Surveillance and diagnosis for hepatocellular carcinoma


  • M. Rasool Aljabiri,

    1. The Sheila Sherlock Hepatobiliary-Pancreatic and Liver Transplantation Unit, The Royal Free Hospital, Hampstead, London, United Kingdom
    Search for more papers by this author
  • Francesca Lodato,

    1. The Sheila Sherlock Hepatobiliary-Pancreatic and Liver Transplantation Unit, The Royal Free Hospital, Hampstead, London, United Kingdom
    Search for more papers by this author
  • Andrew K. Burroughs

    Corresponding author
    1. The Sheila Sherlock Hepatobiliary-Pancreatic and Liver Transplantation Unit, The Royal Free Hospital, Hampstead, London, United Kingdom
    • The Sheila Sherlock Hepatobiliary-pancreatic and Liver Transplantation Unit, Royal Free Hospital, Pond St, Hampstead, London NW3 2QG, UK
    Search for more papers by this author
    • Telephone: 0044-207-472-6229; FAX: 0044-207-472-6226

The most common primary malignancy arising within the liver is hepatocellular carcinoma (HCC) and almost always in the setting of cirrhosis. HCC is the fifth most common solid tumor in the world and accounts for ∼500,000 deaths each year.1–3 Data on the epidemiology and natural history of chronic hepatitis C virus infection show that its frequency has increased over the past 20–30 years in the United States3 and Europe4 (Table 1). HCC is the leading cause of death in patients with cirrhosis in Europe.5 Although only a minority of patients with cirrhosis diagnosed with HCC have tumor amenable to potential curative therapy, all therapies have best results with single nodules ≤3 cm in size,6 including liver transplantation (LT).7, 8

Table 1. Incidence and Risk Factors for HCC in Europe
IncidencePatients with cirrhosisFive-year survival 15–20%
  1. silver staining of nucleolus region (AgNor); proliferating cell nuclear antigen (PCNA).

  First cause of death
PrevalenceChild-Pugh A5%
 Variceal bleeding15%
 Spontaneous bacterial peritonitis20%
Risk factorsClinicalMale gender, older age, AFP, viral etiology, portal hypertension, platelet count
 HistologicalAgNor, PCNA, cell dysplasia, irregular regeneration pattern

This clinical scenario has important implications for diagnosing HCC at an early stage. Although surveillance is practiced in many units, there is only one randomized controlled trial concerning surveillance for HCC in cirrhosis. All others are cohort studies. Two well-designed studies show a higher chance of receiving treatment at an earlier stage,9, 10 strongly supporting current practice. A theoretical cohort of patients with Child class A cirrhosis was studied, and researchers found that with incidence of HCC of 1.5% a year, surveillance resulted in an increase in longevity of approximately 3 months.11 However, if the incidence of HCC was 6%, the increase in survival was approximately 9 months. However, this study did not include transplantation as a treatment option. One study that used a similar analysis that did include LT suggested that surveillance became cost-effective when the incidence of HCC exceeded 1.4% per year. These data suggest that in patients with cirrhosis of varying causes, surveillance may be effective when the risk of HCC exceeds 1.5% per year.6, 10–12.

The single randomized controlled trial of surveillance surveillance every 6 months has shown a survival benefit of 6-monthly surveillance of ultrasound with alfa-fetoprotein (AFP). This study included 18,816 people aged 35–59 years with hepatitis B virus infection or a history of chronic hepatitis in urban Shanghai, China. Participants were randomly allocated to surveillance (9,373) or no surveillance (9,443). Participants were invited to have an AFP test and undergo ultrasound examination every 6 months. When the surveillance group was compared with the control group, the number of HCC cases was 86 vs. 67; subclinical HCC was 52 (60.5%) vs. 0, and small HCC was 39 (45.3%) vs. 0 respectively. Resection was achieved in 40 (46.5%) vs. 5 (7.5%) in whom it was not achieved. The 1-, 3-, and 5-year survival rates were 65.9%, 52.6%, and 46.4% vs. 31.2%, 7.2%, and 0, respectively. Thirty-two people died from HCC in the surveillance group vs. 54 in the control group, and the HCC mortality rate was statistically significantly lower in the surveillance group than in controls, being 83.2 of 100,000 and 131.5 of 100,000, respectively, with a mortality rate ratio of 0.63 (95% confidence interval, 0.41–0.98).13

