Guidelines for the diagnosis and treatmentof hepatocellular carcinoma (HCC) in adults

................................................................................................................ ............ .

1.0 FOREWO RD
This document, on the diagnosis and treatment of patients with hepatocellular carcinoma (HCC), was commissioned by the British Society of Gastroenterology as part of a wider initi ative to develop guidelines for clinicians in several areas of clinical practice. Cancer care has been the subject of increased scrutiny, with the development of care guidelines forming a major part of the strategy to reduce cancer related mortality in the UK. There is a strong suggestion that HCC is a disease which will be seen more frequently over the next few years, mainly as a result of the hepatitis C virus (HCV) epidemic. Previously, HCC has been a relatively rare tumour in the UK and much of the data pertaining to its diagnosis and therapy are derived from stud- ies outside of the UK. Because of the lack of screening programmes and the fact that a significant propo rtion of HCC presents as symptom atic disease in individuals not known to have liver disease, most nonsurgical therapies have been used in patients with advanced disease. There are a significant number of variables known to influence prognosis, with stage of underlying liver disease and tumour size at present ation being the most impo rtant. Controlling for these variables is difficult and these factors have cont ributed to a dea rth of ran- domised controlled trials of treatment for this tumou r. There is however a substantial amount of evidence available which can form the basis of a frame work for diagnosis and management. Guidelines are not rigid protocols and they should not be const rued as interfe ring with local clinical judgement. Hence they do not represent a directive of proscribed routes but a basis on which clinicians can consider the options available more clearl y.

strong suggestion from the UK that HCC is becoming a more common cance r, primarily due to the HCV epidemic. These guidelines relate to adult medical practice; high risk paedi atric conditions predis- posing to HCC and the management of paedi atric patients with HCC will not be considered. Guidelines are proposed on a number of issues: (a) which patients are at high risk for the development of HCC and should be offered surveillance; (b) what investig ations are required definite diagnosis; and to make a

(c) which treatment modality is most approp riate in a given clinical context.

2.0 INTRODUCTION OBJECTIVES

AND

These guidelines cover two areas of clinical practice relating to HCC: firstly, its diagnosis, including surveillance of high risk individuals; and secondly, treatment of the patient where the diagnosis has been made. HCC remains one of the commonest malignant diseases in the world but it has not previously been a leading cause of death in the Western world. There is now conclusive evidence from the USA and a

3.0 FORMULATION GUIDELINES

OF

A system atic review of the relevant liter ature and synthesis of available evidence with later phases of peer group appraisal and then expert review was perfo rmed. Draft proposals were amended at this stage. The strength and evidence used in these guidelines was that recommended by the north of Eng- land evidence based guidelines development project. 3.1 Categories evidence of

accordingl y. Grade A: Requires at least one randomised controlled trial of good quality addressing the topic of recommend ation. Grade B: Requires the availability of clinical studies without randomis ation on the topic. Grade C: Requires evidence from category IV in the absence of directly applicable clinical studie s.

4.0 SUMMARY RECOMMENDATIONS

OF

Ia: Evidence from meta-analysis of randomised controlled trials. Ib: Evidence from at least one randomised trial. IIa: Evidence obtained from at least one well designed controlled study without randomis ation. IIb: Evidence obtained from at least one other type of well designed quasi expe rimental stud y. III: Evidence obtained from well designed nonexpe rimental descriptive studies such as compar ative studies, correlation studies, and case studie s. IV: Evidence obtained from expert committee reports or opin- ions, or clinical experiences of respected autho rities. 3.2 Grading of recommendations Recommend ations are based on the level of evidence presented in suppo rt and are graded

Surveillance using abdominal ultrasound and fetoprotein (AFP) estim ation can detect HCC of a smaller size than those presenting without screening (evidence IIa). The only potentially curative therapies depend on detection of small HCC (evidence IIa). Despite the above, there are no data confi rming that these advantages in detection of earlier lesions produces an impr ovement in long term survival or cost saving (evidence IIa). Surveillance for hepatocellular carcinoma should be consid- ered in the following high risk groups:
...................................................... ...... .
Abbreviations: HCC, hepatocellular carcinoma; HCV, hepatitis C virus; HBV, hepatitis B virus; AFP, -fetoprotein; CT, computed tomography; MRI, magnetic resonance imaging; PEI, percutaneous ethanol injection; HBsAg, hepatitis B surface antigen.

