INTRODUCTION — The diagnosis of biliary neoplasms is usually accomplished through a combination of imaging modalities and tissue sampling. Optimal methods to diagnose these tumors depend upon the clinical setting, the characteristics of the individual tests, and the available resources.
This topic review will focus primarily on the accuracy of endoscopic methods to diagnose biliary neoplasms. General discussions on the diagnosis of the individual tumors are presented separately in the corresponding topic reviews.
ENDOSCOPIC RETROGRADE CHOLANGIOPANCREATOGRAPHY — Endoscopic retrograde cholangiopancreatography (ERCP) permits the visualization of the biliary neoplasms while providing the opportunity to obtain tissue samples and to perform therapeutic maneuvers (such as biliary decompression) when necessary. Thus, it continues to have a central role in the diagnosis and management of biliary neoplasms.
Radiographic appearance — Cholangiographic findings during ERCP can help distinguish benign from malignant neoplasms. A biliary neoplasm typically appears as a stricture during ERCP. However, strictures caused by a benign process (such as pancreatitis and primary sclerosing cholangitis) can have the appearance of malignancy and vice versa. In addition, a neoplasm can develop in a long-standing benign stricture and its recognition may be difficult, such as in patients with primary sclerosing cholangitis and chronic pancreatitis [1]. (See "Primary sclerosing cholangitis in adults: Management", section on 'Cancer screening' and "Overview of the complications of chronic pancreatitis", section on 'Pancreatic cancer'.)
Some of the characteristics of the stricture suggestive of malignancy include a length of more than 10 mm, a ragged contour, and the presence of a fixed filling defect and/or an abrupt transition from relatively normal to the stricture, so called shouldering, typically located above the stricture.
In addition, strictures involving the hilum should raise concerns of malignancy. The majority of hilar strictures are due to either malignancy or extrinsic compression by lymph nodes. Our experience and that reported in the literature supports the view that hilar strictures should be considered malignant until proven otherwise. Exceptions are patients who have recently undergone liver transplantation or have autoimmune cholangiopathy. Tissue confirmation should be sought in the majority of patients. The clinical setting, radiographic appearance, and cholangiographic findings may be sufficient to establish the diagnosis of malignancy in patients in whom tissue sampling is not feasible.
Tissue sampling during ERCP — Tissue sampling of suspected bile duct tumors can be obtained by cytologic brushings, biopsy forceps, bile aspiration, or a combination of all of these modalities [2-24]. In patients in whom a plastic stent has been present, the stent can be spun and the cells obtained can be assessed [25].
Brush cytology — The most commonly used method for tissue sampling during ERCP is cytologic brushing (picture 1 and image 1). Estimates from several studies have suggested that the sensitivity of cytologic brushing ranged from 10 to 67 percent while specificity typically exceeded 90 percent [2-24,26-29]. The following examples from some of the largest series illustrate the range of findings:
●One report focused on 86 patients with biliary strictures of unknown etiology who underwent brush cytology [2]. A final diagnosis was achieved in 78 patients based upon intraoperative findings and histologic investigation, autopsy, or prolonged follow-up. Strictures were malignant in 57 patients and benign in 21 patients. The sensitivity and specificity for brush cytology was 56 and 90 percent, respectively. The overall accuracy was 65 percent. The positive and negative predictive values were 94 and 43 percent, respectively.
●Another study included 148 patients who had undergone brush cytology of the bile or pancreatic ducts and in whom a final diagnosis was available from surgical pathologic examination or long-term clinical follow-up [30]. The overall accuracy of biliary and pancreatic cytology was 64 and 72 percent, respectively.
●A third series included 74 patients with pancreaticobiliary strictures who underwent brush cytology performed on 55 bile and 19 pancreatic duct specimens, respectively [31]. The diagnosis was confirmed by surgery, autopsy, or during follow-up. The sensitivity, specificity, positive and negative predictive values were 56, 100, 100, and 51 percent, respectively with an overall accuracy of 70 percent.