The knowledge that all treatments yield better results when used for smaller and unifocal HCC6 drives current diagnostic practice. Staging of patients with HCC has prognostic significance and determines therapy, including surgery. For both LT and resection, survival is higher for patients with a single nodule <5 cm in size or with 3 nodules each ≤3 cm in diameter.7


In 2001 experts under the aegis of the European Association for the Study of the Liver published a consensus statements to standardize the diagnostic and therapeutic approach to HCC.6 These have evolved, and recent practice guidelines from the American Association for the Study of the Liver (AASLD) have included these data and data from later studies14 (Fig. 1).

Figure 1.

Suggested algorithm for evaluation of a liver mass detected during hepatocellular carcinoma surveillance in a patient with cirrhosis according to the American Association for the Study of Liver Diseases (AASLD) guidelines. From Bruix J, Sherman M, Practice Guidelines Committee, American Association for the Study of Liver Diseases. Management of hepatocellular carcinoma. Hepatology 2005;42:1208–1236.

The European Association for the Study of the Liver (EASL) recommendations defined histological and radiological criteria for the identification of HCC. Surveillance is recommended for patients with cirrhosis in whom potentially effective treatment for HCC can be offered if tumors can be detected early. The available data on tumor growth suggest that the interval from undetectable lesions to those 2 cm diameter is approximately 4–12 months.15, 16 Thus, with the aim of detecting HCC <3 cm, the suggested interval for surveillance is 6 months. This surveillance is also recommended by AASLD guidelines. Many centers shorten this to 3 months while the patient is on a transplant waiting list.

In patients carrying hepatitis B surface antigen who do not have cirrhotic livers, data (M. Sherman, unpublished data) suggest that surveillance becomes cost effective when the incidence of HCC exceeds 0.2% a year.

Cirrhosis is a disease that produces pathological changes, such as regenerative and dysplastic nodules (DN), so it is sometimes impossible to discriminate these from a small HCC. During the evolution from a benign regenerative nodule to low- and high-grade DN and finally to HCC, angiogenesis makes the arterial blood the main blood supply. Therefore, differences in focal lesion perfusion can help to characterize the nodules in a cirrhotic liver.17–20

The EASL consensus statements also categorized patients with cirrhosis and HCC on the basis of size of the suspected nodule. For tumors >2 cm, diagnosis can be made by using 2 noninvasive radiological techniques demonstrating arterial hypervascularity.21–23 When AFP is >400 ng/mL, only one imaging technique was considered sufficient to confirm the diagnosis. By means of only radiological diagnostic techniques and clinical findings, the positive predictive value is >95%.23, 24

However, there was no mention of discordant vascular patterns in the radiological imaging. AASLD guidelines revised these statements, giving diagnostic importance to venous/portal washout, together with arterial hypervascularization, for the diagnosis of HCC. Moreover contrast-enhanced (CE) ultrasonography (US) was added as a noninvasive additional diagnostic technique to spiral computed tomography (CT) and magnetic resonance imaging (MRI), and only one technique showing both arterial hypervascularity and venous washout was needed or an AFP level of >200ng/mL even if the washout pattern is not typical.

In cases of an atypical vascular pattern, a liver biopsy is recommended. However, no mention is made of the diagnostic accuracy of histology for these nodules or for the applicability of biopsy (e.g., difficult location or poor coagulation), which is even more the case when the nodule is <2 cm in diameter. In addition, histological diagnosis may have less accuracy because of better differentiation in smaller nodules.25, 26

When the nodule is 1–2 cm in size, EASL suggests proceeding to needle liver biopsy to confirm the diagnosis, irrespective of the vascular profile at imaging. AASLD has refined this; the guidelines suggest performing 2 imaging techniques and proceeding to liver biopsy only in cases of discordant vascular pattern or atypical vascular pattern between the 2 techniques. For nodules <1 cm in size, EASL recommends serial US examinations every 3 months until the nodule >1 cm, and the same approach is suggested by AASLD guidelines.

