Macroregenerative (Dysplastic) Nodules and ...

4 downloads 0 Views 1MB Size Report
sonL4 called "adenomatous hyperplasia" (AH), which he ... and those with atypia ("type I1 MRNs" or "atypical ... atypical features bc classified as type I lesions.'?.
SEMINARS IN LIVER DISEASE-VOL.

15, NO. 4, 1995

Macroregenerative (Dysplastic) Nodules and Hepatocarcinogenesis: Theoretical and Clinical Considerations

In recent years, a growing literature has supported the concept that large nodules usually found in cirrhotic livers represent premalignant lesions in the setting of chronic liver disease. With the use of advanced imaging techniques in high-risk populations in Japan,'-3nodules suspicious for malignancy have often been identified and resected. Although some resected lesions were found to be small hepatocellular carcinomas (HCCs),)-S others were not. Some of these nonmalignant nodules were devoid of atypia; some had architectural or cytological atypia insufficient for a diagnosis of HCC, though suggestive of a premalignant state; others contained microscopic subnodules of HCC.3-7In follow-up studies from Japan, North America, and Europe, including autopsy and series of explants from liver transplant center~,'O-'~ the occasional finding of microscopic foci of HCC in the nodules was confirmed and significant associations with HCC elsewhere in the same liver were established. Such findings suggested that these nodular lesions are probably a frequent pathway of human hepatocarcinogenesis in a wide array of liver diseases and in diverse populations of patients. Recently, researchers have begun to engage more complex issues concerning these nodules, including elucidation of their biologic behavior and of their relationship to animal models of hepatocarcinogenesis. These newer studies have led to an understanding of the early events of human hepatocarcinogenesis and seem to require revision of existing nomenclatures, diagnostic criteria, and clinical approaches for the lesions. This review is intended to bring the reader up to date on current concepts of how these premalignant lesions develop and how HCC may arise within them. These conceptual developments bear on recent attempts at developing clinically useful

From the Department of Pathology, Tisch Hospital, New York University Medical Center; New York. Reprint requests: Neil D. Theise, M.D., Rm. 461, Dept. of Pathology, Tisch Hospital, New York University Medical Center, 560 First Avenue, New York, NY 10016.

360

nomenclature, so this review will also attempt to evaluate these new systems of nomenclature in light of our current understanding. Finally, the clinical import of such theoretical considerations will be discussed.

DEFINITIONS AND OBSERVATIONS The nodules under discussion consist of hepatocytes and almost always contain some intact, normal-appearing portal tracts. Since they are most often found in the setting of cirrhosis, they seem to correspond to what EdmondsonL4called "adenomatous hyperplasia" (AH), which he viewed as lesions having "limited growth potential." Other terms used more recently include macroregenerative nodule (MRN),8.10-12 dysplastic n ~ d u l ehepatocellu,~ lar p~eudotumor,'~ and adenomatoid hyperplasia.15 AH has been most widely used, although not exclusively, by researchers from Japan. MRN, coined by Furuya et a18 for the first Japanese autopsy study of the nodules, has become the most widely accepted term in publications from the United States and Europe. The terms AH and MRN have been used fairly interchangeably, though each is problematic, as will be discussed further on. Throughout the bulk of this review, the term MRN will be used, following the author's own prior convention. MRNs are defined grossly as large hepatic nodules that are distinct from the surrounding liver parenchyma in color, texture, or the degree to which they bulge from the cut surface of the liver (Fig. 1). The minimum size criterion is somewhat arbitrary, varying among different researchers from 0.5 cm in greatest dimension6 to 0.8 c m ~ ~ . to ~ 6 1.0 cm.8.L0.'2 The size criterion depends on what the individual investigator considers suitable to distinguish the lesions from ordinary cirrhotic nodules. Confirmation that a nodule is in fact an MRN comes with histologic examination and the identification of intact portal structures distributed through the lesion. These portal structures may be reduced compared to a similar area of nondiseased hepatic parenchyma, but they may also be virtually normal in number and distribution. In a

Copyright 0 1995 by Thieme Medical Publishers, Inc., 381 Park Avenue South, New York, NY 10016. All rights reserved.

Downloaded by: Albert Einstein College of Medicine. Copyrighted material.

NEIL D. THEISE, M.D.

