Surgical Management of Biliary Malignancy

I. Intrahepatic Cholangiocarcinoma

Presentation

Cholangiocarcinoma is a malignancy arising from the epithelial lining of the biliary ducts. The disease is further subdivided by anatomic location, namely intrahepatic, perihilar, and extrahepatic. Intrahepatic cholangiocarcinoma (IHCC) has also been referred to as cholangiocellular carcinoma, cholangiolar carcinoma, and peripheral cholangiocarcinoma. In 1990, IHCC was formally defined as malignancy arising from the biliary epithelium proximal to the left and right main ducts,¹ and subsequent studies have verified the utility of this classification for prognosis and management.^2,3 Because IHCC arises in the peripheral biliary tree within the parenchyma of the liver, it is associated with unique diagnostic and therapeutic challenges.

Epidemiology

IHCC accounts for 10% to 15% of all primary liver cancers and is the second most common primary liver malignancy after hepatocellular carcinoma. Despite this, it is a rare cancer and accounts for less than 1% of all gastrointestinal(GI) malignancies. The annual incidence of IHCC varies widely across the globe, ranging from 85/100,000 in Northeastern Thailand to 0.3/100,000 in Israel.⁴ In the US, the overall age-adjusted incidence is 1.31/100,000 per year.⁵

The incidence of IHCC appears to be increasing steadily worldwide, while the incidence of extrahepatic cholangiocarcinoma (EHCC) is decreasing.^5,6 Similar trends are also seen in the US.^7,8 Based on a review of the North American Association of Central Cancer Registries, the age-adjusted incidence of IHCC in men from 1999-2001 was 14.2/1,000,000 and rose to 18.2/1,000,000 from 2011-2013. The incidence also appears to be increasing among women, but at a higher rate.⁷ Although some have raised the possibility that this increase is due to changes in how cancers extending from the hepatic parenchyma to the segmental bile ducts are classified, the consensus is that this represents a true increase in disease incidence.⁹

Globally, IHCC most commonly presents during the seventh decade of life, with >75% of patients over the age of 65 at diagnosis. The disease is uncommon in those under the age of 40 years,^10,11 with the exception of patients with primary sclerosing cholangitis (PSC). In the US, the incidence of IHCC is higher in men than women, with age-adjusted incidence rate of 18.2/1,000,0000 person-years in men and 14.5/1,000,000 person-years in women, but the rate of increase is higher in women than in men.⁷

Risk Factors

A number of risk factors for IHCC have been established, including PSC, hepatobiliary parasite infection, hepatolithiasis, congenital defects in the biliary system, and exposure to certain chemicals. Many of these factors are associated with chronic inflammation in the biliary system. Most factors confer an increased risk for both IHCC and EHCC, suggesting a common etiology of cholangiocarcinogenesis. However, the majority of patients presenting with IHCC have no identifiable risk factors.

Lifestyle Factors

Lifestyle factors have been implicated in the development of IHCC in studies of Asian, European, and North American populations. A recent meta-analysis of 22 non-randomized studies from Western and Asian populations confirmed a moderate increase in risk associated with smoking (odds ratio [OR] 1.17, 95% confidence interval [CI] 1.05-1.31).¹² In a SEER-Medicare database case-control study comparing 625 IHCC cases with 90,834 controls, Shaib et al. identified smoking, alcoholic liver disease, and diabetes as significant risk factors for the development of IHCC.¹¹ A later SEER database case-control study by Welzel et al. compared 535 patients with IHCC to 102,782 disease-free controls and confirmed an increased risk associated with smoking, alcoholic liver disease, and type II diabetes.¹³ This study also identified nonalcoholic liver disease and obesity as risk factors for IHCC. A similar association between obesity and an increased risk of IHCC, but not extrahepatic disease, was found in a subsequent study by Welzel et al.¹⁴ The authors suggested that the rise in obesity and metabolic syndrome may be contributing to the rise in IHCC incidence.

Several studies have identified obesity-related illness as an important risk factor in the development of IHCC. A SEER database analysis of 649 hepatocellular carcinoma cases and 743 IHCC cases found that the presence of metabolic syndrome (as defined as three of the following: central obesity, dyslipidemia, hypertension, and impaired fasting glucose) was significantly associated with an increased risk of IHCC (OR 1.56, 95%CI 1.32-1.83; p<0.0001).¹⁴ Nonalcoholic fatty liver disease (NAFLD) is a common hepatic manifestation of metabolic syndrome, and multiple studies have found an association between NAFLD and IHCC. A recent case-control study of 34 Japanese patients with IHCC found nonalcoholic steatohepatitis to be an independent risk factor for development of IHCC (Kinoshita 2016).¹⁵ This association was confirmed in a systematic review and meta-analysis of 7 case-control studies comprising 5,067 IHCC, 4,035 EHCC, and 129,111 control patients that found NAFLD was an independent risk factor for IHCC (OR 1.97, 95%CI 1.41-2.75, I²=71%).¹⁶

Primary Sclerosing Cholangitis

PSC, a rare cholestatic liver disease characterized by progressive inflammatory destruction of intrahepatic and/or extrahepatic bile ducts, is the most significant risk factor for cholangiocarcinoma in the West. Typically, patients who develop IHCC associated with PSC present at a younger age, usually between 30 and 50 years of age. A Dutch retrospective study of 211 patients diagnosed with PSC over 3 decades found the 20-year risk of cholangiocarcinoma was 9%, but in patients with concomitant PSC and inflammatory bowel disease, the 10-year and 20-year risk for cholangiocarcinoma were 14% and 31%, respectively.¹⁷

The diagnosis of IHCC in patients with PSC can be challenging because of the overlap between PSC and IHCC symptoms and a lack of highly sensitive diagnostic tools. In particular, the imaging findings of developing IHCC may be indistinguishable from the fibrotic structuring caused by underlying PSC. Therefore, delayed IHCC diagnosis is frequent in this patient subset, and it is common for PSC patients to present with advanced IHCC.

Viral Hepatitis

Chronic viral hepatitis has been shown to be a risk factor for IHCC. Several retrospective case-control studies in Europe and Asia have found the prevalence of hepatitis C virus (HCV) or hepatitis B virus (HBV) infection in patients with IHCC to be 2-5 times higher than in those without IHCC.^18–20 In two US retrospective case-control studies using the SEER database, only HCV, not HBV, was found to be a risk factor for IHCC.^11,14 However, a recent meta-analysis of 12 studies from Europe, Asia, and the US identified both viral infections as risk factors.^21,22

A prospective Japanese study of 600 patients with HCV infection found that 2.3% of patients developed IHCC over a 2-decade follow-up period, an incidence approximately 1,000 times higher than in the general population.²³ Another retrospective study using Taiwan’s National Health Insurance Research Database compared 10,062 patients with HBV who received treatment to 10,062 propensity-matched controls who did not receive treatment for their HBV. The cumulative incidence of IHCC was significantly higher in patients with untreated HBV than those who received treatment (4.32%, 95% CI 2.96-5.68 vs. 1.53%, 95% CI 0.73-2.33; p=0.005), suggesting that uncontrolled HBV infection may contribute to the development of IHCC.²⁴

Liver Flukes

Infection with the liver flukes Clonorchis sinensis and Opisthorchis viverrini, foodborne trematode parasites that chronically infect the bile ducts, is a significant risk factor for the development of IHCC.²² C. sinensis and O. viverrini are typically acquired from fish or crayfish, and infection is endemic in China, Korea, and Southeast Asia. Liver flukes are frequently implicated as one of the reasons for the high incidence of IHCC in Southeast Asia, where the incidence is 93.8-317.6/100,000, and over half the population is projected to be infected with O. viverrini²⁵. It is thought that fluke infection leads to chronic inflammation of the biliary tract, creating a microenvironment that supports cholangiocarcinogenesis.

Hepatolithiasis

Hepatolithiasis, the presence of intrahepatic stones, is another risk factor for IHCC. Intrahepatic biliary stones are rare in the US and Europe, with an estimated incidence of 0.6-1.3%, but in Asia, the incidence ranges from 1.7% in Japan to 15% in South Korea and 18.7% in Taiwan.²⁶ The association between hepatolithiasis and IHCC has been well established in both Asian and Western countries. Incidental cholangiocarcinomas are frequently found when patients undergo hepatectomy for hepatolithiasis in Japan and China, with incidences ranging from 3.3-11.6%.^{27, 28} Case-control studies in Europe and Asia also consistently identify hepatolithiasis as a significant risk factor for IHCC.^19,27

The interplay between hepatolithiasis and liver flukes may further contribute to geographic differences in IHCC incidence, as helminth infection is considered a significant risk factor for the development of intrahepatic biliary stones. Concomitant infection with Clonorchis sinensis or Ascaris lumbricoides has been reported in >30% of cases of hepatolithiasis in some studies.²⁹ The degree to which hepatolithiasis and parasitic infection contribute individually to the development of IHCC has not been well-studied.

Chemical Exposure

A number of chemical exposures have been linked to IHCC. The most well-known is Thorotrast (thorium dioxide suspension), a radiocontrast agent commonly used for cerebral angiography until the 1950s. Thorotrast is an alpha-emitter that is retained in the reticuloendothelial system and increases the risk of cholangiocarcinoma 300-fold.³⁰ More recently, employee exposure to 1,2-dichloropropane and/or dichloromethane in a Japanese printing company was linked to subsequent development of cholangiocarcinoma, resulting in the categorization of both compounds as carcinogens by the World Health Organization (WHO).³¹ Several other known carcinogens have been associated with the development of IHCC, including asbestos and vinyl chloride.^32,33

Congenital Abnormalities of the Biliary System

Certain congenital malformations of the biliary tree are associated with increased risk of IHCC, including Caroli’s Disease, congenital hepatic fibrosis, and choledochal cysts. Choledochal cysts are cystic malformations of the biliary tree associated with a 10-15% overall incidence of cholangiocarcinoma. The risk is substantially lower, however, in patients who undergo early surgical resection. In children <10 years of age who receive surgery for a choledochal cyst, the lifetime risk of developing IHCC is 0.7%. In patients who receive treatment after the age of 20, however, the risk is 14%.³⁴ In studies of resected choledocal cysts in a mixed-age population, malignancy was reported in 5-10% of surgical specimens.^35,36 Even after resection, patients with choledochal cysts have an increased risk of developing cholangiocarcinoma compared to patients in the general population.^36,37

Symptoms

IHCC often remains asymptomatic until advanced stages. Several surgical series report up to one-third of patients being asymptomatic at diagnosis.^38,39 When patients are symptomatic, they often report vague symptoms of abdominal pain, fatigue, and loss of appetite. Several surgical series report abdominal pain as the most common presenting symptom in IHCC, in contrast to hilar and distal cholangiocarcinoma where jaundice and gastrointestinal upset are nearly ubiquitous.^38,40

IHCC patients rarely present with symptoms of biliary obstruction, but a small subset of patients (10-15%) may present with jaundice from tumor extension involving the bile ducts at the biliary confluence from tumor thrombus, or mass effect from the primary tumor or metastatic lymph nodes. Jaundice on presentation is generally an ominous sign, as it indicates advanced or anatomically disadvantageous disease; patients with jaundice on presentation are much less likely to have resectable disease. In one study, symptomatic patients were found to have a resectability rate of 25%, compared to 58% in asymptomatic patients (p<0.001).³⁸

Diagnosis and Imaging Work Up

Complete resection remains the only potentially curative option for IHCC, and thus, workup should focus on confirming the diagnosis and appropriately staging patients to determine the appropriate treatment strategy. Furthermore, it is important to determine the extent and location of disease, evaluate the amount of functional liver that will remain after resection, identify any acute medical issues that must be addressed before surgery, and evaluate the patienťs overall physiologic capacity to undergo surgery.

As with any condition, workup of IHCC should begin with a detailed history, thorough physical examination, and evaluation of comorbid conditions. Special attention should be given to signs of chronic liver disease. Liver function should be assessed with serum alanine aminotransferase, aspartate aminotransferase, bilirubin, alkaline phosphatase, albumin, platelet count, and prothrombin time. Viral hepatitis serology is also useful for directing perioperative care and involving a hepatologist if required. Signs of undiagnosed or worsening liver disease should prompt further investigation. A detailed history of prior malignancies is important.

National Comprehensive Cancer Network (NCCN) guidelines recommend patients are evaluated for additional primary malignancies or metastatic disease with contrast-enhanced cross-sectional imaging of the chest, abdomen, and pelvis (including multiphasic liver imaging). Patients are all recommended to undergo colonoscopy and upper endoscopy to rule out primary gastrointestinal malignancy. In patients with relevant anatomy, a mammogram and gynecologic exam should be performed if not done recently.

A tissue diagnosis is usually obtained before starting systemic or palliative treatment, but for patients for whom resection is likely, biopsy confirmation of the diagnosis is not required prior to surgery.⁴¹

Imaging

IHCC often presents as an incidental malignant-appearing liver mass on imaging performed for other indications. When caught early these can be small masses but they more typically present as large and often central masses (Figure 1). It can be challenging to distinguish IHCC from hepatocellular carcinoma or liver metastases, which are much more common. Thus, multiple imaging modalities are often necessary for staging and surgical planning.

Figure 1.

Portal venous phase computed tomography scan of a large central tumor that was found to be Intrahepatic cholangiocarcinoma on workup.

Transabdominal Ultrasound

Transabdominal ultrasound is often the first imaging modality clinicians use to examine right upper quadrant abdominal pain or jaundice. On abdominal ultrasound, IHCC appears as a heterogeneous hypoechoic liver mass. Tumors are often fibrotic and poorly vascularized and therefore demonstrate little internal blood flow on Doppler assessment.⁴² Tumors may be associated with capsular retraction, which is rare in other tumor types.⁴³ Transabdominal ultrasound can identify hepatic venous invasion and portal invasion, but overall detection of vascular invasion is not accurate with this modality.^44,45 Abdominal ultrasound also has poor sensitivity for detection of peritoneal and liver metastases and lymph node disease.⁴⁵

Computed Tomography

Triple-phase computed tomography (CT) is the standard imaging method for the preoperative evaluation of IHCC. The triple-phase imaging protocol includes arterial, portal venous, and delayed phases. Arterial phase images are collected 20-30 seconds after injection of intravenous contrast, portal-venous phase images are obtained 60 seconds after injection, and delayed phase images are acquired 3-10 minutes after injection.

The classic CT findings for IHCC are a well-defined, irregular, heterogeneous mass with peripheral enhancement, though only 50-60% of IHCC have all of these features on presentation.⁴⁶ Virtually all IHCC display delayed enhancement in the portal venous phase due to the dense fibrosis often found in the tumor.⁴⁷ The degree of hypoenhancement in the arterial phase may have prognostic implications, as hypoenhancement has been associated with increased likelihood of biliary, perineural, and lymphatic invasion.⁴⁸

IHCC is a rare finding compared to hepatocellular carcinoma or liver metastases, but can be distinguished from more common liver masses on CT. Hepatocellular carcinoma displays hyperenhancement in the arterial phase with washout in the venous phase, while IHCC demonstrates progressive uptake of contrast during the arterial and venous phases.⁴⁹ IHCC and colorectal liver metastases can both display central hypoenchancement, but IHCC can be differentiated from these metastatic lesions by the presence of distal biliary dilation.

CT can also be used to evaluate the hepatic parenchyma for cirrhosis and steatosis, which have implications for operative outcomes and prognosis. Additionally, there are a number of manual and software-based techniques that utilize cross sectional imaging such as CT or MRI for estimating the volume of the liver remnant after hepatectomy, which can be useful for predicting the risk of postoperative liver failure.⁵⁰

Magnetic Resonance Imaging

IHCC on magnetic resonance imaging (MRI) appears hypointense on T1 imaging relative to the normal liver. On T2 imaging, IHCC appears heterogeneous, with approximately half of tumors displaying central hypointensity and peripheral hyperintensity and pooling of contrast on delayed images.⁵¹ Liver metastases from colorectal cancer can display similar central hypointensity, but this is a rare finding in other liver masses. IHCC can be further distinguished by the presence of intrahepatic biliary dilation peripheral to the mass (more obvious on MRI but also can be seen on CT), which is found in 54% of IHCC patients, but only 3% of patients with colorectal liver metastases.⁵² IHCC can be distinguished from hepatocellular carcinoma by contrast uptake; a recent retrospective study found while no MRI imaging feature was unique to either tumor type, IHCC was more likely to present with peripheral arterial phase enhancement or progressive central enhancement, while hepatocellular carcinoma was more likely to demonstrate washout.⁵³

MRI is also useful for surgical planning, as MRI can delineate vascular involvement of the tumor, and magnetic resonance cholangiopancreatography (MRCP) can be used to assesses biliary tree involvement.

Positron Emission Tomography

Positron-emission tomography (PET) is a common modality for staging gastrointestinal cancers, but data suggest it is of limited value in IHCC after other cross-sectional imaging has been performed. By PET, IHCC appears as fluorodeoxyglucose (FDG)-avid lesions within the liver parenchyma, often with peripheral enhancement matching that found on CT .⁵⁴ In a study of all biliary cancers in the early 2000s, Petrowsky et al. found no benefit for PET-CT compared to CT alone for the detection of primary IHCC or regional lymph node metastases, although PET-CT significantly improved detection of distant metastases.⁵⁵ Kim et al. reported similar results in a study of IHCC alone, showing no improvement in detection of regional lymph node metastases, but significantly better detection of occult distant metastases. These studies found that PET-CT changed patient management in 20-30% of cases.⁵⁴ A more recent systemic analysis of 47 studies of PET-CT in biliary cancers found PET-CT had a sensitivity and specificity of 88.4% and 69.1%, respectively, for regional lymph node metastases and 85.4% and 89.7%, respectively, for distant metastases.⁵⁶ In our practice we do not use PET routinely but it can be used as a problem solving tool when standard cross-sectional imaging raises concerns for metastatic disease.

Morphologic Classification

The Liver Cancer Study Group of Japan defined 3 subtypes of IHCC, mass-forming, periductal infiltrating, and intraductal growth types, based on gross morphological features. The mass-forming subtype is the most common (60-80% of cases)⁵⁷ and appears as a discrete mass, distinct from surrounding liver on imaging, that is firm and nodular on pathology. The mass-forming subtype has been identified as a predictive risk factor for lymph node metastases as compared to other morphologic types.⁵⁸ Periductal infiltrating is the second most common type (15-35%) and is characterized by growth along the bile duct without mass formation. It can present as diffuse bile duct thickening or as irregular bile duct dilation. Intraductal growth is the rarest subtype (10%); it is a slow-growing, papillary subtype that presents as nodular masses within the bile duct with proximal bile duct dilation. In some studies, the intraductal growth subtype has been associated with less aggressive disease as compared to other morphologic subtypes, but other studies have found no difference in survival among subtypes. ^58,59

Biomarkers

Currently, there are no reliable diagnostic serum biomarkers for IHCC. Common serum tumor markers used in other cancers lack sensitivity and specificity for IHCC. The NCCN guidelines recommend considering serum biomarker testing – specifically, cancer antigen (CA) 19-9 and carcinoembryonic antigen (CEA) as baseline tests, and alpha-fetoprotein (AFP) to help differentiate between IHCC, hepatocellular carcinoma and mixed IHCC/hepatocellular carcinoma.⁶⁰

CA 19-9 is an epithelial protein used as a diagnostic and surveillance tool in several upper gastrointestinal cancers, including pancreatic, gastric, and biliary tree cancers. In patients without risk factors for cholangiocarcinoma, the sensitivity of CA 19-9 for the diagnosis of cholangiocarcinoma varies from 53-92% and the specificity from 50-98%.⁶¹ A meta-analysis of 31 studies representing 1,264 patients with cholangiocarcinoma and 2,039 controls found that CA 19-9 had a sensitivity of 72% and specificity of 84% for detecting cholangiocarcinoma on pooled analysis.⁶² Further complicating testing accuracy, approximately 7% of the population are Lewis a-antigen negative and do not produce CA 19-9, and CA 19-9 elevation is associated with several benign conditions, including heavy tea consumption, tobacco smoking, cholangitis, and conditions causing cholestasis.⁶³

The response of CA 19-9 to biliary drainage in cholestasis may help differentiate between benign and malignant etiologies of elevation. A retrospective review of 128 consecutive patients with obstructive jaundice (87 with pancreaticobiliary malignancy and 41 with benign disease) found that the majority of patients had elevated CA 19-9. After biliary drainage, 40/41 patients with benign etiologies experienced a decline in CA 19-9, whereas of patients with underlying malignancy, only 50% saw a drop in CA 19-9 with resolution of biliary obstruction.⁶⁴

CEA is elevated in many GI cancers, although it is only elevated in one-third of patients with cholangiocarcinoma. The sensitivity for detecting IHCC has been found to range from 33-68% and the specificity from 79-100%.⁶¹ AFP is commonly elevated in hepatocellular carcinoma and can be used to help distinguish HCC and IHCC . The sensitivity of negative AFP in distinguishing IHCC from HCC is >90%.⁶⁵ AFP may be mildly elevated in up to 20% of cases of IHCC, and rare cases of IHCC with extremely elevated AFP have been reported.^66,67

Resectability

Similar to other oncologic liver resections, IHCC is considered resectable if the tumor can be fully removed with negative margins while maintaining sufficient functional liver with adequate perfusion and venous and biliary drainage. The patient also must have sufficient physiologic reserve to tolerate the operation. Unfortunately, IHCC is often diagnosed at an advanced stage, and therefore, at least half of patients are not candidates for liver resection at the time of diagnosis.³⁸

The presence of distant metastases, including lymph node metastases outside the regional basin (including the celiac and periaortic nodes), is a contraindication to resection. Some practitioners consider that the presence of multiple intrahepatic tumors is not an absolute contraindication to resection, but it is clear that multifocal disease likely represents intrahepatic metastases and is associated with poor oncologic outcomes after surgery.^38,68,69

Multiple intrahepatic tumors are present in 42% of patients on presentation.⁷⁰ In one recent study, Buettner et al. compared outcomes between patients undergoing resection for solitary and multiple intrahepatic lesions, the median overall survival in patients with 2 tumors was 21.2 months versus 15.3 months in those ≥3 tumors. While not directly comparable and highly susceptible to staging and selection bias, these survival times are longer than the 1-year medial survival usually cited for systemic and locoregional ablative therapies for unresectable IHCC.⁷¹

The distribution and number of tumors in those with multifocal disease may affect survival after surgery, suggesting that a subset of highly selected patients with multifocal disease may derive a benefit from resection. A 2018 study of 259 patients who underwent resection for IHCC found patients with multifocal disease limited to a single liver segment had significantly longer survival after resection than patients with tumors in multiple liver segments (five-year survival: 34.2% and 9.9%, respectively, p=0.001). The overall survival of patients with single-segment multifocal disease was still shorter, however, than that of patients with a solitary tumor (49.4 months, p<0.001).⁷² Furthermore, in the Buettner et al. study of 1013 patients undergoing resection for IHCC, the presence of two lesions in the liver was not a poor prognostic factor for overall survival on multivariate analysis (HR 1.19, 95%CI 0.90-1.57; p=0.229), while the presence of three lesions or more was associated with shorter overall survival (HR 1.97, 95%CI 1.48-2.64; p<0.001).⁷¹

Although resection is associated with prolonged survival in certain subsets of patients with multifocal disease, other treatment modalities have recently been shown to be associated with similar outcomes. A recent study of 116 consecutive patients treated for multifocal IHCC found no associated difference in survival with resection compared to intra-arterial therapy, including chemoembolization, radioembolization, or hepatic artery infusion therapy. Resection was associated with a higher rate of 90-day morbidity when compared to intra-arterial therapy (31.6% vs. 8.5%, p= 0.009).⁷² In general, our practice has been to treat multifocal disease with combinations of chemotherapy (regional and systemic) and intr-arterial therapies. Highly selected patients with limited multi-focal disease more consistent with regional satellite tumors have been offered resection.