After the detection by US of a suspicious nodule, both EASL statements and AASLD guidelines require confirmation by dynamic techniques such as CEUS, spiral CT, or dynamic MRI. However, what may be the single best technique after US is not well established.27 A current algorithm is shown in Figure 2.

Figure 2.

Algorithm for characterization of small nodules in cirrhosis by assessment of vascularity. From Bolondi L, Gaiani S, Celli N, Golfieri R, Grigioni WF, Leoni S, et al. Characterization of small nodules in cirrhosis by assessment of vascularity: the problem of hypovascular hepatocellular carcinoma. Hepatology 2005;142:27–34.


HCC, hepatocellular carcinoma; LT, liver transplantation; AFP, alfa-fetoprotein; AASLD, American Association for the Study of the Liver; EASL, European Association for the Study of the Liver; DN, dysplastic nodule,; CEUS, contrast-enhanced ultrasonography; US, ultrasonography; CT, computed tomography; MRI, magnetic resonance imaging; SPIO, superparamagnetic iron oxide; RES, reticuloendothelial; MELD, Model for End-Stage Liver Disease; CLIP, Cancer of the Liver Italian Program Investigators; BCLC, Barcelona Clinic Liver Cancer.


Serum AFP

Serum AFP is not a surveillance test. If US suggests a nodule, then AFP should be measured. A concentration of 200 ng/mL in the presence of imaging diagnosis of a nodule fulfilling HCC criteria in a patients with cirrhosis is considered diagnostic of HCC. A multicenter survey examined serum AFP in 1,158 HCC patients on the basis of the AFP level at diagnosis. The study found that as a diagnostic tool, serum AFP levels (>20 ng/mL as a cutoff) performed poorly, with a sensitivity only of only 54%.28–31

Des-Gamma-Carboxy Prothrombin

Des-gamma-carboxy prothrombin is also known as prothrombin induced by vitamin K absence II (PIVKA II). It is a potential diagnostic test, but it has not been shown to provide an advantage compared with AFP.28–30

US, Doppler US, and Power Doppler US

When the presence of disease in the explanted liver is used as the gold standard, the sensitivity of US ranges 33–72%,32–35 in part because the livers were explanted months after initial diagnosis. However, US surveillance may miss nodules in cirrhosis. Apart from operator experience36 and equipment, patient body mass and intestinal interposition may influence detection rates, resulting in sensitivity rates of 35–84% in older studies.36 The false-negative nodules were all <2.5 cm in the single study that reported this33; US specificity ranged 92–100% (Table 2).

Table 2. Ultrasound Lesion-by-Lesion Sensitivity (%) and Specificity (%) Compared With Explanted Livers
StudyNo. HCC/No. totalTime from imaging to transplantationSensitivitySpecificity
  1. Abbreviations: HCC, hepatocellular carcinoma; ND, not done; NA, not applicable.

Gambarin-Gelwan et al. 200019/106All <6 mo5894
Kim et al. 200116/52Mean 57 d3392
Rode et al. 20016/43ND4695
Teefey et al. 20039/25ND72NA

At US examination, HCC may appear as round or oval lesions; small tumors usually have smooth and sharp margins. Nodular-type HCC with extranodular growth and multinodular-type HCC may have irregular margins. Small HCC may be hypoechoic, hyperechoic, or even isoechoic compared with the surrounding liver. Approximately 75% of small (≤2 cm) HCC lesions are hypoechoic; the hyperechoic appearance usually reflects fatty changes of the tumor, and its differential diagnosis may be confused with hemangioma. As HCC nodules grow, they tend to develop hypoechoic peripheral rims and become more heterogeneous.

