Primary malignant tumours of the liver

Published on 11/04/2015 by admin

Filed under Surgery

Last modified 22/04/2025

Print this page

This article have been viewed 1986 times

Primary malignant tumours of the liver

Chetana Lim and Olivier Farges

Introduction

With the exception of hepatocellular carcinoma (HCC), which is one of the most common malignancies, primary tumours of the liver are relatively rare in adults. HCC arises from hepatocytes, and cirrhosis is its main aetiological factor. This tumour remains a subject of considerable interest due to its rising incidence and the development of innovative treatments. Intrahepatic cholangiocarcinoma (ICCA) arises from the peripheral intrahepatic biliary radicles and other rare primary tumours arise from mesodermal cells, and include angiosarcoma, epithelioid haemangio-endothelioma and sarcoma.

Hepatocellular carcinoma

HCC accounts for 90% of all primary liver malignancy and its incidence continues to increase. It is the sixth most common neoplasm, accounting for more than 5% of all cancers, and is the third most common cause of cancer-related death. The International Agency for Research on Cancer has estimated in 2008 through its GLOBOCAN series that primary liver cancer caused more than 690 000 deaths worldwide, similar to colon or rectal cancer.¹

HCC usually occurs in male patients, and cirrhosis precedes its development in most cases. Due to better medical management of cirrhosis, survival of cirrhotic patients has steadily increased in recent years, resulting in a greater risk of developing HCC. Cohort studies have reported that in patients with HCC, the death rate due to cancer is 50–60%, while hepatic failure and gastrointestinal bleeding are responsible for approximately 30% and 10% of the deaths, respectively. HCC may now be identified at an early stage, particularly through the screening of high-risk patients.

Control of HCC nodules may be achieved successfully by surgical resection and by percutaneous treatment. The indications for these therapies depend on the morphological features of the tumour and the functional status of the non-tumorous liver. Unfortunately, these treatments are associated with a high incidence of tumour recurrence due to the persistence of the underlying cirrhosis, which represents a preneoplastic condition. Liver transplantation may seem a logical alternative treatment but has its own limitations, including tumour recurrence, the limited availability of grafts, and cost. The most exciting areas of progress are the control of hepatitis B virus (HBV) or hepatitis C virus (HCV), the prevention of carcinogenesis in patients with chronic liver disease, early radiological screening and the development of medical therapies. In the setting of liver surgery, better liver function assessment and understanding of the segmental liver anatomy with more accurate imaging evaluation are the most important factors that have led to a decrease in postoperative mortality. Active follow-up and treatment of recurrence have also contributed to increased 5-year survival rates as high as 70%.²

Incidence of HCC

The world age-adjusted incidence of HCC in men is 14.9 per 100 000, but has geographical variation related to the prevalence of HBV and HCV infections, which are the two main risk factors worldwide and account for more than three-quarters of all cases (Table 5.1). The incidence may be as high as 99 cases per 100 000 in Mongolian men; other high-rate areas include Senegal, Gambia, South Korea, Hong Kong and Japan. By contrast, North and South America, Northern Europe and Oceania are areas with low incidences (less than 5 cases per 100 000); in these areas, HCV is the main risk factor, together with alcohol abuse, non-alcoholic fatty liver disease and obesity. Southern European countries have intermediate rates.

Table 5.1

Age and prevalence of HBV and HCV among patients with HCC in different geographical areas

The rising incidence of HCC was first documented in the USA, where this doubled between the late 1970s and the early 1990s, reaching 3 cases per 100 000. The recent epidemic of HCV infection probably accounts for a large part of this increase. Alternative explanations include ageing of the population, increased detection, improved survival of cirrhotic patients, and the recent epidemic of obesity and type II diabetes.

It has been estimated that HCV began to infect large numbers of young adults in North America and South and Central Europe in the 1960s and 1970s as a result of intravenous drug use. The virus moved into national blood supplies and circulated until a screening test was developed in 1990, after which rates of new infection decreased dramatically. In Canada, Australia, Japan and various European countries, where HCV infection spread earlier than in the USA, a similar trend was observed but in some countries the incidence of HCC is now decreasing. In the USA, the incidence of HCV-related HCC is still increasing and is projected to peak in 2019 if the risk in HCV-infected persons with fibrosis remains stable.

Cirrhosis of other aetiologies

Primary biliary cirrhosis (PBC) has been considered as a low-risk factor for HCC, not only because of its rare incidence but also because it predominantly affects women (with a sex ratio of 9:1). A recent meta-analysis of 12 studies has reported that PBC is significantly associated with an increased risk of HCC (18.8-fold) compared to the general population.¹¹ However, there were several confounding factors in this meta-analysis, such as advanced histological stage of PBC, history of blood transfusion, and smoking or drinking habits that might be associated with an increased probability for HCC development in PBC patients, or may be directly associated with PBC development. In contrast, HCC development in patients with secondary biliary cirrhosis is exceptional if it even exists.

Autoimmune hepatitis has a low risk of HCC development. Potential reasons are the female predominance and the delayed development of cirrhosis through corticosteroid therapy. HCV infection needs to be ruled out as it may induce autoantibodies. Recent data reported that cirrhosis at presentation is an important prognostic risk factor for HCC. In a prospective multicentre cohort study evaluating 193 Japanese patients with autoimmune hepatitis, seven (3.6%) developed HCC during an 8-year period, all of whom had underlying cirrhosis.¹²

Aflatoxin

Aflatoxin B1 has also long been associated with the development of HCC because areas with a large consumption of this toxin coincide with areas with a high incidence of HCC (Asia and sub-Saharan Africa). Aflatoxin is ingested in food as a result of contamination of imperfectly stored staple crops by Aspergillus flavus. It is thought to induce HCC through mutation of the tumour suppressor gene p53. Although some studies suggest that it is an independent risk factor, others suggest that it could be a co-carcinogen only in patients with HBV infection. HCC in this setting frequently develops in a non-cirrhotic liver.

Metabolic liver diseases

An increased risk of HCC is recognised in some other forms of metabolic liver diseases such as α₁-antitrypsin deficiency, porphyria cutanea tarda, tyrosinaemia and hypercitrullinaemia. Patients with glycogenosis type IV, hereditary fructose intolerance and Wilson disease may also develop HCC, but with a lower risk. There is evidence that iron and copper overload in haemochromatosis and Wilson disease generate, respectively, oxygen/nitrogen species and unsaturated aldehydes that cause mutations in the p53 tumour suppressor gene.

Adenoma, contraceptives and androgens

Like adenoma in other locations, hepatocellular adenomas (HCAs) have a risk of malignant transformation and hepatocyte dysplasia is the intermediate step between HCAs and HCC. A recent systematic review estimated the risk to be 4.2%.¹³ This risk and the treatment strategy to prevent it may, however, be refined.

HCAs are most classical in women of child-bearing age and are associated with the prolonged use of contraceptives and oestrogen treatments. Discontinuation of oral contraceptives does not completely avoid the risk of malignant transformation. Malignancy within HCAs measuring less than 4 cm in diameter is exceptional. There is also recent evidence that HCAs may develop in men, especially if there is a background of a metabolic syndrome. The risk of malignant transformation in men is 50% (10 times higher than in women) and malignancy can occur in HCAs as small as 1 cm.¹⁴ Therefore, whereas resection of HCAs larger than 4 cm is recommended in women, all HCAs irrespective of size should be resected (or ablated) in men.

The number of HCAs does not appear to increase the risk of malignant transformation and, in particular, patients with polyadenomatosis are not at increased risk.14–16

Malignant transformation of HCAs has also been linked to the genotype and phenotype of HCAs. It is more prevalent in telangiectatic or atypical HCAs than in steatotic HCAs. Most importantly, the presence of a β-catenin mutation (observed in approximately 10–15% of HCAs) confers a particularly high risk of malignancy.¹⁷

Malignant transformation of HCAs may also occur within known specific aetiologies, such as with type I glycogenosis, anabolic steroid use, androgen treatments and Fanconi disease. Recreational anabolic steroid use is also known to potentially result in the development of adenoma, and malignant transformation to HCC has been reported.

Pathology of HCC and nodular lesions in chronic liver disease

Preneoplastic lesions are morphologically characterised by dysplastic lesions in the form of microscopic dysplastic foci and macroscopic dysplastic nodules (DNs).

Dysplastic foci are microscopic lesions composed of dysplastic hepatocytes of less than 1 mm in size, and occur in chronic liver disease, particularly in cirrhosis. DNs are defined as a nodular region of less than 2 cm in diameter with dysplasia but without definite histological criteria of malignancy. They are divided into low and high grade depending on the degree of cytological or architectural atypia. Low-grade DNs are approximately 1 cm in diameter, slightly yellowish, and have a very low probability of becoming malignant. High-grade DNs are less common but are typically slightly larger nodules (up to 2 cm) characterised by increased cell density with an irregular thin-trabecular pattern and occasionally unpaired arteries. These are often difficult to differentiate from highly differentiated HCCs. They may contain distinct foci of well-differentiated HCC and are therefore considered as precancerous lesions and become malignant in a third of cases. It must, however, be appreciated that lesions smaller than 2 cm may also represent HCC.

HCCs can be subdivided according to their gross morphology, degree of differentiation, vascularity, presence of a surrounding capsule and presence of vascular invasion. All of these criteria have practical implications.

On gross morphology, HCCs can be solitary or multinodular, consisting of either a collection of discrete lesions in different segments that develop synchronously (multicentric HCC), or as one dominant mass and a number of ‘daughter’ nodules (intrahepatic metastases) located in the adjacent segments. Diffuse HCCs are relatively rare at presentation and consist of poorly defined, widely infiltrative masses that present particular diagnostic challenges on imaging. A third type is the infiltrating HCC, which typically is less differentiated with ill-defined margins.

