Volume 4, Issue 1 , Pages 29-35, January 2006
Hepatocellular Carcinoma: Patients With Increasing Alpha-Fetoprotein But No Mass on Ultrasound
Article Outline
- Clinical Scenario
- The Problem
- Management Strategies and Supporting Evidence
- Areas of Uncertainty
- Published Guidelines
- Recommendations
- Suggested Reading
- Copyright
Abbreviations used in this paper: AFP, alpha-fetoprotein , AFP-L3, lens culinaris agglutinin-reactive AFP , CT, computed tomography , EASL, European Association for the Study of Liver Diseases , HBV, hepatitis B virus , HCC, hepatocellular carcinoma , HCV, hepatitis C virus , MRI, magnetic resonance imaging , US, ultrasonography
Clinical Scenario
A 63-year-old man with hepatitis C virus–induced cirrhosis is referred to the clinic for evaluation of an increase in the level of alpha-fetoprotein (AFP) to 400 ng/mL, which was detected in a blood test performed by the primary care physician. The patient is active at work and asymptomatic. He has compensated cirrhosis and well-preserved liver function, with a serum albumin level of 38 g/L, a serum bilirubin level of 1.6 mg/dL, and a prothrombin time of 1.2 international normalized ratio. Physical examination shows a liver span of 4 cm and splenomegaly without ascites. The ultrasound examination shows hepatomegaly without any liver mass, preserved portal blood flow, and splenomegaly.
What should be the diagnostic strategy for this patient? What is the role of monitoring AFP levels as a surveillance strategy for detection of hepatocellular carcinoma (HCC), or as a biomarker for diagnosis of HCC? What is the accuracy of ultrasonography, computed tomography (CT) scan, and magnetic resonance imaging (MRI) in the detection and staging of HCC?
The Problem
HCC is the third most common cause of cancer-related death worldwide. It is believed that the incidence will increase further in the next 2 decades, particularly in Europe and the United States, as a result of the spread of hepatitis C virus (HCV) and hepatitis B virus (HBV) infection during the past century. HCC is currently the leading cause of death in cirrhotic patients. Thus, HCC is an increasingly demanding health care problem.
HCC is generally the long-term result of chronic viral infection. Chronic HBV carriers have a 100-fold relative risk of developing HCC, and the annual incidence increases in HBV-related cirrhotic patients to 2%–6.6%. Other environmental carcinogens such as aflatoxin B1 enhance the neoplastic risk by 3-fold. In developing countries, HCC results from acquired HBV infection at birth or early in life, and universal vaccination of infants has been shown to prevent the development of the tumor significantly. By contrast, in developed countries, HCC arises in a cirrhotic liver because of HCV or alcohol abuse, with an annual incidence of 3%–5%. Because of the lack of availability of HCV vaccinations, HCC development can be prevented only by clearing the virus with effective antiviral combinations. In established cirrhosis, however, the effect of antiviral therapies in preventing the development of HCC is unproven.
The purpose of surveillance is to recognize HCC at an early stage, when the tumor potentially is curable, and, consequently, to decrease tumor-related deaths. Although unquestionable proof is lacking that surveillance significantly decreases cancer-related deaths, it is implemented widely in developed countries because it enables the detection of HCC at early stages when the tumors are amenable to curative treatments. Measurements of AFP levels and performing ultrasonography (US) examination every 6 months are recommended in cirrhotic patients in whom therapy could result in enhanced survival.
The AFP level is the most widely studied and used biologic marker for HCC detection and diagnosis. Its potential use may be confined to 3 different scenarios: first, as a biomarker for early detection of HCC within surveillance programs; second, as a diagnostic tool compared with other markers/imaging techniques; and, third, as a potential biomarker of treatment response. The first 2 scenarios are analyzed later, along with other situations in which AFP levels may be increased. In addition, a review of the accuracy of the state-of-the-art imaging techniques in the diagnosis of HCC is provided, considering the fact that the use of AFP is connected closely to improvements in imaging modalities.