 Males and females with established cirrhosis due to hep atitis B virus (HBV), particularly those with ongoing viral replic ation (evidence grade III, recommend ation grade B). Males and females with esta blished cirrhosis due to HCV (evidence grade III, recommend ation grade B). Males and females with established cirrhosis due to genetic haemochrom atosis (evidence grade III, rec- ommend ation grade B). Males with alcohol related cirrhosis who are abstinent from alcohol or likely to comply with treatment (evidence grade III, recommend ation grade B). Males with cirrhosis due to primary biliary cirrhosis (evidence grade III, recommend ation grade B). The risk of HCC development in cirrhosis due to auto- immune hep atitis, primary sclerosing cholan gitis in both sexes, and alcoholic and primary biliary cirrhosis in women is generally low. Non-ci rrhotic HCCs do occur in viral cirrhosis but the absolute risk is low (evidence IIb). If surveillance is offered, it should be six monthly abdominal ultrasound assessments in combin ation with serum AFP estim ation (evidence grade III, recommend ation grade B). Abdominal ultrasono graphy should be unde rtaken with approp riate dedic ated equipment and by an operator skilled in the assessment of patients with cirrhosis (evidence grade IIb, recommend ation grade B). If surveillance is offered, patients should be aware of the implic ations of early diagnosis and the lack of proven survival benefit. 4.1 Diagnosis of HCC A focal lesion in the liver of a patient with cirrhosis is highly likely to be HCC (evidence grade IIa). Initial assessment should be by spiral computed tomo gra- phy (CT) of the liver (local spread) and thorax (metastases) (evidence grade IIa, recommend ation grade B). Magnetic resonance imaging (MRI) with contrast enhance- ment or angiography with lipiodol injection and follow up CT may increase the accuracy of detection of other liver lesions (evidence grade III, recommend ation grade C). Biopsy is rarely required for diagnosis, and seeding of tumour in the needle tract occurs in 1 3%. Biopsy of poten- tially operable lesions should be avoided where possi ble (evidence grade IIa, recommend ation grade B). 4.2 Treatment of HCC The only proven potentially curative therapy for HCC remains surgical, either hepatic resection or liver transplant ation, and patients with single small HCC (<5 cm) or up to three lesions <3 cm should be referred for assessment for these treatment modalitie s. Liver transplant ation should be considered in any

patient with cirrhosis and a small (5 cm or less single nodule or up to three lesions of 3 cm or less) HCC (evidence grade IIa, recommend ation grade B). Patients with replic ating HBV had a worse outlook due to

3 2 . 5 2 1 . 5 1 0 . 5 0 17 96 18 91 18 96 19 91 19 96

Y e a r

Figure 1 Incidence (per 100 000 population) of hepatocellular carcinoma (HCC) in the USA from 1976 to 1996 (from El-Serag and Mason 1999 1).

(evi- dence grade IIb). Chemoembolis ation can produce tumour necrosis and has been shown to affect survival in highly selected patients with good liver reserve. Chemoembolis ation using lipiodol is effective therapy for pain or bleeding from HCC (evidence grade IIa, recommend ation grade B). Systemic chemotherapy with standard agents has a poor response rate and should only be offered in the context of trials of novel agents (evidence grade I, recommend ation grade A). Hormonal therapy with tamoxifen has shown no survival benefit in controlled trials (evidence grade I, recommen- dation grade A) and is not recommended.

5.0 BACKGROUND
5.1 Epidemiology of HCC HCC causes appr oximately 1500 deaths per year in the UK. There is strong evidence from the USA that the incidence of HCC is rising1: nine cancer registries repo rting via the National Cancer Institute showed a 41% rise in mortality from prima ry liver cell cancer between 1980 and 1995 with a 70% rise in overall incidence (fig 1). Similar although less robust evidence is emer ging in the UK.2 HCC is unusual among human cance rs in that the aetiological agent responsi ble is usually readily identifia ble. The prevalence of HCC worldwide parallels that of viral hep atitis and the majority of cases are associ ated with HBV and HCV. The increase in HCC incidence in the developed world is likely to be a direct result of the HCV epidemic occur- ring some 2030 years after the rise in this infection in target 3 popul atio Alcohol, genetic haemochrom atosis, and ns. rarely primary biliary cirrhosis are associ ated. In the UK, up to 40% of cases present with HCC as the first indic ation of underlying liver disease, in distinction to count ries such as Japan where 80% of HCC are detected at an asymptom atic stage by screen- ing of patients with known cirrhosi s.4

(Child-Pugh A) who are unsuita ble for liver transplanta- tion. Such surgery carries a high risk of postoper ative decompens ation and should be unde rtaken in units with expe rtise in hepatic resection and management of liver fail- ure (evidence grade IIa, recommend ation grade B). 4.3 Non-surgical management Non-sur gical therapy should only be used where surgical therapy is not possi ble. Percutaneous ethanol injection (PEI) has been shown to produce necrosis of small HCC. It is best suited to peripheral lesions, less than 3 cm in diameter (evidence grade IIb, rec- ommend ation grade B). Radiofrequen cy ablation may be a good alte rnative ablative therapy but data are limited HBV recurrence and were previously not considered candidates for transplant ation. Effective antiviral therapy is now available and patients with small HCC, as defined above, should be assessed for transplant ation (evidence grade IIa, recommend ation grade B). Hepatic resection should be considered as primary therapy in any patient with HCC and a noncirrhotic liver (including fibrolamellar variant) (evidence grade IIa, recommend ation grade B). Resection can be carried out in highly selected patients with hepatic cirrhosis and well preserved hepatic function