Regional lymph node (i.e., hepatoduodenal, periduodenal, and peripancreatic nodes) metastasis in IHCC is also associated with worse overall survival after resection^38,68·⁶⁹, but the presence of these metastases is not an absolute contraindication to resection. Several studies have identified five-year survival as high at 26% after resection when metastatic disease is found in regional lymph nodes at surgery, suggesting that the presence of IHCC in regional lymph nodes does not preclude long-term survival.⁷⁴ A recent review of the National Cancer Database (NCDB) identified overall survival after resection in patients with positive regional lymph nodes to be 15 months, but among patients with positive lymph nodes, younger age was a positive predictive factor for survival (HR 1.34, 95%CI 1.19-1.57; p<0.001) and receipt of adjuvant chemotherapy was associated with longer overall survival (19.1 vs. 9.3 months without treatment; p=0.033). These data suggest that among patients with positive lymph nodes, subgroups may exist with better outcomes after surgery.⁷⁵

Preoperative Preparation

Preoperative care for IHCC focuses on minimizing risks of perioperative complications. Comorbid conditions should be well-managed preoperatively. All patients will at the least undergo a partial hepatectomy, so the function of the estimated remaining liver should be assessed and optimized prior to surgery.

Viral Hepatitis

Recent data suggest that treating viral hepatitis before hepatectomy may improve oncologic outcomes. A recent retrospective study of 928 patients with IHCC and HBV infection found that patients who received antiviral therapy prior to resection had a reduced risk of recurrence compared to those with a high viral load who did not receive therapy (HR 0.60, 95%CI 0.46-0.79).⁷⁶ There are no studies on the effect of treating HCV in patients with IHCC, but HCV treatment in hepatocellular carcinoma suggests that treatment may improve outcomes after surgery. A retrospective study of 220 patients with HCV infection who underwent partial hepatectomy for hepatocellular carcinoma found that sustained viral response after treatment with antiviral medication was an independent positive predictor of survival (HR 0.61, 95%CI 0.40-0.94; p=0.024).⁷⁷ Although evidence for treatment of viral hepatitis before surgery for IHCC is limited, the involvement of a hepatologist in treatment decision making during the preoperative period is warranted.

Assessment of Functional Liver Volume

Most patients with IHCC require major hepatectomy due the commonly found large size and central location of these tumors. Therefore, prior to surgery, patients should undergo assessment of the volume and functional capacity of the liver that will remain. The rate of liver failure after partial hepatectomy for IHCC has been found to be 6.5-6.9%.^78,79 The risk of post-resection liver failure is directly related to the volume of functional liver remaining after surgery. The remnant liver volume (RLV), a measure of the percentage of functional liver remaining after surgery compared to the volume of functional liver before surgery, is considered a reliable predictor of liver failure.^80,81 In normal liver, an RLV of <25% has a 90% (95%CI 68-99%) positive predictive value for post-hepatectomy liver failure.⁵⁰ In patients with compromised liver function, an RLV of >40% is generally recommended.⁸¹

CT and MRI can offer accurate estimates of the volume of normal liver parenchyma, tumor volume, as well as information about the quality of the liver parenchyma (e.g., the presence of steatosis, cirrhosis, or atrophy). Formal manual or software-assisted volumetric assessment should be performed prior to hepatectomy. There are also a number of dynamic tests that can quantitatively assess liver function when there is concern that patients may have compromised hepatic reserve. Indocyanin green retention at 15 minutes (ICG-15) is a measurement of hepatic perfusion. A value of <10% retained at 15 minutes is considered normal in most labs, and >8% retained has been associated with postoperative liver dysfunction.⁸²

Portal Vein Embolization

In patients with marginal predicted functional liver remnant, portal vein embolization (PVE) may be employed to initiate hypertrophy in the liver remnant and increase functional capacity. PVE causes redistribution of portal blood flow to the non-embolized segments, promoting compensatory regenerative hyperplasia, which can increase the functional liver remnant and reduce the risk of postoperative liver failure. PVE has been shown to result in as much as a 62% relative increase in the functional liver remnant in 4-6 weeks.⁸³ The rate of regeneration after PVE has been shown to be a predictor of the risk of post-hepatectomy liver failure.⁸⁴

PVE has never been compared to associating liver partition and portal vein ligation for staged hepatectomy (ALPPS), a two-stage procedure in which the portal vein branch is surgically ligated. Retrospective studies comparing PVE to ALPPS have demonstrated equivalent post-procedure morbidity and efficacy inducing hypertrophy.⁸³ The addition of PVE does not introduce risk of additional surgical complications or worse surgical outcomes. When compared to patients with higher predicted FLR who did not require PVE before surgery, patients with lower initial predicted FLR who receive PVE have been shown to have the same rate of R0 resection and complications.⁸⁵

Biliary Drainage

While biliary obstruction is much less common in IHCC than in other biliary malignancies, large or centrally located tumors can cause biliary obstruction and jaundice in a subset of patients. Cholangitis, an uncommon presenting symptom, is an absolute indication for biliary drainage, but the role of biliary drainage for jaundice alone is limited. Historically, all jaundiced patients were offered biliary drainage preoperatively, with the thought that hyperbilirubinemia may cause coagulopathy and contribute to poor perioperative outcomes. In recent years, studies have demonstrated that jaundice is rarely associated with worse surgical outcomes and that preoperative biliary drainage may actually be associated with higher complication rates.^86,87

There are few studies on preoperative biliary drainage in IHCC, but several studies in hilar cholangiocarcinoma suggest drainage may contribute to improved liver regeneration and reduced postoperative complications.^88,89 However, several other small studies have found no difference in surgical outcomes in jaundiced patients who underwent hepatectomy for hilar cholangiocarcinoma without preoperative drainage.^90,91 Preoperative biliary drainage has been shown to increase the risk of infectious complications, likely due to contamination of the biliary tree.⁹² This risk can likely be minimized by using transhepatic drains that do not cross the ampulla. Any drainage procedure introduces the procedural risks of choleperitoneum, bleeding, cholangitis, and seeding of tumor, and should be performed only when required.

A prospective study of 434 patients who underwent surgery for IHCC found that patients who underwent biliary drainage had significantly higher rates of morbidity than patients with jaundice who did not undergo drainage procedures (67.5% vs. 43.5% p=0.058).⁶⁸ A subset of patients, including those with marginal liver function, may still benefit from pre-operative drainage. Two series from Memorial Sloan Kettering Cancer Center did find that while biliary drainage did not affect surgical outcomes overall, it did improve overall survival and reduce postoperative liver failure in patients with <30% functional liver remnant.^{93, 94}.

Operative Approaches

Surgery for select patients has been shown to be associated with prolonged survival and is currently the only potentially curative treatment for IHCC. Even with surgery, however, overall survival remains poor, with most studies estimating an overall survival of 23.4-30.0%.^95–97 The cure-rate is also low after surgery, with a recent study of 584 patients undergoing curative-intent surgery showing a 9.6% cure rate.⁹⁷ Although discouraging, this still compares favorably to the five-year survival for patients with unresectable disease, which is 5-10%.⁷⁰

The goal of surgery in IHCC is to achieve full excision of the tumor with negative margins, stage appropriately with lymphadenectomy, and to maintain adequate FLR. The majority of patients now undergo major hepatectomy and lymphadenectomy, with a smaller portion of patients also undergoing a biliary resection. The decision to perform major versus minor hepatectomy is dictated by tumor location and characteristics. In a retrospective multi-institution study of 1,023 patients undergoing curative-intent resection for IHCC, major hepatic resection was associated with a higher rate of postoperative complications than minor hepatectomies (48.4% vs. 27.2%, p<0.001) and 90-day mortality (7.7% vs. 3.1%, p=0.002). After propensity matching, there was no difference in overall and recurrence-free survival for major and minor hepatectomies.⁹⁸

It has not been well-established whether anatomic resection confers better outcomes than nonanatomic resection in IHCC patients. A retrospective review from 2005 of 659 patients who underwent hepatic resection for IHCC at a single institution found no survival difference between anatomic and nonanatomic resections. A smaller propensity-matched study of 150 patients also found no difference in overall and recurrence-free survival between patients who received anatomic and nonanatomic resections.⁹⁹ Finally, a recent propensity-matched review of 702 cases found that overall, anatomic resection had no advantage over non-anatomic resection but was associated with improved overall and recurrence-free survival in patients with stage IB or II tumors without vascular invasion.¹⁰⁰

Lymphadenectomy

Regional lymph node metastases are a common finding in IHCC, reported in approximately 30% of patients who undergo resection.^40,70 The regional lymph nodes for right-sided IHCC include the periduodenal, peripancreatic, and hilar nodes, whereas the common bile duct cystic duct, portal vein, and hepatic artery nodes are the regional lymph nodes for left-sided IHCC. The periaortic, caval, and celiac lymph nodes are considered distant metastases.¹⁰¹

While recommended for staging and treatment of IHCC, lymphadenectomy with hepatic resection has not been widely performed at Western centers. Surgical series report lymph node dissections in only 45-55% of resections.^75,102,104 A large review of the NCDB in 2018 reported sampling of ≥6 lymph nodes in only 11.6% of resections.¹⁰⁵ The current evidence for a therapeutic advantage of lymphadenectomy is uncertain. Two studies found that patients with 1 or 2 positive regional lymph nodes who underwent lymphadenectomy had comparable survival to patients with NO disease, arguing that lymph node dissection has therapeutic value.¹⁰⁶ Ribero et al. found the therapeutic index, a calculation of potential survival benefit of lymphadenectomy based on the incidence of positive regional lymph nodes and the five-year survival of patients with N1 disease, to be similar to that of mesorectal lymphadenectomy in rectal cancer.⁶⁸ However, a large systemic analysis failed to identify any association between regional lymph node dissection and prolonged survival after surgery for IHCC patients.¹⁰²

There is, however, clear prognostic value of lymph node dissection. Regional lymph node metastases are one of the most important prognostic factors in IHCC and are consistently associated with poor survival after surgery.^75,102–104 In the 7^th edition of the AJCC staging guidelines, recovery of a minimum of 6 lymph nodes was recommended for accurate staging (AJCC). The Americas Hepato-Pancreato-Biliary Association (AHPBA) expert consensus guidelines also recommend lymph node dissection for accurate staging and guidance of adjuvant therapy.⁴¹

Vascular Resection

Vascular involvement is commonly required in IHCC, with studies finding the rate of macrovascular involvement around 10.8-12.0%.^70,107 As such, vascular resection may be required to achieve R0 status in some patients. Resection of the portal vein, inferior vena cava, or a combined resection of the two has not been shown to be associated with poor perioperative or oncologic outcomes^108,109, and invasion of these vessels does not preclude curative surgery. Hepatic artery resection is consistently associated with increased perioperative morbidity and mortality^108,110 and is not generally recommended.

Staging Laparoscopy

Staging laparoscopy is a matter of surgeon preference but is recommended to be considered by the NCCN guidelines and by AHPBA for patients with a high risk of metastases.⁴¹ The goal of diagnostic staging laparoscopy in IHCC is to identify the presence of locally advanced or metastatic disease that precludes resection. Reports of cross-sectional imaging accuracy for predicting resectability vary from 60-90%; however, several studies have found that diagnostic laparoscopy changes management in patients planned for resection in 27-38% of cases.^111,112 Patients with unresectable disease detected on laparoscopy have a shorter length of stay and less postoperative morbidity than patients with unresectable disease detected on laparotomy¹¹² and will have a short delay in the start of systemic treatment. There are certain patient subgroups where staging laparoscopy should be considered.¹¹³ Multiple intrahepatic lesions are considered by many to be equivalent to metastatic disease, not local progression, and these patients are recommended for staging laparoscopy. High CA 19-9 has also been associated with a higher risk of unresectable disease.

Minimally Invasive Surgery

Minimally invasive surgery has become more popular and more often utilized over the past few decades, with increasing use of laparoscopic and robotic techniques for complex hepatobiliary surgeries. Multiple retrospective reviews have demonstrated that laparoscopic liver resection has a similar safety profile to open surgery and may offer some short-term benefits, such as reduced length of stay and requirement for transfusion.¹¹⁴ Studies of outcomes in other cancers have shown equivalent oncologic outcomes between minimally invasive and open surgeries.^115,116 The recent OSLO-COMET trial randomized patients with colorectal liver metastases to undergo a parenchyma-sparing resection with either an open or laparoscopic technique (133 and 147 patients in each arm, respectively). The study found that patients who underwent laparoscopic resection had a shorter length of stay and a lower rate of postoperative complications, with similar operative outcomes.¹¹⁷

Data specific to IHCC are limited. A recent meta-analysis of six studies, including 384 patients, found that patients who underwent laparoscopic hepatectomy for IHCC had a higher rate of R0 resection and comparable perioperative complications and overall survival to those who underwent open hepatectomy.¹¹⁶ In this study, the cohort that underwent laparoscopic procedures had significantly smaller tumors than those undergoing open surgery. However, a recent study found that patients with large (>5 cm) and multiple IHCC tumors had comparable perioperative and long-term outcomes to those with smaller, solitary tumors¹¹⁸, suggesting that size may not be an important factor in the safety of laparoscopic surgery for IHCC. Reports on robotic surgeries for IHCC are limited to case reports. The decision to perform minimally invasive surgery for the resection of IHCC depends on surgeon and institutional experience and capabilities.

Liver Transplantation

The data on liver transplantation for IHCC are mixed, and transplantation is not regularly recommended. In hilar cholangiocarcinoma, transplantation preceded by neoadjuvant chemoradiation has demonstrated a survival association in patients with PSC and those with unresectable disease. However, no studies have identified prolonged survival after liver transplant in IHCC patients.¹¹⁹ The earliest study of transplantation for IHCC compared 20 IHCC patients with either unresectable disease or cirrhosis who underwent orthotopic liver transplant with 34 patients who underwent hepatectomy with curative intent between 1981 and 1994. None of these patients received neoadjuvant chemotherapy or radiotherapy. The study did not demonstrate any survival difference between the orthotopic liver transplant and hepatectomy groups.¹²⁰ A recent systematic review of transplantation for IHCC in patients in the Data Analysis Booklet of the European Liver Transplant Registry (ELTR) (n=186) demonstrated 1-, 5 and 10-year survival rates for transplantation for IHCC of 58%, 29%, and 21%, respectively, significantly worse than the survival after transplantation for hepatocellular carcinoma during the same timeframe (l-,5-, and 10-year survival 74%, 51%, and 42%). Five other single-center studies also found significantly worse survival after transplantation for IHCC than HCC.¹²¹

More recent investigations looking at incidental IHCC in patients undergoing transplant suggest that a subset of patients may benefit from transplantation. Takahashi el al. retrospectively reviewed 1,188 liver transplant recipients from 2003-2014 and identified 13 patients with IHCC in the explanted liver. Compared to a matched cohort of 39 patients with hepatocellular carcinoma, patients with IHCC had significantly higher rates of recurrence at 1 and 3 years. However, in 4 patients with well-differentiated IHCC, there was no recurrence at the time of analysis¹²², suggesting that transplant may be associated with prolonged survival in patients whose tumors display less aggressive biology. Another retrospective analysis of 7,503 patients who underwent liver transplantation at international centers identified 48 patients with IHCC as the only cancer in the explanted liver. These patients were classified into “very early” (single tumor >2 cm) and “advanced” (multifocal disease) disease groups. Patients with “very early disease” had a significantly lower risk of recurrence (1-, 3-, and five-year cumulative risk of recurrence: 7%, 18%, and 18%, respectively) and higher actuarial overall survival at 1,3, and 5 years (93%, 84%, and 65%, respectively) than those with “advanced disease” (79%, 50%, and 45%, respectively) (Sapisochin 2016).¹²³ Taken together, these studies suggest that transplantation may be associated with longer survival in patients with early disease and less aggressive tumor biology. Currently, transplantation does not play a major role for this disease.

Surgery for Recurrent Disease

There are no formal recommendations for resection of recurrent disease after curative-intent surgery. Because recurrence is common and often limited to the liver, re-resection can be performed in a select group of patients. Several small studies confirm this is safe and may be associated with a survival advantage in select patient groups. In a retrospective review of 127 patients who underwent curative-intent resection of IHCC, 60.6% recurred, and of the patients with recurrent disease, 15.5% were deemed eligible for re-resection.¹²⁴ Another study by Souche et al. found that in 125 patients who had undergone curative-intent surgery for IHCC, the recurrence rate was 63.5%, and 13.0% of recurrent patients were deemed eligible for re-resection (adequate health status to tolerate surgery, reasonable expectation of R0 resection with a future liver remnant of 40% total liver volume).¹²⁵

In patients selected for re-resection, outcomes are similar to initial resection. A retrospective review of 72 patients who underwent repeat R0 resection for recurrent IHCC reported operative morbidity and mortality of 18.1% and 1.4%, similar to values reported in other case series for first-time resections for IHCC (mortality 1.2-7.3%, morbidity 32-44%) ^{38,40,68,78,79} Additionally, overall survival after surgery was similar to that in other reports of first-time resections, with a five-year survival of 41.9% (Si 2017).¹²⁶

In comparison with other methods, re-resection has also been associated with prolonged survival. In a multi-institution review of 485 patients who recurred after curative-intent surgery, 88 patients underwent repeat curative-intent surgery. Repeat resection patients had better overall survival than patients who underwent non-curative resection, chemoradiation, or supportive care (median survival: 48.6 vs. 9.7 months, p<0.001), however, these are not comparable patient groups. While these studies do not account for selection of patients for resection versus non-operative management, it does suggest that there are patients for whom re-resection is associated with improved survival.¹²⁷

Operative Morbidity and Mortality

Surgery for IHCC is often extensive and physiologically demanding. Hepatectomy is the mainstay of surgical management for IHCC – surgical series report 52-70% of patients undergo major hepatectomy (resection ≥3 hepatic segments). Furthermore, 20-50% of patients undergo additional procedures, including vascular resection, extrahepatic bile duct resection, and lymphadenectomy.^40,68,78,79

When partial hepatectomy for liver malignancy first entered common practice, resection was associated with significant morbidity and mortality approaching 20% in major hepatectomy (Foster 1977).¹²⁸ Over the past four decades, however, the number of partial hepatectomy cases has increased, and advancements in hemostasis and perioperative care, and improvements in surgical and anesthesia techniques have improved perioperative outcomes. Partial hepatectomy is associated with 20.6-32.8% morbidity and 1.8-5.8% mortality, depending on the extent of resection .¹²⁹ Complication and mortality rates are similar for malignant and benign indications for resection.¹²⁹

The current era of liver surgery is marked by increased attention to parenchymal preservation. In a review of 4,152 liver resections for malignancy over 19 years at Memorial Sloan Kettering Cancer Center, Kingham et al. found that the rate of major hepatectomy decreased from 66% in the first 6 years to 36% in the final 6 years. This reduction in the extent of resection was associated with lower intraoperative blood loss, a lower rate of complications, and a lower rate of 30-day readmission .¹³⁰

In IHCC, perioperative mortality ranges from 1.2-7.3%, although single-institution reviews suggest that it is declining over time.^38,40 Perioperative complications are common, with surgical series reporting morbidity ranging from 32-44%.^40,68,78,79 The most common complications across series include liver dysfunction/insufficiency, intrabdominal fluid collection or abscess, pleural effusion, pneumonia, and sepsis. Both morbidity and mortality rates increase with more extensive resections ^40,68,78,79

Neoadjuvant and Adjuvant Strategies

Neoadjuvant Chemotherapy

Neoadjuvant chemotherapy is not routinely recommended for resectable IHCC. In one multicenter retrospective review of 1,057 patients who underwent hepatectomy for IHCC, only 5.9% of patients received preoperative chemotherapy. These patients had comparable overall survival after curative-intent surgery to those who did not have preoperative chemotherapy.¹³¹

There are case reports of large, locally invasive tumors being downstaged to complete resection even in the setting of multifocal disease and extensive vascular invasion (Hashimoto 2015, Tran 2015).^132–133 A review of 186 consecutive patients diagnosed with IHCC found that 53% of the patients with initially unresectable disease were able to undergo resection. Although these patients more often presented with nodal disease and vascular invasion, their overall survival after surgery was not different from patients who were initially resectable.¹³⁴

To date, there have been no randomized studies of preoperative chemotherapy for downstaging of IHCC or for oncologic outcomes. A randomized controlled trial of neoadjuvant gemcitabine, cisplatin, and nab-paclitaxel for high-risk resectable IHCC is currently recruiting (NCT03579771).

