New Zealand Liver Transplant Unit, School of Medicine, University of Auckland, Auckland, New Zealand
Screening for hepatocellular carcinoma in patients with chronic liver disease has been a controversial issue, despite the devastating outcome associated with delayed diagnosis. The only means to improve outcomes is by earlier diagnosis with regular surveillance of patients at greatest risk for this complication, namely those with cirrhosis and those with chronic hepatitis B infection. Established screening tests are serum alpha fetoprotein measurement and abdominal ultrasound. The optimal screening interval is six months, based on the average tumour doubling time. Recent studies have confirmed that screening does lead to the detection of hepatocellular carcinoma at an early stage when curative therapy is possible. Survival from the time of diagnosis is improved in screen-detected hepatocellular carcinomas, compared to incidentally detected tumours. In the only randomised control study of surveillance for hepatocellular carcinoma in a population with endemic hepatitis B virus infection, screening was also associated with an overall reduction in mortality from hepatocellular carcinoma. Screening for hepatocellular carcinoma does meet the cost-effectiveness threshold in both the cirrhotic and the chronic hepatitis B virus populations, although the inclusion of transplantation in the latter impacts negatively on cost-effectiveness. Screening for hepatocellular carcinoma is justified in both patients with cirrhosis and those with chronic hepatitis B virus infection.
Hepatocellular carcinoma (HCC) is the sixth most common cancer and the third leading cause of cancer death worldwide. More than half the estimated 800,000 cases per annum occur within the Asia-Pacific region, reflecting the prevalence of the major risk factors – chronic hepatitis B virus (HBV) infection, chronic hepatitis C virus (HCV) infection and aflatoxin exposure. Although universal neonatal HBV vaccination is already reducing HBV related cases, the incidence of HCC is expected to treble by 2050 because of the increasing numbers of cases related to the current epidemics of chronic hepatitis C and non-alcoholic fatty liver disease.
HCC is one of the few diseases where annual mortality exceeds incidence. This abysmal prognosis reflects the delayed diagnosis of this condition in most patients. In the absence of regular surveillance, diagnosis follows the presentation with symptoms which reflect advanced disease, either from a large tumour burden (cachexia, bone pain from distant metastases), liver failure from massive liver replacement (jaundice, encephalopathy, ascites, capsular pain) or direct vascular invasion and thrombosis of the hepatic veins, portal vein, vena cava and right atrium (variceal haemorrhage, ascites, lower limb oedema). The only possible therapeutic interventions in such cases are sorafenib or palliative measures. In most series, the median survival in patients with symptomatic HCC is less than three months, with five-year survival at <20%.
In contrast, the natural history of small hepatocellular carcinomas (ie.1-3cm) is vastly different, with five-year survival exceeding 50%. This reflects the suitability of small HCCs for curative therapies, namely surgical resection, liver transplantation, or radiofrequency ablation. However, early-stage HCC is always asymptomatic and can only be detected by imaging. This is the basis on which the argument for routine surveillance for HCC in at-risk populations has grown.
When assessing the benefit of any surveillance program, it is useful to determine whether this would meet all internationally accepted criteria for screening, as outlined by Wilson and Junger in 1968.1,2
1. There should be a high burden of disease in the population.
2. The population at risk for this disease, who would be targeted by the screening program, should be easily identifiable.
3. There should be a readily available screening method, which is associated with low morbidity, but which has a high accuracy for the identification of the screened disease.
4. The natural history of the disease is well understood and includes a preclinical period, during which screening can identify the disease at an early, potentially curable stage.
5. There is available a standardised recall procedure, which is acceptable to the targeted population.
6. There is evidence that earlier diagnosis by screening improves survival in those patients who develop the disease and also in the entire “at-risk’ population.
7. There is evidence that screening meets the cost-effectiveness threshold in that population (realising that this figure may depend on many factors, including direct medical costs, local reimbursement and finally the economic wealth of that country).
In this review, each of these aspects will be discussed to ascertain whether the argument for routine surveillance for HCC in at-risk populations can be recommended.
The incidence of HCC is determined by the prevalence of risk factors in the population. The two major risk factors are chronic hepatitis B infection and cirrhosis. HBV is one of the most important carcinogens in the world today, second only to tobacco as a cause of cancer deaths in males. It is directly responsible for almost 500,000 cases of HCC each year, chiefly in the Asia-Pacific and sub-Saharan Africa. The risk of HCC is not the same among all patients with chronic HBV infection, but is influenced by other factors such as gender, age, smoking,3 aflatoxin exposure,4 viral co-infections (HIV, HCV or HDV),5 family history and stage of liver disease.
