indicative of functional EBNET. In the other two cases the diagnosis was made by biopsies taken during endoscopic retrograde cholangiopancreatography (ERCP) [77, 93].

Immunohistochemistry was available in all included cases. Table 3 includes the diagnostic neuroendocrine markers expressed by EBNETs. The vasoactive peptides produced by these tumors were gastrin (14.1 %), serotonin (12.8 %), pancreatic polypeptide (10.3 %) and somatostatin (6.4 %). More rarely, the EBNETs stained positive for VIP (2.6 %), gastrin released protein (2.6 %), protein gene product (2.6 %), glucagon (1.3 %) and cholecystokinin (1.3 %). In only 7 cases these peptides were active to cause symptoms and the final diagnosis was well differentiated endocrine tumor [20, 52, 75]; mostly gastrinoma [48, 58, 92]. Three patients (4.1 %), died during the perioperative period. Follow-up surveillance data were available for 60 patients (82.2 %) and ranged from 1 to 240 months.

Table 3 Extrahepatic biliary neuroendocrine tumors (EBNETs) immunohistochemistry
Full size table

Discussion

NETs are distinct neoplasms with characteristic histological, clinical and biological properties. NETs consist of multipotential cells with the ability to secrete numerous hormonal substances and vasoactive peptides; serotonin, gastrin, somatostatin, VIP, glucagon and insulin being the most common. The most common sites for primary NETs are the appendix, bronchus, ileum and rectum. NETs was assumed to arise from embryonic neural crest cells, which are known as enterochromaffin or Kultschitsky cells that migrate to the respiratory and gastrointestinal tracts during neonatal development [3]. The origin of these cells from neural crest have been recently questioned [105] based on studies such as that by Fontaine and Le Douarin [106]. Kultschitsky cells are extremely scarce in bile duct mucosa, possibly explaining the rarity of EBNETs [88]. According to the analysis of 13,715 NET cases of all organs of the digestive system by Modlin et al. [107], the incidence of EBNETs was 0.32 %.

A multitude of terms has been used to describe EBNETs including “apudoma”, “argentaffin tumor”, “carcinoid”, “malignant carcinoid”, “atypical carcinoid”, “adenoendocrine carcinoma” and“endocrine cell carcinoma”. Carcinoid is the most common term used to describe these tumors before 1996, when WHO initially by Klöppel et al. [108] and subsequently by Capella et al. [4] has agreed to replace it with the broader term “neuroendocrine tumor”. It includes all endocrine tumors, ranging from well differentiated (traditionally known as carcinoid tumors) to poorly differentiated malignancies with endocrine features (e.g. small cell carcinoma). WHO’s staging system is based on tumor size, number of mitoses per high-power field, vascular or perineural invasion and Ki-67 immunostaning to differentiate between the various grades and potential malignancy of NETs [4]. The WHO recently (2010) revised NETs classification [109]. The new classification emphasize to the NET’s grade. Three NETs categories are defined regarding to the grade (G): G1, mitotic index count < 2 mitoses per high-power fields (HPF) and/or Ki-67 index < 2 %; G2, mitotic index count 2–20 mitoses per HPF and/or Ki-67 index 3-20 %; G3, mitotic index count > 20 mitoses per HPF and/or Ki-67 index > 20 %.

The origin of EBNETs remains to be elucidated. Some authors assumed that these neoplasms derive from pre-existing neuroendocrine cells that are physiologically dispersed throughout the biliary mucosa [88]. Another hypothesis was that these tumors originate from ectopic pancreatic tissue distributed during embryogenesis within the biliary ducts [92, 94]. Other authors incriminated a multipotent stem cell of the biliary mucosa, capable of differentiating into NETs and carcinomas [83]. The theory that EBNETs originated from a multipotent stem cell is supported by the mixed phenotypes of EBNETs, ranging from pure EBNETs to composite tumors and to carcinomas with scattered neuroendocrine cells [57]. Chronic inflammation and intestinal metaplasia are implicated in tumor-genesis because NETs are frequently encountered along with these conditions. Chronic inflammatory conditions of the bile ducts result in intestinal metaplasia; as a result an increase in the number of argentaffin cells is observed. This may predispose the ducts to possible EBNET development [37]. In the present review, cholelithiasis associated with EBNETs is found in 19.2 % of the studied cases. The high incidence of cholelithiasis could be associated with EBNETs pathogenesis.