Adjuvant Chemotherapy

There are few prospective, randomized studies of adjuvant chemotherapy after curative-intent resection for IHCC either to establish which patient populations benefit or to define a standardized regimen. Early randomized studies of mixed biliary cancers did not find an improvement in overall survival in patients who received adjuvant therapy, either 5-FU-based or gemcitabine-based regimens, compared to observation alone^135,137 These studies were underpowered to evaluate IHCC alone, and subsequent studies suggest there may be a survival advantage to adjuvant therapy in subsets of patients. A systematic review and meta-analysis of 20 studies comprised of nearly 6,000 patients with IHCC found a trend toward improved overall survival with adjuvant therapy (p=0.06). The analysis also found an associated prolonged survival for adjuvant chemotherapy in patients with positive resection margins (OR 0.36, 95%CI 0.19-0.68; p=0.002) and node-positive disease (OR 0.49, 95%CI 0.3-0.8; p=0.004)¹³⁹ Similarly, an NCDB analysis of 2,751 patients who underwent surgical resection for IHCC found improved survival associated with chemotherapy in patients with N1 disease, T3/4 tumors, and R1/R2 resections¹³⁸ A subsequent multi-institution retrospective analysis of 1,154 IHCC patients who underwent curative-intent hepatectomy over 25 years also found an associated survival advantage with adjuvant chemotherapy. In this group, adjuvant therapy was associated with longer five-year overall survival in patients with node-positive disease (18.3% therapy vs. 12% observation alone 18%; p=0.050) and in patients with T2-4 disease (37% therapy vs. 30% with observation alone; p=0.005). In this cohort study, 30% of patients (n=347) received adjuvant therapy, and 52% of these patients (n=184) had been treated with gemcitabine-based regimens.¹⁴¹ Retrospective analyses of adjuvant therapy must viewed with significant caution since selection bias (despite matching or multivariable modeling) since selection bias likely accounts for much of the reported outcomes.

The recent randomized prospective trial of adjuvant xeloda versus observation (BILCAP trial) of 447 mixed biliary cancers (84 IHCC) found a significant difference in overall survival in a per-protocol analysis when adjusted for node status, disease grade, and sex (HR 0.71, 95%CI 0.55-0.92; p<0.01).¹⁴⁰ The Prodige-12 trial of 196 patients with mixed biliary tree cancers, 86 of whom had intrahepatic cholangiocarcinoma, showed no survival difference between gemcitabine plus oxaliplatin versus observation.¹⁴¹ A recent randomized, prospective trial evaluating chemotherapy regimens in 410 advanced biliary cancers (241 bile duct cancers) found that combination gemcitabine and cisplatin was associated with longer overall survival (11.7 months, CI95% 9.5-14.3) as compared with gemcitabine alone (8.1 months, CI95% 7.1-8.7) (p<0.001).¹⁴² A randomized, prospective, phase III trial (ATICCA-1, NCT02170090) is currently recruiting to test this combination in the adjuvant setting.

In summary, there is no definitive data to guide recommendations for adjuvant therapy after resection with curative intent in patients with IHCC. Current data suggest that patients with R1 resections, large tumors, and node positive disease may be most likely to benefit from adjuvant chemotherapy. The recommended adjuvant regiment to consider is Xeloda although this is based on per-protocol analysis of a trial of all biliary cancers.

External Beam Radiation Therapy

Data on radiation therapy for IHCC are limited; most studies are small in size and include multiple biliary and hepatic cancers, and no randomized trials have been performed. Some small retrospective studies suggest that external beam radiation therapy (EBRT) may be associated with improved outcomes for patients with unresectable disease. An early study of 75 IHCC patients with unresectable disease or local lymph node metastases found a 36.4% tumor response rate to external beam radiation. EBRT was also associated with prolonged survival in patients with unresectable disease. Among patients with lymph node disease, EBRT to the lymph nodes was associated with longer survival after hepatectomy.¹⁴³ Several subsequent retrospective reviews have confirmed an association between EBRT and longer survival in patients with unresectable disease.^144,145 Additionally, a review of 90 patients with resected IHCC and concurrent regional lymph node metastases found adjuvant radiotherapy was associated with longer median survival than treatment without radiation(19.1 vs. 9.5 months, p=0.011).¹⁴⁶

A recent NCDB review comparing radiotherapy modalities in 141 patients unresectable IHCC found that higher-dose stereotactic body radiation therapy was associated with longer overall survival than chemotherapy in conjunction with lower dose external beam radiation (HR 0.37; 95%CI 0.20-0.68; p = 0.001) or radioembolization (HR 0.40; 95%CI 0.22-0.74; p = 0.003).¹⁴⁷

Intra-Arterial Therapy

The unique vascular supply of the liver, in which the healthy parenchyma is primarily supplied by the portal system and malignant lesions are primarily supplied by branches of the hepatic artery, allows for directed administration of chemoradiation agents to the tumor. Localized radiation can be applied to the tumor through radioembolization with Yttrium-90 (Y90) microspheres. Chemotherapy may be administered directly to the tumor via transarterial chemoembolization (TACE) or transarterial chemoembolization with drug-eluting beads (TACE-DEB), or it may be continuously infused via surgically placed hepatic arterial infusion pump (HAI). These techniques offer additional treatment options to patients with unresectable disease, and in some situations, may be an option for downstaging to resectability.

Radioembolization of IHCC allows for administration of much higher local doses of radiation than can be achieved with external beam radiation. Microspheres containing the radionucleotide Y90 are injected into the branches of the hepatic artery supplying the tumor, allowing for localized radiation. An early prospective study of the safety of Y90 radioembolization in 24 patients with unresectable IHCC found a median overall survival of 14.9 months and survival of greater than 2.5 years in some subgroups (Ibrahim 2008).¹⁴⁸ A larger retrospective study of 96 patients with unresectable IHCC found a median survival of 15.6 months in patients with peripheral disease, and 5 patients were converted to resectability.¹⁴⁹

TACE is the most commonly used intraarterial therapy in IHCC. Briefly, in TACE, contrast and chemotherapeutic agents are introduced into branches of the hepatic artery supplying the tumor. Embolic agents are then introduced proximally, attempting to ensure retention of the chemotherapeutic agents. Devascularization also occurs but is typically a minor contributor to response. To date, no randomized studies of TACE for IHCC have been performed, but several recent studies have examined the use of TACE in unresectable IHCC. Kiefer et al. followed 62 patients after chemoembolization with cisplatinum, doxorubicin, mitomycin-C, ethiodol, and polyvinyl alcohol particles, finding a median survival of 15 months, with significantly longer overall survival in those who also received systemic chemotherapy compared to those who received TACE alone (28 months v. 16 months p = 0.02).¹⁵² In a study of 155 patients with unresectable IHCC, Park et al. found that TACE with cisplatin was associated with longer survival than supportive care (12.2 vs. 3.3 months, p<0.001) .¹⁵¹

Drug-eluting beads are a recent development in chemoembolization, in which the injection of chemotherapeutic agents is eliminated, and patients undergo embolization of hepatic artery branches with chemotherapy-eluting microspheres. No randomized trials comparing TACE-DEB with conventional TACE in IHCC have been performed. One small prospective study did find TACE-DEB was associated with similar progression-free survival to systemic chemotherapy and conventional TACE (3.9 v. 6.2 v. 1.8 months) and associated with longer overall survival than conventional TACE (11.7 v. 5.7 months).¹⁵²

HAI pumps are surgically placed chemotherapy infusion devices with a growing role in the treatment of IHCC. HAI with chemotherapy enables targeting of the tumor with relative sparing of the healthy liver parenchyma. Additionally, because the liver clears certain chemotherapeutic agents via first-pass metabolism, systemic effects of chemotherapy are minimized.¹⁵³ The most common agent for HAI is floxuridine (FUDR), a fluorouracil precursor. Delivery of floxuridine via HAI has been used in adjuvant and palliative settings for the management of metastatic disease and has been shown to convert unresectable liver metastases to resectability.^154,155

While the majority of HAI studies in IHCC have been performed in unresectable patients, several studies have demonstrated that HAI chemotherapy in IHCC is safe, induces tumor response, and may downstage unresectable disease in some cases. An early study at Memorial Sloan Kettering Cancer Center of HAI FUDR in patients with unresectable IHCC found a partial response rate of 54% and overall survival of 29 months. At the conclusion of this trial, 1 patient showed adequate response to be eligible for resection.¹⁵⁶ In a subsequent trial at the same institution of 44 patients with IHCC comparing HAI FUDR alone to HAI FUDR plus bevacizumab, 10 (23%) patients survived ≥3 years and 5 (11%) survived ≥5 years. In this study, 3 patients (1 on FUDR alone and 2 on FUDR plus bevacizumab) were downstaged and able to undergo resection after treatment.¹⁵⁷ A recent phase II trial of 38 patients with unresectable IHCC receiving HAI floxuridine with systemic gemcitabine and oxaliplatin also showed promising results, with a 1-year survival of 89.5%. Fifty-eight percent of enrolled patients achieved a partial radiographic response, and 4 patients had sufficient response to undergo resection.¹⁵⁸ Another study of HAI gemcitabine and oxaliplatin in unresectable IHCC attained resectability in 17% of patients.¹⁵⁹

Although HAI therapy shows promise in small trials for unresectable disease, it is currently only performed at a small number of centers, and there is no established protocol therapy outside of clinical trials. Because the majority of recurrences after resection are intrahepatic⁶⁹, HAI therapy may also be beneficial as an adjuvant treatment, but this has not been explored.

Boehm et al. compared the effectiveness of intra-arterial therapies for IHCC in a recent systematic review and meta-analysis of 20 reports representing 657 patients (Table 1.). The study compared HAI, TACE, DEB-TACE, and Y90. This study found a higher median overall survival with HAI, followed by Y90 radioembolization, TACE, and DEB-TACE. There was also a higher response to therapy with HAI than the other modalities. HAI was also associated with the highest rate of grade III/IV toxicity, according to the NCI common toxicity criteria.¹⁶⁰

Table 1.

Effectiveness of intra-arterial therapies for intrahepatic cholangiocarcinoma

	Overall	Hepatic Arterial Infusion	Yttrium-90 Radioembolization	Transcatheter Arterial Chemoembolization	Drug-eluting Bead Transcatheter Arterial Chemoembolization
Response Rate (95% CI)	28.5% (18.0–39.1)	56.9% (41.0–72.8)	27.4% (17.4–37.5)	17.3% (6.8–27.8)	-
Median Overall Survival months (95% CI)	14.5 (12.48–16.43)	22.8 (9.8–35.8)	13.9 (9.5–18.3)	12.4 (10.9–13.9)	12.3 (11.0—13.5)
Rate of Grade III/IV Complications events/patient (95% CI)	-	0.35 (0.22–0.48)	-	0.26 (0.21–0.32)	0.32 (0.17–0.48)
Rate of Liver Toxicity events/patient (95% CI)	-	0.75 (0.65–0.86)	0.64 (0.55–0.72)	0.09 (0.06–0.12)	0.08 (0.0–0.17)

Adapted from Boehm et al. 2015

Prognosis and Long-Term Outcomes

Even after surgery with curative intent, IHCC has a poor prognosis. Overall survival is short, with series reporting five-year survival ranging from 18-40% after resection.^40,161 Recurrence is common after surgery, with a 46-66% recurrence rate, and the majority of recurrences involve the liver, either intrahepatic alone or intrahepatic in conjunction with an extrahepatic recurrence^125,126

Many evaluations of prognosis and long-term outcomes in IHCC are based on single-center experience and include a relatively small number of patients. The staging system has changed significantly since its original conception, and continued reports of surgical series have contributed to a consensus on a number of prognostic factors. A recent meta-analysis of 57 studies representing 4,756 IHCC patients found lymph node metastases, elevated preoperative CA 19-9, and multifocal disease were independent predictors of poor survival on multivariate analysis (HR 2.0 [95% CI 1.5-2.7], 1.6 [95% CI 1.1-2.2), and 1.57 [95% CI, 1.1-2.1], respectively) and that microscopic vascular invasion was associated with poor overall survival (HR 1.87, 95% CI 1.44-2.42).⁹⁶ Other studies have found that margin status and treatment at an academic center may be associated with outcomes ^{96, 162,163}

Staging

Western centers for hepatobiliary surgery commonly stage tumors with AJCC/Union for International Cancer Control (UICC) staging system (Table 2). In Japan, IHCC is typically staged with the classification described by Yamasaki in 2003 and adopted by the Liver Cancer Study Group of Japan (LCSGJ), which is based on the number and size of tumor and local invasion (Table 3). The AJCC system classifies IHCC as a primary liver tumor, and through the 6^th edition, a common system for staging was applied to all primary liver tumors, including hepatocellular carcinoma and IHCC. Nathan et al. demonstrated that tumor size at presentation was not predictive of outcomes and that the combined primary liver tumor classification did not appropriately stratify T2 and T3 disease in IHCC. Further, they demonstrated that multifocal disease, lymph node metastases, and vascular invasion were associated with poorer outcomes.¹⁶¹ These data resulted in reclassification IHCC into a distinct staging group, and the staging was revised to include vascular invasion and number of tumors in the 7^th edition.

Table 2.

AJCC Staging for intrahepatic cholangiocarcinoma

7th Edition		8th Edition
T1	Solitary tumor without vascular invasion	T1	Solitary tumor without vascular invasion
		T1a	Solitary tumor without vascular invasion, <5cm
		T1b	Solitary tumor without vascular invasion, >5cm
T2	Solitary tumor with intrahepatic vascular invasion, multiple tumors with or without vascular invasion	T2	Solitary tumor with intrahepatic vascular invasion or multiple tumors, with or without vascular invasion
T2a	Solitary tumor with vascular invasion
T2b	Multiple Tumors with or without vascular invasion
T3	Tumor perforating the visceral peritoneum or involving local hepatic structures by direct invasion	T3	Tumor perforating the visceral peritoneum
T4	Tumor with periductal invasion	T4	Tumor involving local extrahepatic structures by direct invasion

Stage Ia-T1a N0M0; Stage Ib-T1b N0M0; Stage II-T2 N0M0; Stage IIIa- T3 N0M0; Stage IIIb- T4 and/or N1,M0; Stage IV- any T, any N, M1. Based on American College of Surgeons. Edge SB, Byrd DR, Compton CC, Fritz AG, Greene FL, Trotti A, (editors). AJCC Cancer Staging Manual. 7th Ed. Springer New York, 2010.

Table 3.

Liver Cancer Study Group of Japan staging system.

T1	Meets all three criteria: -solitary tumor -tumor <2cm -No invasion of Portal Vein, hepatic vein, or serosa	N0	No lymph node Metastases
T2	Meets two of three criteria	N1	Metastasis to any lymph node
T3	Meets one of three criteria
T4	Meets none of the three criteria

Stage I- T1 N0M0; Stage II- T2N0M0; Stage III- T3N0M0; Stage IVa- T4N0M0, anyTN1M0; Stage IVb-any T, any N, M1

The recent AJCC 8^th edition update further modified T-staging. Tumor size >5 cm has been shown to be associated with poorer outcomes, and T1 was stratified into T1a and T1b based on tumor size. T2 was destratified to include patients with vascular invasion and multifocal disease, as these were found to be equivalent predictors of survival. The T3 classification was limited only to cases with peritoneal penetration, and direct invasion was made a criterion for T4 disease. Finally, periductal invasion was removed from the T4 classification. Validation studies of the 8^th edition have shown mixed results. A large SEER database review found the 7^th and 8^th editions to have equivalent prognostic performance, but two retrospective studies of resected patients found equivalent survival between T2 and T3 patients in the 8^th edition classification ³·¹⁶⁴·¹⁶⁵

IHCC Nomograms

In addition to postoperative staging systems, a number of prognostic nomograms incorporating other demographic and clinical information have been developed. The Fudan score is a nomogram based on preoperatively available clinical information, including elevated alkaline phosphatase, elevated CA 19-9, tumor >10 cm, obscure tumor boundary, and presence of multiple tumors. Based on the number of these factors present, patients were stratified into low, intermediate, high, and extremely high-risk groups, with five-year overall survival of 48.6%, 25.6%, 10.3%, and 0.0%, respectively. In a 74-patient validation set, the Fudan score demonstrated prognostic discrimination for overall survival among patients undergoing laparotomy¹⁶⁶

The Wang and Hyder nomograms are both based on a combination of preoperative and pathological information. The Wang nomogram is based on a study of 367 patients from a single Chinese center who underwent partial hepatectomy for IHCC. It incorporates serum biomarkers and tumor characteristics, including CEA, CA 19-9, tumor diameter and number, vascular invasion, lymph node metastasis, direct invasion, and local extrahepatic metastasis. In the validation cohort of 82 patients, the C-index was 0.75 (95%CI 0.68-0.83), which compared favorably to C-indices of the AJCC 7^th edition, AJCCC 6^th edition, and LCSGJ staging system, 0.65, 0.65 and 0.64 respectively (p<0.001).¹⁶⁷

The Hyder nomogram was developed from 514 patients who underwent resection for IHCC at 13 international hepatobiliary referral centers. The factors predictive of survival were age at diagnosis, tumor size, number of tumors, presence of cirrhosis, lymph node metastasis, and macrovascular invasion. The C-index of the nomogram was 0.69 (95%CI 0.62-0.76), which also compared favorably to the AJCC 7^th and AJCC 6^th edition staging systems, with a C-index of 0.59 and 0.54 respectively (Hyder 2014).¹⁶⁸

The three nomograms were all validated in a novel validation set of 188 patients. The Wang nomogram performed the best, with a C-index of 0.72 (95%CI 0.64–0.80), followed by the Hyder nomogram at 0.66 (95%CI 0.57–0.69) and the Fudan Score at 0.55 (95%CI 0.47–0.64). The AJCC 7^th edition had a C-index of 0.63 (95%CI 0.58–0.67), better only than the Fudan score (Doussot 2015).¹⁶⁹ These nomograms have not been compared to the AJCC 8^th edition staging system.

Margins

The importance of margin status for prognosis after surgery remains controversial. Margin status has been extensively studied, with studies reaching different conclusions about the association with recurrence and overall survival. A recent multi-institutional database review of 583 patients who underwent resection for IHCC found R0 resection was associated with improved overall and recurrence-free survival. Spolverato et al. found that 80.6% of patients had resections with negative margins, and a positive margin was an independent predictor of shorter recurrence-free survival and shorter overall survival. Furthermore, when margin width was stratified into 1-4 mm, 5-9 mm, and ≥1 cm, a wider margin was associated with improved survival outcomes.¹⁷⁰

Two large systematic reviews and meta-analyses, however, offer conflicting conclusions on the importance of negative margins for prognosis. Li et al. reviewed 21 studies representing 3,201 patients and found patients with negative margins to have superior overall survival (HR 2.03, 95%CI 1.03-4.01; p=0.041) and progression-free survival (HR 1.86, 95%CI 1.54-2.25; p<0.001) compared to those with microscopic positive margins.¹⁷¹ In contrast, Mavros et al. reviewed 57 studies, of which 11 studies with 583 patients addressed margin status. In the systematic analysis, positive margins were associated with worse overall survival (2 of 11 studies) and worse recurrence-free survival (2 of 11 studies) but were not associated with survival in multivariate analysis.⁹⁶

Academic Centers

In recent years, there has been a trend towards centralization of complex oncologic surgeries to designated cancer centers. An NCDB review of 40,000 patients who underwent resection for cholangiocarcinoma found that the proportion undergoing surgery at academic centers nearly doubled over the past decade. In the same group, surgery at a high-volume center was an independent positive predictor of survival (HR 0.92, 95%CI 0.88–0.97, p<0.001).¹⁶² Other studies also suggest that outcomes for surgical management of IHCC at academic centers may be associated with improved outcomes. An NCDB review of 2,256 patients who underwent hepatectomy for IHCC from 2004-2014 found that facility hepatectomy volume was not associated with differences in long- or short-term outcomes, but that there were significant differences between academic and non-academic centers. Patients treated at academic programs demonstrated lower rate of positive margins, lower 90-day mortality, and improved overall survival (HR 0.78, 95%CI 0.63-0.96).¹⁷² Another NCDB study of 27,120 patients with a histologic diagnosis of IHCC, both resected and unresected, also found that treatment at a community program was a poor prognostic factor for survival (HR 1.24, 95%CI 1.04–1.49; p=0.023). Furthermore, among patients who underwent resection, patients at academic centers were more likely to receive R0 resections than those treated at community programs (72.4% vs. 67.7%, p=0.006).¹⁶³

Conclusion

IHCC incidence is increasing worldwide. Liver flukes and hepatolithiasis are common risk factors in Asia, whereas PSC and obesity/metabolic syndrome are additional risk factors in Western countries. Most patients, however, have no identifiable risk factors. Late presentation is common, primarily due to absent or non-specific symptoms and a lack of diagnostic biomarkers. IHCC treatment utilizes a combination of chemotherapy and surgery. Surgery is usually indicated for patients with single resectable tumors. In patients with multifocal disease or spread to regional lymph nodes, chemotherapy may be useful to help select which patients may be candidates for resection. Studies are needed to improve selection criteria to stratify patients into risk groups.