Microscopically, HCCs exhibit variable degrees of differentiation that are usually stratified into four different histological grades, known as Edmondson grades 1–4, which correspond to well-differentiated, moderately differentiated, poorly differentiated and undifferentiated types. The degree of differentiation typically decreases as the tumour increases in diameter. Very-well-differentiated HCCs can resemble normal hepatocytes and the trabecular structure may reproduce a near normal lobar architecture so that histological diagnosis by biopsy or following resection may be difficult. A number of immunomarkers have been described to selectively identify the malignant nature of these HCCs, not only in resected specimens but also in liver biopsies: glypican 3 (GPC3), heat shock protein 70 (HSP70) and glutamine synthetase (GS). Positive immunomarker staining for any two markers can detect early and well-differentiated HCC in 50–73% of cases, with 100% specificity when the analysis is performed on resected specimens.¹⁸

Vascularisation is a key parameter in differentiating HCC from regenerating nodules. Progression from macroregenerative nodule to low-grade DN, high-grade DN and frank HCC is characterised by loss of visualisation of portal tracts and development of new non-triadal arterial vessels, which become the dominant blood supply in overt HCC lesions. This arterial neoangiogenesis is the landmark pathological feature of HCC diagnosis, and the rationale for chemoembolisation and anti-angiogenic treatment.

A distinct fibrous capsule may surround tumour nodules. This capsule, present in 80% of resected HCCs, has a variable thickness, which may not be complete, and is frequently infiltrated by tumour cells. Capsular microscopic invasion by tumour cells is present in almost one-third of tumours smaller than 2 cm in diameter, as compared with two-thirds of those with a larger diameter.

HCC has a great tendency to spread locally and to invade blood vessels. The rate of portal invasion is higher in the expansive type, in poorly differentiated HCCs and in large tumours. Characteristically, microscopic vascular invasion is seen in 20% of tumours measuring 2 cm in diameter, in 30–60% of cases with nodules measuring 2–5 cm and in up to 60–90% when nodules are more than 5 cm in size. The presence of portal invasion is the most important predictive factor associated with recurrence. The tumour thrombus has its own arterial supply, mainly from the site of the original venous invasion. Once HCC invades the portal vein, tumour thrombi grow rapidly in both directions, and in particular towards the main portal vein. As a consequence, tumour fragments spread throughout the liver as the thrombus crosses segmental branches. Once the tumour thrombus has extended into the main portal vein, there is a high risk of complete thrombosis and increased portal hypertension. This accounts for the frequent presentation with fatal rupture of oesophageal varices, or liver decompensation including ascites (Fig. 5.1), jaundice and encephalopathy. Invasion of hepatic veins is possible, although less frequent. The thrombus eventually extends into the suprahepatic vena cava or the right atrium and is associated with a high risk of lung metastases. Rarely, HCC may invade the biliary tract and give rise to jaundice or haemobilia. Mechanisms of HCC-induced biliary obstruction include:

Figure 5.1 CT scan of a patient with a tumour thrombus originating from an HCC located in the right liver. The thrombus extends in to the main portal vein. Ascites is present.

• intraductal tumour extension;

• obstruction by a fragment of necrotic tumour debris;

• haemorrhage of the tumour resulting in haemobilia;

• metastatic lymph node compression of major bile ducts in the porta hepatis.

The rate of invasion of the portal vein, hepatic vein and bile duct at the time of diagnosis is 15%, 5% and 3%, respectively. However, it is estimated that during the natural history of HCC, approximately 1 in 3 patients will develop portal vein thrombosis.

When present, metastases are most frequently found in the lung. Other locations, in decreasing order of frequency, are: adrenal glands, bones, lymph nodes, meninges, pancreas, brain and kidney. Large tumour size, bilobar disease and poor differentiation are risk factors for metastatic disease.

Clinical presentation

HCC rarely occurs before the age of 40 years and reaches a peak at around 70 years of age. The age-adjusted incidence in women is two to four times less than in men and the difference is most pronounced in medium-risk south European populations and premenopausal women. Reasons for this higher rate in men include differences in exposure to risk factors, higher body mass index and higher levels of androgenic hormones.

There are basically three circumstances of diagnosis: (1) incidental finding during routine screening; (2) incidental finding during investigation of abnormal liver function tests or of another pathological condition; and (3) presence of liver- or cancer-related symptoms, the severity of which depend on the stage of the tumour and the functional status of the non-tumorous liver. In developed countries, a growing number of tumours are discovered incidentally at an asymptomatic stage. As tumours increase in size, they may cause abdominal pain, malaise, weight loss, asthenia, anorexia and fever. These symptoms may be acute as a result of tumour extension or complication.

Spontaneous rupture occurs in 5–15% of patients and is observed particularly in patients with superficial or protruding tumours. The diagnosis should be suspected in patients with known HCC or cirrhosis presenting with acute epigastric pain, as well as in Asian or African men who develop an acute abdomen (Fig. 5.2). Minor rupture manifests as abdominal pain or haemorrhagic ascites, and hypovolaemic shock is only present in about half of the patients. Portal vein invasion may manifest as upper gastrointestinal bleeding or acute ascites, and invasion of hepatic veins or the inferior vena cava may result in pulmonary embolism or sudden death.

Figure 5.2 CT scan of a patient with a ruptured HCC. Note that the rupture is limited at the upper part of the liver. This patient had haemorrhagic ascites.

Clinical symptoms resulting from biliary invasion or haemobilia are present in 2% of patients. Possible paraneoplastic syndromes associated with HCC include polyglobulia, hypercalcaemia and hypoglycaemia. Finally, in patients with underlying liver disease, a sudden onset or worsening ascites or liver decompensation may be the first evidence of HCC formation.

Clinical examination may only reveal large or superficial tumours. There may be clinical signs of cirrhosis, in particular ascites, a collateral circulation, umbilical hernia, hepatomegaly and splenomegaly.

Liver function tests and tumour markers

Liver function tests

Abnormal liver function tests are a non-specific finding and reflect an underlying liver pathology or the presence of a space-occupying lesion. Because most HCCs develop within a cirrhotic liver and since HCCs in normal livers are usually large, entirely normal liver function tests are exceptional. Jaundice is most frequently the result of liver decompensation.

Serum tumour markers

α-Fetoprotein: Serum α-fetoprotein (AFP) is the most widely recognised serum marker of HCC. It is secreted during foetal life but residual levels are very low in adults (0–20 ng/mL). It may increase in patients with an HCC, and serum levels greater than 400 ng/mL can be considered as diagnostic of HCC with 95% confidence. Levels may exceed 10 000 ng/mL in 5–10% of patients with HCC. Very high levels usually correlate with poor differentiation, tumour aggressiveness and vascular invasion. An AFP > 20 ng/mL has a sensitivity of 60% and therefore a surveillance programme using this cut-off value would miss 40% of tumours. If a value of > 200 ng/mL is used, 22% of tumours would be missed. Only 10% of small tumours are associated with raised AFP levels, whereas 30% of patients with chronic active hepatitis without an HCC have a moderately increased AFP. This usually correlates with the degree of histological activity and raised levels of transaminase, and it may therefore fluctuate. Tumours other than HCC can also be associated with increased AFP levels, but these are rare (non-seminal germinal tumours, hepatoid gastric tumours, neuroendocrine tumours).

Others serum tumour markers: Alternative serum markers for HCC, such as des-γ-carboxy prothrombin (DCP) or prothrombin induced by vitamin K absence (PIVKA-II; > 40 mAU/mL) and AFP-L3 (> 15%) have not come into common practice except in Japan, where they are covered under the national health insurance. A prospective study of at-risk patients comparing the accuracy of AFP and DCP in the early detection of HCC showed that the combination of both markers increased the sensitivity from 61% and 74%, for each marker alone, to 91% for both markers combined.¹⁹

Radiological studies

The aims of imaging in the context of HCC are to screen high-risk patients, differentiate HCC from other space-occupying lesions and select the most appropriate treatment.

Differentiation of HCC from other tumours relies on its vascularisation. The most reliable imaging features of an HCC are the presence of hyperarterialisation of the nodule in the early (arterial) phase and washout during the portal or late phase following injection (the tumour becomes hypovascular compared to the adjacent parenchyma). By definition, the term ‘washout’ can only be applied to tumours that are hypervascular in the arterial phase (although this may be very transient).

Critical in choosing the most appropriate treatment are the number of lesions, their size and extent, and the presence of daughter nodules, vascular invasion, extrahepatic spread and underlying liver disease.

These aims may be achieved by ultrasound (US), contrast-enhanced US, computed tomography (CT), magnetic resonance imaging (MRI), angiography or a combination of these.

Imaging aims to:

• screen patients for the development of HCC, and this is best achieved by US;

• differentiate potential HCC from other tumours, and this is best achieved by demonstrating the presence of hypervascularisation during the arterial phase and washout during the portal or late phase.

Ultrasound

US is the first-line investigation for screening because of its low cost, widespread availability and high sensitivity in identifying a focal liver mass. In experienced hands, US may identify 85–95% of lesions measuring 3–5 cm in diameter and 60–80% of lesions measuring 1 cm. Differences in accuracy worldwide may be explained by steatosis rates, heterogeneity of the liver disease and in operator variability. Typically, small HCCs are hypoechoic and homogeneous and cannot be differentiated from regenerating or dysplastic nodules. With increasing size, they may become hypo- or hyperechoic but most importantly heterogeneous. A hypoechoic peripheral rim corresponds to the capsule. The infiltrating type is usually very difficult to identify in a grossly heterogeneous cirrhotic liver. Besides echogenicity, the accuracy of US depends on the dimension and location of the tumour, as well as operator experience. A 1-cm-diameter tumour can be visualised if it is deeply located, whereas the same lesion located on the surface of the liver can be missed. Similarly, tumours located in the upper liver segments or on the edge of the left lateral segment may be missed. Tumours detected at an advanced stage despite surveillance are frequently located at one of these two sites. Obesity may also prevent accurate assessment of the liver (thickened abdominal wall or steatotic liver). Doppler US may demonstrate a feeding artery and/or draining veins. US is also accurate in identifying vascular or biliary invasion and indirect evidence of cirrhosis such as segmental atrophy, splenomegaly, ascites or collateral veins. Tumour thrombosis is associated with enlargement of the vascular lumen, and duplex Doppler may detect an arterial signal. Contrast US is addressed below.