The endocrine nature of EBNETs can’t be usually diagnosed preoperatively because of the absence of detectable serum markers and the usual lack of hormonal symptoms [64]. In only two cases urinary 5-hydroxyindoleacetic acid (5-HIAA) was measured and found to be slightly elevated [9, 75]. Preoperative gastrin serum level was increased in 3 cases [58, 92]. Yet, only in Martignogni et al. [58] case preoperative images assisted in localizing the tumor in biliary ducts. Despite the availability, the technological advances and use of many diagnostic imaging tools, preoperative diagnosis is difficult because of the similarity of findings among biliary malignancies. In the vast majority of the cases, the diagnosis is made intra-or postoperatively by pathology reports. Preoperative diagnosis could be made by endoscopic biopsies, but the incidence of false negative results in brush cytology may be high due to submucosal location of the neoplasm. Noronha et al. [94] in her recent review suggested that accurate preoperative diagnosis can be made by examining brush cytology specimens. To date only two EBNETs were preoperatively diagnosed based on the histological results of biopsies [77, 93]. Maybe, the routine uptake of brush cytology specimen during ERCP, percutaneous transhepatic cholangiography (PTC) or endoscopic ultrasound-guided fine needle aspiration and elaborate examination of the specimens will facilitate preoperative diagnosis.

Differential diagnosis should include adenocarcinoma (80 %), papillary adenocarcinoma (9.3 %), mucinous and mucin-producing adenocarcinoma (4.8 %), in situ carcinoma, squamous cell and adeno-squamous carcinoma, colloid carcinoma, oat-cell carcinoma, anaplastic carcinoma, pleomorphic giant cell tumor and sarcoma [72]. Additionally, benign epithelial tumors such as adenomas, cystadenomas and papillomas, granular cell tumors, rare lesions such asparaganglioma, melanoma, lymphoma and botryoidrhabdomyosarcoma in children should be included in differential diagnosis [94]. When a primary bile duct neoplasm is suspected, differentiation between an unusual bile duct tumor, such EBNET and a cholangiocarcinoma is almost impossible prior to surgery and pathology report. Nevertheless, there are some features that differ between the two tumors and may help in preoperative suspicion of an EBNET; EBNET occurs more frequently in females and younger patients than adenocarcinoma, while aggressive local invasion by the primary tumor is rare in EBNETs, yet present in the majority of cholangiocarcinomas [62, 94]. Metastases are presented in 1/3 of EBNETs compared to 2/3 of all cholangiocarcinomas. Finally, total surgical resection is feasible in the vast majority of NETs, while curative resection of cholangiocarcinoma is feasible in only one-third of all cases [62].

EBNETs are slowly growing tumors and the only curative treatment up to date considered being the aggressive surgical resection. Biological characteristics of EBNETs demand radical surgical removal, in order to achieve complete tumor excision with negative histological margins [6, 62]. Preoperative decompression of biliary tree with stent placement still is controversial. Up to date in 24.5 % EBNET cases a stent was placed preoperatively to decompress the biliary ducts, but in one case described by Ross et al. [59] septic cholangitis developed after stent placement.

