Biliary Tract Malignancy
Clinical Overview and Novel Detection Method
For information on specimen collection and processing, see Biliary Tract Malignancy by FISH Testing.
Contact Dr. Gores
For questions regarding the management or treatment of a patient with this condition, contact Dr. Gores (Monday-Friday) at 507-284-6488.
Click CC to turn on closed captioning.
Published: September 2011Print Record of Viewing
Pancreatobiliary tumors such as cholangiocarcinoma (bile duct cancer) or pancreatic cancer often present as biliary tract obstruction or strictures. Bile duct or hepatobiliary brushing specimens obtained during endoscopic retrograde cholangiopancreatography (ERCP) are used to assess affected patients for possible malignancy using cytologic and molecular techniques. We have developed a statistical model that incorporates routine cytology and FISH test results with clinical factors to predict risk of malignancy in an individual patient.
Presenter: Gregory Gores, MD
- Chair of the Division of Gastroenterology and Hepatology
- Professor of Medicine and Physiology in the College of Medicine at Mayo Clinic
Presenter: Kevin Halling, MD, PhD
- Vice chair for research and development for the Department of Laboratory Medicine and Pathology
- Consultant in the Division of Laboratory Genetics at Mayo Clinic in Rochester, Minnesota
Questions and Feedback
If you have questions about this Clinical Insight presentation, e-mail
Welcome to Mayo Medical Laboratories. These presentations provide short discussions of current topics and may be helpful to you in your practice.
Our presenters for this program are from Mayo Clinic in Rochester, Minnesota. Dr. Gregory Gores is chair of the Division of Gastroenterology and Hepatology and Professor of Medicine and Physiology in the College of Medicine and Dr. Kevin Halling is vice chair for research and development for the Department of Laboratory Medicine and Pathology and consultant in the Division of Laboratory Genetics. Dr. Gores will discuss the clinical aspects of biliary tract malignancy and present case studies and Dr. Halling will explain the laboratory testing available to assist in diagnosis. Dr. Gores will start off this presentation. Thank you both, Dr. Gores and Dr. Halling.
Thank you Sharon, for that kind introduction. I have no disclosures. Dr. Halling is the co-inventor on the patent for the FISH probe set described in this study and receives royalties. He also receives grant funding from Abbott Molecular to develop FISH assays for the detection of tumor cells in cytologic specimens.
Pancreatobiliary Strictures (PBS)
Pancreatobiliary strictures, abbreviated in this slide as PBS, are a common occurrence in clinical practice. Pancreatic biliary strictures may result from malignant or benign processes although malignant strictures are likely more common in clinical practice. Pancreatic cancer and cholangiocarcinoma are the most common etiologies of these strictures. It may be difficult to determine the difference between a small cancer in the head of the pancreas and a distal cholangiocarcinoma given their intimate anatomic association. The etiologies of these pancreatobiliary strictures are often unclear.
Primary Sclerosing Cholangitis (PSC)
However, primary sclerosing cholangitis, abbreviated as PSC in this slide is a relatively common hepatobiliary disorder. It increases the risk of cholangiocarcinoma. This is a chronic inflammatory disorder involving the intra and extra hepatic bile ducts. It leads to biliary tract and hepatic fibrosis, biliary tract strictures, cholestasis, and cirrhosis. PSC patients have a high risk of developing cholangiocarcinoma over their life time. Their lifetime risk is approximately 10% which is a 1000 fold greater risk than the general population. Most PSC patients are carefully followed for this potential complication by their hepatologist or gastroenterologist. If the cancer is found early, liver transplantation represents a chance for cure.
Pancreatobiliary Strictures (PBS)
The distinction between malignant and benign strictures is exceedingly difficult. This is because the tumors are often not seen by cross-sectional imaging studies due to their longitudinal growth along the bile duct rather than radial growth away from the biliary tract. Inflammatory strictures in PSC mimic those of malignant strictures. Obtaining biopsies is difficult due to the location of these strictures and the fact that the tumors are exceedingly desmoplastic.