II. Perihilar Cholangiocarcinoma

Perihilar cholangiocarcinoma (CCA) was originally described in 1965 by Klatskin in a report of 13 cases of adenocarcinoma at the bifurcation of the hepatic duct within the porta hepatis.¹⁷³ It is now recognized that perihilar CCA accounts for 50% of all cholangiocarcinomas (distal, intrahepatic, and perihilar).⁴⁰ Perihilar tumors arise from the extrahepatic bile ducts between the main lobar ducts and the common hepatic duct above the insertion with the cystic duct. Intrahepatic CCA arise proximal to this location, and tumors arising from the common bile duct are termed distal CCA.¹⁰¹ Primary sclerosing cholangitis remains the greatest risk factor for perihilar CCA (lifetime risk 6-20%), whereas numerous other risk factors, including choledochal cysts, liver flukes, cirrhosis, and non-alcoholic steatohepatitis, are associated with increased risk for both intra- and extrahepatic CCA (Table 4). However, most cases of CCA are sporadic, without identified risk factors.

Table 4:

Risk factors for development of perihilar CCA

Category	Risk Factor
Autoimmune	Primary sclerosing cholangitis
Anatomic	Choledochal cysts Chronic hepatolithiasis
Infectious	Liver flukes (Clonorchis sinensis, Opisthorcis viverrini) Biliary ascariasis Hepatic schistosomiasis
Exposures	Thorium dioxide Nitrosamines
Genetic diseases	Lynch syndrome Biliary papillomatosis

The majority of CCA, regardless of where they arise, present with symptoms at a late stage of disease when surgical resection is difficult, or metastases are present. Perihilar CCA is further complicated by the near universal presence of obstructive jaundice which is difficult to treat due to the common involvement of bilateral segmental biliary radicals. Further the proximity of these tumors to major vascular structures, often precludes surgery. As a result, most patients with perihilar CCA die within a year of diagnosis. Even when resection is possible, the 5-year overall survival for perihilar CCA is only 20-50%.^{174, 175, 40}

Perihilar CCA often poses clinical challenges, and thus careful management and early multidisciplinary evaluation are essential in the surgical and medical treatment of this disease. Prompt referral to a high-volume center once a diagnosis of perihilar CCA is suspected should be strongly considered given the complexity of diagnosis, biliary drainage, and overall treatment of this disease.

Presentation

The presenting symptom for almost all patients with perihilar CCA is obstructive jaundice, although vague abdominal pain may precede jaundice if there is unilateral biliary obstruction.¹⁷⁶ Cholangitis is uncommon in the absence of biliary instrumentation. For patients with primary sclerosing cholangitis, new-onset jaundice with a dominant stricture may be a subtle presentation of a new cholangiocarcinoma.

Physical examination is typically nonspecific. Since the level of biliary obstruction is nearly always above the cystic duct, the gallbladder is rarely palpable. Laboratory values are consistent with biliary obstruction, and tumor markers, such as CEA and CA 19-9, may be elevated; however, no blood tests are specific for perihilar CCA.¹⁷⁷ Ca 19-9 may be elevated in as many as 85% of patients with CCA, but the sensitivity and specificity of Ca 19-9 is reduced in a variety of clinical scenarios including primary sclerosing cholangitis, obstructive jaundice and cholangitis. In addition, approximately 10% of patients lack the Lewis antigen and are non-secretors of Ca 91-9.

Histologic Features

Ninety percent of CCA are adenocarcinomas. Mixed hepatocellular-cholangiocarcinoma occur in up to 5% of patients and usually occur within the liver (intrahepatic CCA) and is more common in patients with chronic liver disease. CCA usually have immunohistochemical staining patterns demonstrating CK7 positivity and CK20 negativity. Additional immunohistochemical stains can be used to help distinguish CCA from other tumors in the appropriate clinical situation. There are three macroscopic subtypes of perihilar CCA: nodular (exophytic or mass forming), sclerosing (infiltrative with periductal fibrosis), and papillary (intraductal polypoid).¹⁷⁸ Episodic jaundice may occur with perihilar CCA of the papillary histologic subtype, due to a ball-valve effect from tumor fragments breaking off and passing distally; tumors of this subtype are associated with increased resectability and improved prognosis compared to the more common sclerosing and nodular subtypes.^{179, 180}

Diagnosis and Imaging Workup

The diagnosis of perihilar CCA is often complex and requires the use multiple imaging modalities to characterize the anatomic location of the biliary stricture or mass, assess underlying vasculature, to stage tumors and assess resectability.

Distinguishing benign from malignant biliary lesions can be very difficult, if not impossible. Post inflammatory, post-surgical or autoimmune (i.e. IgG4 cholangiopathy) biliary strictures can have a similar appearance to malignant strictures. Other tumors such as metastases or gallbladder cancer may also mimic perihilar CCA in presentation. In the setting of primary sclerosing cholangitis, the presence of multiple strictures compounds the diagnosis of CCA, even when malignancy is suspected. Unfortunately, tissue diagnosis is often challenging, as sampling difficulty and false negative and false positive results occur with regularity. Applying fluorescence in situ hybridization (FISH), to detect cell aneuploidy can help establish a diagnosis of CCA in up 60% of cases when standard brush cytology is negative.¹⁸¹ Given the difficulty of establishing tissue diagnosis, histologic confirmation of tumor is not essential to surgical treatment decisions when other clinical, laboratory and imaging studies support a diagnosis of perihilar CCA.

High-quality liver imaging is essential for the diagnosis and characterization of perihilar lesions and should precede any drainage or diagnostic procedures when malignancy is suspected. Imaging workup for patients with painless jaundice often begins with ultrasound, but multi-phase, thin-cut, liver contrast computed tomography (CT) is necessary for anatomic assessment of a biliary stricture or tumor, vascular anatomy, and level of biliary obstruction (Figure 2).¹⁸² Magnetic resonance imaging (MRI) with magnetic resonance cholangiopancreatography (MRCP) can be a helpful adjunct to assess location, radial extent and the relationship of the lesion to the adjacent biliary and vascular structures.¹⁸³ Routine positron emission tomography (PET) is not recommended for the evaluation of perihilar CCA due to low diagnostic yield,. However, selective use PET/CT can be helpful for the detection of nodal and distant metastatic disease; especially when imaging suggests this as a possibility.¹⁸⁴

Figure 2.

Newly diagnosed perihilar cholangiocarcinoma. These CT images demonstrate a perihilar CCA affecting the bifurcation and occluding the left biliary tree. Coronal (A) and axial (B) CT images at diagnosis demonstrate narrowing of left portal vein (red arrowhead) and right hepatic artery (white arrowhead). The mass is highlighted with white arrow.

Cross sectional imaging can demonstrate a mass in the perihilar region in the nodular subtype of CCA, or polypoid, expansile lesions in papillary subtypes y. However, for sclerosing tumors, lesions can be very subtle. At times, only biliary dilatation proximal to the malignancy can be seen. In these situations, when the diagnosis may not be clear, additional tests such as direct cholangiography should be considered.

Ultimately, any obstructive lesion in the perihilar lesion should be considered malignant until proven otherwise. We would like to stress that tissue diagnosis is not necessary before proceeding with operative intervention. Occasionally, stone disease can obstruct the biliary tree in this location but good quality CT and/or MRCP can typically see this. In those cases, direct therapeutic cholangiography should confirm the disease. Lastly, in cases where there has not been a histologic diagnosis of malignancy one should consider serum IgG4 levels and if elevated a short course of steroids to see if the obstruction resolves. Autoimmune cholangiopathy is a rare but distinct and potentially diagnosable condition that can mimic perihilar CCA.¹⁸⁵ Often the diagnosis cannot be definitively established, and the surgeon must consider that approximately 5 to 10% of cases may turn out to be benign.

The biliary tree may be approached endoscopically, from ‘below’ or percutaneously from ‘above’ for purposes of biliary decompression, anatomic evaluation and tissue diagnosis. For patients with localized disease who may be surgical candidates, relief of jaundice, if indicated, should focus on decompression of the future liver remnant. Biliary decompression of the hemi-liver that will be resected should be avoided, whenever possible. Biliary drainage of the ipsilateral liver introduces bacteria into the biliary system and increases the potential for cholangitis and venous thrombosis with no therapeutic benefit. Accessing the biliary tree for tissue sampling with either endoscopic retrograde cholangiopancreatography (ERCP) and percutaneous transhepatic cholangiography (PTC) have similar sensitivity and specificity (70%) at making histologic diagnosis.¹⁸⁶ Given comparable diagnostic yields, the most important consideration when choosing between these modalities is minimizing risk of inducing cholangitis. Our preferred approach is to use their percutaneous route (PTC) for diagnosis and biliary drainage, as ERCP is associated with up to a 60% risk of cholangitis after ineffective attempts to access the level of the stricture or adequately relieve the biliary obstruction.¹⁸⁷ Institutional preferences and capabilities may differ, however, and should dictate the route of intervention.

In patients who have unresectable perihilar CCA but are potential candidates for liver transplantation, percutaneous biopsy should be avoided as in current protocols this may obviate the possibility of transplantation. Prompt referral to a specialty center with a perihilar CCA transplantation protocol should be considered in these cases.¹⁸⁸

Resectability and Preoperative Staging

In patients with a probable or confirmed tissue diagnosis of perihilar CCA, initial assessment is focused on determining resectability of the tumor. The principles of determining resectability include the absence of extra-hepatic metastatic disease and the ability to perform an extirpative procedure with R0 intent that preserves adequate liver function while providing adequate biliary drainage to the liver remnant. Preoperative staging for perihilar CCA has traditionally been performed using the Bismuth-Corlette classification system (Figure 4).¹⁸⁹ While this classification system provides a framework to define the anatomic location of the tumor along the biliary tree, assessment of resectability should also take into account the relationship of the tumor to the underlying vasculature (e.g., ability to preserve future liver remnant function). Critical tumor-related factors include the extent of tumor involvement within the biliary tree, degree of vascular involvement, atrophy of involved hepatic lobes, and presence/absence of extrahepatic disease. Atrophy of the involved liver lobe usually indicates longstanding ipsilateral biliary obstruction or occlusion of the ipsilateral portal vein in which results in rapid atrophy.

Figure 4.

Blumgart staging criteria for hilar cholangiocarcinoma. T1 - tumor involving biliary confluence with unilateral extension to second-order biliary radicles. T2 - tumor involving biliary confluence with unilateral extension to second-order biliary radicles and ipsilateral portal vein involvement or ipsilateral hepatic atrophy. T3 - tumor involving biliary confluence with bilateral extension to second-order biliary radicles; or unilateral extension to second-order biliary radicles with contralateral portal vein involvement; or unilateral extension to second-order biliary radicles with contralateral hepatic lobar atrophy; or main or bilateral portal venous involvement.

These factors are incorporated into the Blumgart classification and further refined in the Memorial Sloan Kettering Cancer Center (MSKCC) T-stage criteria (Table 5, Figure 4). The MSKCC system is designed as a preoperative staging system for predicting resectability and outcome in perihilar CCA.^175,94,190 In an analysis of 380 patients at MSKCC, approximately 60% of patients are found to be unresectable after staging. Only 1.3% of patients classified as T3 by the MSKCC system were found to be resectable, compared to 45.4% of T1 and 40.6% of T2 patients.¹⁹⁰ We routinely use this staging system to guide therapy. Patients who are T3 are rarely resectable and are more likely to harbor metastases. As a result, additional testing such as PET imaging or staging laparoscopy is considered in these patients. Also, conversion strategies including neoadjuvant chemotherapy may be considered for T3 patients on a case by case basis.¹⁹¹

Table 5:

MSKCC Preoperative T-Stage Criteria

Stage	Criteria
T1	Tumor involving biliary confluence ± unilateral extension to second-order biliary radicals
T2	T1 and ipsilateral portal vein involvement ± ipsilateral hepatic lobar atrophy
T3	Main or bilateral portal venous involvement, or Tumor involving biliary confluence with  - Bilateral extension to second-order biliary radicles, or  - Unilateral extension to second-order biliary radicles with contralateral portal vein involvement, or  - Unilateral extension to second-order biliary radicles with contralateral hepatic lobar atrophy

The recently updated 8^th edition American Joint Committee on Cancer (AJCC) staging system includes vascular and regional lymph node involvement in the staging schema, but its primary use remains in the postoperative setting, as it is based on pathologic factors (Table 6) as a prognostic tool.

Table 6:

AJCC Staging System for Perihilar Bile Duct Tumors (8^th Edition)

Primary Tumor (T)
TX	Primary tumor cannot be assessed
T0	No evidence of primary tumor
Tis	Carcinoma in situ/high-grade dysplasia
T1	Tumor confined to bile duct, with extension up to muscle layer or fibrous tissue
T2a	Tumor invades surrounding adipose tissue
T2b	Tumor invades adjacent hepatic parenchyma
T3	Tumor invades unilateral branches of portal vein or hepatic artery
T4	Tumor invades main portal vein or its branches bilaterally, or common hepatic artery; or second-order biliary radicals bilaterally; or unilateral second order biliary radicles with contralateral portal vein or hepatic artery involvement
Regional Lymph Nodes (N)
NX	Regional lymph nodes cannot be assessed
N0	No regional lymph node metastasis
N1	1-3 positive lymph nodes typically involving the hilar, cystic duct, common bile duct, hepatic artery, posterior pancreaticoduodenal, and portal vein lymph nodes
N2	4+ positive lymph nodes
Distant Metastasis (M)
M0	No distant metastasis
M1	Distant metastasis

Based on American College of Surgeons. Amin, M.B., Edge, S.B., Greene, F.L., et al. (Eds.) AJCC Cancer Staging Manual. 8th Ed. Springer New York, 2017, and American College of Surgeons. Edge SB, Byrd DR, Compton CC, Fritz AG, Greene FL, Trotti A, (editors). AJCC Cancer Staging Manual. 7th Ed. Springer New York, 2010.

Preoperative Preparation

Once a patient has been deemed an appropriate operative candidate with potentially resectable disease, preparation for surgery requires careful multidisciplinary care and coordination (Figure 5). Biliary drainage is often required given the increased risk of postoperative complications associated with jaundice; particularly in the face of a small future liver remnant. Once a patient is deemed to require preoperative biliary drainage, a preoperative total bilirubin <3 mg/dL is recommended prior to resection, as elevated bilirubin levels are associated with increased postoperative morbidity and mortality.^176,192 A fundamental concept of preoperative biliary drainage for this disease is that adequate biliary drainage is achieved in the future liver remnant (not necessarily the hemiliver to be resected).^193,194 As biliary diversion has been shown to impair intestinal barrier function and lead to fluid and electrolyte abnormalities, efforts should be made to internalize any percutaneously placed biliary drains. Both percutaneous transhepatic cholangiography (PTC) and endoscopic retrograde cholangiopancreatography) ERCP are accepted approaches for biliary drainage, although the optimal approach is not universally agreed upon. For resectable perihilar CCA, our preferred approach for biliary drainage is via PTC. This approach more accurately targets specific biliary branches which may not be accessible by ERCP.¹⁹⁵ An additional benefit of a transhepatic percutaneous approach is the ability to place internal-external drains distal to the biliary obstruction but proximal to the ampulla. By avoiding direct communication with gastrointestinal flora of the duodenum, the risk of ascending cholangitis is likely reduced.¹⁹⁶ Further procedural complications associated with ERCP such as pancreatitis and GI preformation are avoided. However, PTC drainage may increase the possibility of intraabdominal tumor seeding. Biliary drainage has been associated with recurrence along PTC tract in the skin and pleura, and also with carcinomatosis from biliary spillage intraabdominally.^197–199 This can be avoided with ERCP placed stents. ERCP biliary drainage also avoid external drains and minimize fluid losses that are associated with PTC drains that require extenternal drainage. The INTERCPT trial (NCT03172832) – a multi-center, randomized trial of PTC vs. ERCP – was initiated to compare the efficacy of these interventions for biliary drainage in patients with suspected perihilar CCA; Unfortunately the trial has been suspended due to low accrual.²⁰⁰ Thus at this time, either approach is acceptable and the availability of local expertise will guide the best approach in each institution.

Figure 5.

Algorithm illustrating our approach to treatment of patients with perihilar cholangiocarcinoma. FLR, future liver remnant; PVE, portal vein embolization; CXRT, chemoradiation.

An additional consideration is the size of the future liver remnant (FLR). Inadequate future liver remnant size is associated with post hepatectomy liver failure and mortality.²⁰¹ Formal volumetric measurements are performed when the anticipated future liver remnant is small. Hypertrophy of the future liver remnant can be induced with pre-operative portal vein embolization (PVE) or with Associating liver partition and portal vein ligation (ALLPS). Given the high peri-procedural morbidity and mortality of ALLPS for perihilar CCA, we favor PVE to induce FLR hypertrophy (Figure 6). Portal vein embolization directs portal blood to the FLR, resulting in hypertrophy. It is considered for patients with an expected FLR <30% in order to minimize postoperative liver insufficiency.²⁰² PVE is relatively safe and associated with a complication rate of 2.5% and a relative increase in FLR of up to 46% within 8 weeks.²⁰³ The size of the FLR should also be considered when deciding whether to proceed with preoperative biliary drainage. We assessed the role of biliary obstruction in relation to FLR and reported that biliary drainage in those with a functional liver remnant (FLR) <30% decreased hepatic insufficiency and mortality, but biliary drainage in patients with an FLR ≥30% cases did not improve perioperative outcomes.⁹³ Thus, our practice is not to perform biliary drainage in cases where FLR ≥30% and the operation can be performed expeditiously. It should be noted that the preoperative management of perihilar CCA patients is very complex and although we have reviewed some basic guidelines, each case must be individualized in the context of a multi-disciplinary review.

Figure 6.

Preoperative calculation of functional liver remnant after extended R hepatectomy for a perihilar cholangiocarcinoma (indicated with green arrows).

Operative Approach/Strategies

Diagnostic laparoscopy should be considered prior to resection of perihilar CCA, as up to 29% of patients initially thought to be resectable will be found to have metastatic or unresectable disease.¹⁷⁵ But the utility of diagnostic laparoscopy is limited. In a study by Ruys et al, the diagnostic yield of laparoscopy was low as only 14% of patients were able to avoid laparotomy due to findings at laparoscopy. Most patients (75%) required laparotomy to confirm unresectability, mainly due to the limited ability of laparoscopy to evaluate locally advanced disease or nodal metastases.²⁰⁴ Nevertheless, laparoscopy adds minimal morbidity and may avoid unnecessary laparotomy, and may be useful in selected patients at higher risk of metastatic disease (i.e. T3 tumors).²⁰⁵

Complete resection and liver transplantation are the only treatments associated with long-term survival in patients with perihilar CCA. The optimal therapy is margin-negative resection, and attempted resection is indicated for patients with potentially resectable tumors. Liver transplantation may be an appropriate option for select patients with small, unresectable lesions or those patients who have underlying liver disease, such as primary sclerosing cholangitis.¹⁷⁴ The operative approach to liver transplantation for perihilar CCA is beyond the scope of this review.

Resection of perihilar CCA is technically demanding given the central location of these tumors, which are in close proximity to the left and right hepatic inflow (hepatic artery and portal veins) and central biliary system. As a result, even small tumors usually require extended hepatectomy to achieve R0 margin clearance. Further, a small number of tumors which extend caudally to the lower third of the bile duct may require pancreaticoduodenectomy for margin clearance. As a result, only approximately 40% of patients are candidates for resection at diagnosis, and most series demonstrate that only 40-50% of those explored with curative intent are found to be resectable.^190,204 Some series have demonstrated resectability rates as high as 70%.²¹⁸

An R0 resection of perihilar CCA involves resection of the biliary confluence, and the majority of patients require major hepatectomy. The decision between right or left hepatectomy is based on tumor location within the biliary tree, vascular involvement, and lobar atrophy. Since tumors involving the biliary confluence, by definition, involve the central portion of the liver they all require a resection that involves this portion of the liver. Additionally, since vascular involvement is common this nearly always requires a concomitant right or left hepatectomy (which must be extended to involve the central biliary tree and central liver). We also advocate for resection of the caudate lobe as nearly all tumors about the caudate and involve the biliary tributaries arising from the caudate lobe.²⁰⁶ Even T1 tumors should undergo hepatic resection, as biliary resection alone results in increased recurrence and decreased survival.²⁰⁷ The extent of the hepatic resection must be tailored to each case. A left hepatectomy encompasses segment IV and is adequate for left sided tumors. A right hepatectomy does not include segment IV and in the great majority of cases must be extended to at least encompass segment IV. Left trisectionectomy is sometimes required for tumors that only spare the posterior sectoral ducts/vessels. Some groups advocate parenchymal-preserving options, such as central hepatectomy (resection of segments 1, 4b, and 5), and in select patients, this strategy may lead to outcomes similar to those undergoing major hepatectomy.^208,209 Portal vein resection may be necessary to achieve clear margins but is at the cost of increased perioperative morbidity.²¹⁰ Lastly, some cases require combined portal vein and hepatic arterial reconstruction.²¹⁸ These are extremely challenging cases with high rates of morbidity. Any surgeon tackling hilar CCA cases but be prepared for vascular reconstruction.