●In a study of 58 patients with hilar cholangiocarcinoma, a malignancy was demonstrated by brush cytology in 40 percent of patients and biopsy forceps in 53 percent [32]. The combination increased the yield to 60 percent (35 of 58 patients).
●A novel approach to cytology using a brush was suggested in a study in which the brush was cut after obtaining cytologic specimens, immersed in methanol, and then embedded in paraffin [33]. Sections were cut along the axis of the brush until the metal wire was almost reached. Using this technique in 112 patients with a biliary tract stricture had a sensitivity and specificity of 91 and 100 percent, respectively, when the diagnosis was confirmed histologically. In nonsurgical patients, sensitivity and specificity were 96 and 100 percent, respectively, compared with the clinical outcome as the reference standard.
Primary sclerosing cholangitis — Establishing a diagnosis of malignancy can be particularly difficult in patients with primary sclerosing cholangitis (PSC) [34]. Several approaches have been proposed but their effectiveness remains unsettled. (See "Primary sclerosing cholangitis in adults: Clinical manifestations and diagnosis", section on 'Cholangiocarcinoma'.)
●In a study of 61 patients with PSC who underwent liver transplantation, cytologic specimen classifications were compared with bile duct histopathologic findings in explanted livers [35]. The sensitivity and specificity of cytologic low-grade and high-grade dysplasia/adenocarcinoma were 100 and 84 percent, compared with 73 and 95 percent for high-grade dysplasia/adenocarcinoma only.
●In a study of 230 patients with PSC followed for a 6-year period, 10 percent developed pathologically confirmed cholangiocarcinoma [36]. The combination of a serum tumor marker (CA 19-9), when combined with a cross imaging study of the liver, such as ultrasound, tomography, or magnetic resonance, provided a better sensitivity than tumor marker alone, but without improvement in the specificity and either positive or negative predictive values. Adding cholangiography increased specificity and positive predictive value.
Tumor type — The accuracy of brush cytology is influenced by the type of tumor. As a general rule, the sensitivity is lower for tumors that do not originate from the biliary tract [2,26]. Primary biliary cancers usually have associated mucosal changes that facilitate acquisition of cells, whereas pancreatic or other metastatic tumors may only produce biliary stricture by extrinsic compression, leaving the bile duct mucosa intact and thus reducing cell acquisition. In the study discussed above [2], for example, the sensitivity was significantly higher for cholangiocarcinoma compared with pancreatic carcinoma (80 versus 35 percent).
Technique — In addition to the tumor type, the accuracy of brush cytology is influenced by the technique of the endoscopist obtaining the brushings and the cytologist interpreting the results [6,7,37]. False positive results may be more likely in patients with strictures arising from inflammatory processes (such as chronic pancreatitis, possibly PSC, and after manipulation and secondary local trauma, such as after sphincterotomy) [26]. Some reports suggested that the disruption of malignant strictures by dilation enhanced the diagnostic yield of bile cytology [14,38-41]. Repeat brushings are generally indicated with negative results when the clinical or radiographic findings suggest malignancy. The finding of atypical hyperplasia should also raise concern for an underlying malignancy [4].
We advance the brush approximately 10 times over the stricture by moving the brush and the endoscope. When we suspect pancreatic cancer, we brush both the biliary and pancreatic ducts using a standard brush (using a different brush in each duct) although this tends to increase procedure time and possibly the risk of pancreatitis. When using this approach, we typically place a temporary self-extruding pancreatic stent. We have found that all the commercially available brushes have equivalent efficacy in obtaining tissues. Once the samples have been obtained, we cut the brush head off and drop it in fixative immediately after removing it from the patient. In our experience cutting the brush head off appears to be more sensitive (unpublished data) compared with preparing slides and fixing them in the endoscopy suite.