The use of Doppler US or power Doppler US may help establish a diagnosis. HCC lesions usually have florid arterial vascularization, and this is seen as an intralesional flow signal with an arterial spectrum. Large HCC may have a basket pattern, with a fine blood network surrounding the nodule and flowing into it. Doppler examination shows the presence of pulsatile flow with arterial waveform and high resistive and pulsality indexes. Large regenerative nodules and DN do not have arterial intralesional vessels, so these findings support the diagnosis of HCC. The problem is that in small HCC, the sensitivity of Doppler techniques is low, and a typical arterial pulsatile flow arterial pattern is detected in <50% of nodules.37


CEUS permits observation of the enhancement pattern of each liver nodule throughout the vascular phases in real time, like contrast CT and MRI scans. It is performed with intravascular microbubble contrast agents. Images are based on nonlinear acoustic effects of bubbles enhancing the grayscale, with high contrast and good spatial resolution. This technique improves the accuracy of US in the diagnosis of focal liver lesions.38, 39 Contrast agents are safe and well tolerated. At CEUS examination, HCC typically shows a strong enhancement in the arterial phase and a rapid washout during the portal and late phases, during which the nodule remains iso- or hypoechoic. Regenerative nodules and DN do not usually show any uptake during the arterial phase, and their appearance resembles that of the surrounding liver.

It has been found that after use of contrast, HCCs in 46 (90.2%) of 51 patients were identified as hyperechoic lesions during the early arterial or arterial phase, with no correlation with histological differentiation. In the portal phase and the late phase, the echogenicity of HCCs after contrast application was variable. A hypoechoic appearance was noted in 33.3% of HCCs in the portal phase and in 41.2% of HCCs in the late phase. This study concluded that CEUS is highly efficient for the detection of tumor vascularity in HCCs. By using CEUS, most HCCs, regardless of histological differentiation, can be characterized as hypervascular lesions in the early arterial and arterial phase with irregular tumor vessels.40

In another study, comparison was made with CEUS vs. CT in 74 patients with solitary biopsy-proven HCC. For 28 hepatocellular carcinomas ≤20 mm in size, CEUS showed 15 carcinomas (53.6%) as hypervascular, and 3 carcinomas (10.7%) were missed. On CT, 12 (42.9%) of 28 hepatocellular carcinomas appeared hypervascular and 3 (10.7%) were missed, whereas for the 46 hepatocellular carcinomas >20 mm, CEUS showed 42 (91.3%) as hypervascular compared with 35 (76.1%) on CT. Thus, CEUS as good as CT for detecting hypervascularity in HCC.41

The advent of second-generation contrast agents, together with low mechanical index scanning, have led the European Federation for Ultrasound in Medicine and Biology to define guidelines for its use.42 CEUS is recommended to characterize any suspected lesion detected at baseline US in patients with cirrhosis.

However, as for non-CEUS, the sensitivity of the use of US contrast agents decreases with smaller lesions. If spiral CT is considered a gold standard to identify hypervascular lesions, the sensitivity of contrast enhanced US was 97% for nodules >3 cm in diameter, 92% for nodules 2–3 cm in diameter, and 67% for nodules <1 cm in diameter.43 Another disadvantage is that only one lesion or lesions close together can be examined at one time. Therefore, a cost-effective algorithm that indicates which technique to use after routine US still needs to be determined.27

Spiral CT

The introduction of spiral CT has dramatically improved the accuracy of CT in the diagnosis of HCC. CT data acquisition speed is greatly increased so that it is possible to scan the whole liver during a patient's single breath hold, allowing a satisfactory evaluation of different contrast enhancement phases. Moreover, 3-dimensional reconstructions are available with faster data acquisition. The standard spiral CT evaluation for HCC should include unenhanced and contrast enhanced phase of the arterial phase (25–30 seconds after contrast injection), of the portal venous phase (70–80 seconds after contrast injection), and a late phase (180–210 seconds after contrast injection). Because HCC is a vascular tumor that receives almost all its blood supply from arterial vessels, it may be seen well during the arterial phase; the rest of the liver receives the majority of blood supply from the portal vein, and it will not enhance until the contrast reaches the portal system. Fast scans during the different phases of contrast flow through the liver help in increasing the differentiation between tumor and liver.

The technique can be improved by the use of multidetector row-scan technology, which, as a result of the rapid acquisition time, allows visualization of 2 different arterial phases, early and late. The first arterial phase is a CT arteriography that may be of some help, depending on the aim of the examination. Some studies show that a double arterial phase did not improve the detection rate of HCC compared with a single arterial phase,44, 45 whereas it did improve with the addition of a third arterial phase because the middle phase is claimed to be the most sensitive.46

However, despite the tremendous technical progress in this area, CT is still insensitive for detecting small HCC. When considering studies on explanted livers, the sensitivity of spiral CT ranged 54–79%,21, 35, 47, 48 mostly as a result of the lack of detection of smaller lesions (< 2 cm).