For laparotomy, we enter the abdomen through a right subcostal incision with cephalad extension at the midline. After thorough exploration of the abdomen, intraoperative ultrasound is performed to assess for intrahepatic metastases and to confirm the anatomy of the area. Next, we look for any evidence of distant lymphatic spread outside of the porta hepatis. If nodes outside the regional lymph node basin are found to be involved, then resection is contraindicated.

Once resectability is felt to be likely, early transection of the distal bile duct margin at the suprapancreatic common bile duct allows the extrahepatic bile duct to be reflected cephalad. In some cases, dissection of the arterial and portal venous structures can be carried out prior to division of the distal bile duct to ensure resectability. The distal common bile duct margin is sent for frozen-section analysis, and the portal vein and hepatic arteries are dissected free to the hilum. Inflow and outflow control are performed as indicated by the extent of hepatic resection. After resection, proximal margins are confirmed by frozen-section analysis when further resection is feasible, and hepaticojejunostomy with a 60 to 70 cm Roux-en-Y limb is used to restore biliary drainage. Intraoperative bile duct margin assessment can be problematic, and we have found 9% of intraoperative margin negative specimens are positive on final pathology in a retrospective series.²¹¹ While negative margins on the specimen and bile duct led to a disease-specific survival of 54 months, negative margins achieved after re-resection for a positive margin on frozen section led to survival similar to those with primarily negative margins.²¹²

Operative Outcomes

While R0 resection confers 5-year survival rates as high as 40 to 50%, this comes at the cost of high postoperative morbidity and mortality. The most common complications include infection (intra-abdominal or wound-related), bile leak, and liver failure. In one series of 105 patients, postoperative mortality was 14.3%, although patients resected to clear margins had a median OS of 58 months compared to 28 and 19 months for patients with R1 and R2 resection, respectively.¹⁷⁶ A more recent series from a Japanese center demonstrated a 0% 90-day mortality rate and grade III or higher postoperative morbidity in 37% of resected patients, underscoring the importance of an experienced center and careful patient selection.²¹³

Five-year overall survival for all resected patients is 32-38%.^174,94,209 Although R0 margin status is associated with better survival as mentioned above, recurrence rates remain high (50-70%).^209,190,214 Additional factors associated with improved survival include the use of major hepatectomy, well-differentiated tumor histology, and negative nodal status. Portal vein resection, hepatic artery resection, or blood transfusion are factors associated with inferior long-term outcomes.^209,190

Adjuvant Therapy and Other Strategies

Three recent randomized controlled trials have explored the role of adjuvant therapy for biliary cancers, and while results have been largely disappointing, there is now evidence that capecitabine may confer a survival benefit in CCA. The Bile Duct Cancer Adjuvant trial (BCAT) compared gemcitabine vs. observation after resection of perihilar and distal CCA; overall survival (OS) (62.3 v.s 62.8 months) and recurrence-free survival (RFS) (36.0 vs. 39.9 months) were not significantly different between the groups.²¹⁵ Criticisms of this trial centered around poor trial accrual and a patient population that was low risk (90% R0, 66% N0). The PRODIGE 12-ACCORD 18-UNICANCER GI trial compared adjuvant gemcitabine + oxaliplatin vs. observation. This study did not find a difference in OS or RFS, although there was a trend toward improved OS driven largely by the gallbladder cancer patient population. ¹⁴¹ Most recently, the BILCAP trial examined whether adjuvant capecitabine improved OS in resected cholangiocarcinoma and gallbladder cancer.¹⁴⁰ While the intention-to-treat analysis revealed no significant OS difference between patients in the capecitabine cohort vs. those in the observation group (51.1 vs. 36.4 months, p=0.097), a per-protocol analysis did demonstrate significantly prolonged OS (53.0 vs. 36.0 months, p=0.028). Moreover, RFS was significantly improved in the intention-to-treat analysis (24.4 vs. 17.5 months, p=0.033). Based on these results, most patients are now being offered capecitabine for adjuvant treatment. The ongoing, multi-center, randomized ACTICCA-1 trial (NCT02170090) is comparing the efficacy of gemcitabine + cisplatin, a regimen found to be effective in the metastatic setting in the ABC-02 trial, vs. standard of care (capecitabine) after curative-intent resection of biliary tract cancer; the results of this trial will help guide future management.^216,142 It should be stressed that these trials focus on large heterogeneous groups with all types of biliary cancer and do not necessarily apply to specific sub-types of cholangiocarcinoma. It is our hope that future trials will be more disease-specific. There are no randomized trials on adjuvant radiation to support its use, but retrospective series suggest improved local control. Photodynamic therapy, chemoradiation, and irreversible electroporation have also been investigated in the setting of unresectable disease, but a detailed discussion of these techniques is outside of the scope of this review.

Summary

The management of perihilar CCA requires careful workup and requires multidisciplinary expertise. Early referral, prior to biliary drainage, to an experienced center is required for best results.²¹⁷ In patients with resectable disease, the selective use of biliary drainage should be reserved for the FLR, as drainage of the liver segment that will be resected may cause complications that delay optimal treatment. The fundamental goal of perihilar CCA management is to achieve R0 resection for patients with resectable disease or transplantation for those with localized but unresectable disease. While adjuvant chemotherapy has not been shown to significantly prolong OS or RFS, the BILCAP trial does provide rationale for adjuvant capecitabine; ongoing trials should help define the optimal adjuvant treatment approach for these patients.

III. Mid and Distal Bile Duct Cancers

Cholangiocarcinomas are primary bile duct cancers that arise from the epithelial lining of the biliary ductal system. This chapter will focus on cholangiocarcinomas that arise within the mid and distal bile duct, which is defined as the segment of common bile duct from the insertion of the cystic duct (proximally) to the ampulla of Vater (distally). Because the American Cancer Society consolidates extrahepatic biliary malignancies to include both primary gallbladder cancers and extrahepatic bile duct cancers, the true incidence of distal cholangiocarcinomas is challenging to define. In the United States, extrahepatic biliary malignancies accounted for 12,000 new diagnoses and 4,000 deaths in 2019; however, the majority of these cases likely reflect primary gallbladder cancers, with the remainder a mix of hilar and distal cholangiocarcinomas.²¹⁹ Among Western populations, known risk factors include primary sclerosing cholangitis (PSC; lifetime risk 10-15%)^220–222 and congenital choledochal cysts (lifetime risk 5-30%),²²³ as well as several genetic disorders (lynch syndrome, BAP1 tumor predisposition syndrome, cystic fibrosis, and multiple biliary papillomatosis).^224–227 While the vast majority (>90%) of cholangiocarcinomas are adenocarcinomas, the primary location of these histologically similar tumors accounts for their diverse presentations, treatment options, and outcomes.¹⁷⁹

Presentation

Unlike intrahepatic cholangiocarcinomas, which often present when very large in size and with nonspecific symptoms, distal bile duct cancers almost uniformly present with obstructive jaundice and are identified when they are relatively small in size. In a series of 294 patients who underwent surgical exploration for cholangiocarcinoma at The Johns Hopkins Hospital, the most common presenting symptom for distal bile duct cancers was jaundice (87%), followed by weight loss (30%) and abdominal pain (27%).²²⁸ Similarly, in a series of 177 patients undergoing operative intervention for extrahepatic cholangiocarcinoma at Mayo Clinic—the majority of which were mid or distal bile duct tumors—presenting symptoms tended to be related to biliary obstruction, including pruritus (66%) and pain (49%); other common symptoms included weight loss (56%), anorexia (50%), and fatigue (40%).²²⁹ Cholangitis is a relatively infrequent symptom at presentation, despite the fact that obstruction of the bile duct is almost universally present.

Aside from jaundice, the remainder of the physical examination is likely to be nonspecific. While hepatomegaly has been reported in 40% of patients, a palpable mass and ascites are both rare physical findings, as these lesions are typically small, and peritoneal dissemination and/or portal vein occlusion is uncommon.²²⁹ Laboratory findings at presentations reflect the degree of biliary obstruction and any concomitant hepatocellular dysfunction. In a series of 257 patients undergoing curative intent resection for extrahepatic cholangiocarcinoma at Memorial Sloan Kettering Cancer Center (113 patients with distal bile duct adenocarcinoma), 60% had a preoperative total bilirubin ≥2.5 mg/dL.²³⁰ In a series from the Mayo Clinic, the mean total bilirubin and alkaline phosphatase at presentation were reported at 5.9 ± 1.0 mg/dL and 324 ± 29 (IU/L), respectively.²²⁹ Aspartate aminotransferase and alanine aminotransferase were also found to be moderately elevated (mean 89 ± 16 and 138 ± 27 IU/L, respectively).²²⁹ Other common laboratory values, including hematocrit, leukocyte count, albumin, and creatinine are unlikely to be significantly altered.²²⁹

Finally, with the increasing utilization of ultrasonography and cross-sectioning imaging, the initial presentation for a subset of patients may follow radiographic abnormalities identified on studies obtained for related—or potentially unrelated—indications. If the study indication is jaundice, then a right upper quadrant ultrasound may be the initial radiographic study, which is sensitive for identifying biliary dilation and a point of caliber change, but less sensitive for identifying a mass or distinguishing malignant ductal abnormalities from other etiologies.²³¹ In general, an abrupt caliber change is suggestive of malignant etiology, while a more gradual tapering may suggest a benign stricture.²³² Computed tomography (CT) and magnetic resonance imaging (MRI)/magnetic resonance cholangiopancreatography (MRCP) may similarly demonstrate intrahepatic and extrahepatic biliary dilation, with a potentially greater ability to differentiate malignant from benign biliary stricturing.^233,234 In addition, these cross-sectional modalities have the potential to identify distant metastatic spread and changes within the associated lymph node basins, although the sensitivity for distinguishing reactive lymphadenopathy from malignant involvement—as extrapolated from a study of patients with intrahepatic cholangiocarcinoma—may be unreliable.²³⁵ Any of these findings should prompt referral to a multi-disciplinary hepatopancreatobiliary unit. The nuances of these and other imaging modalities in the proper diagnosis and staging workup are discussed in greater detail below.

Diagnosis/Imaging Workup

In the absence of a definitive tissue diagnosis, the differential diagnosis for a patient presenting with obstructive jaundice or radiographic biliary dilation is broad, and includes choledocholithiasis, primary sclerosing cholangitis, benign biliary stricture, autoimmune sclerosing cholangitis or pancreatitis, gallbladder cancer, pancreatic head cancer, ampullary masses, or other metastatic cancers causing extrinsic biliary compression. Differentiating these potential etiologies can be achieved via a combination of laboratory, radiographic, and, when necessary, endoscopic studies. However, discrimination between distal bile duct cancer and cancer of the head of the pancreas is challenging, and often is only defined on operative pathology.

All patients presenting with biliary obstruction should undergo a thorough laboratory workup that includes liver function tests. Tumor markers such as carcinoembryonic antigen (CEA) and carbohydrate antigen 19-9 (CA 19-9) should be considered as well. CEA represents a group of glycoproteins typically produced by gastrointestinal tissues in fetal development. Elevation in CEA is therefore abnormal, but nonspecific for cholangiocarcinoma, as elevations may occur in a number of benign inflammatory conditions, as well as other gastrointestinal malignancies such as colorectal cancer, gastric cancer, and pancreatic cancer.^236,237 Similarly, CA 19-9, which is a sialyl-Lewis antigen, is a cell surface carbohydrate that can be elevated in cholangiocarcinoma as well as multiple benign and malignant conditions, including pancreatic cancer, benign biliary obstruction, colorectal cancer, and esophageal cancer.²³⁸ In one comparative study of 333 high-risk patients with PSC at the University of Pittsburgh, the median serum CEA and CA 19-9 were 8.2 ng/mL and 377.1 U/mL, respectively, in those found to have cholangiocarcinoma, compared to 2.9 ng/mL and 39.1 U/mL among patients without cholangiocarcinoma. When cutoff values of 5.2 ng/mL for CEA and 180 U/mL for CA 19-9 were applied, this resulted in a sensitivity and specificity of 62.5% and 78.4% for CEA and 75.0% and 97.3% for CA 19-9. Positive predictive values were 38.5% and 85.7%, respectively.²³⁹

Importantly, CA 19-9 is unreliable in the setting of cholangitis or obstructive jaundice,²⁴⁰ and is not expressed at any level in approximately 10% of patients who are Lewis antigen negative.²⁴¹ CA 19-9 is only a reliable tumor marker once the jaundice has resolved .As mentioned above, liver function tests will frequently demonstrate an obstructive pattern if jaundice is the presenting symptom, and these baseline liver function tests should be obtained both to guide the need for preoperative biliary drainage, and also to assess baseline hepatocellular dysfunction. Finally, if there is clinical concern for autoimmune pancreatitis or sclerosing cholangitis, serum immunoglobulin G subclass 4 (IgG4) should be obtained, although a subset of patients with cholangiocarcinoma—particularly in the setting of PSC—may also have elevated serum IgG4 levels.²⁴²

Contrast-enhanced MRI/MRCP or CT are the initial diagnostic modalities of choice in the workup of distal cholangiocarcinoma, as these noninvasive studies may identify some tumors without introducing the potential risk of endoscopic retrograde cholangiopancreatography (ERCP) or percutaneous transhepatic cholangiography (PTC) (Figure 7). It is worth stressing that these studies should be obtained before any invasive biliary procedure is performed since these procedures will cause inflammatory changes and obscure tumor-related findings. MRCP, in particular, is useful for generating three-dimensional reconstruction of the biliary tree both above and below the level of obstruction. In a prospective Turkish study of 295 patients with pancreatobiliary disease, MRCP had 100% sensitivity and specificity for bile duct cancers relative to the subsequent pathologic assessment of surgically resected specimens.²⁴³ Other studies with fewer patients have estimated the sensitivity and specificity of MRI/MRCP for diagnosing biliary malignancies at 81-94% and 85-100%, respectively.^{183,244–246} When a mass can be appreciated on MRI, the lesion tends to be hypointense on T1 -weighted images with higher signal intensity on T2-weighted images.⁵¹ While CT can similarly identify an abrupt ductal cutoff suggestive of an obstructing tumor, evaluate abnormalities in the regional lymph node basins, and assess the abdominal viscera and soft tissues for distant metastatic spread, it typically does not offer the three-dimensional reconstruction made possible by MRCP, nor can it readily differentiate an obstructing stone in the bile duct.^233,235,247 MRCP is particularly helpful to diagnose choledocholithiasis as the cause of distal bile duct obstruction. The most common radiologic finding is that of distal bile duct obstruction with either a mass or thickening of the distal bile duct. Unlike pancreatic cancer the pancreatic duct is often not dilated. Positron emission tomography (PET) is not generally helpful in determining the presence of malignancy nor in defining the primary tumor stage; however, it has been found to potentially help in the identification of occult metastatic disease and involved regional lymph nodes.^248,249 A recent study has reported that the findings of PET can alter clinical management in approximately 25% of patients.²⁵⁰ Finally, with respect to staging, all patients deemed appropriate for resection based on abdominal imaging should also undergo some form of chest imaging, such as CT or PET CT, to evaluate for isolated thoracic metastases.

Figure 7.

Typical imaging characteristics of distal cholangiocarcinoma on computed tomography (CT) and magnetic resonance imaging (MRI). In this 67-year old male with a newly diagnosed, localized distal cholangiocarcinoma, the coronal images of both contrast-enhanced CT in the venous phase (top left) and T2-weighted MRI (top right) demonstrate tapering of a dilated common bile duct within the pancreatic head, without an obvious mass or dilation of the main pancreatic duct. On CT, often only subtle thickening of the presumed mass originating from the intrapancreatic bile duct can be appreciated, such as this contrast-enhanced axial CT in the arterial phase (bottom left). Alternatively, on a post-contrast, T1-weighted MRI, the mass demonstrates increased signal intensity (bottom right). In all images, the white arrows represent the location of the tumor.

In general, invasive cholangiography via ERCP or PTC should be reserved for patients with unresectable or metastatic disease, those who require preoperative biliary drainage (see Section IV on preoperative preparation), or cases in which a diagnostic dilemma requires either tissue acquisition or further evaluation of the bile duct prior to operation. Examples of the latter include biliary obstructions in patients at high risk for benign strictures, such as those who have undergone prior bile duct surgery, those who have a history of common bile duct stones, or patients who carry a diagnosis of PSC. Tissue may be obtained via epithelial brushings (cytology only), during endoscopic ultrasound guided fine needle aspiration (EUS-FNA), or by image-guided percutaneous biopsy. For distal cholangiocarcinomas, endoscopic ultrasound (EUS) is particularly useful for primary tumor staging and evaluating regional lymph nodes. The National Comprehensive Cancer Network (NCCN) emphasizes that all cases should undergo multidisciplinary review by an experienced team of radiologists and surgeons in order to determine resectability, consider alternative treatment modalities, and assess the need for more invasive diagnostic studies.²⁵¹

In general, the workup for a patient with distal bile duct obstruction should include high quality cross-sectional imaging prior to invasive procedures. Although a histologic diagnosis is reassuring, a distal bile duct stricture in a patient without a known reason to have a benign stricture should be considered malignant. A negative biopsy cannot rule out malignancy and surgery often must proceed without a histologic diagnosis and with the understanding that a small proportion of cases will ultimately be diagnosed as benign on final pathology. Lastly, in patients with a stricture and no histologic diagnosis, one should consider obtaining IgG4 levels. If elevated and autoimmune pancreatitis/cholangiopathy are suspected short course steroids are indicated and may resolve the stricture ruling out a malignancy and sparing someone an unnecessary option.

Resectability

Broadly, for all cholangiocarcinomas, the contraindications to resection include 1) distant visceral or peritoneal metastases, 2) non-regional lymph node metastases (e.g. aortocaval or paraceliac lymph nodes), and in certain centers, 3) main portal vein or hepatic artery involvement. While true resectability may often be ascertained only at the time of operation, the resection rate among patients undergoing exploration for histologically confirmed cholangiocarcinoma tends to be highest for distal tumors (96%, compared to 40 - 70% and approximately 66% for perihilar and intrahepatic cholangiocarcinomas, respectively). The rate of achieving a margin negative resection is also considerably higher among patients with distal tumors (78%) compared to perihilar and intrahepatic cholangiocarcinomas.⁴¹¹ Consequently, patients with distal cholangiocarcinomas tend to have the most favorable survival outcomes.²⁵²

Determining the resectability of distal cholangiocarcinomas follows the same general principles of other periampullary tumors.²⁵³ These tumors tend to arise from the intrapancreatic portion of the common bile duct, and thus most require pancreaticoduodenectomy, although in certain cases segmental resection of the bile duct can be attempted with intraoperative frozen section margin evaluation.²⁵⁴ Unlike pancreatic adenocarcinoma—where resectability is narrowly defined based on tumor involvement with the superior mesenteric and celiac branch vessels—the growth pattern of distal cholangiocarcinomas generally does not involve these vascular structures to the point of unresectability.

The United States Extrahepatic Biliary Malignancy Consortium analyzed 176 patients with distal cholangiocarcinoma who underwent curative-intent resection at one of 10 high-volume hepatopancreatobiliary centers.²⁵⁵ Despite the fact that 71% of patients had tumors that invaded adjacent organs, only 4.5% required portal vein reconstruction, signifying that direct vascular involvement is relatively rare. A similar rate of portal vein resection (6.8%) was reported by the Japanese multiinstitutional Nagoya Surgical Oncology Group, which described 453 consecutive patients who had undergone pancreaticoduodenectomy for distal cholangiocarcinoma.²⁵⁶ The need for portal vein reconstruction is associated with several poor prognostic factors, including lymphatic invasion, perineural invasion, and lymph node metastases; as such, long-term survival is nearly uniformly poor following portal vein reconstruction for distal cholangiocarcinoma, and should be considered a relative contraindication to resection.^256,257

Interestingly, until recently, the 7^th American Joint Committee on Cancer (AJCC) staging system for distal cholangiocarcinomas assessed tumor stage exclusively based on extent of tumor invasion, with T4 tumors invading the celiac or superior mesenteric arteries and T3 tumors invading adjacent organs without arterial involvement; portal vein involvement was not included in this staging system.²⁵⁸ The United States Extrahepatic Biliary Malignancy Consortium found that tumor stage, as defined at that time, failed to stratify patients into distinct survival outcomes.²⁵⁵ The 8^th AJCC staging system has subsequently altered the definition of tumor stage to more precisely measure depth of invasion,¹⁰¹ and early validation studies suggest that this new schema may have improved prognostic capabilities (Table 7).^259,260

Table 7.