Forceps biopsy — Tissue samples can also be obtained during ERCP using a biopsy forceps. Flexible and stiff forceps are available, but the former is preferred since it makes intubation easier. While a prior sphincterotomy is not mandatory, it certainly facilitates ductal intubation and, in many patients undergoing ductal biopsies, a sphincterotomy would have been performed for drainage or stent placement.
The sensitivity of forceps biopsy ranged from 43 to 88 percent in various reports, with specificity comparable to brush cytology (ie, >90 percent) [2-24,42,43]. Whether forceps biopsy alone or in combination with brush cytology is more accurate than brush cytology alone is unclear. One study suggested that the combination of the technique increased the sensitivity by approximately 15 to 25 percent (from 47 to 65 to 70 percent sensitivity) compared with either method alone [9]. Put another way, this suggests that an endoscopist would have to perform both techniques in four to six patients to detect one tumor that would not have been diagnosed by either method alone. By contrast, another report suggested that the sensitivity of the two techniques was similar (approximately 53 percent) and that the benefit of combining the two techniques (sensitivity of 61 percent) was too small to justify the routine use of the combination approach [42]. There was one bile duct perforation related to forceps biopsy that required surgical oversewing, otherwise both techniques were performed safely. We generally perform both procedures when strictures are located in the intra-ampullary tract, in patients with complete ductal obstruction, or when previous brush cytology alone was negative.
An alternative approach has been described in which a biopsy forceps is used to obtain specimens from the lower edge of a stricture [44]. These specimens are then smeared between two glass slides, with the goal of forming a monolayer of cells. The slides are stained using a rapid Papanicolaou method and interpreted by a cytopathologist. The procedure is then repeated several times. In a study that included 133 patients who underwent this technique, with a goal of reviewing at least 10 specimens, the technique had a sensitivity of 72 percent and a specificity of 100 percent for diagnosing malignancy [44].
Endoscopic fine-needle aspiration — A combination of brush cytology and endobiliary biopsy with endoscopic fine-needle aspiration was more sensitive (73 to 77 percent) than either method alone in at least three reports [11-13]. One study suggested that combining stricture dilation, cytology, and fine-needle aspiration substantially improved the accuracy for diagnosis of malignant strictures caused by gallbladder or pancreatic cancer compared with cytology alone [38].
Bile aspiration — Aspiration of bile juice is the simplest method of obtaining a specimen for the evaluation of a stricture. However, sensitivities have been disappointing, ranging from 6 to 32 percent [2,11,27]. Some studies suggest that the sensitivity could be improved by dilating the stricture or performing brush cytology prior to bile acquisition [14,27]. Because of the better yield of brush cytology, bile aspiration alone is rarely used. Bile aspiration may add to the sensitivity for patients undergoing cholangioscopy with targeted biopsies [45].
Scraping cytology — Another variation of cytologic examination was proposed in which a guidewire was inserted into pancreatic ductal strictures and used to scrape the stricture; pancreatic juice above the strictures was also collected [46]. In a report of 86 patients with pancreatic ductal strictures (71 malignant), sensitivity and specificity were 93 and 100 percent, respectively [46].
An endoscopic device with scraping loops was found to improve the rates of cancer detection in a study of 119 patients with malignant strictures: 51 versus 65 versus 75 percent for biopsy forceps, scraping tool, and combined sampling, respectively [47].
Tumor markers — The expression of tumor markers by biliary neoplasms provided the rationale for studies evaluating the efficacy of their detection during cytologic sampling. A number of markers and techniques have been developed including digital image analysis, K-ras testing and fluorescence in situ hybridization, with sensitivities ranging from 14 to 60 percent [48]:
p53 and K-ras — Several studies have evaluated detection of mutations in K-ras and p53 in bile duct specimens [15,16,20]. One study included 43 patients with primary sclerosing cholangitis in whom 47 brush cytology specimens were obtained from a dominant stricture [15]. Of these, 27 specimens were analyzed for mutations p53 or K-ras. The results were compared with the histologic diagnosis or clinical status at least two years later. The sensitivity, specificity, positive, and negative predictive values of brush cytology alone were 60, 89, 59, and 89 percent, respectively. Adding the results of p53 and K-ras analysis did not improve the results.