Another important issue is the specificity of this technique. It is well recognized that a number of benign lesions can simulate HCC on CT. The advent of arterial phase imaging has identified a series of new HCC pseudo-lesions that can be mistaken for enhancing tumors. Specificity of spiral CT ranges 93–97% in explant studies21, 35, 47 (Table 3).

Table 3. Spiral CT Lesion-by-Lesion Sensitivity (%), Positive Predictive Value (%) and Specificity (%) Compared With Explanted Livers
StudyNo. HCC/No. totalOverall sensitivitySensitivity for lesions <1cmSensitivity for lesions 1–2 cmSensitivity for lesions >2 cmPositive predictive valueSpecificity
  1. Abbreviations: HCC, hepatocellular carcinoma; NA, not applicable.

Lim et al. 200015/4171NANANANA96
Rode et al. 200154NANANANANA93
Burrel et al. 200329/5061106510087NA
Valls et al. 200451/8579NA619488NA

Among the hypervascular lesions that can simulate HCC are pseudoaneurysms, arteriovenous malformations, hemangioma, and DNs. Hypovascular lesions simulating HCC are hemangiomas, confluent fibrosis, and regenerative nodules. In addition, patients with cirrhosis may have transient focal enhancement of the liver during the arterial phase, mostly because of arterial-portal shunting. However, these images are often peripheral, are wedge shaped and not round, and sometimes have disappeared by the time of follow-up examination. These features help exclude HCC. The portal phase acquisition is particularly important because HCCs rapidly wash out during this phase, whereas hemangioma or vascular malformation show contrast enhancement.

The most problematic lesions mimicking HCC are DNs that may show arterial-phase enhancement, a problem that cannot be resolved, even when the most modern imaging technologies are used. Another technique that could improve CT performance is the injection of contrast through the hepatic artery while performing CT. This method has revealed 66% more tumors in patients with HCC compared with triphasic helical contrast CT,49 but its use is limited by cost and the mortality and morbidity related to the angiographic procedure. However, it can be easily performed during chemoembolization or embolization therapy when the catheter in the hepatic artery is needed to perform the therapeutic procedure.

Lipiodol CT requires the administration of contrast media through the hepatic artery. It is limited by the need to perform angiography and requires the patient to be rescanned at least 14 days after the injection. This allows the normal liver to wash out the contrast while the HCC retains it. However, retention of lipiodol is not specific for HCC alone, with only one study reporting that lipiodol-CT is more specific than spiral CT in detecting small, less differentiated nodules.50 Moreover, it hinders assessment of residual vascularity after locoregional therapies. Most centers have abandoned its use.


The use of MRI in the diagnosis of HCC has expanded greatly over the last few years as a result of the development of new hardware and software that have allowed the rapid acquisition of very detailed images, thus reducing the artifacts that result from the patients' respiration during the procedure. Volumetric sequences allow the reconstruction of 3-dimensional images with excellent spatial and temporal resolution. Moreover, the introduction of liver-specific contrast agents has been introduced to clinical practice, thus enabling a perfusional investigation of liver nodules.

The standard examination for the detection of HCC should include the following: T1-weighted fast spoiled gradient-echo sequences with fat suppression; respiration-triggered or breath-held T2-weighted fat spin echo sequences with fat suppression; and serial dynamic T1-weighted fast spoiled gradient-echo sequences after injection of a gadolinium chelate.17

Once again, because HCC is a heterogeneous tumor, these characteristics are reflected in an MRI examination. For example, HCC may have variable stromal and intracellular components, such as fat, glycogen, or ions, that may cause different appearance in T1- and T2-weighted images. The signal may appear hypo-, iso-, or hyperintense in T1-weighted images and iso- or hyperintense in T2-weighted images.