5.1.1 Factors influencing the risk of HCC development

The risk of HCC development is much greater in men for the majority of aetiolo gies.5 This is independent of the fact that males are more likely to develop chronic HBV carriage than

Incidence of HC

Diagnosis and treatment of HCC in adults

women. HCV may be a relative exception to this with a male to female ratio of 1.2:1 compared with 1.9:1 for HBV. The reasons for this are unclea r.6
( 2 ) A g e

The average age of HCC development in the UK is 66 years, which probably reflects the long term nature of most underlying liver diseases producing tumour development. This tumour is rare below the age of 45 years in areas with low lev- els of HBV infection. In high HBV prevalence areas, HCC has a bimodal age dist ribution with peaks at ages 45 and 65. 7
(3) The presence of hepatic cirrhosis

Cirrhosis is present in the vast majority of patients with HCC in the UK and Europe: estim ates vary between 90% and 95%.8 9 It is unclear if cirrhosis per se is biologically impo rtant in the tumo rigenic pathway, or if tumour development and fibrogenesis take place concu rrently but with fibrosis taking a sho rter time period. Noncirrhotic HCCs occur in young p atients (fibrolamellar variant) and in the elderly (apparent de novo HCC). Fibrolamellar HCC has an equal sex incidence and an average age at diagnosis of 30 years.10 Non-ci rrhotic HCC does occur in patients with viral liver disease, particularly HBV11 where direct viral inte gration into host DNA may play a role. 12 Noncirrhotic HCC is described in HCV13 and haemochrom atosis 14 but is rare.
(4) Aetiology of liver disease

Alcoholic cirrhosis carries an increased risk of HCC development; this risk is dif- ficult to quantify as mortality from continued alcohol consumption and cardiovascular disease is very high in this group. The available data suggest that abstinence from alcohol does not protect against HCC development, and that tumour development is seen in 14% of male cirrhotics per year, a similar level to that produced by HBV or HCV infection. 23 24 The rate of HCC development in women with alcohol related cirrhosis is more difficult to establish but seems significantly lower with few reports in the liter ature. 24 Primary biliary cirrhosis does carry a risk of HCC development but studies quanti ating risk in this group are rare. However, the available data suggest that women, even with esta blished cirrhosis, have a low risk but males have a similar risk to patients with alcohol related cirrhosi s.25 26 Wilsons disease is rare, but an increasing number of patients are surviving into adulthood with preexisting hepatic cirrhosis. HCC is well described despite adequ ate copper chel ating therap y, although the true incidence is difficult to esta blish. 27

There is a considera ble variation in the risk of HCC development in follow up studies of patients with cirrhosis of different aetiologies. Viral infection, either HBV or HCV, carries a high risk, with cirrhotic patients with either infection having appr oximately a 35% per year risk of HCC development. 15 16 In some studies the risk is even higher, up to 12% per year in HBV infected patients, 17 but this may represent patient selection; those with more severe liver disease may be at greater risk. In HCV infection there is compelling evidence that HCC develop- ment occurs with higher frequen cy at a very advanced stage of underlying liver disease18: up to 30% of patients undergoing liver transplant ation for end stage HCV cirrhosis have undetected HCC found in the explanted liver.19 In non-viral cirrhosis, again a great divergence of risk of HCC is seen with aetiology. Patients with cirrhosis due to genetic haemochrom atosis who were iron loaded at presenta- tion had a very high risk of HCC development (79% per year). 20 The risk falls with venesection but not to baseline lev- els (1 3% per year). 21 In contrast, patients with cirrhosis of autoim mune hep atitis have a very low risk of HCC develop- ment. Descriptions of HCC in the liter ature in this group are rare, despite a substantial number of cirrhotic individuals under long term follow up. Those which do exist suggest HCV coinfection may be an impo rtant facto r.22

5.2 Natural history of HCC HCCs develop as small nodule s. The majority of their growth takes place in an asymptom atic phase which may be years in length. Estim ated doubling times of HCC vary between one and 19 months, 28 29 with a median of six month s. There has been a suggestion that tumou rs with certain defined aetio- logies may have more aggressive behaviour but there are no conclusive data to suppo rt this. There are data as to survival in untre ated patients with HCC, showing that the major factors influencing overall survival are severity of underlying liver dysfunction and tumour size at initial detection. Between 50% and 90% of patients with Child-Pugh A cirrhosis will survive a year untre ated compared with only 20% with Child- Pugh C.2931 Small HCCs at present ation have relatively long tumour doubling times, and overall survival with tumou rs of less than 5 cm was 81100% at one year and 1721% at three years with no therap y.29 32 This suggests that if earlier diagnosis can be made, the oppo rtunity for inte rvention may be greater. 5.3 Potential screening tests for HCC AFP, a normal serum protein synthesised by fetal liver cells and yolk sac cells, is the most widely studied screening test used as a tumour marker for HCC. The normal range for AFP is 1020 ng/ml and a level >400 ng/ml is usually regarded as diagnosti c. Two thirds of HCCs less than 4 cm however have AFP levels less than 200 ng/ml and up to 20% of HCC do not produce AFP, even when very large. 33 Modific ations of AFP with differing carbo hydrate structures may occur in HCC and can be detected by altered patterns of lectin binding. These altered profiles have led to the development of alte rnative diagnostic tests 34 but none are widely available or have been shown to markedly enhance diagnostic ability over AFP. Desgamma-carb oxy prothrombin has been used as an alte rnative tumour marker for HCC; 67% of HCC have elevated levels but only 8% of small