Similar conclusions were reached in another study assessing the diagnostic value of p53 immunoreactivity in 143 brush cytology specimens from patients with various pancreatobiliary strictures [16]. Sixty-three of 89 (71 percent) malignant strictures were identified cytologically while 45 cases (51 percent) were p53 immunoreactive. Atypical cytology was found cytologically in 19 specimens (13.3 percent), of which p53 expression was positive in 12 cases. Of these 12 cases, four were carcinomas and eight benign. Thus, adding p53 immunostaining did not increase diagnostic accuracy. A third report found no value of K-ras analysis in distinguishing chronic pancreatitis from pancreatic cancer [49].
Telomerase RNA — In situ hybridization for telomerase RNA (found in 85 to 90 percent of malignancies) was evaluated in a study involving 18 patients with benign and malignant biliary strictures, three of whom had cholangiocarcinoma and one who had gallbladder cancer [17]. Routine cytology was positive in five of eight malignant strictures while telomerase RNA was positive in six (sensitivity 75 percent). Interestingly, both techniques were positive in only three patients.
A second study found that 13 of 16 specimens of malignant tissue had telomerase activity compared with none of the nonmalignant specimens [18]. In addition, p53 over expression was noted by immunostaining in 9 of 16 specimens with malignancy. Combining both telomerase activity and p53 immunoreactivity resulted in sensitivity and specificity of 100 percent. These findings await confirmation.
Measures of DNA proliferation — DNA proliferation can be assessed with a variety of methods such as digital image analysis (DIA) and fluorescence in situ hybridization (FISH) [7,21,48,50,51]. These techniques show promise in improving the specificity of cytology. DIA quantifies the amount of cellular DNA by measuring the intensity of nuclei stained with a dye that binds to nuclear DNA. FISH uses fluorescently labeled DNA probe test to detect chromosomal abnormalities in cells. FISH increased the sensitivity of cytology analysis from 32 percent to 51 percent without reducing specificity [52]. (See "Tools for genetics and genomics: Cytogenetics and molecular genetics".)
DIA and FISH were evaluated in a study of 233 consecutive patients undergoing ERCP for a pancreatobiliary stricture. The patients underwent standard cytology, DIA, and FISH [50]. The test performance was similar across groups. Standard cytology had low sensitivity (4 to 20 percent) but 100 percent specificity. In patients with negative cytology, FISH increased sensitivity while preserving specificity. The sensitivity and specificity of DIA was intermediate between routine cytology and FISH.
In a study of 235 patients with primary sclerosing cholangitis (PSC) who underwent FISH, 120 (51 percent) of the patients were positive on FISH testing, but only 40 (33 percent) had cholangiocarcinoma [51]. Patients with polysomy were more likely to have cholangiocarcinoma (26 of 47 patients, 55 percent), especially if a dominant stricture was present (19 of 26 patients, 73 percent). The sensitivity and specificity of polysomy for detecting cholangiocarcinoma were 46 and 88 percent, respectively. The use of FISH in patients with PSC was also examined in a meta-analysis of eight studies with 828 patients. The sensitivity and specificity of FISH for diagnosing cholangiocarcinoma were 68 and 70 percent, respectively.
Adding DNA measurements as an adjunct to brush cytology in patients with biliary strictures has been suggested to increase diagnostic yield. In a study including 159 patients with bile duct strictures undergoing ERCP, the sensitivity and specificity of DNA aneuploidy for tumor detection were 43 and 96 percent, respectively [53]. Adding DNA analysis to brush cytology did not increase the sensitivity.