Gadolinium chelates are distributed in extracellular space, and their enhancing effect depends on the blood supply during the early dynamic phase and on the interstitial matrix during the delayed phase. The new fast MRI techniques allow the acquisition of multiple intravenous contrast-enhanced dynamic images of the liver, including the arterial phase, the portal peak phase, and a delayed phase. As with CT, on MRI, most HCC are characterized by an enhancement in the arterial phase and rapid washout during the portal phase (Table 4). Regenerative nodules and DNs are usually not hypervascular, but as for CT, some benign lesions, especially high-grade DN, may have increased arterial blood supply that may render diagnosis difficult. In addition, cirrhotic livers often contain arterioportal shunts that may produce false-positive interpretations.51

Table 4. Dynamic MRI Lesion-by Lesion Sensitivity (%), Positive Predictive Value (%), and Specificity (%) Compared With Explant Livers*
StudyPatient N (HCC/total)Overall sensitivitySensitivity for lesions <1 cmSensitivity for lesions 1-2 cmSensitivity for lesions >2 cmPositive predictive valueSpecificity
  • Abbreviations: MRI, magnetic resonance imaging; HCC, hepatocellular carcinoma; NA, not applicable.

  • *

    MRI protocol included dynamic imaging plus liver specific imaging after administration of reticuloendothelial-targeted agent.

Rode et al. 20016/43777180100NA57
Krinsky et al. 200110/7153NANANANA83
Krinsky et al. 200124/243345210087NA
Burrel et al. 200329/5076348910090NA
Teefey et al. 20039/2577NANANA74NA
Bhartia et al. 200314/317838929154NA

Dynamic MRI can also be performed with other tissue-specific agents—for example, superparamagnetic iron oxide (SPIO), which is taken up by Kupffer cells of the reticuloendothelial system (RES-targeted contrast agents) and which is absent in HCC. The agents produce a decrease in the intensity of the liver tissue, especially on T2-weighted images, because HCC nodules show little or no uptake of the contrast. Gadolinium and SPIO agents could be coupled,52 but the usefulness of this technique is still being debated.53

Despite substantial progress, MRI still does a poor job detecting small HCC; the sensitivity of dynamic MRI ranges 33–77% in explant studies.34, 35, 47, 54 It is 78% when dynamic imaging plus RES-targeted agents are used.55 When considering only HCC <1 cm, sensitivity drops to 4–71%. Nevertheless, dynamic MRI has been shown to be better than helical CT in the detection of nodules 1–2 cm in diameter.47 In this study, 50 patients with HCC undergoing LT were analyzed. Twenty-nine of them had HCC; the pathologic examination was the gold standard. On a per-nodule basis, sensitivity of MRI was superior to triphasic helical CT (76% vs. 61%, P = 0.001). Sensitivity for detection of additional nodules decreased with size, and MRI was better than CT for nodules 10–20 mm in diameter (84% vs. 47%, P = 0.016). The increased detection of nodules <2 cm in size needs to be evaluated in the various clinical contexts–for example, in selection for LT, where it may not change the risk of recurrence. The reference standard imaging in terms of allocation protocols in LT needs to be defined.56


The improvements in imaging techniques have allowed the identification of smaller nodules within the cirrhotic liver, but often the nature of these small nodules, and most importantly the relationship between these nodules and the development of HCC, is uncertain. It is well accepted that cirrhosis itself represents a preneoplastic condition, but in carcinogenesis, the exact point in time between a lesion that still has some growth regulation and one that has lost growth regulation is difficult to define. The only morphological change that is considered to be intermediate to HCC is dysplasia,57 although foci of dysplasia may take place within benign regenerative nodules.

Moreover, it is not clear whether all DNs will inevitably evolve in HCC. Some reports suggest that high-grade DNs should be considered as precursors to HCC in most cases.58 In clinical practice, a histological diagnosis of high-grade DNs poses several difficulties in the interpretation of this finding because a conservative therapeutic approach could affect the outcome.

More difficulties arise considering that in a DN, foci of HCC can exist, but biopsy may miss sampling the malignant portion of the lesion. In the setting of cirrhosis, the picture is further complicated by vascular lesions that are not malignant, such as arterioportal shunts or other perfusion abnormalities.