(<2 cm) HCCs have abno rmal levels35 and it has not gained wide acceptance. Using AFP test- ing also produces false positives; levels in the range 20250 ng/ml are frequently seen in regener ating nodules in viral cirrhosi s.36 A rising AFP over time, even if the level does not reach 400 ng/ml, is virtually diagnostic of HCC. Ultrasound can detect large HCCs with high sensitivity and specificity. It is less able to reliably identify smaller lesions, which are required if better therapy is to be offered. Expertise of the operator and dedic ated equipment seem impo rtant in enhancing results; where this is available ultrasound detects 8595% of lesions 35 cm in diameter and can achieve 6080% sensitivity in the detection of lesions of 1 cm. 37 38 In the UK at present, detection of lesions below 2 cm by ultrasound is uncommon. Combining AFP and ultrasound improves detection rates. Ultrasound screening was initially suggested at six monthly intervals on the basis of tumour doubling times. There is some evidence that more frequent examin ations may enhance sen- sitivity but at the expense of more false positive test s.39 40 Patients with a negative ultrasound and an elevated but not diagnostic level of AFP appear to be at high risk and more fre- quent ultrasound examin ation in this group, probably three monthl y, may have a higher yield. 38 5.4 Screening studies in HCC There are no randomised controlled studies of screening for HCC development in cirrhosis of any aetiology. It is highly unlikely that any such randomised study could be unde rtaken now as surveillance of patients with cirrhosis is widely accepted and it would be almost impossible to recruit patients to a no screening arm of such a study. In the absence of such data, practice has been based on non-randomised studies either screening at risk popul ations or from clinic based series. There are a number of series demonstr ating the ability of either AFP alone as a screening investig ation or, more

Diagnosis and treatment of HCC in adults

liver mass on u trasound AFP raised norma AFP

logical assessment of the mass are required. If investig ations suggest HCC, then again biopsy of non-tumour liver will dete rmine the surgical approach. Radiological imaging can exclude benign liver lesions with a high degree of sensitivity and specificit y.51 52 54 Only in situ ations where considera ble doubt exists will biopsy of the lesion be required. Radiological imaging with ultrasound, CT, and angiograp hy usually unde rstages HCC.54 55 56 This however has little signifi- cance in the selection of patients for liver transplant ation as despite this unde rstaging, studies have shown that prognosis relates to preoper ative radiological assessment of the tumour rather than the actual stage at examin ation of the removed organ. 55 56 With hepatic resection however, the situation is (<2 cm diameter) on ultrasound, the level of diagnostic certainty over a diagnosis of HCC is lower; probably 75% of such nodules turn out to be HCC.52 Again, other radiological techniques or a raised AFP may esta blish a definitive diagnosi s. If not, either a repeat examin ation to show enlargement of the lesion, percu- taneous fine needle aspir ation, or biopsy may be indic ated. 53 The risk of seeding of HCC does not appear to be related to tumour size54 and if surgical therapy is possible biopsy should be avoided. In a patient not known to be cirrhotic, usually where the first present ation is with a liver mass, the initial investig ation should be AFP. If raised in the absence of a testicular prima ry, this confirms the diagnosi s. If the lesion is potentially operable, then biopsy of the non- tumour liver may be required to dete rmine the best treatment option. If AFP is normal, a search for other causes (non-li ver primary) and further radio-

Assess for surgery Diagnostic of HCC

CT/MRI/ lipiodo angiography

Major diagnostic doubt

Biopsy Figure 2 Investigating a mass in a cirrhotic liver. HCC, hepatocellular carcinoma; AFP, fetoprotein; CT, computed tomography; MRI, magnetic resonance imaging.