A cost-effectiveness analysis has demonstrated that screening with six monthly alpha fetoprotein (AFP) and ultrasound is justified in any patient with chronic HBV infection, with an estimated annual incidence of HCC greater than 0.2%.6 This would include all HBeAg+ males and females, HBeAg negative males over the age of 40, HBeAg negative females over the age of 50, cirrhotics and anyone with a single first degree or at least two second degree relatives who have had a confirmed diagnosis of HCC.7,8,9 Recent sub-analyses of the REVEAL study (a prospective cohort study of 3653 Taiwanese patients with chronic HBV infection), would suggest that these baseline predictors may be replaced by quantitative serum HBV DNA measurements, since there is a direct correlation between HBV DNA titre and HCC risk. The risk exceeds 1% per annum for all patients with baseline HBV DNA >104 copies/ml.10 Other studies have suggested that maintained virologic suppression with long-term antiviral therapy may reduce HCC risk.11 Finally, it is generally accepted that baseline and serial HBV DNA levels are the best predictors of risk for HCC, but affordable, reproducible HBV DNA assays are not widely available for screening those populations in the Asia-Pacific and Africa with endemic HBV infection. Epidemiologic studies have also observed an increased risk of HCC in Asian patients infected with HBV genotype C compared to B.12,13
In HBsAg negative patients, surveillance for HCC is justified for those with an annual risk of HCC exceeding 1.5%.6,14 In this regard, established cirrhosis is a risk factor for HCC, although the link between fibrogenesis, regeneration and hepatocarcinogenesis in non-HBV cirrhosis remains poorly understood. The risk is highest in those patients with cirrhosis secondary to HCV, alcohol, non-alcoholic fatty liver disease and haemochromatosis. Although HCV core protein may have direct oncogenic effects, the few anecdotal reports of HCC developing in non-cirrhotic patients probably represent under-staging of liver fibrosis because of the sampling error of liver biopsy. The incidence of HCC has doubled in the west since 1983, despite a falling prevalence of HBV infection in these countries, reflecting the impact of the recent HCV epidemic.15,16,17 The incidence of HCV related HCCs is projected to treble in western countries during the next 20 years.18
The incidence of HCC is moderate (1-1.5%) in patients with autoimmune cirrhosis.19 In contrast, the incidence is low (<1%) in patients with alpha-1-antitrypsin deficiency, primary sclerosing cholangitis and primary biliary cirrhosis and the benefit of HCC surveillance in these patients is uncertain.20 One group at particularly high risk of developing hepatocellular carcinoma are those awaiting liver transplantation for decompensated cirrhosis.21
Patients with chronic hepatitis B infection or non-HBV cirrhosis are at highest risk for developing HCC and therefore should be considered for HCC surveillance. Chronic HBV infection is asymptomatic in the early stages and can only be detected through screening programs. Although national screening programs could be justified in countries with endemic HBV infection, especially those in the Asia-Pacific region, these are rare because of the considerable resources required, not only to screen, but then to provide long-term follow-up of all identified carriers. Instead, opportunistic testing is encouraged through public awareness and primary care campaigns targeting those from high-risk ethnic groups. Successful HBV screening programs exist in New Zealand, Shanghai, Taiwan and Alaska. HCC surveillance in patients with non-HCV cirrhosis is facilitated by the fact that most identified cirrhotics are already under regular secondary care follow-up at least six-monthly.
Optimal surveillance tests are serum AFP measurements and abdominal ultrasound examinations. Both are safe, non-invasive and reasonably inexpensive. Unfortunately, serum AFP is troubled by a lack of sensitivity (ranging between 39 and 61%). Almost one third of HCCs are not associated with elevated serum AFP or tissue expression of AFP, reflecting dedifferentiation of the tumour. Serum AFP also lacks specificity (ranging between 75 and 91%). Extrahepatic production of AFP may also occur in placenta or embryonic tumours and in population screening; the most common source of elevated AFP is pregnancy. In addition, AFP may be produced within the liver in the presence of active liver regeneration and is actually a useful prognostic marker in patients with acute hepatic failure.
The accuracy of AFP for the detection of HCC is influenced by the value of AFP adopted as the cut-off for normality. In a case-control study of 170 patients with HCC and 170 matched patients without HCC, an AFP cut-off of 20g/L had a sensitivity of 69% and a specificity of 89%. A cut-off of 100g/L increased specificity to 99%, but reduced sensitivity to only 31%. A cut-off of 400g/L did not increase specificity any further, but reduced sensitivity to 17%.22 In this particular study, the accuracy of AFP for the detection of HCC was lower in patients with chronic hepatitis B infection than in those with HCV-cirrhosis. The overall specificity of AFP for HCC is lower in the HBV population than in non-HBV cirrhotics because of the higher incidence of acute hepatitis flares in chronic hepatitis B. For these reasons, the recently updated American Association Study of Liver Diseases guidelines have dropped serum AFP measurements as recommended screening for HCC.23 Several new serum markers have been advocated in an attempt to improve the accuracy of non-invasive screening for HCC (Lectin-bound AFP (AFP-L3; DES-g-carboxy prothrombin; Protein-induced Vit K Antagonist-II (PIVKA-II); P53 antibodies; Glypican-3; serum Osteopontin; Golgi Protein 73; a-1-fucidase).24 Unfortunately, while many of these are more specific than AFP, this is at the expense of sensitivity. As a result, the accuracy of these serum markers is poor and none have yet been adopted into clinical practice.