Computed tomography

CT is more accurate than US in identifying HCCs and their lobar or segmental distribution, particularly with the development of helical and multislice spiral scanners. Spiral CT is undertaken without contrast and during arterial (25–50 s), portal (60–65 s) and equilibrium (130–180 s) phases after contrast administration. In addition, it is useful for identifying features of underlying cirrhosis, accurately measuring liver and tumour volumes, and assessing extrahepatic tumour spread. HCCs are usually hypodense and spontaneous hyperdensity is usually associated with iron overload or fatty infiltration, which is seen in 2–20% of patients. Specific features are early uptake of contrast and a mosaic shape pattern. During the portal phase, the density diminishes sharply and results in washout (tumour is hypodense compared to adjacent parenchyma) during the late phase (Fig. 5.3). HCCs may show variable vascularity depending on tumour grade and some are poorly vascularised. The capsule, when present, is best seen during the portal or late phase as an enhanced thickening at the periphery (delayed vascular enhancement is characteristic of fibrosis). Vascular invasion of segmental branches may also be identified. Intratumoral arterioportal fistula may develop and present as early enhancement of portal branches or as a triangular area distal to the tumour with contrast enhancement different from the adjacent parenchyma. Nonetheless, such fistulas are seen frequently in cirrhotic patients without HCC as infracentimetric hypervascular subcapsular lesions.

Figure 5.3 Typical vascular kinetics of an HCC. There is early uptake of contrast at the arterial phase (a) that becomes isodense during the portal phase (b), with washout during the late phase (c).

Magnetic resonance imaging

MRI tends to be more accurate than other imaging techniques in differentiating HCC from other liver tumours, especially those > 2 cm in diameter. As for CT, the technique of MRI should be accurate with T1- and T2-weighted images and with early, intermediate and late phases following contrast injection of gadolinium. The characteristics of an HCC are the mosaic shape structure and the presence of a capsule. Tumours are hypointense on T1-weighted images and hyperintense on T2-weighted images, but these characteristics are present in only 54% of patients; 16% of HCCs demonstrate hypointensity on both T1 and T2 images. Hyperintensity on T1-weighted images is also possible, and associated with fatty, copper or glycogen infiltration of the tumour. The kinetics of vascular enhancement following injection of contrast are the same as during CT, with early uptake and late washout. Recently, liver-specific magnetic resonance contrast medium such as Gd-EOB-DTPA that accumulates in Kuppfer cells (due to phagocytosis) or in hepatic cells has increased the accuracy of MRI, but has not yet come into common practice except in Eastern countries.

Contrast-enhanced ultrasound

Contrast-enhanced US (CEUS) is the most recent technique to assess vascularisation of tumours. A contrast agent (stabilised microbubbles) is administered intravenously via a bolus injection followed by saline flush. Enhancement patterns are typically described during the arterial (10–20 s postinjection), portal venous (30–80 s) and late phase (120–360 s). Whereas US microbubbles are confined to the vascular spaces, contrast agents for CT and MRI are rapidly cleared from the blood into the extracellular space. The sensitivity of CEUS to detect arterial enhancement is greater than that of CT or MRI because of the continuous monitoring of the images. Washout is slower for well-differentiated than for poorly differentiated tumours. However, it is subject to the same limitations as other US modes: if the baseline scan is unsatisfactory, the CEUS study will also be unsatisfactory. The advent of the second-generation US contrast agent Sonazoid, approved exclusively in Japan in 2007, has made Sonazoid-CEUS more effective for screening and staging than CEUS using other vascular agents such as SonoVue. Sonazoid contrast agent is taken up by Kupffer cells in the postvascular phase or Kupffer phase (starting 10 min postinjection) and provides extremely stable Kupffer images suitable for repeated scanning from 10 to about 120 min after injection.²⁰

Other imaging

Angiography: Although the diagnostic usefulness of angiography has been considerably reduced, it is still widely used as part of arterial chemoembolisation. Arteriography shows early vascular uptake (blush) and, if used, lipiodol injection is retained selectively for a prolonged period by the tumour. On subsequent CT, the retained radiodense lipiodol reveals the tumour as a high-density area. Uptake within the liver is not specific for HCC, since all hypervascular liver tumours, including focal nodular hyperplasia, adenoma, angioma and metastases, will retain Lipiodol. False-negative results may be observed with an avascular, necrotic or fibrotic HCC.

Positron emission tomography: The contribution of fluorodeoxyglucose (FDG)-positron emission tomography (PET) in the diagnosis of HCC remains limited because of low sensitivity (60%). FDG-PET can detect poorly differentiated but not-well differentiated HCC due to similarities between the metabolism of FDG in normal hepatocytes and hepatocellular cells. Recently, [¹⁸ F]fluorocholine (FCH), a PET tracer of lipid metabolism, has been shown to be significantly more sensitive than [¹⁸ F]FDG at detecting HCC, particularly in well-differentiated tumours.²¹ Interestingly, in metastatic HCC, FDG-PET has a high sensitivity and is more suitable for the detection of bone metastases than CT or bone scintigraphy.²²

Accuracy of imaging techniques

CT and MRI with contrast enhancement have the highest diagnostic accuracy (> 80%) and the techniques can be combined. However, pathological analysis of explanted livers from transplant patients shows that both techniques lose accuracy in assessing extension of the disease. For any technique, additional intrahepatic tumours, especially those less than 1 cm, are not diagnosed preoperatively in 30% of cases. MRI angiography with 2-mm sections is currently considered the most accurate technique, with a sensitivity of 100% for nodules more than 20 mm, 89% for nodules 10–20 mm and 34% for nodules < 10 mm.

Requirement for and reliability of histological assessment

Pathological confirmation of HCC can be obtained by cytology, histology or a combination of these with increasing accuracy. The accuracy of pathological assessment is increased if a sample of non-tumorous tissue is available for comparison. Liver biopsy is limited by the potential for haemorrhage and pain, and may occasionally be responsible for neoplastic seeding and vascular spread. The reported incidence of needle tract seeding is 1–5%. Tumour involvement is generally limited to subcutaneous tissues, has a slow progression and it is possible to perform local excision without apparent impact on survival. Even if the false-positive rate is low, the risk of needle tract seeding is balanced by the risk of pursuing an aggressive treatment such as resection or transplantation in a patient without malignancy. Every attempt should be made not to puncture the nodule directly but to access the nodule through a thick area of normal liver. As described below, several studies have shown that expert pathological diagnosis of HCC can be reinforced by staining for GPC3, HSP70 and GS, particularly in biopsies of small lesions that are not clearly HCC.

There is a significant false-negative rate for fine-needle biopsy, especially in small lesions, lesions that are difficult to access or those developing on the background of a multinodular parenchyma. A negative result should therefore never rule out malignancy.

Diagnosis of HCC

The standard for the diagnosis of HCC is histology. This is particularly true for tumours measuring 3 cm or less or when active treatment is required. Ideally, these samples should be associated with a biopsy of the non-tumorous liver and be made available for research, with patient consent. Non-invasive diagnosis (using radiological imaging alone) requires rigorous technique and interpretation.

The first attempt to standardise the diagnostic criteria was in 2000 by the European Association for the Study of Liver Disease (EASLD). Since the last publication of the AASLD practice guidelines for the management of HCC in 2005, several studies have reported that AFP determination lacks adequate sensitivity and specificity for effective surveillance and diagnosis.¹⁹ The advocated strategy in the 2011 updated version of the diagnostic criteria²³ was based on imaging techniques and/or biopsy as follows:

• For nodules less than 1 cm found on US, it was considered that other imaging techniques would be unlikely to reliably confirm the diagnosis. Since the accuracy of liver biopsy for such small lesions and the likelihood of HCC are low, it was felt reasonable to repeat an US at 3- to 6-month intervals until the lesion disappeared, enlarged or displayed characteristics of HCC. If there has been no growth over a period of up to 2 years, routine surveillance can be resumed.

• For nodules larger than 1 cm found on US screening of a cirrhotic liver, diagnosis of HCC can be established by one contrast-enhanced imaging technique (multidetector CT or dynamic contrast-enhanced MRI). The specific imaging pattern of HCC is defined by intense contrast uptake during the arterial phase followed by contrast washout during the venous or delayed phases. The value of these non-invasive criteria for HCC in cirrhosis has been confirmed prospectively.24–26 These typical imaging features have a specificity and predictive positive value of approximately 100% and sensitivity of 71%.

• If the findings are not characteristic or the vascular features are not typical, and in other clinical settings (e.g. absence of cirrhosis), a diagnostic biopsy was recommended, although it was acknowledged that a negative biopsy did not exclude the diagnosis.

Subsequent to these recommendations, several studies have reported that CEUS may give a false-positive HCC diagnosis and cannot selectively differentiate intrahepatic cholangiocarcinoma from HCC. This technique has therefore been withdrawn from the diagnostic algorithm proposed by the AASLD.

Current guidelines ²³ for the positive diagnosis of HCC are:

• For nodules > 1 cm with cirrhosis, early uptake and delayed washout on a single dynamic imaging study (triphasic CT or MRI with gadolinium) is considered characteristic of HCC.

• For nodules > 1 cm without cirrhosis or if the vascular features are not typical, a diagnostic biopsy is recommended.

• Biopsy of small lesions should be evaluated by expert pathologists. Tissue that is not clearly an HCC should be stained with CD34, CK7, GPC3, HSP70 and GS to improve diagnostic accuracy.

• If the biopsy is negative for HCC, it is recommended that the lesion should be re-evaluated using US at 3- to 6-month intervals. If the lesion enlarges but remains atypical for HCC, a repeat biopsy is recommended.

Natural history of HCC and staging systems

Traditionally, the natural history of HCC is considered to be particularly grim, with a median survival of 6 months in symptomatic patients. However, the 3-year survival of asymptomatic untreated patients who are not end-stage at the time of presentation may be as high as 50%. These observations have important implications for patients with an HCC diagnosed at an early stage, particularly in patients with preserved liver function.