commonl y, the combin ation of AFP and ultrasound. The larg- est study of screening is in the Alaskan popul ation with a high HBV carriage rate. Screening was unde rtaken in the total popul ation with hepatitis B surface antigen (HBsAg) positiv- ity, irrespective of viral replic ation. The results of this study show that from 1982 to 1998, 18 299 AFP estim ations were unde rtaken in 2230 HBsAg positive individual s. Twenty patients developed HCC, five were inopera ble at present ation, 14 had resections, but six recu rred. 41 A similar study of patients with HBV, only 4% of whom had proven cirrhosis, detected 14 cancers in 1069 patients screened, with six curative surgical procedures unde rtaken. 42 Prospective studies of patients with viral cirrhosis have been carried out using ultrasound and AFP measurements and showed that 6487% of detected tumou rs were single and 43 75% were 3 cm or less in diamete r.4346 In these studies, 2966% of the detected can- cers were treated surgically with an attempt at cure. These studies are not directly compara ble with the situ ation in the UK as few centres had liver transplant ation available as a therapeutic option. 5.5 Diagnostic tests for HCC When a patient presents with a liver mass, irrespective of screenin g, there is a requirement to make a diagnosis and to stage the disease. The commonest clinical scenario is a patient with a mass discovered on ultrasound where AFP may or may not be raised. If the patient is known to have pre-existing cir- rhosis and the mass is greater than 2 cm in diamete r, there is a greater than 95% chance that the lesion is a HCC.47 48 If AFP is raised, this confirms the diagnosis and further investig ation is only required to esta blish the most approp riate therapy (fig 2). If AFP is normal, further radiological imaging (CT, MRI, or lipiodol angiography with follow up CT) will usually allow a confident diagnosis to be made 4951 and proceed to assessment of treatment without the need for biopsy. In the few cases where real diagnostic doubt persists, biopsy may be indic ated. In the uncommon situ ation of discovery of a small mass

more difficult as a more extensive tumour will not be treatable by a more extensive resection. The development of ultrasound contrast agents and the improving technology of MRI, using iron or gadolinium contrast agents, may enhance the ability to detect satellite or second distant lesions within the 58 liver.57 The exact role of this investig ation in treatment planning is unce rtain and while these may produce major changes in the diagnostic radiology of HCC in the near future, at present their role should be regarded as expe rimental. 5.6 Liver transplantation or surgery for HCC in cirrhosis There are no randomised controlled trials compa ring the out- come of surgical resection and liver transplant ation for HCC. Both techniques are primarily suited to small volume unifocal disease, and only a small propo rtion of patients with HCC in the UK will be suitable for either of these potentially curative treatment s. The decision as to which therapy is approp riate will depend on availability of resources and individual tumour characte ristics. Early results of liver transplant ation for HCC were poor 5961 with five year survival figures well below 50%, mainly due to tumour recu rrence. It is now clear that this was the result of poor selection of patients for transplant ation. It is well estab- lished that patients with single lesions of 5 cm diameter or up to three lesions of less than 3 cm in the absence of vascular invasion, as defined by imaging, have an almost zero recurrence rate for HCC and the prognosis after transplanta- tion is the same as for a similar underlying liver disease with- out HCC.62 63 64 Resection

of HCC is a viable option, with short term survival figures very similar to transplant ation. 60 61 6569 After three years of follow up however, there is a clear advan- tage for transplant ation in terms of tumour free survival. 66 69 Resection is only suitable for patients with excellent liver function (Child Pugh A) because of the high risk of hep atic decompens ation. Perioper ative mortality in expe rienced cen- tres remains bet ween 6% and 20% depending on the extent of the resection and the severity of preoper ative liver impai rment. 7073 The majority of this early mortality is due to liver failure. The residual liver after resection continues to have a malignant potential. Recurrence rates of 5060% after five years of follow up after resection are usual 74 75 and the majority of this recurrence is intrahep atic, representing either satellite nodules or do novo second tumour development. Small satellite nodules are not usually detected by preopera- tive imaging and where resection is unde rtaken the use of intraoper ative ultrasound may allow detection and better resection mar gins.76 Fibrolamellar HCC has a very different biology and arises in non-ci rrhotic liver. Surgical resection for this tumour is there- fore less likely to produce liver failure. The overall survival for fibrolamellar hepatoma at five years is 2536%.77 78 Because it arises without preexisting liver disease, fibrolamellar hep atoma usually presents with symptoms, and therefore even though resection may be unde rtaken there is often vas- cular or diaphragm atic involvement. Survival rates following resection vary from 65% to 12.5%.79 80 Liver transplant ation has