Compared to AFP, ultrasonography is more sensitive (78%) but less specific (71%) for the detection of HCC.25 Although one study has suggested that six monthly AFP testing alone may be effective for the detection of early HCC,26 more recent studies suggest that a combination of serum AFP levels and abdominal ultrasound is a more accurate means of screening.27,28 Although CT is more sensitive and specific than ultrasonography and is the investigation of choice for the diagnosis of suspected HCC, repeated abdominal scans are associated with a significant cumulative radiation exposure (three abdominal CT scans have the radiation exposure of 60 milliGrays, equivalent to that of an atomic bomb survivor).29 Six monthly surveillance CT scans would be associated with a real increase in lifetime cancer risk. Repeated MRI is not associated with cumulative radiation exposure and is therefore not associated with this risk. However, it is expensive and not widely available for mass screening.30
Although the recently updated American Association Study of Liver Diseases guidelines23 have stated that surveillance for HCC should be with six-monthly ultrasonography and that AFP should only be used when ultrasound is not readily available. However, the accuracy of AFP as a screening test for HCC is higher in populations with a high proportion of non-cirrhotic cases (related to the prevalent HBV genotype). For this reason and reasons of cost and availability, most screening programs in countries with endemic HBV infection still utilise serum AFP in addition to ultrasound examinations.31,32
For HCC, cure is highest when the tumour is detected at an early stage prior to vascular invasion and extrahepatic spread. Cross-sectional studies have demonstrated a direct correlation between size of tumour and risk of vascular invasion. In a histopathologic study of more than 1000 explants following transplantation for HCC, vascular invasion could be demonstrated in more than 40% of tumours larger than 3cm, more than 50% in those larger than 5cm and more than 60% of those larger than 6.5cm.33 Radiofrequency ablation, resection and liver transplantation will achieve greater than 50% five-year survival in patients diagnosed with early-stage HCC (as defined by the Barcelona Clinic Liver Cancer classification ie. single tumour or up to three tumours <3cm).34
Studies of serial imaging of small untreated HCCs demonstrate that the median doubling time is six months. The estimated time interval for the HCC to grow from 1-3cm is 18 months.35,36 From these data, a screening interval of six months would seem optimal. When compared to tumours detected in a group of patients screened with annual surveillance, HCCs detected in patients undergoing six monthly ultrasound were more likely to be <3cm (76% v 42%) and within Milan criteria for transplantation (69% v 60%).37 However, a subsequent study in haemophiliacs with HCV found no difference.38 It should be noted that this study was flawed in that all HCV infected patients were included, not just patients with established cirrhosis. Hence the incidence of HCC in both groups was extremely low (<1% over 6 years), resulting in a type 2 error.
Most cirrhotic patients who are undergoing surveillance for HCC have the initial diagnosis of cirrhosis made during secondary care assessment. As a consequence of this diagnosis, most are receiving six monthly clinic visits, which include HCC surveillance, according to regional guidelines: European Association of the Study of the Liver, Asia-Pacific Association of the Study of Liver Diseases and American Association Study of Liver Diseases. In contrast, in countries with endemic HBV infection, less than 10% of patients will ever meet criteria for referral to secondary care. Even in those with a diagnosis of chronic HBV infection, local economic factors, lack of reimbursement and access to medical care prevent take-up of HCC surveillance. The only successful HCC surveillance programs in populations with endemic HBV are those funded and co-ordinated by the state.26,32,39
Multiple studies have demonstrated that surveillance for HCC in high-risk populations results in the detection of smaller tumours at an earlier, potentially curable stage, and subsequently an increased likelihood of suitability for curative surgical therapies, either radio-frequency ablation, resection or liver transplantation.26,40,41,42 These studies also demonstrate increased survival from the time of diagnosis of HCC, in patients with screen-detected HCC compared to patients with non-screen-detected HCC. The survival benefit is maximised when liver transplantation is available. This benefit is maintained after correction for lead-time bias of up to four years.
To answer the question as to whether screening for HCC will reduce HCC mortality in the overall target population, there needs to be a study comparing screening versus not screening for HCC in this population. The only randomised control study of screening for HCC was conducted in 19,000 HBV carriers in Shanghai. Screening reduced HCC mortality by 40% (from 132/100,000 per annum to 83/100,000 per annum). This was despite lack of availability of transplantation and poor access to resection.32 Attempts to reproduce this study in a western population have failed due to subject refusal to be randomised to an unscreened group.43 Of note, the 2003 European Association of the Study of the Liver HCC consensus noted a lack of randomised control studies of screening versus not screening, but concluded that such studies were now unethical given that effective therapy for small HCC tumours is available.