Changes in the American Joint Committee on Cancer (AJCC) staging of distal cholangiocarcinoma between the 7^th and 8^th Editions

	AJCC 7th Edition	AJCC 8th Edition
Primary Tumor Stage
TX	Primary tumor cannot be assessed	Primary tumor cannot be assessed
Tis	Carcinoma in situ	Carcinoma in situ/high-grade dysplasia
T1	Tumor confined to the bile duct histologically	Tumor invades the bile duct wall with a depth less than 5mm
T2	Tumor invades beyond the wall of bile duct	Tumor invades the bile duct wall with a depth of 5-12mm
T3	Tumor invades the gallbladder, pancreas, duodenum, or other adjacent organs without involvement of the celiac axis or superior mesenteric artery	Tumor invades the bile duct wall with a depth greater than 12mm
T4	Tumor involves celiac axis or superior mesenteric artery	Tumor involves the celiac axis, superior mesenteric artery, and/or common hepatic artery
Regional Lymph Node Stage
NX	Regional lymph nodes cannot be assessed	Regional lymph nodes cannot be assessed
N0	No regional lymph node metastasis	No regional lymph node metastasis
N1	Regional lymph node metastasis	Metastasis in 1-3 regional lymph nodes
N2	N/A	Metastasis in 4 or more regional lymph nodes

Based on American College of Surgeons. Edge SB, Byrd DR, Compton CC, Fritz AG, Greene FL, Trotti A, (editors). AJCC Cancer Staging Manual. 7th Ed. Springer New York, 2010, and American College of Surgeons. Amin, M.B., Edge, S.B., Greene, F.L., et al. (Eds.) AJCC Cancer Staging Manual. 8th Ed. Springer New York, 2017.

Preoperative Preparation

Beyond the preoperative assessment of potential resectability, the primary consideration before operative exploration is in respect to the need for biliary decompression. There exists considerable debate regarding the benefit of relieving biliary obstruction preoperatively, and thus the decision is often surgeon and center specific. This situation should be differentiated from the patient with biliary sepsis, in which urgent biliary decompression is necessary for source control, although as previously mentioned, cholangitis is an infrequent presenting symptom.

Multiple randomized controlled trials have compared outcomes following pancreaticoduodenectomy with or without preoperative biliary drainage among jaundiced patients with pancreatic cancer and other periampullary malignancies, and strategies for patients with distal cholangiocarcinoma have largely been extrapolated based on these studies.^86,261–266 Only a single early randomized trial demonstrated improved perioperative outcomes associated with preoperative percutaneous transhepatic biliary drainage (predominantly renal-related outcomes),²⁶¹ with the remaining randomized studies demonstrating no benefit or even detrimental outcomes.^86,262–266 A 2010 multicenter Dutch trial represents the largest and most contemporary study evaluating this controversy, randomizing 202 patients (relatively healthy patients, no vascular involvement and bilirubin less than 14mg/dl) with obstructive jaundice to immediate surgery versus preoperative biliary drainage, primarily via ERCP-assisted stent placement. In this study, serious complications were substantially higher in the biliary drainage group (74% vs. 39%, p<0.001), although no difference in mortality was observed. Interestingly, the morbidity was largely due to ERCP related complications rather than surgical complications.⁸⁶ A 2012 Cochrane Review concluded that there was insufficient evidence to support preoperative biliary drainage, and that its routine use may increase serious adverse event rates.²⁶⁷

In our practice, patients with a potential new diagnosis of distal cholangiocarcinoma are discussed at a weekly tumor board consisting of hepatopancreatobiliary specialists in surgery, radiology, and gastroenterology. When imaging is consistent with a malignant obstruction (mass, no concern for choledocholithiasis, or autoimmune pancreatitis/cholangitis), preoperative biliary decompression is generally avoided. If the patient is deeply jaundiced (bilirubin >15 mg/dL) or is going to require additional diagnostic workup or time before surgery, we generally will recommend placement of an ERCP-assisted metal stent. These stents are durable, provide rapid normalization of the bilirubin, and have not been shown to decrease or impede resecatability.²⁶⁸ Routine biliary decompression is not justified, and the NCCN guidelines emphasize that the decision for preoperative biliary drainage should be conducted in a multidisciplinary fashion.²⁵¹

Operative Approach/Strategies

In selected cases, operative exploration should begin with a staging laparoscopy, which can identify unresectable disease in nearly one-fifth of patients deemed resectable by preoperative imaging. An analysis from the United States Extrahepatic Biliary Malignancy Consortium recently evaluated 1,090 cases of potentially resectable extrahepatic biliary malignancies, including 250 cases in which a staging laparoscopy was performed and 840 cases that began with a laparotomy.²⁶⁹ Among patients with distal bile duct tumors, 13% of patients who underwent staging laparoscopy were found to have occult metastatic disease, thus preventing an unnecessary laparotomy. Among patients who underwent immediate laparotomy, 8% were ultimately deemed unresectable on exploration. When this procedure is performed, evaluation should include inspection of all peritoneal surfaces, with particular attention given to the falciform ligament, liver, diaphragm, porta hepatis, gastrohepatic omentum, and proximal small bowel mesentery; laparoscopic ultrasound may also have a role, particularly in identifying liver metastases that do not extend to the liver surface.²⁷⁰

The vast majority of patients with distal cholangiocarcinomas will require a pancreaticoduodenectomy to achieve a margin negative resection. Occasionally, margin negative bile duct resections are feasible when anatomically just above the pancreatic head, but this is only possible in approximately 10% of all patients with extrahepatic cholangiocarcinomas^230,252. Often, tumors amenable to segmental bile duct resection arise from an intraductal papillary neoplasm of the bile duct. This rare but recognized lesion may be characterized by an exophytic, focal growth pattern that facilitates a less radical resection.^271–273 Symptoms suggestive of this growth pattern include intermittent jaundice or cholangitis, thought to be secondary to the mucin production or shedding of tumor pieces that may temporarily impede bile flow.^273,274

For patients requiring a pancreaticoduodenectomy, we typically begin by performing a wide Kocher maneuver, mobilizing the duodenum and head of the pancreas off the great vessels to a point lateral of the left renal vein. This provides a tactile sense of resectability based on tumor involvement with the mesenteric vessels, and access to evaluate any suspicious aortocaval lymph nodes. We then enter the lesser sac, and identify the infrapancreatic superior mesenteric vein, where a retropancreatic tunnel is created. At this point, if we feel that the tumor can likely be separated from the vein, we perform the portal dissection and determine the proximal extent of disease and our biliary transection point. The lymph nodes of the hepatic artery and the porta hepatis are stripped from their adjacent structures. The gastroduodenal artery is divided, the bile duct is encircled, and a standard cholecystectomy is performed. The common bile duct is typically divided just proximal to the cystic duct insertion or at a point where the bile duct is grossly normal.

Frozen section assessment of the proximal biliary margin is controversial, however in the setting of presumed cholangiocarcinoma (versus pancreatic adenocarcinoma), it should be strongly considered. When performed, however, the surgeon should recognize that there is a reported 25% risk of false-positive or false-negative results.²⁷⁵ In cases where we have a high suspicion of proximal extension, the bile duct is divided at the hepatic duct bifurcation. When the margin is positive at this location, we would recommend re-resection to a negative margin. Occasionally, the proximal spread will be such that a liver resection is required in order to obtain a negative margin. These are very difficult situations, especially if not anticipated preoperatively. While in highly selected cases it is reasonable to consider a combined pancreatoduodectomy and liver resection the high morbidity and mortality must be carefully considered.

Following standard division of the antrum (or duodenum) and pancreatic neck, the superior mesenteric and portal vein are dissected off the pancreas and widely mobilized. The uncinate is then divided just lateral to the superior mesenteric artery border. A duct-to-mucosa pancreatojejunostomy is created, a hepaticojejunostomy is performed and a hand-sewn, antecolic gastrojejunostomy is brought up as a loop approximately 50 centimeters distal to our pancreatobiliary anastomoses. Placement of operative drains for pancreatoduodenectomy is controversial however it should be noted that typically for distal bile duct cancers the pancreatic duct is small and non-dilated. Although personal preference is the rule, these are often situations to consider drainage.

An appropriate regional lymphadenectomy is essential for accurate staging of all cholangiocarcinomas. For distal tumors, this involves lymph nodes along the common bile duct, common hepatic artery, and portal vein; among the majority of patients requiring pancreaticoduodenectomy, a complete lymphadenectomy should also by definition include peripyloric, pancreaticoduodenal, and superior mesenteric nodes. Two studies have evaluated an appropriate threshold for lymph node assessment following distal cholangiocarcinoma resection. A 2007 analysis of the Surveillance, Epidemiology, and End Results (SEER) database identified improved survival for patients with 10 or more lymph nodes examined.²⁷⁶ Subsequently, a retrospective analysis at the Memorial Sloan Kettering Cancer Center was designed to identify the number of lymph nodes needed to accurately stage patients as node negative. Results from this study indicated that a threshold of 11 lymph nodes was required for accurate staging, and that among patients staged as node negative, further evaluation did not improve the survival of the group deemed “node negative”.²³⁰

Finally, minimally invasive approaches to pancreaticoduodenectomy have gained traction in certain centers for patients with both benign and malignant conditions. In general, acceptable outcomes with minimally invasive pancreaticoduodenectomy have only been reported from high-volume centers with a dedicated team for this approach.²⁷⁷ Three randomized trials have evaluated minimally invasive versus open approaches with attention to short-term perioperative and oncologic outcomes.^278–280 Of these, the largest and most recent was a multiinstitutional Dutch study which randomized 105 patients at high-volume hepatopancreatobiliary centers to undergo either laparoscopic or open pancreaticoduodenectomy.²⁸⁰ In this study, all surgeons were required to complete a dedicated training program in laparoscopic pancreaticoduodenectomy, and randomization accounted for both surgeon case volume and preoperative risk of pancreatic leak. The trial was terminated early due to excess mortality in the laparoscopic arm (10% vs. 2%). Among enrolled patients, a laparoscopic approach was also associated with higher rates of serious complications (50% vs. 39%). The other two randomized trials of 32 and 66 patients, respectively, did not identify differences in pancreatectomy-specific morbidity or short-term mortality. While the role of minimally invasive approaches for patients with distal cholangiocarcinoma is still evolving, currently these techniques should be utilized by highly experienced surgeons, within structured training programs, or under investigative protocols.

Operative Morbidity and Mortality

The morbidity and mortality associated with pancreaticoduodenectomy have been well described, particularly among patients with pancreatic adenocarcinoma.^281–290 Among high-volume institutions and surgeons—defined as greater than 16 and four resections per year, respectively—mortality rates for pancreatic resection are reported under 5%, with an inverse relationship between mortality and volume.^281,282 When performed at centers and by surgeons with the highest volume, perioperative mortality of 2% or less is typically reported.^283–286 The most common serious complications following pancreaticoduodenectomy are delayed gastric emptying, pancreatic leak, surgical site infection, and hemorrhage related to leak.^288–290 However, it is important to consider this morbidity profile specifically among patients with distal bile duct tumors rather than primary pancreas cancers, as these patients are more likely to have normal pancreatic duct size and gland textures,²⁹¹ which are established risk factors for postoperative pancreatic fistula.²⁹²

Only one large series has directly compared postoperative outcomes following pancreaticoduodenectomy for patients with distal bile duct versus pancreatic cancers.²⁹³ Among 710 patients who underwent pancreaticoduodenectomy (7.6% with cholangiocarcinoma), rates of postoperative morbidity were substantially higher among patients with bile duct cancers, including pancreatic fistula (48.1% vs. 19.4%; and 44.3% vs 13.5% with grade B or C leaks), intraabdominal abscess (37.0% vs. 18.9%), and postoperative hemorrhage (18.5% vs. 10.1%). While comparative morbidity data for patients with distal cholangiocarcinoma are limited, it is important to note that in this study, the reported outcomes for patients with pancreas cancer were highly similar to other contemporary series, indicating that there likely is a real difference in outcomes for these distinct cancer types.^284,286,287 Moreover, these results are in keeping with another large, multiinstitutional series of patients undergoing pancreaticoduodenectomy for distal bile duct cancer, which reported postoperative pancreatic fistulae in 42.5% of patients, including 26.3% of patients with grade B or C leaks; in addition, the authors reported high rates of delayed gastric emptying (36.3%), postoperative hemorrhage (16.7%), reoperation (18.1%), and in-hospital morality (7.9%).²⁹⁴ Perhaps most importantly, distal cholangiocarcinoma patients with these complications, particularly postoperative pancreatic fistula, were significantly less likely to receive postoperative chemotherapy (68% vs. 100%) and had significantly reduced disease-specific survival (HR 1.67). While the authors reported no difference in postoperative death (1.8% vs. 1.4%), this outcome was not explicitly defined. The well-established relationship between postoperative mortality, complications, and a failure to rescue suggests that these patients are best treated at centers capable of rapidly identifying and intervening on this morbidity profile.²⁹⁵

Finally, for patients who do undergo isolated bile duct resection, the most common morbidity is related to the biliary anastomosis. In the largest study specifically comparing segmental bile duct resection to pancreaticoduodenectomy, the incidence of bile leak was similar (6% vs. 11%), although these rates were considerably higher than reported in other large series of patients undergoing pancreatobiliary resection.^284,285 Finally, while infrequently reported, both the enterobiliary and pancreaticojejunal anastomoses are at risk for late stricture formation.^296,297 Another important late complication is new-onset diabetes, which may occur in up to 16% of patients undergoing pancreaticoduodenectomy, regardless of the indication for surgery.²⁹⁸

Adjuvant and Neoadjuvant Strategies (Table 8)

Table 8.

Randomized controlled trials assessing systemic chemotherapy in patients with extrahepatic, including distal cholangiocarcinoma.

Trial Name, Location, Total Patients (distal cholangiocarcinoma)	Treatment Arms	Median RFS (months)	Median OS (months)
PRODIGE 12-ACCORD, France, 196 (55)	Adjuvant gemcitabine (1,000 mg/m2) and oxaliplatin (85 mg/m2) vs. observation alone following R0/R1 resection	30.4 vs. 18.5 (HR 1.08, p=0.48)	75.8 vs. 50.8 (HR 1.08, p=0.74)
BCAT, Japan, 226 (123)	Adjuvant gemcitabine (1,000 mg/m2) vs. observation alone following R0/R1 resection	36.0 vs. 39.9 (HR 0.93, p=0.69)	62.3 vs. 63.8 (HR 1.01, p=0.96)
BILCAP, United Kingdom, 447 (156)	Adjuvant capecitabine (1250 mg/m2) vs. observation alone following R0/R1 resection	24.4 vs. 17.5 (HR 0.75, p=0.033)	51.1 vs. 36.4 (HR 0.81, p=0.10)

Given the high rates of local and distal recurrence following complete resection, adjuvant strategies employing either chemotherapy or chemoradiation therapy have been utilized in an attempt to improve outcomes. Historically, the decision for adjuvant treatment has been guided by retrospective series, which largely provided conflicting or highly biased evidence in support of these strategies.^137,299,300 Within these retrospective studies, the patients with the highest associated benefit from adjuvant treatment appeared to be those with node-positive and/or margin-positive disease.¹³⁷ To date, evidence regarding postoperative radiation therapy for distal cholangiocarcinoma relies entirely on these retrospective analyses.³⁰⁰

Recently, three randomized trials were completed evaluating adjuvant chemotherapy for resected biliary tract cancers.^140,141,215 None specifically distinguished distal cholangiocarcinomas from more proximal bile duct cancers or gallbladder cancers, and only one study demonstrated a clinical benefit for postoperative treatment. The PRODIGE 12-ACCORD 18 trial was a randomized, multicenter French trial comparing intravenous gemcitabine and oxaliplatin with observation following complete resection of biliary tract cancers.¹⁴¹ Of 196 randomized patients, 55 had distal cholangiocarcinoma. The Bile Duct Cancer Adjuvant Trial (BCAT) was a randomized, multicenter Japanese trial among patients who had undergone complete resection of intrahepatic or extrahepatic cholangiocarcinoma, comparing intravenous gemcitabine to observation alone. Of 226 randomized patients, 123 patients had distal cholangiocarcinoma. In neither of these two trials were differences observed in either recurrence free or overall survival between groups.

The only study to demonstrate a benefit for adjuvant therapy was the BILCAP trial, a randomized, multicenter British trial comparing adjuvant oral capecitabine with observation following complete resection of cholangiocarcinoma or gallbladder cancer.¹⁴⁰ From 2006 to 2014, 447 patients were enrolled and randomized, with a median follow-up time of five years. In the intention-to-treat analysis, median overall survival was higher in the capecitabine group (51.1 vs. 36.4 months), although this did not reach statistical significance. However, in a prespecified per-protocol analysis, which excluded ineligible patients and patients who failed to complete at least one cycle of capecitabine, a statistically significant survival benefit was appreciated for adjuvant chemotherapy (HR 0.75, p=0.028). Among all enrolled patients, 156 had distal cholangiocarcinoma, and the survival benefit among these patients was similar to the total population, although this did not reach statistical significance in subgroup analysis (HR 0.70, 95% CI 0.47-1.06). Interestingly, in subgroup analysis, there did not appear to be a particular benefit for patients with node-positive disease or margin-positive resections relative to their comparison groups.

Taken together, these studies lend weak support to the use of adjuvant chemotherapy, with unclear guidance on the patients most likely to derive a survival benefit; at present, postoperative oral capecitabine is the only regimen supported by randomized data. Currently, the NCCN guidelines suggest that patients with node-positive or margin-positive disease should receive postoperative fluoropyrimidine- or gemcitabine-based chemotherapy, or fluoropyrimidine-based chemoradiation, acknowledging that these recommendations are largely extrapolated from studies of patients with advanced disease.²⁵¹ These regimens are also recommended for patients following node-negative, margin-negative complete resection, although the NCCN guidelines state that these patients may also be followed with surveillance alone.

Unlike hilar cholangiocarcinoma, for which neoadjuvant strategies have been derived largely as a bridge to transplantation,^301,302 there exists extremely limited data to support preoperative chemotherapy or chemoradiation therapy for patients with distal bile duct tumors.^303,304 These studies suggest that a minority of patients may have a pathologic complete response following treatment. Thus, in our practice, neoadjuvant chemotherapy or chemoradiation therapy is reserved for the patient with distal cholangiocarcinoma deemed locally unresectable on initial review. Only patients with a strong radiographic treatment response, subsequently considered potentially resectable on multidisciplinary review, would then be a candidate for attempt at curative resection. It must be stressed that while these tumors may require a similar operative approach, distal cholangiocarcinomas are biologically distinct from pancreatic adenocarcinomas,²⁹⁹ for which there is now increasing evidence in support of neoadjuvant strategies.^306,307 We strongly agree with the NCCN’s emphasis that all patients with distal cholangiocarcinoma undergoing adjuvant or neoadjuvant therapy should ideally be treated in the setting of a clinical trial.

Long-Term Outcomes/Prognostic Factors

Following complete gross resection, five-year survival rates are reported between 27-48% for patients with distal cholangiocarcinoma.^{40,100,102,252,294,309,311–313}. The most important prognostic factors affecting overall survival following resection are margin status and lymph node status.^{40,252,294,308,309,313–315} Other factors which have been reported to be associated with survival include histologic differentiation^{40,252,294,309} perineural invasion^312,313 the need for perioperative blood transfusion, and other perioperative morbidity.²⁹⁴ A recent meta-analysis evaluating 2,063 patients from 23 studies over a 15 year contemporary period largely confirmed the findings of these prior series, establishing lymph node metastases, margin status, perineural invasion, and histologic differentiation as the dominant predictors of five-year survival.³¹² Notably, age, gender, and the use of postoperative adjuvant chemotherapy were not consistently associated with survival outcomes.

A multiinstitutional study by the Nagoya Surgical Oncology Group recently evaluated patterns of recurrence following curative intent resection, estimating a cumulative probability of recurrence at 54.3% over five years, with a median time to recurrence of three years among all patients.³¹³ For patients with documented disease recurrence, the majority (78.9%) presented initially with distant spread, typically to the liver (43.2%), peritoneal surface (16.9%), or retroperitoneum (11.7%). Locoregional recurrence occurred in 25.8% of cases; interestingly, for patients who did recur locally, few were also found to have distant metastatic spread, yet overall survival was no different between groups. Predictors of disease recurrence included perineural invasion, lymph node metastasis, and extent of tumor invasion; again, adjuvant chemotherapy was not associated with recurrence-free survival. Taken together, these data highlight the gap that still exists between surgical advances allowing for the safe resection of distal bile duct tumors, and augmentation strategies that can offer meaningful improvements in long-term survival. Ultimately, treatment options beyond surgery are necessary to achieve either a cure or durable, long-term survival for the vast majority of patients who initially present with otherwise resectable disease.