The aim of staging systems is to predict outcome. This can either be used to anticipate prognosis or, more recently, for selection of treatment. Survival of patients with HCC is mainly influenced by the morphological spread of tumour, the presence and severity of cancer-related symptoms, and the severity and evolution of the underlying cirrhosis. The most recent systems attempt to integrate all three groups of parameters.

• Although staging was assessed initially by the TNM classification, the pathological staging of HCC has evolved in Eastern (Liver Cancer Study Group of Japan) and Western (American Joint Committee on Cancer, International Union Against Cancer) countries. These take into account the number of tumours, vascular invasion and tumour size (Table 5.2). A limitation is that they are based on pathological findings and can only be applied accurately (retrospectively) in operated patients.

Table 5.2

Comparison of the tumour (T) staging in the Liver Cancer Study Group of Japan (LCSGJ) and American Joint Committee on Cancer (AJCC) staging systems

AJCC/UICC and LCSGJ TNM classification of HCC.

• Liver damage induced by underlying liver disease has traditionally been assessed by the Child–Pugh score. This was, however, designed to assess the functional reserve of cirrhotic patients undergoing portocaval shunt surgery and is not entirely appropriate for HCC patients in whom therapeutic options may include liver transplantation and liver resection.

Several groups have attempted to combine these features within integrated staging systems. There are currently six such systems, designated as the CLIP (from Italy), GRETCH (from France), BCLC (from Spain), CUPI (from China), JSS and JIS (from Japan) scores. It is beyond the scope of this chapter to detail all of them (Table 5.3). It should be appreciated that these scores have been computed retrospectively by multivariate analysis of a specific patient population and not all have been externally validated.

Table 5.3

Main variables retained in prognostic models

The numbers refer to the score given to each variable. A total score is obtained by adding each individual score. In the CLIP score, the median survivals according to the score in the initial ³⁰ and prospective validations³² were: score 0, 36–42 months; score 1, 22–32 months; score 2, 8–16 months; score 3, 4–7 months, score 4 or above, 1–3 months.

^*2N = twice normal.

Currently, the BCLC system is widely accepted, as it includes variables linked to tumour stage and function, physical status and cancer-related symptoms, and it combines each stage (very early stage: 0; early stage: A; intermediate stage: B; advanced stage: C; and end stage: D) with a treatment algorithm. It has been externally validated and has been recently supported by both the AASLD and EASL. However, validation in Eastern countries has not been achieved to date. One area of concern regards the definition of intermediate and advanced HCC; most Asian experts agree that early stage means that HCC can be controlled by curative treatment, but advanced stage (which includes portal vein invasion and distant metastatis) is hard to define in the BCLC system as it can be divided into two other different groups: locally advanced with portal vein invasion and advanced with extrahepatic metastasis. At present, most Asian countries have their own HCC staging system with different constituent variables.

Screening for HCC

Screening is used routinely in countries where effective therapeutic interventions are available. HCC fulfils most of the criteria required for a surveillance or screening programme to be justified. HCC is common in highly endemic areas (and its incidence is growing in others) and it is associated with a high mortality. Furthermore, survival is extremely poor by the time patients present with symptoms related to the tumour, and the population at risk is clearly defined (in particular – but not exclusively – patients with HCV- and HBV-related cirrhosis, especially when they are male and over 60 years of age). Acceptable screening tests with low morbidity and high efficacy exist that allow the tumour to be recognised in the latent/early stage. Finally, effective treatments exist in selected patients.

The two most common tests used for screening of HCC are US and serum AFP measurements, although many clinicians consider the latter investigation to be of little value for screening. However, a progressive increase of AFP in patients who have a normal AFP at baseline should prompt a CT or MRI scan if US is negative. It should be underlined that US is most difficult in obese patients with fatty liver disease and cirrhosis, but no alternative strategy for surveillance has been adequately tested.

No clear evidence is available to determine the optimal interval for periodic screening. Tumour doubling times vary widely, with an average of 200 days. It has been estimated that the time taken for an undetectable lesion to grow to 2 cm is about 4–12 months, and that it takes 5 months for the most rapidly growing HCC to reach 3 cm. Because treatments are most effective for tumours < 3 cm, screening programmes are usually performed at 6-monthly intervals. The efficacy of screening to improve the prognosis of HCC has mainly been demonstrated in China on HBV carriers.²⁷ These results need to be validated in other geographical areas. Until then, most rely on a 6-month interval (3–4 months in Japan) in high-risk patients. It has been reported that surveillance is cost-effective if the expected HCC risks exceeds 1.5% per year in patients with HCV and 0.2% per year in patients with HBV.²⁸

There are limitations to screening programmes. Of patients presenting with HCC, 20–50% have previously undiagnosed cirrhosis and therefore escape surveillance. Access to medical care and compliance is a limitation in highly endemic areas, with 50% of patients with alcoholic cirrhosis defaulting from surveillance over 5 years. US is highly operator dependent, and the cost and invasiveness of CT and MRI make them unsuitable for screening. However, these latter modalities are particularly suited in patients with irregular background liver parenchyma or obesity. Physicians should also take into account the presence of comorbid disease, severity of liver disease and available treatment options when deciding whether or not to screen a cirrhotic patient. Screening of Child C patients in particular is inappropriate if they are not potential liver transplant candidates.

Surveillance of at-risk patients is being used increasingly at 6-monthly intervals with US to detect HCC at an early stage.

High-risk groups consist of those with established cirrhosis due to HBV, HCV and haemochromatosis. Male patients with alcohol-related cirrhosis abstaining from alcohol or likely to comply with treatment should also be considered. US is recommended as a screening tool, whereas CT and MRI are most useful in confirming the diagnosis.

Treatment options

There is a wide range of treatment options for HCC (liver transplantation, liver resection, ablation, chemoembolisation, systemic treatments) and the decision should therefore be taken in a multidisciplinary team meeting involving a hepatic–pancreatic–biliary surgeon, interventional radiologist, oncologist and hepatologist, using predefined guidelines.

Liver transplantation, liver resection and ablation are traditionally defined as curative treatments. However, when underlying liver disease is present (typically cirrhosis), only transplantation is curative by simultaneously treating the aetiology of HCC. Recurrence is essentially invariable with all other treatments.

In patients without cirrhosis, liver resection is the ideal treatment but this group accounts for only a small proportion of patients with HCC. In cirrhotic patients, management is more challenging and should take into account tumour extension, status of the non-tumoral liver and general condition of the patient.

HCC in normal livers

The treatment of choice in patients with no or minimal coexisting fibrosis is partial liver resection. The non-tumorous liver has a high regenerating capacity, allowing even major hepatectomies to be performed. Perioperative mortality and morbidity are less than 1% and 15%, respectively. Five-year survival is greater than 50%.²⁹ These results may, however, vary according to the population studied. Patients with a metabolic syndrome in particular are at increased risk of postoperative mortality. Lymphadenectomy is recommended as the prevalence of lymph node metastases is approximately 15%, compared to less than 5% in cirrhotic patients. Adjuvant chemotherapy is not recommended. Regular follow-up with throraco-abdominal CT at 6-monthly intervals is recommended, as early detection and treatment of recurrence may improve survival.

There is very little place for other invasive treatments. Percutaneous ablation, as a rule, has no role due to the usually large tumour size at diagnosis. Liver transplantation is associated with a perioperative mortality of 10%, a need for long-term immunosuppression and long-term results not significantly different from those of resection. In a recent multicentre study based on a collaboration of 38 European transplant centres, only 105 patients transplanted for an HCC occurring in a normal liver were identified.³⁰ Transplantation had been performed as the primary treatment because partial liver resection was precluded by anatomical factors or the need to preserve a sufficient volume of liver remnant, or as rescue treatment for intrahepatic tumour recurrence not amenable to repeat resection. The 5-year survival rate was 59% in patients without macrovascular or lymph node invasion, irrespective of tumour size and differentiation.

Liver resection of HCC in cirrhotic patients

Liver resection:

Main limitations: Resection of HCC has four limitations in cirrhotic patients: (i) the tumour is multifocal in 20–60% of patients at the time of diagnosis and liver resection can normally only be considered in patients with unifocal tumours; (ii) cirrhosis is an important risk factor for the development of postoperative complications; (iii) oncological resections dictate wide margins whereas the diseased underlying liver usually requires parenchymal sparing; (iv) recurrence is invariable as cirrhosis persists.

Risk of surgery and patient selection: The risk of hepatectomy is increased in cirrhotic patients due to coagulation defects, portal hypertension, liver failure and impaired regeneration. In-hospital death was 10% in the 1990s (even higher in some subgroups) but has decreased since then as a result of improved patient selection, operative technique and perioperative management. Although some very large series report no mortality, the average mortality rates in national surveys or registries are 4–6% and are therefore higher than in non-cirrhotic patients or after resection of other malignancies.

Hepatectomy, as a rule, should only be performed in Child–Pugh A cirrhotic patients. Child–Pugh B or C patients are at a prohibitive risk of early liver failure even after a minor hepatectomy or mere laparotomy. Child–Pugh A patients may, however, still be at increased risk of postoperative liver failure, in particular after major resections, due to impaired ability to regenerate. This correlates with the fibrosis grade, although it is only in patients with extensive fibrosis or cirrhosis that this impairment has clinical impact. Typically, following a major liver resection, there is an increase in prothrombin time (peak on postoperative day 1) and an increase in serum bilirubin (peak on postoperative days 3–5) that tend to normalise within 5–7 days. Recovery of both tests is, in contrast, delayed or absent in cirrhotic patients. When the prothrombin time is less than 50% of normal and serum bilirubin is greater than 50 μmol/L on postoperative day 5, the risk of postoperative mortality is close to 50%.