Bile Duct Brushings
Obtaining bile duct brush cytologic specimens is the usual approach for distinguishing malignant from benign pancreatobiliary strictures. Bile duct strictures are obtained for cytology during endoscopic retrograde cholangiopancreatography or ERCP. The bile duct brushing cytology is important since it is sometimes able to provide a positive diagnosis for cases in which biopsies are negative or often inadequate. These specimens are obtained by passing an endoscope orally down to the level of the ampulla. The instrument used in this technique is a side-viewing instrument, allowing access to the ampulla. The ampulla is cannulated and a wire brush can be passed up the appropriate duct for obtaining cytologic specimens via the brush approach.
Current Clinical Practice
Therefore, in our current clinical practice since 2003, all ERCP obtained brush specimens underwent routine or conventional cytology plus fluorescent in situ hybridization, abbreviated as FISH. Dr. Kevin Halling will speak to you about the assessment of the cytologic specimens and the use of fluorescent in situ hybridization in this patient population.
Factors That Influence Sensitivity Of Cytology Specimens
Thank you, Dr. Gores. As Dr. Gores explained, the Mayo Clinic Molecular Cytology laboratory currently performs 2 tests to help determine if a patient has a pancreatobiliary malignancy and they are conventional cytology and FISH testing. I will explain how this testing is performed and interpreted in more detail in the coming slides but before I do that I want to mention that factors that can influence the sensitivity, specificity, and positive and negative predictive value of these tests include sampling error, interpretation error, and the clinical scenario. With regard to sampling, the cytology and FISH tests require that the neoplastic cells be present in the specimen for the tests to be able to detect the tumor. Consequently, false-negative results can result if the brush does not adequately sample the lesion.
Once the cytology laboratory receives a biliary tract brushing specimen they remove the cells from the brush and place them on a slide. These cells are then stained with a Pap stain and evaluated by a cytopathologist. As is shown on this slide, the cytopathologist interprets the cells on the slide as showing features that are either positive for malignancy, suspicious for malignancy, atypical (meaning that there are cells that are not normal but which cannot be definitely ascribed to a neoplastic process), or negative for malignancy.
Cancer Is A Genetic Disorder That Results From Mutations
It is important to remember that cancer is a genetic disorder that is characterized by chromosomal abnormalities such as aneuploidy (as is shown in the tumor karyotype on the left) and molecular genetic abnormalities (as is shown in the DNA sequence of the KRAS oncogene of a tumor on the right). These mutations promote tumor formation and progression.
Detecting Tumor Cells in Cytology Specimens Using FISH
Because most solid tumors exhibit aneuploidy (that is an abnormal number of chromosomes) and normal cells do not show chromosomal abnormalities, a technique that is known as FISH can be used to help determine if tumor cells are present in a specimen. FISH stands for fluorescence in situ hybridization and is a technique that utilizes fluorescently labeled probes to various DNA targets to determine if a cell has chromosomal abnormalities such as aneuploidy.
Types of FISH Probes
There are 2 basic types of FISH probes: chromosome enumeration probes and locus specific probes. Chromosome enumeration probes attach to the centromeres of chromosomes and are used to determine how many copies of a chromosome are present in a cell. Locus specific probes are designed to attach to genes of interest, for example the P16 tumor suppressor gene or the HER2 oncogene, and are used to determine if a cell shows deletion or amplification of those genes.
Our lab has clinically validated a FISH probe set that can be used to enhance the detection of pancreatobiliary malignancies. This probe consists of 4 probes. There are 3 chromosome enumeration probes to chromosomes 3, 7, and 17 which are labeled with red, green, and aqua fluorophores respectively and a single locus specific probe to the P16 tumor suppressor gene on the short arm of chromosome 9 which is labeled with a yellow fluorophore. As is shown in this slide, normal cells will exhibit 2 copies for each of the 4 probes. Cells with abnormal FISH signal patterns are either categorized as showing polysomy (which is shown on the right) or trisomy (which is shown in the middle). Polysomic cells have extra copies for at least 2 of the four probes. Trisomic cells have 3 copies of one of the 4 probes (generally 3 copies of the chromosome 7 probe) and 2 copies of the other three probes. We have found that polysomic cells are most indicative of the presence of a malignancy. Trisomic cells increase the chance the patient may have a malignancy compared to patients that have only normal cells by FISH. But trisomic cells are also commonly found in patients that are not found to have a neoplasm and thus trisomy is not specific for malignancy.