One study evaluated the usefulness of detecting and measuring aberrantly methylated DNA as a marker of pancreatic and periampullary cancers in 130 patients with biliary strictures [54]. Methylation analysis was performed on endoscopically obtained brush samples from the biliary and pancreatic ducts from 130 individuals with biliary tract strictures: 41 with pancreatic ductal adenocarcinoma, 10 with biliary tract cancers, 13 with other periampullary neoplasms, and 66 with non-neoplastic strictures, including 27 with primary sclerosing cholangitis and 39 with other benign strictures. Real-time quantitative methylation-specific polymerase chain reaction was measured in brush DNA. At least one gene positive for methylation by quantitative methylation-specific polymerase chain reaction was observed in 73 percent of patients with pancreatic cancer, 80 percent of those with cholangiocarcinoma compared with 14 percent of those with benign strictures.
The genomic assessment of neoplasms that may involve the bile duct has expanded to include a variety of oncogenes and tumor suppressor genes, including TP53, CDKN2A, and SMAD4 (in addition to K-ras) [55,56].
Advances in next generation sequencing have improved the accuracy of diagnosing biliary malignancies. Next generation sequencing has evaluated 28 genes that were commonly mutated, amplified or deleted in biliary tract malignancies. In a study of 252 patients, the sensitivity and specificity of this method was 73 percent and 100 percent, respectively [57]. In addition, it improved the sensitivity of cytology brushings and biopsies from 35 percent and 52 percent to 77 percent and 83 percent, respectively.
Cholangioscopy during ERCP — Miniature endoscopes to visualize the bile (cholangioscopy) or pancreatic (pancreatoscopy) ducts can be used intraoperatively, during ERCP, or with percutaneous transhepatic cholangiography. They have an increasingly important role in the diagnosis and nonsurgical treatment of biliary and pancreatic diseases. The use of cholangioscopy in the diagnosis of biliary malignancy has increased in the last decade and at endoscopy centers worldwide. Cholangioscopy has established itself as a significant contributor in the visual and tissue diagnosis of benign or malignant strictures, as well as in determining the local extent of disease. Data have suggested that visual diagnosis has surpassed the sensitivity of tissue diagnosis [58]. In addition, improvements in the characterization and classification of the visualized changes have increased the overall sensitivity and specificity of cholangioscopy [59]. (See "Cholangioscopy and pancreatoscopy".)
Optical coherence tomography (OCT) has been studied with cholangioscopy to evaluate biliary strictures and may increase the sensitivity for diagnosing malignant strictures. OCT uses infrared light reflectance to produce high-resolution cross-sectional tissue images [60]. By providing in-vivo cross-sectional imaging, OCT detects epithelial changes that may improve sensitivity for diagnosing malignant strictures and provide standardized criteria predictive of malignancy. However, the role of OCT for biliary disease remains uncertain, and this is discussed separately. (See "Optical coherence tomography in the gastrointestinal tract", section on 'OCT of the bile and pancreatic ducts'.)
In one study, intraductal OCT was used in 37 patients with biliary strictures to detect malignancy [61]. OCT criteria for malignancy included unrecognizable layer architecture and presence of large, non-reflective areas compatible with tumor vessels. Thirty-five of the patients were able to undergo OCT (the remaining two could not due to tight strictures). Of those 35 patients, 19 had malignant strictures and 16 had benign strictures. When both OCT criteria were met, the sensitivity for OCT was 53 percent, with a specificity of 100 percent. If only one criterion was met, the sensitivity was 79 percent, with a specificity of 69 percent. When the results of OCT were combined with the results from biopsies and/or brushings, the sensitivity increased to 84 percent, with a specificity of 69 percent (the sensitivity and specificity for biopsies/brushings alone were 79 and 69 percent, respectively).
Confocal laser endomicroscopy has also been used in the evaluation of bile duct strictures. (See "Confocal laser endomicroscopy and endocytoscopy", section on 'Bile duct strictures'.)