Management Strategies and Supporting Evidence
Surveillance Programs and Alpha-Fetoprotein Levels
HCC satisfies the established criteria to recommend a surveillance program for a specific entity: (1) the disease is common with a clear-cut population at risk, (2) there must be a noninvasive and accurate diagnostic strategy, (3) there must be an established effective therapy, and (4) the screening program should be cost effective.
Target population
Surveillance programs for early detection of HCC should target a well-selected population at risk. Several epidemiologic studies have shown that the main risk factors for HCC are older age, male sex, and cirrhosis of any cause. The heterogeneous geographic distribution of HCC reflects different prevalences of risk factors among separate areas of the world. Therefore, restrictive criteria of age, stage of liver disease, or baseline conditions that would preclude radical therapies should be assumed when surveillance programs are organized.
Screening tests
The most-used screening tests for HCC are AFP determination and US. The available data on tumor growth suggest that the time length for an undetectable lesion to be detectable (diameter, ≃2 cm) is about 4–12 months. Thus, the suggested interval for surveillance in patients with cirrhosis has been set at 6 months. Patients with particularly high risk do not warrant a more intense surveillance schedule because higher risk does not mean faster tumor growth.
Alpha-fetoprotein as a biomarker for hepatocellular carcinoma surveillance
AFP is a poor screening test because it has a sensitivity of 40%–65%, a specificity of 62%–94%, and a positive predictive value of 9%–32% using a cut-off level of 20 ng/mL, the threshold used to trigger further testing (Table 1). Case-control studies mostly have addressed AFP as a diagnostic test. In 1 study, the Youden index (sensitivity + specificity − 1) for this cut-off value was found to be .45, which is particularly low. As a result of these data, monitoring of AFP levels as a surveillance strategy has been challenged, and it currently is agreed that it no longer should be used for HCC surveillance. On the contrary, persistently increased AFP levels have been shown to be a risk factor for HCC development, and thus this biomarker still is useful to define patients at risk. The value of the glycosylated fractions of AFP (AFP-L3), glypican-3, descarboxy prothrombin, or other biomarkers has yet to be established unequivocally.
Table 1. Accuracy of AFP, DCP, and AFP-L3 in the Diagnosis of HCC in Cross-Sectional and Prospective Studies
| Study | Study design | Marker | Cutoff level (ng/mL) | Sensitivity | Specificity |
|---|---|---|---|---|---|
| Sato, 1993 | Prospective | AFP-L3 | 15% ofAFPa | 55 | 68 |
| Sherman, 1995 | Prospective | AFP | 20 | 64 | 91 |
| Peng, 1999 | Case control | AFP | 20 | 65 | 87 |
| 200 | 45 | 100 | |||
| Cedrone, 2000 | Prospective | AFP | 20 | 55 | 88 |
| Ishii, 2000 | Prospective | DCP | 60 | 41 | 90 |
| AFP | 20 | 40 | 62 | ||
| Tong, 2001 | Prospective | AFP | 20 | 41 | 94 |
| Trevisani et al, 2001 | Case control | AFP | 20 | 60 | 91 |
| 200 | 22 | 99 | |||
| Nguyen, 2002 | Case control | AFP | 20 | 63 | 80 |
a AFP-L3: cut-off level reflects % of total AFP, in patients with AFP > 30 ng/mL. |
Ultrasonography
US is a much better surveillance tool than AFP. As a screening test in hepatitis B surface antigen carriers, US has a sensitivity of 71% and a specificity of 93%, but its positive predictive value is only 14%. Newer techniques such as contrast-enhanced US currently are not suitable for surveillance and the use of CT or MRI may be too expensive and/or invasive. An important aspect of surveillance is the expertise of the operator performing the US examinations because the efficacy of surveillance by US could be diminished by poor application of an otherwise effective resource.
Effective therapy
Around 40%–80% of patients with HCC detected in the setting of screening programs present with early tumors, and at least half of them may be treated with resection, transplantation, and percutaneous ablation. These treatments provide survival rates that clearly are better than their untreated counterparts (5-y survival rates, 40%–70% vs <20%), and thus are considered effective in well-selected patients.
