As radiological techniques used to screen patients for hepatocellular carcinoma (HCC) have improved, smaller and smaller liver nodules are being identified. The smaller the nodule seen, the more difficult it is to be certain of the relationship between the nodule and the development of HCC. Determining whether these small nodules are cancer is challenging. Deciding whether they should be treated or not is at present a matter of expert opinion, rather than evidence based.
Abbreviations
HCC, hepatocellular carcinoma; CT, computed tomography; EASL, European Association for Study of the Liver.
Carcinogenesis is a process in which there is no clear morphological boundary between a lesion that is still subject to growth regulation and one that that is no longer subject to regulation and can proliferate independently of its neighbors. Thus, perhaps it should not come as a surprise that it is so difficult to make this distinction when we attempt to define the nature of small nodules seen on ultrasound screening for HCC. Further complicating the picture is the presence of numerous nodules that have nothing to do with malignancy, namely the cirrhotic nodule. Some of these nodules have arterial blood flow because the sinusoids are arterialized or because the portal veins have been damaged or destroyed.1 On vascular imaging, such as triphasic computed tomography (CT) scanning, these may show arterial enhancement, but typically they do not show venous washout.
Like all other cancers, HCC evolves through series of steps in which clones of cells emerge that are less and less subject to growth regulation. Although there are probably many steps that occur at a genetic level, the only discrete morphological change that has been identified as an intermediate in the malignant pathway is the small cell dysplastic change.2 This may occur as a focus a few cells in size, or enlarge into a nodule (high-grade dysplastic nodule). That at least some of these lesions become malignant seems clear from the finding of a “nodule-in-nodule” appearance. This occurs when a nodule that is morphologically neoplastic develops within a dysplastic nodule.3, 4 However, it is not clear that all dysplastic nodules inevitably become cancer. In fact, it is likely that some remain stable for some time, and others even regress completely.5 Small cell dysplastic change is, in general, a marker of increased HCC risk. The HCC may develop in a focus of small cell change (dysplastic focus or nodule) or may develop elsewhere in the liver. Simply finding a dysplastic nodule does not mean that HCC will inevitably follow in that nodule.
In addition to these changes Japanese pathologists have described two morphological variants of liver nodule that they consider to be established but early malignancy.6 The so-called “early HCC” is a nodule with poorly defined margins, in which the cells are morphologically close to normal but in which some features associated with malignancy can be found. Most often this is invasion of the portal space, and less often of vascular structures, by hepatocyte-like cells. The hepatocytes in the nodule often contain fat and are supplied by the portal vein rather than the hepatic artery.7, 8 On ultrasound these lesions are bright because they contain fat, and on CT they are hypoattenuating. Because they are fed by the portal vein, they do not show the typical arterial enhancement and venous washout that characterize HCC.9 In contrast, the so-called “small HCC,” thought to be a more advanced form of HCC than the “early HCC,” is surrounded by a capsule and is fed by the hepatic artery.9 The hepatocytes express many more features of malignancy, such as an altered nucleo-cytoplasmic ratio and nuclear atypia. The distinguishing features of these two lesions have been described from resected specimens. These lesions may show a halo effect on ultrasound, and on CT they show the typical arterial enhancement and venous washout. Needle biopsy of early HCC, however, rarely shows the typical features of the lesion, because the diagnostic features are very focal. Even small HCC can be missed on a biopsy. It should be emphasized that the natural history of these small lesions is unknown because they were described in resected specimens, so their natural history could not be observed. It is presumed (but not certain) that these lesions are malignant. It seems likely that the “small HCC” is indeed malignant based on its description, and invasion of the portal space in the “early HCC” would also suggest malignancy, when present.
In 2001 the European Association for Study of the Liver (EASL) convened a consensus conference on the management of hepatocellular carcinoma.10 With regard to HCC diagnosis the consensus document states that lesions smaller than 1 cm do not need to be biopsied but can simply be observed to see whether they enlarge over time. In a liver with cirrhosis, lesions larger than 2 cm that were shown to be vascular using two imaging techniques can also be considered HCC and do not need to be biopsied. The group recommended that lesions between 1 and 2 cm should be biopsied. The authors recognized that these recommendations were based on expert opinion and that there were no studies to support these recommendations. Nonetheless, these seemed appropriate at the time. The recommendations were silent regarding what to do with nodules larger than 2 cm that did not meet the defined radiological criteria.