Additional selection criteria for surgery have therefore been proposed for Child–Pugh A patients. In Japan, the indocyanine green (ICG) test is usually used. After injection of 0.5 mg ICG/kg body weight, retention of ICG is measured in peripheral blood, in particular 15 min after the injection (ICG-R15). Normal values of ICG-R15 are 10%. In cirrhotic patients, minor resections can be performed when it is 22% or less, but major resections only if it less than 14–17%. In contrast, in Europe and the USA, selection mainly relies on the absence of significant portal hypertension or cytolysis. This requires that patients have no evidence of oesophageal varices, splenomegaly, portosystemic shunts (including a patent umbilical vein) or ascites (even on imaging studies), and that they have a platelet count greater than 100 × 10⁹/L. Some groups even advocate that invasive measurement of the hepatic vein–portal vein gradient should be less than 10 mmHg. Several studies have shown that a normal serum bilirubin and the absence of clinically significant portal hypertension are the best predictors of good outcomes after resection.³¹ Recently, the MELD score and thrombocytopenia, irrespective of Child–Pugh grade and tumour features, have been shown to be associated not only with postoperative mortality and morbidity but also with long-term survival.³²^,³³

In cirrhotic patients who have a single resectable lesion, hepatectomy should only be performed if patients have well-preserved liver function, normal bilirubin and hepatic vein pressure < 10 mmHg.

There has been considerable interest during the past 5 years in the optimal management of the remnant liver. This includes: (i) more selective use of inflow occlusion; (ii) avoiding excessive mobilisation of the liver; and (iii) measuring the future remnant liver volume (RLV) using CT reconstruction. In patients with chronic liver disease, an RLV of approximately 40% of the total liver volume is required before a major hepatectomy is performed. When this is not the case, preoperative portal vein embolisation (PVE) is indicated as a means of increasing the RLV and, perhaps more importantly, preoperatively testing the ability of the liver to regenerate. When right hepatectomy is contemplated (the most frequent circumstance when there is a risk of small RLV), the right portal vein is percutaneously injected, under ultrasound guidance, with glue or ethanol. This should induce atrophy of the right liver within 2–6 weeks and a compensatory hypertrophy of the left RLV. Due to its efficacy, PVE (alone or in association with transarterial chemoembolisation) has become almost routine before a right hepatectomy in cirrhotic patients.³⁴ The absence of hypertrophy of the left liver following a successful right PVE means that the liver is unable to regenerate and that hepatectomy is contraindicated. There is also increasing evidence that parenchymal size alone does not necessarily reflect function and there is therefore interest in the functional evaluation of the remnant liver volume.

When major hepatectomy is contemplated, preoperative portal vein embolisation (PVE) of the lobe to be resected should be performed to test the ability of the future remnant liver to regenerate.

Technique: There is increasing evidence that both anatomical resections (as opposed to tumorectomies) and wide (as opposed to limited) margins may improve long-term survival without increasing the perioperative risk. The rationale is tumour spread through microvascular invasion, the incidence and extent of which is related to tumour diameter and degree of tumour differentiation. Several retrospective studies have reported an approximately 20% improvement in overall and disease-free survival following anatomical compared to limited resections.³⁵ The impact of the margin width has been evaluated in a prospective controlled trial.³⁶ A 2-cm margin was associated with a 75% 5-year survival as compared to 49% for 1-cm margins. Both concepts are not exclusive and should be taken into account, especially in tumours with diameters between 2 and 5 cm.

The concept of anatomical resections is to remove both the tumour and the adjacent segments that have the same portal tributaries and to achieve wide margins. Margins greater than 1–2 cm should be achieved to ensure that potential satellite nodules are also resected.

There is increasing interest in laparoscopic resections for HCC.³⁷ Although not formally proven yet, it may have the advantage of less intraoperative bleeding, postoperative complications, postoperative analgesic drug consumption and a shorter hospitalisation time. More specifically in the context of cirrhotic patients, it may also reduce the risk of postoperative ascites and its consequences, as well as facilitate subsequent liver transplantation if required, because of fewer adhesions.³⁸

Outcome after resection: The largest series from the Liver Cancer Study Group in Japan has reported 1-, 3-, 5- and 10-year survival rates of 87%, 66%, 48% and 21%, respectively in 11 631 cirrhotic patients treated by hepatic resection between 1992 and 2003. Comparable results have been reported by other groups worldwide, with no differences between Western and Asian studies. Independent predictors of survival are age, degree of liver damage, AFP level, tumour diameter, number of nodules, vascular invasion and surgical margins. Survival rates as high as 68% at 5 years have been reported in Child grade A patients with well-encapsulated tumours of 2 cm diameter or less. These figures continue to improve, even when patients with larger tumours are included. Active treatment of recurrences has been a major reason for this improvement.

Treatment of recurrence: Tumour recurrence is the major cause of death following resection of HCC in the cirrhotic patient. Its incidence is 40% within the first year, 60% at 3 years and approximately 80% at 5 years. However, it is invariable if follow-up is extended beyond 10 years as the precursor condition (cirrhosis) persists after surgery. It is frequently difficult to differentiate true recurrence from de novo tumours. The former tend to occur within the first 2 years and their main risk factors are vascular invasion, poor histological differentiation, presence of satellites and number of nodules. De novo recurrent tumours occur later and the main risk factors are the same as those of a primary HCC. Molecular analysis suggests that their respective proportions are 60–70% and 30–40%. Recurrence within the liver is multifocal in 50% of patients and is associated with distant metastasis in 15%, especially in the lungs, adrenal gland or bones. Extrahepatic recurrence without simultaneous intrahepatic recurrence is infrequent in cirrhotic livers. Anatomical resection and a tumour-free margin of 2 cm are associated with improved survival.

Evidence that neoadjuvant or adjuvant treatments reduce the risk of recurrence is currently lacking and these treatments are therefore not recommended. This applies to preoperative chemoembolisation, neoadjuvant or adjuvant chemotherapy, internal radiation with ¹³¹I-labelled Lipiodol, adoptive immunotherapy, retinoic acid or interferon, although some of these strategies initially showed promising results. An updated follow-up of the only randomised trial of adjuvant ¹³¹I-labelled Lipiodol has shown that the improved overall and disease-free survivals in the treatment group persisted until the seventh postoperative year.³⁹ This study was characterised by a very high proportion of HBV-related HCC (88%). Three recent meta-analyses of published studies^40–42 favour the use of interferon to reduce the risk of HCC recurrence; however, the quality of the studies was low due to heterogeneity of the patient populations, interferon used, duration of the treatment regimen and whether results were independent of the effect of viral suppression. No study has confirmed the potential efficacy of retinoic acid. Anti-angiogenic treatments are currently being evaluated.

The most effective strategy to prevent HCC recurrence is liver transplantation in selected patients (see below). However, there are two other important strategies. The first is management of the underlying chronic liver disease, as this improves prognosis and it is possible that it also reduces tumour recurrence. The second is to actively screen operated patients and actively treat recurrences if they are confined to the liver, by repeat surgery, ablation, chemoembolisation or liver transplantation.

At present, the best way to improve survival is to monitor resected patients regularly, as some may benefit from treatment of the recurrence if it is confined to the liver.

Liver transplantation (LT):

Rationale: HCC is the only tumour for which transplantation plays a significant role, and this is the most attractive therapeutic option because it removes both detectable and undetectable tumour nodules together with all the preneoplastic lesions that are present in the cirrhotic liver. In addition, it simultaneously treats the underlying cirrhosis and prevents the development of postoperative or distant complications associated with portal hypertension and liver failure.

Patient selection: LT is not readily available in most high endemic areas of HCC. Even when available, there is donor shortage; LT can therefore only be performed in a fraction (less than 5% in most Western countries) of HCC patients. HCC patients are considered potential candidates for LT if their anticipated survival is approximately the same as those patients transplanted for other indications. This may be achieved if strict selection criteria are applied; otherwise, HCC patients are at high risk of death from tumour recurrence. These include an HCC: (i) confined to the liver (i.e. no extrahepatic disease, including lymph nodes); (ii) without vascular extension; and (iii) with limited tumour burden.

Tumour burden was initially defined as a single tumour less than 5 cm in diameter or the presence of two or three tumours less than 3 cm in dameter (the so-called Milan criteria).⁴³ With the adoption of these criteria, the 5-year survival after LT ranges between 60% and 75%. There have subsequently been concerns that these criteria were too restrictive, which led to the proposal of expanded criteria. The best known and validated of these are the UCSF criteria – a single tumour less than 6.5 cm in diameter, or three or fewer tumours, the largest of which is less than 4.5 cm with the sum of the tumour diameters being less than 8 cm.⁴⁴ Others take into account poor tumour differentiation or a high (or rapidly increasing) AFP serum concentration. An international consensus conference held in 2012 recommended a limited expansion of the listing criteria beyond the standard Milan criteria.⁴⁵ Predicting tumour biology through molecular profiling rather than tumour morphology is the aim of current research in this field.

Treatment on the waiting list: The average time from listing to transplantation in Europe and the USA is usually greater than 12 months. Up to 25% of patients may be excluded from the waiting list due to disease progression. Three approaches have been developed to avoid these drop-outs:

• Living-donor liver transplantation (LDLT) is an alternative source of grafts but has its own drawbacks, including the inherent risk for the donor, the risk of small-for-size grafts and the fact that only 25–30% of transplant candidates have a potential donor. It has the advantage of being performed rapidly, so avoiding drop-out on the waiting lists. The number of LDLT in Western countries has, however, decreased recently and the trend is to favour cadaveric transplantation through changes in allocation policies.

• New rules of graft allocation have been implemented, initially in the USA and subsequently in Europe. In the USA, the Model of End-stage Liver Disease (MELD) organ allocation policy implemented in 2002 has given priority to candidates with HCC within the Milan criteria. Waiting times have shortened, obviating the need for LDLT. Similar policies have been applied in other countries such as France and the UK.

• Treatment of the tumour by resection, ablation or chemoembolisation is widely used while the patient is on the waiting list to avoid tumour progression beyond the oncological criteria. There is some evidence that these treatments may reduce drop-out rates on the waiting lists, but it is not clear if the outcome is the same for patients within or beyond the Milan criteria. The impact of these treatments on downstaging and post-transplantation survival is similarly uncertain. A specific advantage of resection over ablation or chemoembolisation is that it provides pathological details of the tumours. However, it is still unclear if the presence of poor prognostic factors should encourage or discourage transplantation.

• Liver transplantation is an effective option for HCC patients who fulfil the Milan criteria.