ENDOSCOPIC ULTRASOUND AND INTRADUCTAL PROBES — Endosonography of the biliary tree has not proved as useful as in other parts of the gastrointestinal tract. The technique has only limited ability to characterize the inner layer of the bile duct, to detect invasion of peripheral structures, and to assess the luminal contents of the gallbladder [62,63]. In addition, extensive experience is required in the performance of echoendoscopy of the bile duct. The addition of fine-needle aspiration biopsy may increase the yield [64]:
●In a study of 24 patients with proximal biliary strictures and inability to make a diagnosis by brushings, EUS visualized a mass in 23 (96 percent) patients [64]. EUS-guided fine-needle aspiration (FNA) demonstrated malignancy in 17 of 24 (71 percent) of patients. The overall sensitivity, specificity, positive predictive value, negative predictive value, and accuracy of EUS-FNA were 77 percent, 100 percent, 100 percent, 29 percent, and 79 percent, respectively.
●Similar results were obtained in a study 81 patients with cholangiocarcinoma confirmed on surgical pathology who had undergone EUS [65]. EUS identified the tumor in 76 patients (94 percent), a rate higher than what was seen with triphasic computed tomography (30 percent), or magnetic resonance imaging (42 percent). EUS-FNA was performed in 74 of the patients (91 percent) and diagnosed cholangiocarcinoma in 54 patients for a sensitivity of 73 percent. The sensitivity was higher for distal lesions than for proximal lesions (81 versus 59 percent).
Thus, EUS FNA is a sensitive method for the diagnosis of proximal biliary strictures after negative brushings, but a negative biopsy does not exclude malignancy. (See "Endoscopic ultrasound-guided fine needle aspiration in the gastrointestinal tract".)
A forward-viewing linear echoendoscope (FVL-EUS) with FNA was used in four patients with hilar biliary strictures [66]. In all four patients, it was possible to visualize and perform FNA and a definitive tissue diagnosis was obtained in all of the patients. This echoendoscope appears promising in the evaluation of patients with hilar biliary strictures.
Intraductal ultrasonography — Intraductal ultrasonography using high frequency transducers has been used to characterize biliary strictures and to distinguish between benign and malignant disease. Criteria that suggest the presence of malignancy include a hypoechoic mass, especially if infiltrating surrounding tissues, heterogeneity of the internal echo, notching or irregularity of the outer border, a papillary surface, or disruption of the normal bile duct structure. Limitations of this technique include its lack of widespread availability, inability to detect mild invasion to the perimuscular loose connective tissue, and its inability to detect lymph node or distal metastases. (See "Intraductal ultrasound of the pancreaticobiliary ductal system".)
An illustrative prospective study evaluated cholangiography, routine cytology, digital image analysis (DIA), and fluorescence in situ hybridization (FISH) of tissue samples, and intraductal ultrasound in 86 patients with indeterminate biliary strictures [48]. The combination of DIA, FISH, and intraductal ultrasound increased the yield several-fold compared with routine cytology and histology alone.
SUMMARY AND RECOMMENDATIONS
●The diagnosis of biliary tract malignancies is based upon the combination of clinical presentation, laboratory data, imaging, and tissue sampling techniques. The choice among these depends in part upon the tumor location, available expertise, and the clinical settings. Combination approaches are usually required including endosonography, endoscopic retrograde cholangiopancreatography (ERCP) and/or magnetic retrograde cholangiopancreatography, and combination of brush cytology, biopsy, and aspiration. The available cholangioscopes (ie, miniature endoscopes to visualize the bile duct) have further facilitated the diagnosis of malignant biliary strictures by offering targeted biopsies and also by providing visual characterization of the stricture. (See 'Cholangioscopy during ERCP' above.)
●For patients with a suspected biliary neoplasm, tissue sampling can be obtained by cytologic brushings, biopsy forceps, bile aspiration, or a combination of these modalities during ERCP. (See 'Tissue sampling during ERCP' above.)
●Despite the available techniques, the diagnosis may remain elusive. In some cases, the diagnosis can be confirmed histologically only after surgery.
ACKNOWLEDGMENT — The authors and UpToDate thank Dr. Susana Escalante-Glorsky, who contributed to earlier versions of this topic review.