Sensitivity and Specificity of Various Tests for Detection of Pancreatobiliary Strictures Malignancy
As I mentioned in the previous slide, our lab has clinically validated the FISH probe set that we use for detecting pancreatobiliary malignancy. We did this by determining the clinical sensitivity and specificity of the assay in a large cohort of patients that were being evaluated for pancreatobiliary malignancy. We also simultaneously determined the clinical performance characteristics of conventional cytology and DIA for the detection of pancreatobiliary malignancy. On the left, this figure shows that if only positive cytology diagnoses were considered consistent with malignancy that cytology had a sensitivity of about 20% and a specificity of 100%. If both positive and suspicious cytology interpretations were considered consistent with malignancy then the sensitivity of cytology was about 38% and its specificity was about 97%. If positive, suspicious and atypical cytology interpretations were all considered consistent with malignancy then the sensitivity of cytology increased to about 50% but the specificity dropped to about 85%. We no longer use DIA so I will not talk about that. But this slide also shows that if only polysomic FISH diagnoses were considered consistent with malignancy that the sensitivity of FISH was about 42% with a specificity close to 100%. If both polysomic and trisomic FISH findings were considered consistent with malignancy the sensitivity jumped to 62% but the specificity dropped to about 79%. This further demonstrates the lack of specificity that trisomy has for the presence of malignancy.
What Do We Do With These Data?
Based on data from our validation studies, our biostatisticians have developed a multivariable model that allows us to determine the likelihood that a patient that is being evaluated for pancreatobiliary malignancy has a tumor based on several parameters which include the patient’s age, their PSC status, and their cytology and FISH results.
We enter this information into something we refer to as the “Bile Duct Calculator” which is shown on the next slide.
Bile Duct Pathology Calculator
For example, if we had a 36 year old patient with PSC who has a suspicious cytology result and a polysomic FISH result we would enter that information into our calculator.
And it would generate the information shown on the next slide.
Representative Test Report Using Bile Duct Calculator
In this slide you can see that the likelihood that this 36-year-old patient with PSC, a suspicious cytology result, and a polysomic FISH result has a pancreatobiliary tumor (based on our patient cohort) is 98% with 95% confidence intervals extending from 87% to 100%. Based on this we state that these results are consistent with malignancy. It should be noted that these predictions are based on data generated from a Mayo Clinic cohort of 498 patients in which the prevalence of malignancy was 46%.
Bile Duct Calculator
This slide shows that we will interpret any case in which the likelihood of malignancy is predicted to be >98% as consistent with malignancy but that if the prediction of malignancy is between 85 and 98% we will call it suspicious for malignancy. If there are 1 or more abnormal test results (for example an atypical cytology result but a negative FISH result) and the calculator predicts a risk of <85% then we will interpret the case as equivocal for the presence of tumor. At this time I am going to hand things back to Dr. Gores who will describe several typical cases in which the combination of conventional cytology and FISH were used to help to determine if a patient had a pancreatobiliary tumor. Thank you.
Impact of Disease Prevalence on Positive Predictive Value (PPV)
Before I review several case series utilizing FISH studies for the diagnosis of pancreatobiliary strictures, I would like to emphasize the impact of disease prevalence on positive predictive value. The positive predictive value is highly dependent upon disease prevalence of the population studied. In the validation of FISH studies for the diagnosis of biliary and pancreatic malignancies, the disease prevalence was about 45%. You can see that as the disease prevalence drops, the positive predictive value decreases concomitantly. Therefore, it is difficult to extrapolate these studies performed in a diagnostic setting where the suspicion of cancer is high, to those performed in a surveillance setting where the disease prevalence is quite low.