been perfo rmed for fibrolamellar hepatoma with five year sur- vival rates of 2849%.81 82 Tumour recurrence is however relatively common and resection remains the mainst ay of therapy for this rare tumou r, particularly in parts of the world where donor organ shortages exist. Appr oximately 7% of hepatocellular cancers, excluding the fibrolamellar variant, arise without cirrhosis. As with fibro- lamellar tumou rs these usually present with symptoms at an advanced stage. Survival overall tends to be longer than with tumou rs arising in cirrhosis, and five year survival after surgi- cal resection is appr oximately 25%.83 Liver transplant ation has been unde rtaken for tumou rs which are too extensive for resection but the results are poor. A review of 77 patients undergoing transplantaion for non-ci rrhotic HCC showed a five year survival of 11%.84 Such poor survival figures suggest that resection should remain the primary therapy and transplant ation will rarely be offered for this indic ation. In patients with cirrhosis, both resection and transplanta- tion probably have a role. In areas of the world where organ don ation rates cannot supply existing demand, resection is likely to be widely used. Transplant ation probably offers the best chance of cure for patients with small tumou rs and cirrhosis, and is therefore the treatment of choice, even in Childs A cirrhosis. In the UK at present, most patients requir- ing transplant ation are able to undergo this treatment without major delay and it should therefore be considered in all suitable cases. Any patient with a single small tumour (5 cm or less) should be assessed for surgery in a centre where resection or transplant ation is available. 5.7 Ablative therapy A number of non-sur gical therapies are in clinical use for HCC; percutaneous ablative therapies are well described, most com- monly using ethanol injection. Radiofrequen cy ablation is a relatively new technique where high frequen cy ultrasound probes are placed into a liver mass, usually under ultrasound control. Series show that tumour necrosis can be produced and that morbidity and mortality are low for both technique s. There are no randomised controlled trails of alcohol injection versus radiofrequen cy ablation or of either technique versus surgical resection, and assessment of the liter ature relies on compa risons of series of patients treated by the different tech- nique s. Although PEI has not been subjected to randomised controlled trials there is a considera ble liter ature on its use in HCC. In large series, complete response rates of 75% in tumou rs less than 3 cm in diameter have been repo rted, with five year survival rates of 3575%. 8588 These studies have gen- erally been restricted to patients with good underlying liver function. The largest series with 746 patients showed 79% and 47% three and five year survival, respectively, in patients with Childs A cirrhosis (239 patients) and 63% and 29% in Childs B cirrhosis (149 patients). 88 This study was rest ricted to tumou rs less than 5 cm in diamete r. The results of series from Asia and Europe report similar survival rates. One problem in inte rpret ation of the outcome in these series is that

biopsy or fine needle aspir ation diagnosis of HCC was not obtained in all series. This clearly could introduce a major error in series with a substantial number of lesions less than 2 cm in diameter where the diagnosis may be unce rtain. Treatment of larger and multiple lesions is possible, often requi ring repeated sessions and a general anaestheti c, but recurrence occurs in more than 50% at one year, and only 10% of 34 cm lesions were completely ablated. 89 Treatment is technically very difficult in lesions affecting the posterior segments of the liver.90 Compli- cations are uncommon, but seeding in the needle tract occurs in 3%91 and serious bile duct injury in 1%.86 92 The largest series to report complic ations 92 included 1066 patient s. The mean number of sessions needed to destroy a HCC nodule was 6.7. One death (0.09%) and 34 complic ations (3.2%) were

iii

repo rted, and eight episodes of bleeding and seven cases of tumour seeding occurred. Pain following injection requi ring cessation of therapy occurred in 3.2%. Compa rison with other techniques is difficult but most published series where compa risons with historical series were made showed little difference in survival rates between resection, transplant ation, and percutaneous alcohol injection in tumou rs less than 3 cm in diamete r. A study of 260 <5 cm tumou rs in Child A cirrhosis showed a three year survival of 79% for surgery and 71% for PEI compared with 26% for no treatment. 93 Similar results have been repo rted from other centre s.94 Most centres still regard surgery as the best proven therap y, providing a chance of cure, but PEI represents the best therapy for patients with small inopera ble HCCs. Expe rimental studies have been unde rtaken using agents other than ethanol as a tumour dama ging agent (cis- platinum, cold acetic acid) but to date none has shown convincing advantage s.95 96 Radiofrequen cy ablation of HCC is a relatively newly described technique using a probe placed into the tumour mass, usually percutaneousl y.97100 It uses high frequen cy ultrasound to gener ate heat at the probe tip which can destr oy tissue. A single probe can destroy lesions of up to 3 cm and a multiple tipped probe has been used to target lesions of up to 6 cm in diamete r. In a single series of 149 tumour nodules treated either percutaneously or at open oper ation, with an average tumour diameter of 3.5 cm, the local recurrence rate at 19 months was 3.6%, with all nodules showing initial complete ablation. 101 Distant metastases or a second tumour developed in 46%. Larger tumou rs can be treated by radio- frequen cy ablation; the largest series is 126 HCCs greater than 3 cm in diamete r. Complete necrosis was produced in 47%.102 There are few data on long term outcome in larger HCCs. A compa rison of 112 patients treated by percutaneous alco- hol injection or radiofrequen cy ablation showed that 47 of 52 treated by