• Decreasing drop-out rates on the waiting list rely on changes in graft allocation policies and on treatment of the HCC while the patient is on the waiting list if this time exceeds 6 months.

• Living donor transplantation can be proposed for HCC if the waiting list is expected to be so long that there is a high risk of drop-out because of tumour progression.

Transarterial chemoembolisation (TACE):

Technique: HCC, in contrast to the liver parenchyma, receives almost 100% of its blood supply from the artery. When the feeding artery is obstructed, the tumour experiences an ischaemic insult that results in extensive necrosis. With the development of more supraselective embolisations, greater attention is paid to accessory arteries that may contribute to tumour vascularity, such as the diaphragmatic or mammary arteries, that should also be embolised to achieve adequate control. Injection of iodised oil has been combined to improve the efficacy of embolisation. Iodised oil (Lipiodol), which is hyperdense on CT, is cleared from the normal hepatic parenchyma but retained in malignant tumours for periods ranging from several weeks to over a year. This accumulation, which is not associated with significant adverse effects, may be used for targeting cytotoxic drugs and increasing their concentration in tumour cells. Recently drug-eluting beads (DC-Beads) loaded with doxorubicin have been developed. This technique is much more expensive than conventional TACE, but preliminary results show superior treatment response and delayed tumour progression.⁴⁶ Combination of TACE and anti-angiogenic treatments is under evaluation.⁴⁷

Contraindications: TACE should not be performed in patients with liver decompensation, biliary obstruction, bilioenteric anastomosis and impaired kidney function. Portal vein thrombosis is also a contraindication unless it is limited to a liver section only and TACE can be performed in a highly selective manner on a limited tumour volume.

Morbidity and mortality: Mortality is less than 1% if these contraindications are applied. Overall, more than 75% of patients develop a postembolisation syndrome characterised by fever, abdominal pain, nausea and raised serum transaminase levels. These symptoms, which are not prevented by antibiotics or anti-inflammatory drugs, are self-limiting and last for less than 1 week. More severe complications occur in less than 5% of patients and include, in decreasing order of frequency: cholecystitis or gallbladder infarction, gastric or duodenal wall necrosis, and acute pancreatitis. These, along with the postembolisation syndrome, have become less frequent with the use of supraselective embolisation. Hepatic abscess formation is rare, occurring in 0.3%, but is associated with a high mortality. The main risk factors are a previous history of bilioenteric anastomosis, large tumours and portal vein thrombosis.

Monitoring: The efficacy of TACE is assessed by CT (usually at 1 month) as the disappearance of the arterial vascular supply to the tumour and a decrease in its diameter. These features do not necessarily evolve in parallel. A decrease in tumour size may, for example, be associated with persistent vascularisation (i.e. residual tumour), whereas compact Lipiodol uptake without residual vascularisation may indicate complete tumour necrosis despite no significant decrease in size (Fig. 5.4).

Figure 5.4 HCC treated with microbead chemoembolisation. (a) Before treatment. (b) Two weeks after chemoembolisation: note the presence of necrosis. (c) Two years after chemoembolisation: the image of the tumour remains, but it is avascular, indicating complete local control.

Efficacy: There is grade A evidence that TACE improves survival.⁴⁸ One of the largest studies is a prospective Japanese nationwide survey reporting median and 1-, 3-, 5- and 7-year survivals of 34 months, 82%, 47%, 26% and 16%, respectively, with a TACE-related mortality of 0.5%. Independent predictors of survival were, in decreasing order of influence: the degree of liver damage, portal vein invasion, maximum tumour size and number of lesions, and AFP levels.

• TACE is one of the two non-curative treatments options (with sorafenib, as described above) that improves survival.

• TACE should be recommended as first-line palliative treatment for non-surgical patients with compensated Child–Pugh A and with large or multifocal HCC, without portal vein thrombosis or extrahepatic metastasis.

Percutaneous local ablative therapy:

Technique: Locoregional therapies are percutaneous treatment modalities that allow the injection of a damaging agent or the application of an energy source directly into the tumour. Damaging agents include chemicals such as ethanol (percutanenous ethanol injection, PEI) or acetic acid. Energy sources either aim at increasing temperature by radiofrequency, microwave or interstitial laser photocoagulation or, alternatively, at decreasing temperature (cryoablation). Irreversible electroporation is a new non-thermal ablation therapy that uses a high-voltage direct electrical current to create nanopores in the cellular membrane and results in cell death via apoptosis. Radiofrequency ablation (RFA) has emerged as the most effective of these techniques. It exploits the conversion of electromagnetic energy into heat via a needle electrode (15–18 G) positioned in the tumour, while patients are made into an electric circuit by grounding pads applied to their thighs. The radiofrequency emitted from the tip causes ionic agitation and frictional heat, which leads to cell death from coagulation necrosis. The objective is to maintain a temperature of 55–100 °C throughout the entire target volume for a sufficient period of time. Monitoring the impedance is important because excessive heating results in tissue charring, increased tissue impedance and decreased energy absorption.

The most widely used ablative techniques are PEI and RFA. All randomised controlled trials comparing PEI and RFA have suggested that the actuarial probability of local recurrence was significantly lower with RFA compared to PEI, and that RFA required fewer treatment sessions to achieve comparable antitumoral effects.⁴⁹^,⁵⁰

Advantages and drawbacks: These ablative methods are minimally invasive, preserve the uninvolved liver parenchyma, have no systemic side-effects, and avoid the mortality and morbidity of major hepatic surgery. On the other hand, only tumours less than 5 cm are likely to be treated successfully but the smaller the diameter, the greater the probability of complete local control. The presence of multiple tumours (more than three) is also a limitation because of the need for repeated punctures. In addition, multiple tumours are either the result of multifocal carcinogenesis or vascular extension, and therefore a focal treatment is unlikely to be very effective. Obviously, a common requirement is also the need to clearly visualise the tumour by US and access it safely. Hence, isoechoic HCC or tumours located in the upper part of segments 4, 7 and 8 or at the edge of the left lateral section if it extends behind the spleen may occasionally be unsuitable for treatment. Finally, whichever technique is used, the needle should not enter the tumour directly but pass through the hepatic parenchyma so as to prevent intraperitoneal bleeding or seeding of tumour cells. This may prove impossible for some superficial or protruding tumours. Recently, an experienced group reported that up to one-third of patients who were theoretically good candidates for ablation could not be treated due to non-visibility of the HCC on US, risk of thermal injury or absence of a safe path.⁵¹

Contraindications and limitations: Contraindications to ablation procedures include gross ascites that favours intraperitoneal bleeding, coagulopathy that cannot be corrected, previous history of bilioenteric anastomosis or endoscopic sphincterotomy associated with bile bacterial contamination and therefore a risk of abscess formation. Additional contraindications (more specific to radiofrequency or microwave rather than ethanol injection) come from the proximity of the tumour to the colon, duodenum, stomach or biliary confluence, which may be injured or perforated by the heating process. RFA, unlike microwave ablation, is as a rule contraindicated in patients with a pacemaker. The efficacy of RFA also seems to be more impacted by the proximity of a vascular pedicle (the so-called cooling effect) than microwave ablation. Whereas PEI is a quick and very cheap procedure performed under light sedation, RFA is more costly, prolonged (20–90 minutes) and painful, and therefore generally performed under general anaesthesia. Microwave ablation is also performed under general anaesthesia but the procedure is quicker.

Mortality following ablation is less than 1% and morbidity less than 10%. The most frequent complications are pleural effusion and segmental intrahepatic dilatation, which have no or limited impact. Severe complications include abscess formation, perforation of adjacent organs and intraperitoneal bleeding. Tumour seeding is 5% or less. Risk factors include subcapsular location and poor histological differentiation of the tumour. Coagulating the needle tract while removing the needle may reduce this risk.

Methods and margins: Ablation should not only target the tumour but also aim to achieve a safety margin so as to control satellite nodules. The incidence of these satellite nodules, as well as their distance from the main tumour, increases as the main tumour enlarges. Both incidence and distance also increase for poorly, compared to well differentiated, tumours. This safety margin should be 5 mm at least; hence, for an HCC measuring 3 cm in diameter, the diameter of the ablation should be 4 cm. This is best achieved with thermal rather than chemical ablation. Methods to further improve tumour and margin control include multipolar ablation (several probes are placed around the tumour) and combining ablation with TACE.⁵² Treatment response is assessed by CT or MRI no earlier than 1 month after the procedure. RFA may result in a rim of fibrotic tissue (hypervascular on late-phase MRI or CT) at the periphery of the tumour and should not be mistaken for residual tumour tissue. Follow-up thereafter relies on imaging studies at 3-monthly intervals to ensure that there is no recurrence of contrast enhancement.

Indication: Percutaneous ablative therapies have initially been performed in patients who were unsuitable for resectional surgery and both the EASLD and AASLD have recommended this strategy. It has thereafter been used as neoadjuvant treatment in liver transplant candidates and for treatment of recurrences after liver resection.

As the results of ablation improve, due to improved technology and patient selection, it may also be considered as an alternative to surgery or even as a first-line treatment in selected situations. A large multicentre phase 2 trial reported a 97% sustained complete response and a 68% actuarial 5-year survival following ablation in patients with HCC of 2 cm or less.⁵³ The results of two randomised controlled trials comparing ablation and resection in patients with early HCC demonstrated no difference.⁵⁴^,⁵⁵

RFA has now demonstrated its effectiveness to a point where it is considered by some centres as the first-line treatment for single nodules less than 2 cm in diameter.

However, meta-analyses still favour surgery compared to ablation in terms of 3-year survival and local control.⁵⁶^,⁵⁷ One additional concern is that both in the USA⁵⁸ and in Italy⁵⁹ there has been a recent temporal trend of increased use of ablation as a treatment for HCC with a simultaneous decrease in survival following this treatment, unlike what has been observed for other treatments. These observations suggest that the extension of indications for ablation should be strictly evaluated.

Other palliative treatments

Conventional systemic chemotherapy: Systemic chemotherapy has had very limited value in the past as only a very small number of patients obtained partial response or meaningful palliation using conventional drugs. Therefore, there is no rationale for using chemotherapy in patients with unresectable HCC outside of clinical trials.