Cost effectiveness
The economic benefits of surveillance programs have been assessed based on decision modeling and prospective studies. Biannual screening using AFP and US may provide substantial benefit (mean gain in life expectancy, 3–9 mo) only in well-targeted subgroups of patients (Child–Pugh class A). The cost effectiveness of these strategies ranged from $26,000 to $55,000 for each additional quality-adjusted life-year gained, even when only considering surgical resection as the sole radical option. These cost-effectiveness ratios are lower than the standard threshold quoted as cost effective ($50,000/quality-adjusted life-year), and compare favorably with other medical interventions considered reasonable, such as hemodialysis ($42,000/quality-adjusted life-year). In a recent prospective study, however, it was established that the cost per treatable HCC was $17,934, and the cost per year of life saved was as high as $112,993. Further studies assessing this issue are needed.
In summary, the use of AFP levels in screening programs for HCC is questioned. HCC is identified only as a result of an increase of AFP levels without any evidence of liver mass by US in a marginal number of cases. An exception to this might be the report of a 16-year population-based study of Alaskan natives who were HBV carriers without access to regular US facilities. Abnormal AFP levels were identified in 5% of the population and one third of those finally were diagnosed with HCC.
Alpha-Fetoprotein as a Diagnostic Test
Traditionally, the diagnosis of HCC is established on the basis of histologic or cytologic proof. Limitations to biopsy examination have been well documented for small and large tumors. In small early tumors, the rate of false-negative results may be as high as 40%, and some of the tumors are difficult to reach under US or CT-scan guidance because of their location. In larger tumors, the risk for complications, either intraperitoneal or intrahepatic hemorrhage or needle-track seeding, has been documented clearly in about 3% of cases. These high values certainly could be minimized using fine needles and puncturing the tumor through nontumorous parenchyma but, even in experienced hands, a small rate of serious complications remains. The establishment of a firm diagnosis of well-differentiated HCC with a small histologic sample can be difficult, and may require additional samples or cytologic smears, with each of these procedures carrying a risk. These limitations have led to the use of noninvasive criteria, which chiefly rely on imaging techniques, although biomarkers eventually might play a role.
Alpha-fetoprotein levels
In cirrhotic patients with chronic hepatitis infection, high AFP levels may reflect both exacerbations of the infection that may last for months, particularly HBV infection, and the presence of HCC. The value of AFP levels as a diagnostic test have been tested in case-control and prospective studies. Traditionally, a cut-off level of 400 ng/mL, or even 500 ng/mL in a cirrhotic patient with a given liver mass, was considered diagnostic of HCC. As a consequence, a cut-off level of 400 ng/mL was proposed as noninvasive diagnostic criteria in the European Association for the Study of Liver Diseases (EASL) guidelines. More recently, some studies have shown that a cut-off level of 200 ng/mL led to a sensitivity of 20%–45%, a specificity of 99%–100%, and a very high positive likelihood ratio of around 30. By using receiving operating characteristic curves, it was suggested that the cut-off level of 100 ng/mL had a high specificity value compared with other levels (98%), but with a low sensitivity (22%) (Figure 1). Thus, the conclusions of these studies are that AFP levels of 100–200 ng/mL constitute a useful confirmatory test in patients with a given liver mass.
Figure 1.
Receiver operating characteristic curve of serum AFP level to discriminate between cirrhotic patients with and without HCC. Reprinted with permission from Trevisani et al,6 © 2001 The European Association for the Study of the Liver.
Accuracy of imaging techniques in the diagnosis of hepatocellular carcinoma
Although US is a useful tool for screening, it is operator dependent and thus the accuracy of the results may vary throughout centers. Nevertheless, only a small number of lesions remain undetected even in expert hands. This is because some lesions may be difficult to interpret according to patient characteristics (ie, obesity), tumor characteristics (isoechoic) or location (subdiaphragmatic area), and characteristics of the nontumorous parenchyma (heterogeneous pattern in steatosis or macronodular cirrhosis).