radiofrequen cy ablation had complete tumour necrosis with a median of 1.2 treatment sessions versus 48 of 60 having complete ablation by alcohol injection with 4.8 ses- sions required. 103 The autho rs suggested that radiofrequen cy ablation was more effective but also had a higher complic ation rate. Unpu blished data suggest a possible higher rate of tumour seeding. Randomised trials of alcohol injection versus radiofrequen cy ablation are required. 5.8 Embolisation/chemoembolisati on Chemoembolis ation has been widely used as primary therapy for inopera ble HCC. The liter ature is difficult to inte rpret and compare, as the techniques used differ substantially and the patient groups treated are frequently those with very advanced disease where the risk of therapy as well as potential benefits may be greatest. Initial interest in radiological techniques producing tumour devascula risation developed in the 1970s.104 There is good evi- dence that it is effective at reducing tumour size105 106 and treating pain or bleeding from HCC.107 108 In all of the six initial randomised controlled trials of chemoembolis ation as prima ry treatment for HCC109114 none show any increase in survival, although tumour shrinkage was seen. All of these trials included patients with predominantly large tumou rs and severe underlying liver disease, which may have masked any beneficial effect. There is evidence from noncontrolled series that small HCCs are more likely to respond to chemoembolis ation. 106 This has recently been confi rmed in a trial of repe ated chemoembolis ation using lipiodol and doxo- rubicin versus arterial embolis ation without chemotherapy in patients with small tumou rs and good liver function. 115 In the 38 patients treated with chemoembolis ation, survival was 63% at two years versus 50% (n=34) in the embolis ation group, and 27% (n=35) in the untre ated arm. This study esta blishes the role of chemoembolis ation in the treatment of HCC but it

will only be applicable to a relatively small group of patients; 903 patients were screened for the trial to enrol 107. 112 This has been confi rmed in a further randomised trial. 116 It is also clear that patients responding to repe ated chemoembolis ation are likely to be the same group with a significant chance of response to ablative therap y, and randomised trials of these modalities are now required. Side effects of chemoembolis ation are those of the chemo- therapeutic agent used (usually doxorubicin) in addition to the complic ations of arterial embolis ation, pain, fever, hep atic decompens ation, and rarely infarction of organs other than the liver.117 118 Serious complic ations occur in 35% of treated patient s. A small number of studies have combined ethanol injection with chemoembolis ation. 119121 There is as yet no evi- dence available to suppo rt thi s. Hormona l manipul ation with tamoxife n has been the subject of randomise d clinical trials. Initial data suggestin g a positive effect on survival in patient s with inopera ble HCC122 123 has not been confi rmed in larger randomise d studie s.12412 6 Other agents with hormona l targets , stilbestrol , and flutamid e have been used in HCC but there is no evidence of effectiveness. Some autho rs have suggeste d that hormona l manipul ation may be effective where the oestroge n receptor status is known 127129 ; this is an area where clinical trials are required. Chemotherapy given intravenously has a very limited role in the treatment of HCC. The best single agent is doxorubicin with response rates of 1015%.130 131 More aggressive combina- tion chemotherapy regimens show no impr ovement in response rates and may even produce a reduction in survival of treated patient s.132 133 Any agents used in HCC should be given in the context of clinical trials. Interferon therapy may have a role in the prevention of HCC in hep atitis C cirrhosis. There is a scientific rationale for this therapy as interferon alpha has a broad range of antitumour activity and is known to be effective therapy for some haema- tological malignancie s. Initial data from both Japan and Europe showed a lower risk of HCC in cohorts of patients with hep atitis C cirrhosis who were given interferon therapy com- pared with those who were not treated. 134137 This effect was irrespective of the antiviral effects of interferon alpha, and was seen with treatment dur ation of only three months. These studies were not randomised controlled trials and have inher- ent selection bias. There is other evidence showing no effect of interferon on tumour development rates 138 139 and such tumour preventative therapy in patients with cirrhosis can only be currently recommended as part of clinical trials. Two other approaches to prevent tumour development have been used, retinoids and adaptive immunotherap y. Both of these ap- proaches have been used in the context of prevention of second tumour development after initial tumour resection or ablation. Adaptive immunotherap y, using primed peripheral lympho cytes, showed a significant increase in tumour free survival. 140 Retinoids and compounds involved in the vitamin A metabolic pathway are known to be differenti ation inducing agents with hypoprolifer ative effects. A single

study using retinol showed a 20% reduction in second tumour develop- ment in patients who had been treated with percutaneous alcohol injection. 141 Further studies are required in these area s. Interferon has been used as treatment for HCC rather than for the underlying viral infection. A randomised study using high doses of interferon alpha showed an improved survival142 but a more recent randomised controlled trial of more conventional doses of interferon alpha showed no impr ovement in survival and a high rate of side effects.143 A single randomised trial of octreotide therapy has suggested a survival benefit in HCC.144 Fifty eight patients were randomised to no therapy or octreotide 250 g twice daily. Median survival in the treated group was 13 months versus four in the control group. There was also a significant effect on AFP levels. A subsequent study of 70 patients treated with long acting octreotide showed no significant impr ove-

iii

ment in either survival, quality of life, or AFP levels.145 Further studies using octreotide or its longer acting analogues should be unde rtaken.
7

6.0 ACKNOWLEDGEMEN TS
The contents of these guidelines were reviewed and revised by Dr J OGrady, Dr M Lombard, Mr N Heaton, Professor OJ Garden, Dr D Patch, Professor M Bassendine, and Dr A Burroughs on behalf of the British Society of Gastroenterology Liver Section Committee. This guideline is due for review in 2005.
.............. ...... .