Bolondi and his colleagues have addressed this issue in their article published in this edition of HEPATOLOGY.11 They examined the relationship between the radiological appearance of nodules found on HCC screening and their histological diagnosis. They stratified the results by lesion size according to the EASL criteria. All lesions that were larger than 2 cm turned out to be HCC. The majority (84%) exhibited typical radiology and were confirmed by biopsy. The other 16% did not meet the EASL radiological criteria and were diagnosed only by biopsy. Thus, the study confirms the EASL recommendations for lesions larger than 2 cm. If the lesion is hypervascular by 2 radiological techniques a biopsy is not necessary.However, if the lesion does not meet these criteria biopsy must be performed. For lesions between 1 and 2 cm in size only 71% were diagnosed as HCC on biopsy, either at the first biopsy, or, if the first biopsy was negative, a subsequent biopsy after interval growth confirmed the diagnosis. Again, this confirms the EASL recommendations that lesions between 1 and 2 cm should be biopsied. Note that in this study, as in others,5 a negative biopsy did not exclude the diagnosis of HCC. The message is that all patients with a liver mass and a biopsy that is negative for HCC must be followed at frequent intervals to document a change in lesion size that might require another biopsy.
When the EASL consensus document was written the importance of the “washout” in the portal venous phase of a triphasic CT scan was perhaps not fully appreciated.9, 12 “Washout” is the appearance of a hypoattenuating area in the portal venous phase of a triphasic CT liver scan that was hyperattenuating compared with the rest of the liver in the arterial phase. During the arterial phase the HCC, which is fed by the hepatic artery only, enhances more strongly compared with the surrounding liver. This is because the blood in the liver is both arterial and portal, so that the contrast agent, which is being delivered exclusively by the hepatic artery at this stage, is diluted by the portal blood, while the arterial blood in the tumor is undiluted. During the venous phase the arterial blood feeding the tumor now contains diluted contrast or very little contrast, while the liver enhances more brightly that the tumor because it is fed by venous blood with a relatively high concentration of contrast, as well as arterial blood. Thus, in the arterial phase the lesion is hyperattenuating (“bright”) compared with the rest of the liver, and in the venous phase the tumor is hypoattenuating compared with the rest of the liver. This finding is highly specific for HCC. Even in lesions between 1 and 2 cm the presence of “washout” indicates HCC. Thus, the indications for biopsy can be extended further. A nodule found on screening of a liver with cirrhosis should be biopsied if it is larger than 1 cm, if it does not show features of arterial enhancement with washout, and if it is clearly not some other lesion, such as a hemangioma. It should be remembered, however, that context is important. A similar finding in a patient at low risk for HCC may require a different algorithm for investigation, and a more cautious approach may be warranted.
How important is it to identify HCC at such small sizes? Does it matter if the HCC is identified at 1.5 cm in size or at 3 cm? There are few data to answer this question, but Sala et al.13 have shown that the smaller the lesion that is treated by radiofrequency ablation, the more likely it is that complete radiological obliteration can be achieved. Survival was also better after ablation of smaller lesions, but lead-time bias was not accounted for. Thus, there would appear to be merit in finding HCC as small as possible. However, the paradox is that the smaller the HCC, the more difficult it is to diagnose. There are fewer and fewer features of HCC as the lesions are smaller and smaller. It becomes more and more difficult to separate dysplasia from HCC. It may be argued that finding a high-grade dysplastic nodule is tantamount to finding HCC and that all these lesions should also be ablated. This argument depends on how many high-grade dysplastic nodules ultimately develop into HCC. If the proportion is high, and if a large proportion of nodules that are not HCC initially become HCC within a few years, a case can be made for obliterating all such nodules without the benefit of a biopsy. The counterargument is that imaging studies with CT scanning and magnetic resonance imaging have shown that up to 50% of lesions that are smaller than 2 cm and vascular but do not exhibit “washout” either disappear with time or become smaller and do not become HCC.14, 15 Thus, ablating all vascular nodules between 1 and 2 cm will result in ablation of numerous nonmalignant lesions.
The use of molecular tools may be helpful in the diagnosis of the small early lesions. If it is possible to identify genes expressed early in carcinogenesis in tissue samples it may be possible to make an early diagnosis of HCC. To date, however, there is insufficient information to assess any single marker. HCC has been shown to express beta-catenin,16, 17 p53,18, 19 glypican-3,20, 21 and many other markers. Beta-catenin seems to be expressed later in tumor development, as it is not present in dysplastic nodules.16 The timing of upregulation of p53 expression may also be a later phenomenon, although some small HCCs also express p53.18, 19 Glypican-3 is expressed early in carcinogenesis and may be a serological marker as well as a molecular marker of early HCC.20, 21 Few of the studies of protein expression in HCC have carefully differentiated between early and later HCC. However, a recent study using microarray technology looked at various nodules, including nonmalignant nodules with cirrhosis, low- and high-grade dysplastic nodules, and early and later HCC, as well as normal liver.22 The investigators were able to identify genes that were differentially regulated in HCC compared with dysplastic nodules, and they were able to use this information to predict whether a nodule was dysplastic or neoplastic. This analysis was performed on resected specimens, so that the problem of sampling error on a needle biopsy may still reduce the accuracy of diagnosis, even if the changes in gene transcription are highly predictive of malignancy.
The challenge upon us now is to find noninvasive methods of distinguishing malignant and pre-malignant lesions from benign ones. Until these are available biopsy of small lesions will remain an important part of the investigation of lesions found on HCC screening.