The best techniques for the diagnosis of HCC are helical CT and MRI with contrast enhancement, which have an accuracy exceeding 80%–90%. Characteristically, HCC lesions show an uptake of contrast in the arterial phase with wash-out in late phases. These techniques have replaced angiography and CT hepatic angiography completely. Lipiodol CT is not reliable because of the high rates of false-positive results, and positron emission tomography has low accuracy. The accuracy of CT and MRI, however, decreases with regard to the extent of the disease. Studies investigating liver explants showed sensitivities for HCC detection with helical CT or MRI, on a per-nodule basis, ranging from 52%–76% (Table 2). In a recent comparison with pathologic examination of explanted livers, MRI angiography was more precise in the detection of nodules between 1–2 cm than CT scan (89% vs 65%). However, 20%–30% of intrahepatic tumors, particularly those less than 10 mm, still remain undiagnosed preoperatively with any technique.
Table 2. Accuracy of Dynamic Spiral CT and Gadolinium-Enhanced Dynamic MRI in the Detection of HCC
| Study | No lesions/patients | Sensitivity for HCC | |||
|---|---|---|---|---|---|
| All | <1 cm | 1–2 cm | >2 cm | ||
| Krinsky, 2001 | 19/71 | CT: 10/19(53%) | 1/3(33%) | 6/12(50%) | 3/4(75%) |
| De Ledinghen, 2002 | 54/34 | CT: 28/54 (52%) | 5/15(33%) | [23/39(59%)] | |
| MRI: 33/54 (61%) | 4/15(26%) | [29/39(74%] | |||
| Burrel et al, 2003 | 76/29 | CT: 43/70(61%) | 2/20(10%) | 17/26(65%) | 25/25(100%) |
| MRI: 58/76 (76%) | 8/23(34%) | 25/28(89%) | 24/24(100%) | ||
New advancements in both CT and MRI imaging may improve tumor detection further. Ideally, state-of-the-art MRI should be able to obtain slices of 2.5 mm using a 3-dimensional dynamic sequence in a single breath-hold, preferentially using a gadolinium-enhanced angiographic technique. Several contrast agents have been tested for HCC and are reviewed elsewhere. There is no evidence that any of them provide better results than the technique described earlier. However, the so-called double-contrast MRI technique (with superparamagnetic iron oxide and gadolinium) is particularly promising, and might improve the accuracy of conventional MRI further. Multidetector-row CT scan also is expected to enhance the detection of additional small tumors less than 1 cm in size. The accuracy of these new technologies has to be defined using histopathologic examination of explant livers as the gold standard.
Therefore, the state-of-the-art imaging techniques, whether associated or not with high levels of AFP, allow recognition of the main HCC nodule in almost all of the cases. Exceptions to that rule are diffuse tumors, and also small tumors with atypical vascularization. On the other hand, precise staging/disease extension can be established in only 70%–80% of cases.
Other causes of increased levels of alpha-fetoprotein
High titers of AFP (>100 ng/mL) have been described in acute liver failure, flairs of chronic hepatitis, pregnancy, massive metastatic liver involvement, and in the presence of germ-cell tumors.
Areas of Uncertainty
New Biomarkers for Detection of Early Hepatocellular Carcinoma
AFP has arrived at the end of its use as a biomarker for early detection of HCC. There is a lack, however, of alternative biomarkers for this neoplasm. The Early Detection Network of the National Cancer Institute has established a 5-phase process to define the current status of biomarkers. Phase 1 includes preclinical exploratory studies, either with serum or tissue. Some markers have been identified at this level, such as glypican-3, golgi protein 73, HCC-associate gene 1, insulin growth factor-1, phospholipases A2, plasma glutamate carboxypeptidase, and heat shock protein 70, among others. Genomic studies are underway to expand and refine the list of potential biomarkers for HCC. Only 2 markers have been explored in other phases in the setting of cross-sectional and prospective studies: des-gamma carboxy prothrombin and AFP-L3 (Table 1). Overall, the results show that descarboxy prothrombin, an abnormal prothrombin protein generated by malignant cells, is a slightly better marker than AFP, but confirmatory studies are needed to prove the value of this biomarker in early tumors. Results of AFP-L3 are more difficult to interpret because in most of the studies an abnormal level of AFP was required for AFP-L3 measurement.