Anti-angiogenic targeted therapies:

• A recent trial using molecularly targeted agents has, for the first time, demonstrated an improved overall survival in this disease and sets a new standard for the first-line treatment of advanced HCC. These new agents target angiogenesis and epidermal growth factor (EGF) receptor pathways.

• Sorafenib is therefore recommended as a first-line palliative option in patients not eligible for resection, liver transplantation, percutaneous ablation or TACE, if they still have preserved liver function.

Sorafenib (Nexavar®) exerts an anti-angiogenic effect by targeting the tyrosine kinases vascular endothelial growth factor (VEGF) receptors 2 and 3, and the platelet-derived growth factor receptor β. In an initial phase 3 trial, the median overall survival of Child–Pugh A cirrhotic patients with histologically proven and advanced HCC was 10.7 months in the treated group versus 7.9 months in the placebo double-blinded controlled arm of the study (P = 0.00058), and the median times to tumour progression were 24 weeks and 12 weeks, respectively (P = 0.000007).⁶⁰ This efficacy in advanced HCC (unresectable or metastatic) has been confirmed in an Asian randomised placebo-controlled trial that included mostly patients with HBV-related HCC⁶¹ and in a large phase 4 study with more than 3000 patients.⁶² Side-effects included diarrhoea (39%), hand–foot syndrome (21%), anorexia (14%) and alopecia (14%). The antitumour effect, the pharmacokinetic profile and safety profile were similar in Child–Pugh A and B. These results have established sorafenib as the standard of treatment for advanced HCC in Child A (or B) patients.²³ Several trials assessing strategies such as combination or sequential treatments are under way.

Other agents with comparable action pathways that have been evaluated in phase 2 trials include bevacizumab and sunitinib. Anti-EGF receptor agents such as tarceva and cetuximab also show promising results. Contraindications to these treatments include coronary artery disease, cardiac failure, systemic hypertension and Child B or C cirrhosis.

Radioembolisation: External beam radiation therapy has been of limited value in treating HCC because the normal liver parenchyma is very radiosensitive. Greater interest has therefore been given to injecting radioisotopes such as ¹³¹I-iodised oil or ⁹⁰Y-labelled microspheres directly into the hepatic artery (radioembolisation), which offers the advantage of increased delivery within the tumour and decreased toxicity. The former agent has an efficacy comparable to that of chemoembolisation in patients with HCC not complicated by portal thrombosis but is superior in patients with tumour portal extension. The use of ⁹⁰Y microspheres is more recent and has been shown in a phase 2 trial to be safe and effective, in particular in patients with portal vein thrombosis.⁶³ These results have been reproduced in three recent studies, but without randomised controlled trials comparing ⁹⁰Y-labelled microspheres, TACE or other established treatments, defining the role of this expensive treatment in clinical practice is not possible.

Other treatments: Anti-androgenic, anti-oestrogenic and somatostatin analogues, once proposed, are currently considered ineffective.²³

New treatments such as radioembolisation and antiangiogenics are promising, but costly. The additional efficacy of radioembolisation over TACE, or of anti-angiogenics over no treatment, is required in a context of limited financial resources.

Defining a treatment strategy

Uncomplicated HCC associated with chronic liver disease

Treatment algorithms need to take account of availability of treatments.

• Liver transplantation, when available, is considered first and attention is therefore paid to the extent of liver disease, patient age and presence or absence of associated conditions. If a long waiting time (> 6 months) is expected, resection, ablation or TACE are considered prior to liver transplantation.

• If transplantation is not available or not indicated, resection should be considered. Limiting factors are the number of nodules (ideally there should be only one) and the severity of underlying liver disease (patients should be Child–Pugh A and have neither cytolysis, portal hypertension nor impaired ICG tests). If a right hepatectomy is considered it should be preceded by PVE (with or without TACE).

• If resection is not considered due to the severity of the underlying liver disease and the nodule is single (or if there are less than three nodules), ablation is the treatment of choice provided the tumour is less than 3–5 cm. For single tumours 2 cm or less, RFA is becoming a first-line treatment, as an alternative to resection.

• If neither resection nor RFA is considered, TACE is performed provided there is no ascites or liver failure (and in particular that the serum bilirubin is less than 50 μmol/L) and that the tumour burden is not too extensive (no vascular invasion or extrahepatic metastases).

• Remaining patients are currently considered for anti-angiogenic treatments provided there is neither liver failure nor vascular disease.

According to this algorithm, it may be considered that the proportion of HCC patients who are candidates for transplantation is less than 5%, for resection 10–15%, for ablation 15–20% and for TACE 30–40%.

Treatment of complicated HCC

HCC with macroscopic portal vein invasion: This is a contraindication for liver transplantation and ablative treatments. Traditionally, TACE was also contraindicated (because of the risk of liver necrosis); today it is occasionally performed provided the thrombus is limited to a liver section or less and that embolisation is highly selective, with reduced doses and partial (rather than total) arterial occlusion as the end-point. If thrombus does not extend into the main portal vein, surgical resection can be considered. Radioembolisation and anti-angiogenic therapy is otherwise indicated.

HCC with macroscopic invasion of hepatic veins: This seems to carry an even worse prognosis as the tumour thrombus will extend into the inferior vena cava. When the thrombus is confined to the hepatic vein, resection if possible can be proposed. There is, however, a very high risk of pulmonary metastases developing within 6–12 months of surgery. Extension into the inferior vena cava or the right atrium is usually beyond any treatment.

Ruptured HCC: This should be actively treated unless it occurs as a terminal presentation in patients with multiple tumours, portal thrombosis and end-stage liver failure. The primary aim of treatment is to stop bleeding, ideally by arterial embolisation. Subsequent hepatectomy can be associated with long-term survival. Indeed, (i) bleeding is not necessarily due to tumour rupture, but occasionally due to rupture of an artery at the junction of the tumour and the adjacent parenchyma, and (ii) even if the tumour has ruptured, this is not always associated with peritoneal seeding of tumour cells.

Fibrolamellar carcinoma (FLC)

FLC is a rare variant of HCC, defined as well-differentiated polygonal hepatic tumour cells with an eosinophilic granular cytoplasm surrounded by a fibrous lamellar stroma. It is most frequently observed in the Western hemisphere, where it accounts for approximately 1% of all HCCs. These tumours occur at a younger age than HCC (20–35 years), preferentially in women, and classically do not arise on a background of chronic liver disease.

• FLCs are usually large at the time of diagnosis (8–10 cm), and the common revealing symptoms are a palpable mass, abdominal pain, weight loss, malaise and anorexia.

• The prognosis is better than that of HCC overall. Five-year survival following resection is 50–75%.⁶⁴

• Resection is preferred to transplantation as the latter has very little or no place.

On imaging, FLC presents as a large solitary hypervascular heterogeneous liver mass with a central hypodense region due to central necrosis or fibrosis. On MRI, the central scar has low attenuation on T2 images, whereas the central scar of focal nodular hyperplasia has high attenuation. They have well-defined margins and calcification is present in 68%. Histology demonstrates deeply eosinophilic, polygonal neoplastic cells surrounded by a dense, layered fibrous stroma.

AFP levels are raised in less than 10% of patients.⁶⁴ Lymph node invasion within the hepatic pedicle is frequent (60%) and if resection is considered, simultaneous lymphadenectomy is recommended. There is a significant risk of recurrence, not only within the liver but also as lymph node or distant metastases. Close long-term follow-up is mandatory since recurrence and death beyond 5 years are common. Repeat surgery is a reasonable option in this younger patient population due to the relatively indolent course of the disease and the relative inefficacy of non-surgical treatments.

A recent study has suggested that true FLC should be differentiated from mixed FLC–HCC, defined as conventional HCC displaying some distinct area with FLC features.⁶⁵

Intrahepatic cholangiocarcinoma (ICCA)

ICCA, also known as peripheral cholangiocarcinoma, is the second most frequent primary tumour of the liver after HCC. Tumours arise from the peripheral intrahepatic biliary radicles, which differentiates them from hilar (Klatskin) tumours and common bile duct cholangiocarcinoma.

Until the very end of the 1980s, there was no immunohistological marker that could pinpoint the biliary origin of adenocarcinoma, and ICCAs were therefore probably frequently considered as being liver metastases of an adenocarcinoma of unknown origin. The diagnosis is currently ascertained through immunostaining that shows that they are CK7 positive and CK20 negative (colorectal metastases are CK7 negative and CK20 positive).

This tumour has a poor prognosis overall and resection, sometimes at any cost, was the only therapeutic option. However, the recent implementation of a specific staging system and evidence that chemotherapy is effective pave the way for improved management.

Incidence

In the Western world, the incidence of ICCA is 0.3–3 per 100 000, 10 times less than HCC. Recent reports suggest the incidence is increasing, particularly in the USA, UK, France, Italy, Japan and Australia. Although this increase may be real, it is probably mainly explained by improved identification of this tumour and changing rules on how they should be coded according to the International Classification of Diseases for Oncology (ICD).⁶⁶

Risk factors

The traditional risk factors for cholangiocarcinoma include chronic biliary inflammation such as primary sclerosing cholangitis, chronic choledocholithiasis, hepatolithiasis, parasitic biliary infestation, Caroli’s disease and choledochal cyst. However, in most patients with ICCA (more than 95%) none of these risk factors can be identified. The exception occurs in some areas of Asia and in particular north-eastern Thailand, where the parasite Opisthorcis viverrini is particularly prevalent.

New risk factors are emerging, including chronic non-alcoholic liver disease, HBV infection, HCV infection, diabetes and the metabolic syndrome.⁶⁷ However, in contrast to HCC, most ICCAs develop without a background of liver disease. In surgical series, 75% of patients have normal livers, 16% have chronic hepatitis/liver fibrosis and 9% have cirrhosis.

Classification and staging

The Liver Cancer Study Group of Japan proposed a gross classification of ICCAs into three types based on macroscopic findings: mass forming, which is by far the commonest type (75% in Asian series and probably more in the West); periductal infiltrating, which spreads along the bile ducts; and intraductal growth type with intraluminal spread. However, tumours may have mixed components, in particular a combination of mass forming and periductal infiltrating.