Primary tumors should be removed and local lymphadenectomy should be performed. If the tumor is deemed inoperable, debulking should be attempted [50]. Likewise, solitary liver metastases should be resected. The use of sophisticated devices in liver surgery, like radiofrequency devices, facilitated hepatectomies or allowed radiofrequency ablation in order to destroy the metastases [92]. When the distal bile duct is affected, surgery should consist of resection of the head of pancreas along with the whole of duodenum - with or without pyloric preservation – in order to achieve adequate radicality [17, 42, 59, 83]. If the tumor is localized in the middle or proximal bile duct, then an en bloc excision of the bile ducts - from the main hepatic ducts down to the upper margin of pancreas and adjoining lymph nodes– is necessary to preserve the patient’s prospect of cure. The reconstruction should be made with Roux-en-Y hepaticojejunostomy or hepaticoduodenostomy [83]. Various types of hepatectomies should be attempted in tumors occurring in the hepatics bile ducts or hilar to achieve complete resection [46, 70, 72, 77]. Liver transplantation should be considered and fully justified, especially in young patients, when the tumor had a local-regional extension without distal metastases and traditional curative resection is not possible. Ki-67 index must be fully evaluated prior to attempt liver transplantation. Currently, this is the only curative option for these patients [72, 78].

There are numerous treatment options for patients with advanced EBNETs when surgery is not feasible. Medical treatment includes systemic chemotherapies, targeted therapies, somatostatin analogs, liver-directed therapies such as chemoembolization or thermoablation, and peptide receptor radionuclide therapy [110].

On pathology examination, the tumor’s macroscopical features were nodular, infiltrating or polypoid. Most of them had argyrophilic cells without argentaffin cytoplasmic granules [72]. Histological features are similar to those of the intestinal NETs. They tend to grow in cords, nests or trabeculae and usually invade the ductal wall. Moreover, perineural and lymphovascular invasion is common [66]. From the immunohistochemical point of view, EBNET cells show positive staining for chromogranin, NSE, cytokeratins and synaptophysin. The biliary system is of foregut origin and, therefore, it is not surprising that immunoreactivity for serotonin, somatostatin and gastrin has been demonstrated for some of them [41, 52, 94]. Ki-67 index is also an important aspect of the baseline workup of EBNETs [94]. Well-differentiated EBNETs show a Ki-67 index of 2 % or less, while those of uncertain malignant potential show a Ki-67 index greater than 2 % [94]. In most of the cases the tumors are non-functional and clinically silent.

EBNETs have great variety in their aggressiveness. Those with typical neuroendocrine differentiation and minimal atypia (previously known as carcinoids) tend to be indolent in their behavior, whereas atypical neuroendocrine tumors may have more poorly differentiated or aggressive characteristics and worse prognosis. Atypical neuroendocrine tumors may have some histological features of adenocarcinomas [88]. Although malignant EBNETs are of an aggressive nature, they also tend to be less aggressive than the cholangiocarcinoma [62].

It is difficult to assess the prognosis of these tumors, since the cases are rare and long-term follow-up is often unavailable. In view of available data, these tumors seem to have better prognosis than bile ducts carcinomas after radical surgical treatment [77]. Although the size of the tumor, the presence of lympovascular invasion and the quantitative assessment of Ki-67 reactive cells help to determine prognosis, there are no absolute criteria for judging the malignant potential of EBNETs [7]. Noronha et al. [93] in her review suggests that the best predictor of malignant behaviour seems to be the size of the primary tumour, but based on our research this conclusion is unclear. 63.63 % of EBNETs (7/11) with size less than 1 cm had metastases, while the correspondence percentage for tumours ranging between 1 and 2 cm is 27.6 % (8/29) and that of tumours with size over 2 cm is 28.6 % (8/28). The WHO recently revised classification base mainly on the NET grade (mitotic index and ki-67 proliferation index). Based on our review, few cases of EBNETs display data concerning mitotic and Ki-67 index [6, 75]. We assume that the ranking of the EBNETs based on the grade will explain this observation, nevertheless these data are unavailable. Despite the small number of EBNETs appearing in the literature, the long-term survival rate appears to be significantly better than in other types of biliary malignancies.

In conclusion, EBNETs are difficult to diagnose preoperatively because of their rarity, the absence of detectable serum markers and the usual lack of hormonal symptoms. Mitotic and Ki-67 index should be evaluated in all EBNETs to assess the grade of the tumors. Up to-date, the only curative treatment with good long-term results, for EBNETs, is aggressive surgical resection.