IV. Gallbladder Carcinoma

Introduction

Gallbladder cancer (GBC) is the most common biliary tract malignancy. It can be suspected prior to surgery, identified intraoperatively, or recognized incidentally following cholecystectomy. Although the prognosis for GBC varies with its presentation, it remains one of the deadliest solid malignancies worldwide.³¹⁶ Several geographic, demographic, and clinical risk factors are associated with GBC; however, no patient group has been identified with sufficient risk to warrant screening. Unfortunately, the prognosis of symptomatic GBC remains dismal despite advancements in surgical and systemic therapies.³¹⁷

Epidemiology

The prevalence of GBC varies by geography and ethnicity. Incidence is highest in South America and Asia and lowest in the United States and Western Europe. The highest incidence is reported for women in Northern India and Chile.³¹⁸ While prophylactic cholecystectomy for asymptomatic cholelithiasis is sometimes performed within these populations, the effectiveness of this practice in preventing gallbladder cancer death is unclear.³¹⁹ In the United States, the incidence of GBC is 1.6/100,000, or 4400 cases per year.⁷ This accounts for more than 2000 deaths annually.³²⁰ Hispanic and black populations are at higher risk than whites, however, the American population at highest risk are Native Americans in the New Mexico region, with an annual incidence of 8.9/100,000.³²¹

The median age at presentation for GBC is 67 years, and incidence increases with age.^320,322 Based on differences in median age at presentation, the proposed rate of progression from dysplastic gallbladder polyp to carcinoma in-situ is estimated to be 5 years, while the transition from carcinoma in-situ to invasive cancer is 10 years.³²³ However, because the majority of gallbladder cancers discovered in the United States are not found in the context of an underlying polyp, this timeline may only apply to a minority of cases. Gallbladder cancer disproportionately affects women more than men, with an incidence ratio of 1.7:1. This is atypical of other biliary tract cancers, such as cholangiocarcinoma and ampulla of Vater cancer, for which men are at higher risk.⁷

Risk factors for GBC include cholelithiasis, obesity, and primary sclerosing cholangitis (PSC). Gallstones are present in 85% of patients with GBC, and cancer prevalence increases with larger stone size and number.³²⁴ Compared to patients with stones less than 1 cm in diameter, the odds ratio of GBC for patients with gallstones > 3 cm is 10.1, translating to 4% risk over 20 years.³²⁵ The association between obesity and GBC is difficult to separate from the confounding risk of gallstone formation. Relative risk in the obese population (compared to people with normal BMI) is 1.8 for men and 2.1 for women.³²⁶ Primary sclerosing cholangitis is a chronic inflammatory syndrome that is associated with cholangiocarcinoma. The estimated lifetime risk of GBC in patients with PSC is 3-14%.³²⁷ Due to this increased risk, European consensus guidelines recommend cholecystectomy for gallbladder polyps of any size in the PSC population.³²⁸ In one retrospective series, a polyp size threshold of 0.8 cm was found to yield 100% sensitivity for gallbladder neoplasia among patients with PSC.³²⁹ Accordingly, guidelines from the American College of Gastroenterology recommend cholecystectomy for polyps greater than 0.8 cm.³³⁰ Guidelines for the management of polyps among patients without PSC will be discussed in subsequent sections.

Presentation

Symptomatic Presentation

Gallbladder cancer presents in one of two broad scenarios: 1) suspected prior to surgery, and 2) as an incidental finding after cholecystectomy. Unfortunately, GBC is usually asymptomatic during early stages, contributing to delayed diagnosis and poor prognosis. When present, symptoms are difficult to distinguish from those of cholelithiasis, including right upper quadrant pain that is generally more constant than that typical of biliary colic. In advanced stages, mass effect and systemic symptoms evolve, such as jaundice, anorexia, and weight loss (Figure 8). In this setting, symptoms are similar to those of hilar cholangiocarcinoma. Most patients with GBC who present with jaundice have disseminated disease. In a series from Memorial Sloan Kettering, 34% of patients with GBC presented with jaundice. Among these, only 7% underwent resection with curative intent, and all died of disease within 3 years.³¹⁷

Figure 8.

Blumgart Figure 49.7 (Part V: Biliary Tract Disease, II – neoplastic, C – malignant tumors, Ch 49 GBC). Percutaneous transhepatic cholangiogram showing malignant obstruction of the bile duct confluence. Distortion of the segment V duct (arrow) due to direct invasion from gallbladder cancer.

While most cases of GBC occur in the setting of cholelithiasis, the rate of incidental GBC in routine cholecystectomy specimens is less than 2%.^331–333 Among incidentally-discovered GBC, 34% are stage T1a, 14% T1b, and 41% T2.³³⁴ Preoperative factors associated with incidental GBC include: older age, female sex, and recurrent or chronic cholecystitis.³³⁵ Perioperative risk factors associated with the risk of cancer include prolonged operating time, conversion to open surgery, and macroscopically abnormal gallbladder features (i.e., wall thickening, obvious tumor, inflammation). Of note, the positive predictive value of intraoperative surgeon suspicion for GBC was 31%, emphasizing the role for gross examination of suspicious gallbladder specimens in the operating room. Importantly, patients with an incidentally-diagnosed GBC following cholecystectomy who then go on to receive a curative extended resection fare no worse than those who undergo an up-front en bloc resection, although this comparison is fraught with bias.³³⁶

Radiographic Presentation

In otherwise asymptomatic patients, GBC may be suspected due to abnormal ultrasound findings. Ultrasound features concerning for GBC include thickened, echogenic, or discontinuous mucosa, doppler blood flow through a segment of irregular mucosa, or an intraluminal mass.³³⁷ The majority of GBC’s are located in the gallbladder fundus.³³⁸ When located in the gallbladder neck, cystic duct obstruction can present early as gallbladder hydrops.

Calcification of the gallbladder wall—so-called porcelain gallbladder—is a marker of chronic cholecystitis. Although porcelain gallbladder was once considered to be associated with a high risk of malignancy, more recent series indicate GBC prevalence of < 6%.^339,340 Stippled or focal calcification portends a greater risk of malignancy than diffuse intramural calcification, but the overall risk remains low.³⁴¹ In a retrospective series of 90 patients with porcelain gallbladder initially managed with observation, no patient progressed to cancer over an average of 3.2 years follow-up.³⁴² However, selection biases should be considered. Patients with porcelain gallbladder have a significant risk of malignancy exceeding that of the general population and should be considered for cholecystectomy if otherwise healthy. Patients who present with pain or other symptoms and a porcelain gallbladder warrant resection.

Gallbladder polyps are present in 5-10% of the adult population.^343,344 Most polyps possess no malignant potential, including cholesterol polyps, lipomas, hyperplastic polyps, hemangiomas, and adenomyomas. Less than 10% of polypoid lesions are adenomas, which are considered premalignant.^345,346 Unless large or obstructive, adenomatous polyps of the gallbladder are usually asymptomatic and often found incidentally during imaging for suspected cholelithiasis. In a series from Mayo Clinic of 130 cholecystectomy patients who had polypoid gallbladder lesions on preoperative ultrasound, 31 harbored features suspicious for neoplasia: > 10 mm, vascularity, or invasive features. Twenty-nine of these suspicious polyps had neoplastic changes (dysplasia or invasion) on final pathology.³⁴⁷ Current guidelines recommend resection for gallbladder polyps that are ≥ 10 mm in diameter or have vascular stalks.³⁴⁸ Small polyps without high-risk features may be safely observed. The appropriate duration of surveillance is not well-defined; in general, a polyp that has remained stable for 2-3 years seems unlikely to require further follow-up. In a series from Memorial Sloan Kettering of 417 patients with polypoid gallbladder lesions, only 7% had polyps > 10 mm in diameter. Among 80 patients who underwent cholecystectomy, adenomas were found in 10%, carcinoma in-situ was found in one patient, and none had invasive malignancy.³⁴⁵

Diagnosis/Work-Up

Diagnosis

Frequently, when a diagnosis of GBC is suspected, a right upper quadrant ultrasound has already been performed. Other times, the diagnosis of gallbladder cancer arises following pathologic evaluation of a cholecystectomy specimen. Gallbladder cancer may also be suspected due to mass effect on surrounding structures, presenting with biliary obstruction and advanced malignancy. Findings concerning for GBC on ultrasound include discontinuous mucosal thickening, porcelain gallbladder, mural calcification, or a mass protruding into—or even obscuring—the gallbladder fossa.³⁴⁹ In more advanced cases, the gallbladder-hepatic interface may be obscured. A gallbladder mass can be differentiated from stones by its fixation to the gallbladder wall, however, smaller exophytic carcinomas may be difficult to distinguish from benign polyps. The use of endoscopic ultrasound to aid in GBC diagnosis has fallen out of favor, as modern high-resolution ultrasound is less invasive and has comparable sensitivity and specificity.³⁵⁰

No single serum marker is both sensitive and specific for GBC. In an early case-control study, serum carcino-embryonic antigen (CEA) was associated with 50% sensitivity and 93% specificity for GBC, compared to 79% and 79% for carbohydrate antigen 19-9 (CA 19-9).³⁵¹ In the role of differentiating GBC from benign gallbladder pathology and healthy patients, CA 19-9 and CA 242 outperforms CEA and CA 19-9, with 72% and 64% sensitivity and 96% and 99% specificity, respectively.³⁵²

When GBC is suspected in the preoperative setting, contrast-enhanced computerized tomography (CT) of the chest, abdomen and pelvis is invaluable to characterize primary tumor features and stage the patient. In the pre-contrast phase, GBC is typically hypodense to the liver. With contrast-enhancement, GBC may appear as a heterogeneously-enhancing mass with poorly-defined borders, or as asymmetric gallbladder wall thickening³⁵³ (Figure 8). Unlike hepatocellular carcinoma, GBC tends to retain contrast during the portal venous phase and exhibits slower washout. While not required to establish a diagnosis, magnetic resonance imaging (MRI) with diffusion weighting can be useful to differentiate GBC from similar-appearing conditions such as xanthogranulomatous cholecystitis and adenomyomatosis. Xanthogranulomatous cholecystitis can present as a suspicious mass but is characterized by less diffusion restriction than GBC, and has reduced signal intensity in out-of-phase images due to the presence of foamy histiocytes, fibrosis, and giant cells.³⁵⁴ Adenomyomatosis may be distinguished by the presence of Rokitansky-Aschoff sinuses on T2-weighted series. On ultrasound, these sinuses appear as small anechoic cystic spaces, and may be associated with hyperechoic “comet-tail” artifacts created by cholesterine crystals.

For resectable GBC with an in-situ gallbladder, a tissue diagnosis is not required prior to operative intervention and is discouraged due to the risk of intraperitoneal spread. The corollary to this is that a surgeon considering initial resection of a suspected GBC must be prepared to perform an extended resection for carcinoma during the index exploration. In such situations, the patient should be prepared for the possibility of a benign diagnosis following resection with curative intent. For difficult cases such as possible xanthogranulomatous cholecystitis, an intraoperative biopsy with frozen-section analysis can be considered to confirm a diagnosis of carcinoma before committing to a formal resection and lymphadenectomy.

Histology

In descending order, the most prevalent histologies in GBC are adenocarcinoma (90%), adenosquamous or squamous carcinoma (4%), neuroendocrine (3%), sarcoma (1.6%), and melanoma (1%).³²² Among adenocarcinomas, most are of the pancreaticobiliary type; others include intestinal, mucinous, clear cell, and undifferentiated. When suspecting GBC, the importance of review by an experienced gastrointestinal pathologist cannot be overstated. Early stage, well-differentiated GBC may be difficult to distinguish from Rokitansky-Aschoff sinuses, as both have the appearance of glandular structures extending into the gallbladder wall or even beyond the muscular layer. In particular, hyperechoic cholesterine crystal artifacts can project beyond the gallbladder wall, giving the impression of local invasion. In these cases, desmoplasia, non-perpendicular gland orientation, cellular atypia, mitoses, and necrosis favor the diagnosis of carcinoma.³⁵⁵ The immunohistochemical profile of GBC is characterized by positive staining for cytokeratin 7 (CK7). Other commonly-expressed markers include CK20, monoclonal carcinoembryonic antigen-monoclonal (mCEA), carbohydrate antigen (CA 19-9), S100P, IMP3, MUC1, and MUC5AC.³⁵⁶

Staging

Gallbladder cancer is staged via the TNM Classification of Malignant Tumors (Table 9). The T stage of the primary tumor is based on depth of invasion. Tis indicates carcinoma in situ, T1 indicates invasion into the lamina propria (t1a) or muscular layer (t1b), T2 indicates perimuscular invasion without adjacent visceral invasion, T3 indicates invasion beyond the serosa with or without adjacent visceral invasion, and T4 indicates invasion of more than one adjacent organ or invasion into the portal vein or hepatic artery. Nodal staging is categorized as: N0 for no regional lymph node metastases, N1 for 1-3 regional metastatic nodes, and N2 for 4 or more regional metastatic nodes. Regional lymph nodes for GBC are distributed as follows: hilar, cystic duct, pericholedochal, posterosuperior pancreaticoduodenal, retroportal, right celiac, and hepatic artery. Node metastases beyond these areas are considered M1 disease.

Table 9:

AJCC 8^th Edition Staging for Gallbladder Cancer

	Primary Tumor (pT)
TX	Primary tumor cannot be assessed
T0	No evidence of primary tumor
Tis	Carcinoma in situ
T1	Tumor invades lamina propria or muscularis T1a: tumor invades lamina propria T1b: tumor invades muscularis
T2	Tumor invades perimuscular connective tissue T2a: tumor invades on peritoneal side, without serosal invasion T2b: tumor invades on hepatic side, without parenchymal invasion
T3	Tumor invades visceral peritoneum, liver parenchyma, extrahepatic bile ducts, or adjacent organ(s)
T4	Tumor invades the main portal vein or hepatic artery, or invades two or more adjacent organs
	Regional Lymph Nodes (pN)
NX	Regional lymph nodes cannot be assessed
N0	No regional node metastases
N1	Metastasis in 1-3 regional nodes
N2	Metastasis in 4 or more regional nodes
	Distant Metastases (pM)
M0	No distant metastasis
M1	Distant metastasis
	Staging
0	Tis	N0	M0
I	T1	N0	M0
IIA	T2a	N0	M0
IIB	T2b	N0	M0
IIIA	T3	N0	M0
IIIB	T1-3	N1	M0
IVA	T4	N0-1	M0
IVB	T1-4 T1-4	N2 N0-2	M0 M1

Based on American College of Surgeons. Amin, M.B., Edge, S.B., Greene, F.L., et al. (Eds.) AJCC Cancer Staging Manual. 8th Ed. Springer New York, 2017.

The most recent edition of the American Joint Committee on Cancer (AJCC) staging system separated T2 into T2a and T2b subcategories based on the location of the tumor on the peritoneal or hepatic side of the gallbladder, respectively. This is based in large part on an international cohort study of 252 patients in which patients with hepatic-side T2 tumors had higher rates of neurovascular invasion, nodal metastases, hepatic recurrence, and lower rate of overall survival.³⁵⁷ However, whether the anatomic location of GBC is truly indicative of different physiologic behavior is unclear. In a series of 192 patients, Lee and colleagues reported that, for patients who underwent appropriate hepatic resection, 5-year overall survival for hepatic-side and peritoneal-side T2 GBC’s were 80.3% and 70.5%, respectively.³⁵⁸

Preoperative staging of the primary tumor is best accomplished with contrast-enhanced CT of the chest, abdomen and pelvis, which can determine gross adjacent organ invasion and vascular or biliary involvement. In an analysis of CT and pathologic tumor staging across 100 patients with GBC, overall staging accuracy was 71%. For discordant cases, CT more frequently over-stages (18%) than under-stages (11%) the primary tumor.³⁵⁹ When there is clinical concern for locally-advanced disease, magnetic resonance imaging with cholangiopancreatography (MRCP) may better define vascular or biliary infiltration. Sensitivity of MRI with MRCP is 88-100% for direct hepatic invasion and 80% for bile duct invasion.^360,361

The accuracy of cross-sectional imaging for the staging of nodal and distant metastatic disease varies widely. Despite advances in CT and MRI technology, roughly 20% of patients considered resectable by preoperative imaging are found to have unresectable disease in the operating room.^269,362 Sensitivity of CT and MRI for lymph node metastases in GBC is between 25-93% and 59-92%, respectively.^369,361,363 For high risk findings such as a large primary tumor, equivocal nodes, or peritoneal abnormalities, positron emission tomography combined with CT (PET-CT) may increase specificity for metastatic disease. In one retrospective series, the yield of PET-CT for detecting unresectable disease was 39% and increased with advancing T-stage.³⁶⁴ Compared to CT, PET-CT impacted management in 17% of patients. However, the sensitivity of PET-CT was only moderate at 57% overall and lowest for peritoneal disease (28%). The likelihood of finding unresectable disease by PET-CT for a T1 primary tumor is exceeding low; therefore, patients with early-stage primary tumors may be spared this step. The final step in the work-up of GBC prior to extended resection is consideration of staging laparoscopy (SL). Staging laparoscopy reveals disseminated disease in 18-34% of patients with GBC and is associated with lower odds of noncurative exploration.^269,365 Addition of laparoscopic ultrasound may further increase SL yield. In a series of patients with radiographically-resectable hepatobiliary malignancies, SL performed for GBC detected roughly 50% of unresectable cases, with greater yield than for cholangiocarcinomas and hepatocellular carcinoma.¹¹² The added value of staging laparoscopy is dependent on stage and clinical presentation, with yields ranging from 11% for early-stage GBC to 25% for locally-advanced cases.³⁶² For cases of incidental GBC discovered on cholecystectomy, the yield of SL may be as low as 4%³⁶⁶, reflecting the prevalence of T1-2 tumors in this population. Given the relatively low morbidity of the procedure and potential for treatment alteration, we recommend staging laparoscopy for all patients with GBC in the setting of an in-situ gallbladder, and for patients with resected, incidental GBC with serosal invasion.

Resectability

Evaluation for resectability in patients with GBC must take into account the technical feasibility of locoregional resection and indicators of disease biology. The goals of resection are to render a patient free of disease, provide potential for long term survival, gather prognostic information, and guide subsequent therapy. Definitive resection for GBC ranges from laparoscopic cholecystectomy alone for T1a to en bloc hepatic parenchymal resection for T1b-2 to multiple organ resection for T3-4 tumors. The technical feasibility of resection is tied to surgeon expertise and reconstruction options. However, more relevant in the determination of resectability is the likelihood that extended resection would improve long-term outcomes. In general, due to the proclivity of GBC to recur at distant sites, stage and tumor biology are dominant drivers of survival. Therefore, the decision to resect depends upon the implications of locoregional disease on long-term prognosis.

Bile Duct Involvement

In a recent multi-institutional series of 400 GBC cases, 27% presented with jaundice. Jaundiced patients experienced a higher reoperation rate and worse survival (14 vs 32 months).³⁶⁷ Definitive resection in these patients universally mandates bile duct resection and is associated with positive surgical margins in > 40% of cases.³⁶⁸ In a series of 82 patients with GBC presenting with jaundice, cross-sectional imaging and SL ruled out curative resection in 45 patients. Of the 37 who underwent exploration, peritoneal metastases, liver metastases, major vascular invasion, and extensive retroperitoneal disease precluded resection in all but six. All six patients recurred within 6 months, and none survived beyond 3 years.³¹⁷ In a meta-analysis of 2089 patients, jaundice was associated with worse overall survival, lower resectability, more extensive resection, and higher postoperative morbidity.³⁶⁹ Because it is an indicator of aggressive biology and microscopic distant disease, many consider jaundice to be a relative contraindication to resection. However, it is possible to attain long-term survival in a carefully selected subset of jaundiced patients. One small series of 37 patients with GBC and jaundice reported 5-year survival of 27% following resection.³⁷⁰ Among jaundiced patients, the presence of low CA19-9 and absence of lymphovascular invasion may prompt consideration of resection.³⁶⁷ This is particularly true in the small subset of patients for whom jaundice is driven by concomitant symptomatic gallstone disease and is not necessarily reflective of advanced stage.

Among patients with unresectable disease who present with biliary obstruction, percutaneous and endoscopic interventions have largely replaced hepaticojejunostomy for palliation. Although less invasive than a major operation, these interventions are nevertheless associated with significant morbidity. Compared with percutaneous decompression, endoscopic drainage is associated with higher procedure-related complications, including cholangitis or pancreatitis in up to 50%.³⁷¹ Even percutaneous drainage should be reserved for patients with intractable symptoms that severely limit quality of life. In one series of 102 patients who underwent percutaneous biliary drainage for malignant obstruction, 30-day mortality was 10% and median survival of was than 5 months.³⁷² While specific symptoms such as pruritus may be relieved, the overall impact on quality of life from these interventions remains unclear. Relief of jaundice is often required for the safe delivery of chemotherapy and must be carefully considered in these cases.