Whereas previously ICCAs were staged using a similar system as HCCs, the AJCC implemented a specific classification for ICCAs in 2010.⁶⁸ The T staging takes into account number of tumours and vascular invasion (the presence of either defines T2), rather than tumour size. The reason for this is that it is very unusual to diagnose ICCA early and size does not independently impact survival in published surgical series. T3 tumours are those perforating the visceral peritoneum or involving local extrahepatic structures by direct invasion, although this is fairly rare. The T staging also aims to take into account the periductal-infiltrating pattern of ICCA and, when present, defines T4. However, this infiltrating pattern may be difficult to identify on imaging studies or even on pathological specimens and there is no standardised definition yet. Lymph node involvement has a major impact on survival and, when present, defines TNM stage III. Prevalence of lymph node extension is high and therefore lymphadenectomy should be routinely performed to achieve accurate staging. Median survival of patients with stage I is greater than 5 years (but these patients are very rare), whereas that of patients with stage II is 53 months and that of patients with stage III is 16 months.⁶⁹

Pathology and progression analysis

Two distinct conditions that precede invasive cholangiocarcinoma have been identified. The first is a flat or micropapillary growth of atypical biliary epithelium, which has been called biliary dysplasia or biliary intraepithelial neoplasia. The second is an intraductal papillary neoplasm of the bile duct characterised by the prominent papillary growth of atypical biliary epithelium with distinct fibrovascular cores and frequent mucin over-production. These preneoplastic conditions have mainly been analysed in hepatolithiasis and are observed more frequently in large bile ducts as hilar tumours than in small septal–interlobular bile ducts such as with ICCA.⁷⁰ The dysplasia–carcinoma sequence therefore appears more obvious for hilar lesions than peripheral lesions. This suggests that an alternative source of ICCA could be the canals of Hering or hepatic progenitor cells, which are a target cell population for carcinogenesis in chronic liver disease.

Clinical presentation and laboratory tests

As a rule, ICCA tends to be diagnosed at an advanced stage because the tumour remains clinically silent for a long time. Symptoms, when present, include abdominal pain, malaise, night sweats, asthenia, nausea and weight loss. When they appear, the tumour is frequently unresectable.

ICCAs typically appear with equal frequency in men or women aged 55–75 years. Liver function tests are non-specific even though an increase in liver enzymes (in particular γ-glutamyl transferase) may be the only initial finding in some patients. Although ICCA by definition excludes tumours arising from the biliary confluence or first-order branches, jaundice may be present if the tumour compresses or invades the biliary confluence.

Serum markers lack sensitivity and specificity. Carcinoembryonic antigen (CEA) exceeds 20 ng/mL in 15% and carbohydrate antigen (CA) 19-9 is greater than 300 U/mL in 40% of cases. AFP exceeds 200 ng/mL in only 6% of patients.

Imaging studies

The main characteristic of mass-forming ICCA is that it is a fibrous tumour and therefore displays no enhancement on the arterial phase and delayed enhancement during the late phase. This may be seen both on CT and MRI. On MRI, lesions are hypointense on T1-weighted images and moderately to markedly hyperintense on T2-weighted images (Fig. 5.5). They are typically large, non-encapsulated, heterogeneous, associated with narrowing of adjacent portal veins and retraction of the liver capsule. As the tumour grows, satellite nodules frequently develop in the vicinity of the tumour, and subsequently in the contralateral lobe (Fig. 5.6). When superficial, these satellite nodules may not be visible on imaging. There is a high propensity for lymph node invasion (present in 40% of resected patients if lymphadenectomy is performed routinely), but imaging studies only have a sensitivity of 50% and a specificity of 75% to predict this.

Figure 5.5 Vascular kinetics of a small cholangiocarcinoma on MRI (arrowed). Note that the lesion is spontaneously hypointense (a), that the uptake of vascular contrast is more pronounced at the late phase (b) than at the arterial phase (c), and that there is a retraction of the capsule.

Figure 5.6 CT scan of a patient with an intrahepatic/peripheral cholangiocarcinoma. Note the presence of typical satellite nodules at the periphery of the tumour (a), the absence of vascular uptake (b) and the retraction of the capsule (c).

Diagnosis

The main differential diagnoses of ICCA are other fibrous tumours and in particular metastases from colorectal cancer. Both tumours may easily be confused on imaging studies. The diagnosis relies on a biopsy that shows an adenocarcinoma of biliary phenotype (CK7 positive, CK20 negative). Colorectal metastases are, in contrast, CK7 negative and CK20 positive.

Treatment

Surgical resection is the only curative treatment. Unlike HCC, there is currently no place for liver transplantation.

As the tumour is usually diagnosed at an advanced stage, has ill-defined borders and occasionally extends to major portal branches or hepatic veins, surgery is frequently extensive. A major hepatectomy is required in 75–80%, extended to segment 1 in 30% of cases and including the common bile duct in 20% of cases to achieve a complete resection. This surgery is therefore associated with significant postoperative mortality. This is estimated to be 6%, higher than following surgery for colorectal metastases and almost comparable to that of surgery for HCC despite the usual absence of chronic liver disease.

There is a significant risk (20–30%) that, despite adequate preoperative imaging, contraindications to a curative resection are identified at laparotomy. Staging laparoscopy has been advocated, but is also associated with high false-negative rates and, as a consequence, patients should be warned preoperatively about this possibility. Furthermore, approximately 25% of resected patients will have an R1 or R2 resection. Survival following an R2 resection is usually comparable to, and occasionally worse than, that of non-resected patients. Median survival following an R1 resection is typically 12 months and the 3-year survival is nil.

According to the series published over the past decade, the 1-, 3- and 5-year survival rates following resection of ICCA are 67%, 38% and 27%, respectively. There are few data on survival beyond 5 years. Variables that influence postoperative survival most are the presence of lymph node invasion and an R1 resection.⁷¹ Intraductal growth-type ICCAs are rare but have a better long-term prognosis. Infiltrating-type ICCAs have a worse prognosis than the mass-forming type due to spread along Glisson’s capsule and high incidence of lymph node involvement.

There is little evidence that these figures have improved over the past 10 years. However, the recent demonstration that systemic chemotherapy may be effective in unresectable patients opens the possibility of combining surgery with either adjuvant and/or neoadjuvant chemotherapy.⁷²

There is growing interest in the management of ICCA. Improvement in outcome will require that additional treatments other than surgery are evaluated.

Epithelioid haemangioendothelioma (EHE)

EHEs are neoplasms of vascular origin that arise predominantly from soft tissues, bones and visceral organs, in particular the lung and the liver. Hepatic EHE develops from the endothelial cells lining the sinusoids and progresses along the sinusoids and vascular pedicles. It is extremely rare (no more than 200 cases have been reported), with an incidence of less than 1 per million population. It does not arise on a background of liver disease and there is no identified causative factor. Mean age at presentation is 42 years, with a female to male ratio of 3:2.⁷³ Half present with right upper quadrant pain, a quarter incidentally, and the remainder with severe symptoms such as ascites, jaundice, weakness and weight loss. Liver failure as a result of massive infiltration has been described.

These tumours are usually discovered at an advanced stage; almost 90% are multifocal and then usually involve both lobes. Approximately one-third of patients have extrahepatic spread to regional lymph nodes, peritoneum, lung and spleen.

Although the diagnosis is obvious when appropriate immunohistochemical staining is performed on tumour samples, it is frequently misdiagnosed on other investigations. Laboratory parameters are non-specific and tumour markers are normal. On imaging studies, the lesions are frequently confused with cholangiocarcinoma, metastatic carcinoma, sclerosing angioma or inflammatory pseudotumours. They are usually hypoechoic or heterogeneous on US, hypodense on CT with peripheral and/or central marginal enhancement on the arterial phase becoming isodense during the later phase, and may display a halo or target pattern of enhancement. On MRI, they are hypointense on T1-weighted images and heterogeneously hyperintense on T2-weighted images, with similar contrast enhancement to that seen on CT. Multiplicity of lesions (especially if coalescent), their subcapsular location with liver capsule retraction, and the presence of calcification (10–30%) or central necrotic and haemorrhagic areas should raise the suspicion of the diagnosis, especially in young patients. Histology shows a tumour composed of epithelioid and dendritic cells in variable proportions, with a propensity for invasion of hepatic and portal veins, an overall ill-defined growth pattern and infiltrative margins. These features are difficult to identify or differentiate from other tumours on a percutaneous biopsy sample, but immunostaining for factor VIII-related antigens is highly specific, demonstrating endothelial differentiation. Most tumours also stain positive for CD34 and CD31 endothelial markers. Epithelial markers including cytokeratins are negative.

The natural history of this tumour is highly variable. Although exceptional, prolonged survival of more than 10 years has been reported without treatment, and both partial and complete spontaneous tumour regression has even been described. On the other hand, some patients die within 2 weeks of diagnosis and 20% are dead within 1 year. Overall, only 20–40% survive more than 5 years.⁷³ Because of the rarity of this tumour and its highly variable course, there is no widely accepted therapeutic strategy.

Partial hepatectomy is rarely feasible due to the invariable multifocal involvement of the liver. Palliative resection is not advocated as some have raised concerns that liver regeneration could promote a flare-up of residual tumours. Reports of favourable outcome with an estimated 5-year survival of 75% probably represent a highly selected subgroup.⁷³

The place of liver transplantation has recently been clarified by a multi-institutional analysis.⁷⁴ In 59 patients reported to the European Liver Transplant Registry, impressive 5- and 10-year survival rates of 83% and 74%, respectively, were reported. Invasion of lymph nodes and presence of restricted extrahepatic involvement had limited impact on survival and should therefore not be considered as contraindications to transplantation. The current shortage of liver grafts and the prolonged waiting time may dictate that liver transplantation is only indicated in highly selected patients. Experience with locoregional or systemic chemotherapy is small and of limited value, especially as first-line therapy. Neoadjuvant combination therapies using anti-VEGF antibodies, however, deserve investigation.