Nodule Biopsy

The introduction of radiological techniques that study the perfusional pattern of nodules has enormously improved the accuracy of diagnosis, and the vascular study of lesions has become the main parameter used for the radiological diagnosis of HCC. Consequently, the use of liver biopsy has declined during the past few years, reflecting the EASL and AASLD recommendations. This is clearly an advantage considering that liver biopsy can also be related to a median risk of seeding of 1–2%,25, 59 even if associated with percutaneous ablative techniques25 and can manifest after transplantation.25, 59

Furthermore, when a biopsy sample is negative for HCC, the presence of HCC cannot be excluded because there is a 10% risk of false negativity, especially in the presence of small nodules (<2 cm).25, 59 However, a prospective study has shown that 78.5% of nodules 1–3 cm in diameter with only one positive imaging technique in terms of hypervascularity suggestive for HCC was found to be true HCC at biopsy.60 When looking at the rate of hypervascularity demonstrated by 2 separate radiological techniques leading to the diagnosis of HCC, this rate was significantly lower in the group of nodules 1–2 cm in size than in the group of nodules 2–3 cm in size (44% vs. 84%, P < 0.001). Among the group of nodules 1–2 cm in size with discordant detection of hypervascularity by 2 techniques, a definite diagnosis of HCC was made by histology in 48%.60

AASLD guidelines61 recommend performing liver biopsy in cases of discordant imaging technique for nodules >1 cm and <2 cm. On the basis of previously published data, this would be the case for almost half of these nodules because 44% of small nodules had discordant imaging hypervascularity60 (Fig. 2).

The problem of hypovascular nodules (20% of cases) in this study is another important issue.60 All cases occurred in the group of nodules 1–2 cm in size, and 36% were later proven to be HCC. These data support the current approach suggested by AASLD guidelines.61 However, these recommendations for biopsy open up a problem of the clinical applicability of biopsy being performed for such small nodules. Four factors affect this: the safety of performing a biopsy for anatomical reasons, or ascites or coagulopathy, or a combination thereof; the reliability of obtaining tissue from the right place; the uncertainty of the diagnosis histologically; and the risk of seeding.

In the setting of LT, the need of a precise diagnosis can be dictated by the allocation system that is adopted. In the United States and in some European LT centers, the Model for End-Stage Liver Disease (MELD) HCC scoring system is used, in which a single nodule ≥2 cm in size or multiple nodules increases the priority on the waiting list.28 Here the issue is the need of a precise diagnosis to justify increased priority, regardless of baseline liver function. It is interesting to note that explant studies in the United States have shown the lack of histological confirmation in one-third of patients diagnosed with HCC before transplantation,62 which appears higher than European studies.63 This situation may have been driven by the wish to increase the individual patient's chance of transplantation.

The risk of seeding with HCC appears to be greater when diagnostic biopsy is performed alone compared with therapeutic percutaneous procedures.59

Stigliano et al.59 performed a literature search for the period between 1983 and 2007 on seeding. In published articles that gave the number of the total cohort that underwent biopsy, the median risk of seeding was 2.29% (range, 0–11%) for biopsy group; 1.4% (1.15–1.85%) for percutaneous ethanol injection when used with biopsy and 0.61% (0–5.56%) for radiofrequency ablation without biopsy; and 0.72% (0–10%) for liver nodules that were sampled and ablated. They concluded that the risk of seeding has emerged to be greater when a diagnostic biopsy is performed alone, compared with performing a therapeutic percutaneous procedure. This risk was particularly relevant in patients being considered for LT. In general, HCC has a higher risk of seeding (0.5–1%) than other tumors compared with pancreatic tumors (0.003–0.017%) and other abdominal tumors (0–0.03%). The advent of molecular techniques that are based on tissue analysis may change the risk-benefit profile associated with nodule biopsy.64


It has been considered that appropriate staging system for HCC should include 4 related aspects: tumor stage, degree of liver function impairment, patient's general condition, and treatment efficacy.6

When advanced underlying liver disease (cirrhosis) determines the likely prognosis, primary tumor factors (T stage) may become irrelevant in terms of prognosis. In these cases, other staging systems (Child-Pugh, Okuda staging, Cancer of the Liver Italian Program Investigators [CLIP], or Barcelona Clinic Liver Cancer [BCLC]) provide much better prognostic estimation. On the other hand, for LT assessment, the Milan criteria and University of California, San Francisco (UCSF), criteria are mostly used.