Surveillance Programs in Hepatocellular Carcinoma: Recall Policy
One of the main uncertainties for surveillance programs is what recall policy should be applied. Suspicion of HCC is raised on detecting a hepatic nodule of 10 mm or greater by US or significant increases in the AFP level. Recall procedures in the presence of a nodule in a cirrhotic liver will vary according to the diameter of the lesion (Figure 2). A firm diagnosis of HCC still is difficult to reach when the nodule is less than 1 cm in diameter, and short-term follow-up evaluation is suitable in this setting. On follow-up imaging, the nodule either could vanish, particularly in the case of heterogeneous steatosis, persist in cases of monoclonal macronodule (dysplastic nodule) or HCC, or even show limited growth, which is almost synonymous with HCC. Nodules sized 1–2 cm in diameter should be evaluated further with either CT scan, MRI, or both, although the diagnosis may require histology. Conversely, in nodules greater than 2 cm the diagnosis of HCC is highly probable, and a firm diagnosis can be established by noninvasive criteria, as described later. Suspicion of HCC based on increases in AFP level is more complex. This last criterion is defined by doubling baseline AFP levels of 20 ng/mL or greater if previously normal, or AFP levels greater than 40 ng/mL if previously unknown. These patients will be explored further by spiral CT and/or US-guided biopsy examination. Patients presenting with a detectable hepatic nodule less than 10 mm are followed-up in 3 months to detect further growth or monitor its stability. On detecting tumor growth or an increase in AFP concentration, HCC suspicion is confirmed by additional tests.
Figure 2.
Surveillance and recall strategy for HCC. Endorsed by the EASL panel of experts on HCC. *Available for curative treatments if diagnosed with HCC. **AFP levels to be defined. ***Pathologic confirmation or noninvasive criteria. Reprinted with permission from Bruix et al,2 © 2001 The European Association for the Study of the Liver.
Diagnosis confirmation
The confirmatory diagnosis of HCC is based on cytohistologic findings. According to the EASL guidelines, noninvasive criteria are established in cirrhotic patients by the concomitant finding of 2 imaging techniques showing a nodule greater than 2 cm with arterial hypervascularization, or by a single positive imaging technique associated with AFP level greater than 400 ng/mL. For that purpose, acceptable imaging techniques are US, spiral CT, angiography, and MRI.
Published Guidelines
There is no specific guideline published to assess high levels of AFP. There are 2 published guidelines on surveillance and management of HCC patients, 1 from EASL and 1 from the National Cancer Institute in the United States. The guidelines of the American Association for the Study of Liver Diseases have just been published supporting surveillance programs in cirrhotic patients only by ultrasonography every 6 months, without the need of AFP measurements.
European Association for the Study of Liver Diseases Guidelines
The EASL expert panel proposed the following surveillance recall and diagnostic strategy (Figure 2). In nodules less than 1 cm, which are malignant in less than half of patients, reliable HCC diagnosis is difficult. Thus, close follow-up evaluation is recommended. In nodules of 1–2 cm, HCC diagnosis requires positive cytology or histology. However, there is a 30%–40% false-negative rate with fine-needle biopsy examination. A negative result, therefore, does not rule out malignancy. As described previously, noninvasive diagnostic criteria to be applied solely in cirrhotic patients with tumors greater than 2 cm were proposed.
National Cancer Institute Guidelines
The National Cancer Institute guidelines for HCC currently do not endorse any surveillance or recall policy. The guidelines state that no surveillance strategy has been shown to decrease cancer-related death, and thus additional data are needed.
Recommendations
General Recommendations
AFP levels no longer should be used as a screening test because of the low accuracy of this marker. The usefulness of other markers such as descarboxy prothrombin or fractions of AFP (AFP-L3) has not been proven and further research is needed.
AFP levels can be useful as a confirmatory test for the presence of HCC once a liver mass is detected in a cirrhotic patient. The accepted cut-off value for diagnostic purposes is greater than 400 ng/mL in cirrhotic patients with a liver mass. This cut-off value may be reduced in the future to 100 ng/mL.