10

11

Authors affiliation
S D Ryder, Queens Medical Centre, Nottingham NG7 2UH, UK Correspon dence to: Dr S D Ryder; stephen.ryder@mail.qmcuhtr.trent.nhs.uk Accepted publication 30 November 2002 for
15 12 13 14

7.0 REFERENC ES
1 El-Serag HB, Mason AC. Rising incidence of hepatocellular carcinoma in the United States. N Engl J Med 1999; 340 :745 50. 2 Taylor-Robinson SD, Foster GR, Arora S, et al. Increase in primary liver cell cancer in the UK, 1979 1994 . Lancet 1997; 350 :1142 3. 3 Okuda K, Fujimoto I, Hanai A, et al. Changing incidence of hepatocellular carcinoma in Japan. Cancer R es 1987; 47:4967 72. 4 Okuda K. Clinical presentation and natural history of hepatocellular carcinoma and other liver cancers. In: Okuda K, Tabor E, eds. Liver Cancer. New York: Churchill Livingstone, 1997:1 12. 5 Bosch FX, Munoz N. Hepatocellular carcinoma in the world: epidemiological questions. In: Tabor E, Di Bisceglie AM, Purcell RH , eds. Etiology, Pathology and Treatmentof Hepatocellular Carcinoma. The Woodlands, Texas: Portfolio, 1991. 6 Lee CM, L u SN, Changchien CS, et al. Age, gender, and

16

17

18 19

20

local geographic variations of viral etiology of hepatocellular carcinoma in a hyperendemic area for hepatitis B virus infection. Cancer 1999; 86:1143 50. Lee HS, Han CJ, Kim CY. Predominant etiologic association of hepatitis C virus with hepatocellular carcinoma compared with hepatitis B virus in elderly patients in a hepatitis B endemic area. Cancer 1993; 72:2564 7. The Cancer of the Liver Italian Program (CLIP) Investigators. A new prognostic system for hepatocellular carcinoma: A retrospective study of 435 patients. Hepatology 1998; 28:751 5. Grando-Lemaire V, Guettier C, Chevret S, et al. Hepatocellular carcinoma without cirrhosis in the West: epidemiological factors and histopathology of the nontumorous liver. Groupe dEtude et de Traitement du Carcinome Hepatocellulaire. J Hepatol 1999; 31:508 13. Craig JR, P e te rs R L, Edmondson HA, et al. Fibrolamellar carcinoma of the liver: a tumour of adolescents and young adults with distinct clinicopathological features. Cancer 1980; 46:372 9. Okuda K, Nakashima T, Sakamoto K. Hepatocellular carcinoma arising in noncirrhotic and highly cirrhotic livers: a comparative study of histopathology and frequency of hepatitis B markers. Cancer 1982; 49 :450 5. Chen DS, Hoyer BH, Nelson J . Detection and properties of hepatitis B viral DNA in liver tissues from patients with hepatocellular carcinoma. Hepatology 1982; 2:42 -6S. De Mitri MS, Puossin K, Baccarini P, et al. HCV-associated liver cancer without cirrhosis. Lancet 1995; 345 :413 15. Goh J, Callagy G, McEntee G, et al. Hepatocellular carcinoma arising in the absence of cirrhosis in genetic haemochromatosis: three case reports and review of literature. E u r J Gastroenterol Hepatol 1999; 11:915 19. Sakuma K, Saito N, Kasai M, et al. Relative risks of death due to liver disease among Japanese male adults having various statusesfor hepatitis B s and e antigen/antibody in serum: a prospective study. Hepatology 1988; 8: 1642 6. Tsukuma H, Hiyami T, Tanaka S, et al. Risk factors for hepatocellular carcinoma among patients with chronic liver disease. N E ngl J Med 1993; 328 :1797 801. Tagger A, Donato F, Ribero ML, et al. Case-control study on hepatitis C virus (HCV) as a risk factor for hepatocellular carcinoma: the role of HCV genotypes and the synergism with hepatitis B virus and alcohol. Brescia HCC Study. Int J Cancer 1999; 81:695 9. Shiratori Y, Shiina S, Imamura M. Characteristic difference of hepatocellular carcinoma between hepatitis B and C-viral infection in Japan. Hepatology 1995; 22:1027 33. Faivre J, Forman D, E steve J, et al. Survival of patients with primary liver cancer, pancreatic cancer and biliary tract cancer in Europe. EUROCAREWorking Group. E u r J Cancer 1998; 34:1103 9. Yang Q, McDonnell SM, Khoury MJ, et al. Hemochromatosis associated mortality in the United States from 1979 to 1992: an analysis of Multiple-Cause Mortality Data. Ann Intern Med 1998; 129 :946 53.