Lymph Node Distribution

Lymph node involvement is a strong negative prognostic indicator for gallbladder cancer. Within the AJCC 8^th edition staging system, the presence of 1-3 malignant regional lymph nodes is considered stage IIIA, with an associated 5-year survival of roughly 25%.³⁷³ The presence of four or more cancer-bearing regional lymph nodes is indicative of N2, or stage IVB, disease. The anatomic distribution of lymph node involvement also has significant prognostic implications. Lymphatic drainage for GBC spreads initially from the cystic duct to the periportal stations. Periportal drainage includes the hepatic artery and pancreaticoduodenal basins. Parallel routes may also traverse to the celiac station. In general, all regional lymphatic pathways culminate at the aortocaval station, which is considered M1 disease. In a Japanese series of 55 patients with GBC, those with nodal metastases outside the hepatoduodenal ligament who underwent resection fared no better than stage IV patients treated with primary systemic therapy.³⁷⁴ However, within the regional node basin, the anatomic distribution of nodal metastases appears to have no relationship with survival.³⁷⁵ Therefore, while regional portal lymphadenectomy is standard, aortocaval node involvement is indicative of unresectable disease. Routine sampling of the aortocaval node basin (station 16b1) prior to proceeding with extended resection has been advocated by some, as up to 19% of these patients may be spared a non-beneficial and potentially morbid operation.³⁷⁶

Adjacent Organ Involvement

Invasion of GBC into surrounding organs in close proximity to the non-diseased gallbladder does not preclude extended resection but does portend a poor prognosis. Adjacent organ invasion indicates T3 or T4 disease, depending on the number of organs involved. In the absence of lymph node metastases, these correspond to stage IIIA or IVA, respectively. As an extension of the commonly-adopted en bloc segment IVb/V resection, more extensive hepatic resections for locally advanced GBC may be considered for appropriate surgical candidates with localized disease. Provided that an R0 resection is attained, there appears to be no difference in survival between patients with locally-advanced GBC who require extended hepatectomy versus those undergoing more limited resections.³⁷⁷ More recent series have verified that, for locally-advanced GBC, adjacent organ resection is not itself a negative prognostic indicator; resected patients may anticipate 5-year survival of up to 20%.^368,378 Among these patients, tumor stage (especially nodal status) and the ability to attain an R0 resection are strongly associated with survival.

Not all adjacent organs are created equal. In a retrospective cohort of locally-advanced GBC managed with resection, there were no 3-year survivors among patients who required colectomy, gastrectomy, or pancreaticoduodenectomy.³⁷⁸ Combined hepatopancreatoduodenectomy (HPD) in this context is associated with postoperative morbidity rates well over 50%. In one series, patients who underwent HPD had higher rates of postoperative morbidity and mortality compared to those who required a major hepatectomy.³⁷⁹ Equally important, median survival of patients who underwent HPD was only 10 months, compared to 32 months for those who underwent major hepatectomy. A meta-analysis of 152 patients who underwent HPD for GBC reported an overall postoperative mortality of 10% and 5-year survival of only 10%.³⁸⁰ Because the natural history of GBC is characterized by distant recurrence, neoadjuvant approaches for locally-advanced GBC may help select for patients that could benefit from extended resection.

Preoperative Preparation

The preoperative preparation for resection of GBC depends on the extent of anticipated resection and the clinical scenario. For a T1b-T3 primary tumor that does not require major hepatectomy beyond a segment IVb/V resection, vascular or biliary reconstruction is unlikely. If an extended right or left hepatectomy is required, then optimization of the residual liver remnant is critical. Gallbladder cancers with local invasion into the biliary tree can present with obstructive jaundice. Because of the high rate of positive biliary margins in these cases³⁶⁸, careful planning of resection and reconstruction is facilitated by MRCP. Similar to other hepatopancreatobdiary cancers, patients without obstructive jaundice should not undergo routine biliary stenting, as stent-related complications occur in >20% of cases and may delay definitive therapy.^381,382 If a decompressive stent is required due to biliary obstruction, a large-bore (≥9 Fr) stent that does not traverse the sphincter of Oddi is preferable as these have a lower risk of re-obstruction.^383,384

Except for early-stage GBC’s detected incidentally following cholecystectomy, most cases of suspected GBC proceed to surgery without a tissue diagnosis. Endoscopic cannulation of the biliary tree solely for the purpose of obtaining bile cytology is not indicated but may be pursued as an adjunct if endoscopic stenting is required. While transpapillary cytology possesses a specificity approximating 100%, its moderate sensitivity (44-78%) means that a negative cytology result does not rule out GBC.^385,386

Xanthogranulomatous cholecystitis is particularly difficult to differentiate from GBC on preoperative imaging. For these equivocal cases, a tissue diagnosis prior to committing to an extended resection may be attained through either endoscopic ultrasound with fine-needle aspiration (EUS-FNA) or intraoperative biopsy with frozen section. In experienced hands, the accuracy of EUS-FNA for diagnosing gallbladder pathology approaches 80%³⁸⁷, however, sampling adequacy is user-dependent. While anecdotal reports of needle-tract seeding exist, the risk of isolated needle tract recurrence without systemic disease is low.^388,389 Nevertheless, luminal fluid assessments have shown that malignant cell translocation can occur following EUS-FNA in a minority of cases.³⁹⁰ By contrast, intraoperative core needle biopsy may be obtained through a minimalist incision, is highly accurate and avoids a second general anesthetic exposure.

Operative Approach

Anatomy

Anatomic considerations of the gallbladder, its lymphatic drainage, and surrounding organs dictate the approach for extended resection. The gallbladder rests at the junction of segments IVb and V, along the plane of division between the left and right lobes of the liver. The cystic plate is a fibrous lining between the liver and the muscularis layer of the gallbladder wall. There is no peritoneal lining between the gallbladder and the liver parenchyma. During cholecystectomy for benign disease, to avoid hepatic parenchymal injury, the plane of dissection is between the cystic plate and the gallbladder muscularis. However, the gallbladder wall is characterized by a narrow lamina propria and only one muscular layer. In the setting of GBC, penetration into the muscularis exposes the gallbladder’s lymphatic system and provides opportunities for malignant dissemination. Tumors in the infundibulum or neck of the gallbladder frequently involve the cystic duct or common bile duct and are more likely to result in biliary obstruction than tumors of the gallbladder fundus. While lymphatic drainage is variable, the primary lymphatic basins of the gallbladder are the cystic and choledochal nodes, followed by nodes of the retroportal and posterior-superior pancreaticoduodenal regions. Downstream drainage leads to the celiac, superior mesenteric, and aortocaval node basins.³⁹¹

Extended Resection

The goal of extended resection is to obtain an RO margin. Classically, for GBC with a primary stage of T lb-T2, this entails en bloc resection of the gallbladder with segments IVb and V of the liver. When GBC is discovered incidentally following cholecystectomy, the likelihood of residual disease in the liver bed increases with increasing T-stage, ranging from 0% to >50%.^392,393 Thus, in the absence of an in-situ gallbladder, a second-stage resection of the liver bed within segments IVb and V is recommended for all tumors beyond T1a. Laparoscopic cholecystectomy is curative in >90% of patients who present with T1a GBC³⁹⁴, and a meta-analysis comparing simple cholecystectomy and extended resection for early-stage GBC noted worse survival in patients who underwent extended resection for T1a GBC³⁹⁵. Management of T1b tumors is more controversial. Although laparoscopic cholecystectomy may be curative in >85% of patients with T1b GBC³⁹⁶, retrospective series have found lymph node metastases in >12% of patients with T1b tumors.³⁹² Due to its proclivity for distant metastasis, extended resection has the greatest likelihood of impacting survival for patients that have a high risk for residual locoregional disease after cholecystectomy, but low risk for concurrent distant disease—i.e., T2-T3 patients. This philosophy is best supported by data from a large Japanese registry of patients who underwent laparoscopic cholecystectomy for GBC. Only among those with T2-T3 tumors was subsequent extended resection associated with better survival.³³⁴ A large retrospective German registry review also noted an association between extended resection and survival for patients with T2 cancer.³⁹⁷ In an analysis of the Surveillance, Epidemiology, and End Results (SEER) registry, Coburn and colleagues noted that, even though en bloc resection was associated with greater survival for early-stage tumors, only a minority of patients with GBC ultimately undergo en bloc hepatic resection.³⁹⁸

Despite these findings, the selection process for identifying patients appropriate for extended resection remains imperfect. A collaborative study between the United States and Chile reported significantly worse survival among patients with residual disease relative to those without residual disease. This relationship persisted even when the residual disease group was restricted to those who were NO and had an R0 re-resection.³⁹⁹ In a retrospective cohort from Memorial Sloan Kettering, residual disease at any site was associated with worse disease-free survival and disease-specific survival.⁴⁰⁰ In a multi-institutional dataset of 217 patients who underwent extended resection for GBC, residual disease was strongly associated with recurrence. However, nearly 1/3 of patients who underwent an R0 re-resection still recurred, and an R0 margin was not associated with likelihood of recurrence. In the past, empirical major hepatectomy (i.e., extended right hepatectomy) was commonly performed for GBC. However, more recent data suggest that extended hepatectomy beyond what is necessary to achieve an R0 margin is not associated with improved survival and carries untoward morbidity.³⁶⁸ As such, formal IVb/V segmentectomy and non-anatomic resections of the gallbladder bed are both viable options for extended resection of T1b-T2 tumors, and some T3 tumors. The optimal timing of a second-stage extended resection following laparoscopic cholecystectomy is between 4-8 weeks⁴⁰¹, allowing enough time for postoperative inflammation and fibrosis to subside while avoiding unnecessary delay before definitive therapy.

The main theoretical advantages of anatomic segmentectomy is reduced blood loss and lower risk of postoperative bile leak. Potential oncologic benefits such as inclusion of microscopic tumor deposits and avoidance of microscopically-positive margins are less relevant in GBC as the plane of tumor invasion is relatively predictable. The anatomic approach to segment IVb involves inflow control from within the liver substance, with initial parenchymal division to the right of the umbilical fissure. Alternatively, IVb pedicles can be accessed from within the umbilical fissure and controlled prior to parenchymal division. Further dissection toward the right portal pedicle will allow ligation of the middle hepatic vein. The segment V inflow pedicles, which are usually multiple and lack a single dominant branch, should be carefully dissected. Care should be taken to avoid injury to the main right anterior pedicle and branches supplying segment VIII. As the transection planes approach the base of the cystic plate, one must stay above the structures of the hepatic hilum. In a non-anatomic resection, a minimum 2 cm gross margin is marked surrounding the cystic plate, and injury to the distal middle hepatic vein, which often abuts the gallbladder fossa, should be avoided.

Resection of the primary tumor is performed in conjunction with regional lymphadenectomy of the primary and secondary lymphatic basins for accurate staging (cystic, choledochal, retroportal, and posterior-superior pancreaticoduodenal). Roughly 1/3 of patients who undergo extended resection for GBC possess lymph node metastases, and the prognostic value of nodal staging may be improved by the resection of at least 6 regional nodes.⁴⁰² Unfortunately, adequate lymphadenectomy for early-stage GBC is relatively uncommon at low-volume centers³⁹⁸, particularly if a second-stage resection is not pursued for incidentally-discovered cases.

Bile Duct Resection

To ensure an RO resection, intraoperative frozen section of the cystic duct margin should be submitted in all cases, but in particular for tumors involving the gallbladder neck or infundibulum. If the base of the cystic duct is grossly or microscopically involved with tumor, further resection of the cystic duct or extension to include an extrahepatic bile duct resection should be performed with the distal transection plane at the level of the duodenum. This allows all node-bearing tissue surrounding the portal structures to be swept up with the specimen prior to reconstruction. However, empiric bile duct resection in the absence of tumor invasion is not associated with long-term outcomes.^368,393 Moreover, bile duct resection for the sake of increasing node harvest is also unsubstantiated, as common bile duct excision does not increase overall lymph node yield and increases the morbidity of the operation.³⁹²

Port Site Resection

At the height of enthusiasm for laparoscopic surgery, an international survey of surgeons with a combined experience of 409 incidental GBC’s resected through laparoscopic cholecystectomy reported a 17% incidence of port-site recurrence.⁴⁰³ A subsequent systematic review of case reports estimated that the incidence of port-site recurrences is 14%, occurring at a median of 7 months following laparoscopic cholecystectomy.⁴⁰⁴ However, these findings were quoted from an era preceding laparoscopic extraction bags, and it has since become clear that extraction sites were at higher-risk for port-site metastases than non-extraction sites.⁴⁰⁵ Nevertheless, these findings prompted two decades of debate regarding the utility of routine port-site resection for incidentally-discovered GBC.

More recent data suggest that the incidence of port-site metastases is closer to 10%⁴⁰⁶, and that the incidence of clinically-relevant port-site metastases is likely much lower. In a multicenter French series of 54 patients who underwent port-site excision, only one patient (2%) with a T3 tumor was found to have a port-site metastasis on resection, and this patient died of disseminated peritoneal metastases.⁴⁰⁷ A series of 69 patients who underwent port-site excision from Memorial Sloan Kettering Cancer Center (MSKCC) reported that 19% of excisions harbored metastases, however, these were limited to patients with T2-T3 tumors and port-site resection was not associated with survival. Furthermore, nearly all patients with port site metastases harbored underlying peritoneal metastases.⁴⁰⁸ Corroborating this finding is a multi-institutional cohort of 193 patients in which 47 underwent port site resection. Although nearly one-fourth of patients harbored port-site metastases, port-site resection was not associated with survival.⁴⁰⁵ These findings reinforce the philosophy that, when a cancer’s prognosis is determined predominantly by its predilection for distant metastasis, aggressive interventions directed at local disease control are unlikely to impact long-term outcomes. In the current era, routine port-site excision is not recommended, but may be considered for patients with known gallbladder spillage or extraction bag rupture during laparoscopic cholecystectomy. Unfortunately, the presence of port site metastases is a sign of disseminated peritoneal spread.

Surgical Morbidity

Postoperative morbidity and mortality following resection for GBC are primarily related to resection of adjacent organs and the biliary tree. In experienced hands, en bloc resection of segments IVb and V without major hepatectomy or biliary reconstruction is a low-morbidity operation. In a series of patients who underwent extended resection for GBC at MSKCC, postoperative mortality was 5% and morbidity was 53%. Half of these complications were grade 3-4, and morbidity rates were much higher when resection included major hepatectomy or bile duct excision. All mortalities occurred among patients who underwent a major hepatic resection.³⁶⁸ The most common complications were intra-abdominal abscess and bile leak, nearly all related to patients who underwent bile duct resection. It is not surprising that patients with GBC who present with jaundice experience three-fold greater risk of postoperative complications.³⁶⁹ While en bloc resection of segments IVb and V rarely, if ever, results in hepatic insufficiency, patients requiring a major hepatectomy due to involvement of major inflow vessels can develop liver failure if baseline hepatic function and remnant volume are not carefully considered. For locally-advanced GBC (T3-T4), morbidity and mortality rates as high as 63% and 11% have been reported.³⁷⁸ One of the more extreme versions of adjacent organ resection is HPD, which expands the range of potential complications to include pancreatic fistula and delayed gastric emptying. A recent meta-analysis encompassing nearly 400 patients who underwent HPD reported in-hospital mortality of over 10% and overall morbidity of nearly 80%.³⁸⁰

Prognosis and Adjuvant Therapy

Early reports of GBC depicted a cancer that was almost uniformly fatal, with less than 5% of patients eligible for curative resection and median survival of less than 5 months.^409,410 With the popularization of laparoscopic cholecystectomy, the majority of patients with GBC are now diagnosed incidentally and at earlier stages. Based on SEER data, while stages I and II GBC have roughly equivalent long-term survival, there is a sharp drop in prognosis once the tumor invades beyond the serosa.⁴¹¹ In the modern era, roughly 50% of patients with early-stage (I-II) GBC and less than 10% of those with late-stage (III-IV) disease survive beyond 5 years.⁴¹² In a recent validation study of the AJCC staging system, survival of all surgically-treated patients was 22% at 5 years, with a median survival of only 8 months for locally-advanced GBC.⁴¹³

Although surgery is the only curative therapy for GBC, poor long-term survival among surgically-treated patients indicates a persistent need for adjuvant and neoadjuvant options (Table 10). Early experience with post-resection mitomycin C and 5-fluorouracil (5-FU) for resected gallbladder, bile duct, and pancreatic carcinomas demonstrated significant improvement in 5-year overall survival of 26% vs 14% for GBC.¹³⁵ However, these data were derived from an unplanned post-hoc subset, and findings were only significant on per-protocol analysis. Further, more than half of the patients in this study underwent non-curative resections, and more than half had stage IV disease. Thus, the validity of these results in a true adjuvant setting remains unclear. The PRODIGE 12/ACCORD 18 trial randomized patients to adjuvant gemcitabine and oxaliplatin or observation following R0 or R1 resection for biliary tract cancers. The trial enrolled 194 patients, including 38 patients with GBC. The trial was negative overall. On subgroup analysis, patients with GBC who received adjuvant treatment experienced worse disease-free and overall survival than those who did not.¹⁴¹ Recently, the BILCAP trial randomized patients with resected biliary tract cancers (including 79 with GBC) to adjuvant capecitabine or observation. Reminiscent of the prior outcomes with adjuvant 5-FU, capecitabine was associated with improved survival across the full study population only via per-protocol analysis.¹⁴⁰ There was no significant difference between arms in the GBC subgroup on intent-to-treat analysis. Nevertheless, based on these results, capecitabine has become the current standard of care in the adjuvant treatment of GBC. The ongoing ACTICCA-1 trial, originally designed to compare gemcitabine-cisplatin versus observation, was modified to incorporate capecitabine as the control.²¹⁶

Table 10:

Gallbladder cancer adjuvant therapy phase III trials

Trial	Year	N	Patients	Treatment Arms	Principal Findings
Takada, et al.	2002	112	Stage II-IV (R0-2)	Arm 1: Periop MMC + Adjuvant 5-FU Arm 2: Obs	Favors adjuvant (p < 0.05)* Arm 1: 5Y OS 26% , DFS 20.3% Arm 2: 5Y OS 14.4%, DFS 11.6%
PRODIGE 12	2019	38	Stage I-III (R0-1)	Arm 1: Adjuvant gemcitabine-oxaliplatin Arm 2: Obs	Favors observation (p < 0.05) HR 3.4 (OS) HR 2.6 (DFS)
BILCAP	2019	79	Stage II-III (R0-1)	Arm 1 : Adjuvant capecitabine Arm 2: Obs	No difference HR 0.84 (OS)
ACTICCA-1	2015		Stage I-III (R0-1)	Arm 1: Adjuvant gemcitabine-cisplatin Arm 2: Obs (initial), capecitabine (revised)	Ongoing

The natural history of GBC is characterized by a proclivity for distant metastases, and only 15% of recurrences are limited to locoregional sites²¹⁴ (Table 11). As such, adjuvant radiation therapy is not routinely indicated. In a multi-institutional cohort of 112 patients who underwent R0 or R1 resection for GBC, 61% received adjuvant radiation. There was no association between radiation and survival, and although radiation was associated with lower locoregional recurrence, the majority of these recurrences coincided with distant metastases.⁴¹⁴

Table 11:

Recurrence following resection of gallbladder carcinoma

Author	Year	N	Stage	Recurrence Rate	Time to Recurrence (months)	Local Recurrence	Distant Recurrence	Local + Distant Recurrence
Jarnagin	2003	80	Stage I-III	66.3%	11.5	15.0%	72.5%	12.5%
Yagi	2006	63	Stage I-IV	41.3%	NA	23.1%	76.9%	0.0%
Park	2007	77	Stage II	56.7%	21.1	2.6%	97.4%	0.0%
Choi	2010	83	T2	57.8%	32.5	NA	NA	NA
Kim	2010	166	Stage I-III	31.9%	8.0	37.7%	39.6%	22.6%
Margonis	2016	217	Stage I-III	35.0%	9.5	15.8%	65.8%	18.4%
Jung	2016	88	T2	18.2%	NA	12.5%	56.3%	31.3%

Because most patients who recur do so at distant sites, there is substantial motivation to identify neoadjuvant options for GBC. For initially unresectable cases, neoadjuvant therapy may induce a partial response and improve surgical candidacy. For resectable, stage III patients, response to neoadjuvant agents can guide adjuvant therapy, although this is an unproven strategy. Finally, patients who progress during neoadjuvant treatment avoid an unnecessary operation. Due to moderate efficacy in the locally-advanced and metastatic settings through the ABC-02 trial¹⁴², cisplatin plus gemcitabine has thus far been the most commonly-adopted neoadjuvant regimen for GBC. In a retrospective cohort of patients with locally-advanced or clinically node-positive GBC initially treated with gemcitabine-based therapy, 23% progressed, 51% had stable disease, and 26% experienced partial response. Ultimately, only 14% of patients underwent definitive resection. Overall survival within the resected subset was 51 months, approximating that of the adjuvant group in the BILCAP trial.⁴¹⁵ In a meta-analysis encompassing 434 patients with stage III-IV GBC treated predominantly with neoadjuvant gemcitabine and cisplatin, two-thirds of patients experienced objective response or stable disease, and one-third ultimately underwent R0 resection. Among resected patients, median survival ranged from 18.5 to 50.1 months.⁴¹⁶

Conclusion

Gallbladder cancer is the most common biliary tract cancer and is characterized by an aggressive natural history. While T1a tumors are adequately treated with laparoscopic cholecystectomy, resection of the hepatic bed and regional lymphadenectomy is indicated for tumors invading beyond the lamina propria (T1b+). While adjacent organ resection, including major hepatectomy and bile duct excision, may be considered for appropriate surgical candidates, overly aggressive attempts at locoregional control such as hepatopancreatoduodenectomy are unlikely to confer benefit given the propensity of GBC for distant recurrence. As such, further progress in long-term survival for GBC will ultimately depend on advancements in adjuvant and neoadjuvant strategies.

Figure 3.

Bismuth-Corlette classification scheme. From Blumgart’s surgery of the liver, biliary tract, and pancreas.

Figure 9.

Blumgart Figure 49.5 (Part V: Biliary Tract Disease, II – neoplastic, C – malignant tumors, Ch 49 GBC). Computerized tomography showing a locally-advanced gallbladder carcinoma extending into segments 4B and 5 of the liver and abutting the duodenum.