US at 6-month intervals is the ideal tool to be used for surveillance of HCC in cirrhotic patients, and allows the detection of 80%–90% of cases. The major flaw of this technique is that it is operator dependent.
Noninvasive criteria for HCC diagnosis in cirrhotic patients have been proposed: 2 imaging techniques showing characteristics of a hypervascularized mass of greater than 2 cm in diameter in the arterial phase.
CT scan and MRI angiography are the best techniques for the study of HCC disease extension, once the malignant nature of a suspicious nodule has been confirmed. These techniques have replaced angiography and CT hepatic angiography. Lipiodol CT and positron emission tomography are not reliable.
Specific Recommendations for the Case
In the setting of cirrhosis, AFP levels of 400 ng/mL point to the diagnosis of HCC, which should be detected by US in the majority of patients. If this is not the case, then the following situations are possible (Figure 3):
Figure 3.
Diagnostic algorithm for patients with increased AFP levels (>400 ng/mL) and normal US. *Diagnosis obtained in most of the patients. **Almost never needed currently. Aids in diffuse HCC diagnosis. ***Anecdotic use. Aids in diffuse HCC diagnosis. mets, metastases.
The approach in these cirrhotic patients would be to perform contrast CT scan and/or MRI angiography to detect atypical or infiltrating HCC. If both investigations fail to detect an HCC, even when repeated at 3 months, digital angiography might add some clues in unusual cases. Finally, transjugular biopsy examination to discard a diffuse HCC should be considered in the anecdotic cases in which all of the earlier-described factors are negative. This patient presented with CT scan and MRI findings consistent with an infiltrating HCC.
Suggested Reading
- 1. . Hepatocellular carcinoma . Lancet . 2003;362:1907–1917
- 2. Clinical management of hepatocellular carcinoma . Conclusions of the Barcelona-2000 EASL Conference. J Hepatol . 2001;35:421–430
- 3. . Hepatocellular carcinoma (present status and future prospects) . J Hepatol . 2003;38:S136–S149
- 4. . Alpha fetoprotein (an obituary) . J Hepatol . 2001;34:603–605
- 5. . Test characteristics of alpha-fetoprotein for detecting hepatocellular carcinoma in patients with hepatitis C . Ann Intern Med . 2003;139:46–50
- 6. Serum alpha-fetoprotein for diagnosis of hepatocellular carcinoma in patients with chronic liver disease (influence of HbsAg and anti-HCV status) . J Hepatol . 2001;34:570–575
- 7. Magnetic resonance imaging of liver tumors . J Hepatol . 2004;40:162–171
- 8. Prospective analysis of risk factors for hepatocellular carcinoma in patients with liver cirrhosis . Hepatology . 2003;37:520–527
- 9. . HCC surveillance (who is the target population?) . Hepatology . 2003;37:507–509
- 10. . Screening for hepatocellular carcinoma in chronic carriers of hepatitis B virus (incidence and prevalence of hepatocellular carcinoma in a North American urban population) . Hepatology . 1995;22:432–438
- 11. . New markers for hepatocellular carcinoma . Gastroenterology . 2004;127:S113–S119
- 12. MRI angiography is superior to triphasic helical CT for detection of small HCC prior to liver transplantation. Comparison with pathologic correlation . Hepatology . 2003;38:1034–1042
- 13. National Cancer Institute. US National Institute of Health. Available at: http://www.cancer.gov.
- 14. . Management of hepatocellular carcinoma . Hepatology . 2005;42:1208–1236
J. M. Llovet, Professor of Research–Institut Catala de Recerca Avançada (ICREA), is supported by the ICREA and the Instituto de Salud Carlos III (Fondo de Investigaciones Sanitarias 2002–2005, PI02/0596).
PII: S1542-3565(05)00864-5
© 2006 American Gastroenterological Association. Published by Elsevier Inc. All rights reserved.
Volume 4, Issue 1 , Pages 29-35, January 2006