Our first case is a 33-year-old male with multiple medical problems including recurrent urinary tract infections, polynephritis, and chronic anemia; he is status post small bowel resection for obstruction with resulting short bowel syndrome, and chronic abdominal pain. The patient presented with symptoms of biliary colic. A CT scan demonstrated gallstones with impacted stones in the distal common bile duct. An ERCP was performed and the stones extracted. Biliary brush cytology was sent for routine cytology and FISH studies.
Case 1 Cytologic Findings
These are the cytologic findings — which were interpreted as demonstrating markedly atypical cells, neoplasm cannot be excluded.
The FISH results were negative, however.
Case 1 Test Report Using Bile Duct Calculator
Thus, using the bile duct calculator, the patient’s likelihood of having malignancy would be less than 85%.
Our interpretation of these results are, therefore, that the patient has equivocal findings which should be correlated with clinical presentations to asses the risk for biliary tract neoplasia.
In follow-up, the patient had recurrent biliary tract symptoms. These resulted in a cholecystectomy. Pathologic examination revealed chronic cholecystitis and cholelithiasis with the absence of neoplasm.
Case 2 is a 60-year-old female, previously healthy, who presents with painless obstructive jaundice. A CT exam reveals findings that suggest an impacted biliary stone. In contrast to the CT findings, the ERCP reveals a hilar stricture. When the ERCP is performed, the biliary brush specimen is collected for routine cytology and FISH studies.
Case 2 Cytologic Findings
Note the marked atypical cells on cytologic assessment.
The FISH studies in this case demonstrated polysomy.
Case 2: Test Report Using Bile Duct Calculator
The test report using the bile duct calculator put the predicted value at 99%.
In a patient with atypical cytology and polysomy on FISH at this age and without PSC the positive predictive value is 99%. The results are consistent with malignancy.
On follow-up, additional imaging studies revealed an occlusion of a left branch of the portal vein with splenic hilar varices, hilar adenopathy and ascites. The abdominal fluid was positive for malignancy. The patient underwent chemotherapy, but ultimately succumbed to this disease 14 months after the diagnosis.
Our third and final case is a 36-year-old male with a history of ulcerative colitis and PSC who is screened annually for cholangiocarcinoma. His imagining studies by ERCP demonstrated bile duct strictures, but without defined mass. Conventional cytology demonstrated atypical cells. The patient was referred to the Mayo Clinic for further evaluation. A repeat ERCP was performed so we could obtain FISH studies.
Case 3 Cytologic Findings
The conventional cytologic findings were suspicious for malignancy.
What would you do? Repeat an ERCP in 3 months, repeat the ERCP now, act on suspicious cytology, or wait until cancer declares itself.
The FISH results demonstrated polysomy.
Case 3: Test Report Using Bile Duct Calculator
The positive predicted value for this case in a 36-year-old gentlemen with PSC, a dominant stricture, and polysomy is 98%.
These results are consistent with malignancy.
Indeed, the patient underwent neoadjuvant chemo-irradiation and liver transplantation for potential cure of cholangiocarcinoma. He underwent live donor transplantation in 2007. The explanted liver demonstrates stage 2 PSC with multifocal high and low grade dysplasia, but no residual invasive cancer. His chances of long-term survival are excellent.
In summary, FISH and biliary cytology increase sensitivity while maintaining a degree of specificity for detection of malignancy.
Addition of ancillary tests such as FISH for polysomy cell detection has improved sensitivity to identify malignancy in brush cytology from pancreatobiliary specimens. This test aides and helps modulate equivocal cytologic results providing 1 more piece of information that helps the clinician make a treatment and follow-up decision. They can also be reassuring if the cytology results run counter from the clinical impression. The enhanced sensitivity allows for an earlier diagnosis of potential cancer and therefore, cure. This approach also allows for an individualized medical approach with a statistical calculation of risk. Experience also helps to better understand limitations of test modalities over time.
If you have a patient with a high probability of malignancy, and have questions regarding treatment or management, please feel free to call me at 507-284-6488 to discuss options.