Cost-effectiveness has remained the most controversial issue concerning screening for HCC because cost-effectiveness is dependent on a number of factors: prevalence of HCC in the screened population; direct costs of screening and medical intervention; and local cost-effectiveness threshold – what is accepted as cost-effective in one country may not be considered so in another.
In countries with a high prevalence of HCC where the primary treatment modality is resection, the cost for detection of each treatable HCC is $12,000 and $26,000 for each year of life saved (all costs in US dollars).6,27 In western countries where the prevalence of HCC is low, the cost is between $18,000 per treatable HCC identified and between $25,000 and $50,000 per year of life saved.44-48 These figures are comparable to those for cervical cancer screening ($38,000), breast cancer ($30,000) and colonic cancer ($25,000). The availability of liver transplantation increases the survival benefit (from 0.5 QALY to 0.9 QALY), but significantly reduces the cost-effectiveness (from $30,000 per QALY to $60,000 per QALY). The widening gap between deceased donor supply and demand has resulted in increasing waiting times for liver transplantation and increased waiting list drop-off of patients with HCC (as high as 30% at 12 months).49,50 Despite initiatives such as the introduction of tumour Model of End-Stage Liver Disease (MELD) prioritisation for deceased donor organ allocation and the growth of live-donor liver transplantation, it is likely that transplantation will become a less available therapeutic option for screen detected HCC.
HCC surveillance with six monthly combination serum AFP measurement and abdominal ultrasound examinations is widely recommended in patients with cirrhosis and those with chronic HBV infection. HCC surveillance in these at-risk populations results in earlier detection of smaller HCC, increases the possibility of curative therapy, reduces mortality, reduces overall HCC related mortality in the screened population and meets the cost-effectiveness threshold. There is an ethical obligation to provide effective screening and treatment for HCC in patients at high risk for this complication. It is hoped that in the future, better screening methods and more widespread availability of liver transplantation and adjuvant anti-tumour therapies, which prevent recurrent HCC following resection or radiofrequency ablation, will further improve survival. The cost-effectiveness ratios of screening however, are likely to remain high.
12. Yang H, Chen P, Yeh S, Jen CL, You SL, Chen DS, et al. Risk of hepatocellular carcinoma associated with genotypes and mutants of hepatitis B virus: a community-based prospective cohort study. J Hepatol. 2006;44:S13.
19. Yeoman A, Al-Chalabi T, Karani J, Quaglia A, Devlin J, Mieli-Vergani G, et al. Evaluation of risk factors in the development of hepatocellular carcinoma in autoimmune hepatitis: Implications for follow-up and screening. Hepatology. 2008;48:863-70.
24. Gebo K, Chander G, Jenckes M, Ghanem KG, Herlong HF, Torbenson MS, et al. Screening tests for hepatocellular carcinoma in patients with chronic hepatitis C: a systematic review. Hepatology. 2002;36:S84-92.
33. Pawlik T, Delman K, Vauthey J, Nagorney DM, Ng IO, Ikai I, et al. Tumor size predicts vascular invasion and histologic grade: Implications for selection of surgical treatment for hepatocellular. Liver Transpl. 2005;11:1086-92.
38. Santagostino E, Colombo M, Rivi M, Grazia Rumi M, Rocino A, Linari S, et al. 6-month versus a 12-month surveillance for hepatocellular carcinoma in 559 hemophiliacs infected with the hepatitis C virus. Blood. 2003;102:78-82.
39. Robinson T, Bullen C, Humphries W, Hornell J, Moyes C, et al. The New Zealand Hepatitis B Screening Programme: screening coverage and prevalence of chronic hepatitis B infection. NZ Med J. 2005;118:1211.
45. Bolondi L, Sofia S, Siringo S, Gaiani S, Casali A, Zironi G, et al. Surveillance programme in cirrhotic patients for the early diagnosis and treatment for HCC: a cost-effectiveness analysis. Gut. 2001;48:251-9.
49. McCall J, Marui Y, Gane E, Marui Y, McCall J, Gane E, Holden A, Duncan D, Yeong ML, et al. Liver transplantation for hepatocellular carcinoma: a prospective intent-to-treat analysis. NZ Med J. 2005;118:1217.
50. Maddala YK, Stadheim L, Andrews JC, Burgart LJ, Rosen CB, Kremers WK, et al. Drop-out rates of patients with hepatocellular. Cancer listed for liver transplantation: outcome with chemoembolization. Liver Transpl. 2004;10:449–455.