BCLC Staging Classification

The BCLC classification combines several parameters, retrospectively derived, such as tumor factors (size, nodularity), liver function, and treatment modalities. This staging system is useful in selecting treatment options, particularly for patients with early-stage HCC (stage A). The major limitation of the BCLC staging system is the lack of an external validation6 (Fig. 3).

Figure 3.

Treatment schedule according to the BCLC staging classification. From Llovet et al. 1999; Bruix and Llovet 2002; and Llovet JM, Burroughs AK, Bruix J. Hepatocellular carcinoma. Lancet 2003;362:1907–1901. Llovet JM, Bau C, Bruix J. Prognosis of hepatocellular carcinoma: the BCLC staging classification. Semin Liv Dis 1999;19:329–338.

Milan Criteria for LT

The selection of patients by the Milan criteria is based solely on CT imaging criteria: no >3 tumor nodules; no nodule >5 cm in diameter; absence of macroscopic portal vein invasion; and absence of recognizable extrahepatic disease (lymphadenopathy may complicate assessment in patients with human immunodeficiency virus). However, it does not fully account for the biology of HCC. The most important feature not reflected in the current selection criteria is microscopic vascular invasion and differentiation. However, microvascular invasion and differentiation are related to size of nodule, which remains a good surrogate marker.59

The Milan criteria have been adopted as the bases for selecting patients with cirrhosis and early HCC, and for awarding them extra points within the MELD classification used for organ allocation prioritization.

A number of expanded criteria have been proposed.65–69 Among them, the UCSF criteria70 (single lesion ≤6.5 cm in diameter, or 2–3 lesions ≤4.5 cm with total tumor diameter ≤8 cm) reflect a modest expansion of tumor size limits beyond Milan criteria and have been independently tested in several studies.71–74 The principal issue here is whether expansion of Milan criteria is feasible—that is, whether there is no increased risk of recurrence after transplantation–on the basis of the increase in maximum diameter of the largest nodule, the total diameter of the nodule, or both. Which expansion is safe is not well documented.

In our own series,56 maximum tumor diameter both before transplantation and at the time of explant is the best surrogate marker for recurrence. The best cutoff for no recurrence is tumor diameter of ≤3.5 cm on explant (3.0 cm on pretransplantation imaging).


Current imaging techniques have substantially improved the diagnostic accuracy for HCC in patients with cirrhosis, reducing the need for liver biopsy and its attendant immediate complications, as well as the risk of seeding. This is mostly because of the development of dynamic techniques that study the vascular pattern of nodules. However, the performance of these techniques is poor when nodules are <2 cm in size; such nodules may also be hypovascular. In the era of LT, the need for a correct diagnosis has become imperative in most cases, particularly if priority is extended to these patients over and above the background severity of their liver disease. Sometimes liver biopsy may need to be performed, but diagnostic algorithms still need to be worked out at practical level because single nodules < 2 cm in diameter have in general a far better prognosis and virtually no risk of recurrence after LT. There is a risk of seeding after biopsy, and this must be taken into account during diagnosis and treatment.

In patients with cirrhosis, nodules <1 cm in size suspected of being malignant should be followed up at short-term intervals of no more than 3 months because the use of dynamic techniques is only sufficiently accurate when the nodules are larger. With nodules 1–2 cm in size, if 2 dynamic techniques are not concordant in demonstrating hypervascularity and result in washout, liver biopsy should be performed if precise diagnosis would change the immediate therapeutic option for the patient. However, biopsy could be delayed or avoided depending on the timing of potential therapy such as LT (e.g., listing and reimaging).25, 59 However, another therapeutic option is to treat the nodule as if it is HCC, even in the absence of a histological diagnosis, because the risk of seeding is not increased by adding ablative therapy to the biopsy.25, 59 For nodules >2 cm, a single dynamic technique is sufficient to diagnose HCC if the hypervascular pattern is typical. In the presence of an atypical vascular pattern, the presence of AFP >200 ng/mL is diagnostic.