FIG. 1. A type I MRN (or "low-grade dysplastic nodule"; arrow) in a liver with cirrhosis due to chronic hepatitis C. (Bar = 1.0 cm.)

lesion consisting of well-differentiated hepatocytes, the presence of portal structures confirms that the lesion is not actually HCC or adenoma. Other lesions that must be excluded histologically include focal nodular hyperplasia, which contains a central scar with vascular abnormalities and malformed or absent bile ducts, and anoxic pseudolobular necrosis, i.e., a hemorrhagic regenerative nodule that may bulge from the cut surface of the liver. MRNs have been found in a wide variety of chronic liver including processes that are hepatitic (hepatitis B, C, and autoimmune hepatitis), cholangitic (primary biliary cirrhosis, primary sclerosing cholangitis), metabolic (a1-antitrypsindeficiency, primary hemochromatosis), and toxic (alcoholic liver injury). The fact that these lesions were initially solely reported in Japanese patients was the consequence of meticulous diagnostic screening programs in that country. Subsequent careful examination of resected livers in transplant centers in the United States'@12and Europe'"'' made it clear that MRNs are found in most populations affected by chronic liver disease, and that a significant association with HCC exists in them (Table 1). Typically, livers with MRNs contain a small number of these nodules, rarely more than 10, although there are

TABLE 1. Incidence of MRNs in Cirrhotic Livers Location

Source

No. of Cirrhotic Livers

No. of Livers with MRNs

TokushimaX New York"' San Francisco" KanazawaY New York'?

Autopsies Explants Explants Autopsies Explants

315 44 110 209 155

46 (14%) 11 (25%) 17 (15%) 45 (21%) 32 (22%)

exceptions, which will be discussed further ~ n . ~MRNs -'~ may be subclassified into those without cytological or architectural atypic ("type I MRNs" or "ordinary A H ) and those with atypia ("type I1 MRNs" or "atypical AH")."I2 It is not possible at this time radiographically to distinguish small HCCs from MRNs or type I MRNs from type I1 MRNs with complete specificity, nor is it usually possible to reliably make such distinctions on the basis of gross morphology. Histologic examination, either by biopsy or examination of a resected specimen, is required for accurate classification.

Histologic Features of Type I MRNs Type I MRNs are very well-defined nodules surrounded by a condensed rim of fibrous tissue similar to that surrounding cirrhotic nodules (Fig. 2). The nodules are thus not truly encapsulated. Portal tracts, present in virtually all MRNS, are most often uniformly distributed in type I MRNs and may even be distributed in a virtually normal fashion with regularly intervening terminal hepatic venules. In some nodules, portal structures may be caught up in fibrous septa that partially subdivide the nodule. The hepatocytes of type I MRNs tend to be of comparable size to hepatocytes outside the lesion. The hepatocytes may display changes characteristic of the underlying liver disease affecting the surrounding liver, such as fatty change, Mallory bodies, or increased iron or copper deposition. These changes are distributed in the MRN as they are in surrounding cirrhotic nodules. Occasionally, an MRN in a nonsiderotic liver may contain increased iron, or an MRN in an otherwise siderotic liver will be iron free; it would still be classified as type I.

Downloaded by: Albert Einstein College of Medicine. Copyrighted material.

MACROREGENERATIVE (DYSPLASTIC) NODULES-THEISE

15, NO. 4, 1995

FIG. 2. A type I MRN (or "low-grade dysplastic nodule") in a liverwith cirrhosis due t o chronic hepatitis C. The nodule is not truly encapsulated, but is surrounded by dense fibrous tissue similar to that surrounding small, neighboring cirrhotic nodules. Fibrous islands and septa within the nodule containing normal-appearing portal structures are distributed throughout the lesion. (Chromotrobe aniline blue, ~3.5).

Until recently, large liver cell dysplasia (large LCD) has generally been considered a feature defining an MRN as being type 11, i.e., atypical. However, recent studies of large LCD in cirrhosis and statistical evaluation of the association of this feature in MRNs and HCC"'." indicate the likelihood that large LCD is usually a reactive, not a premalignant, change. Therefore, it has been recommended that MRNs containing large LCD without other atypical features bc classified as type I lesions.'?

Histologic Features of Type II MRNs Type TI MRNs are defined by the presence of cytological and/or architectural atypia. They are usually well circumscribed and surrounded by a condensed rim of fibrous tissue, like type I lesions, although some may merge focally with adjacent liver parenchyma.'"he atypical features in type I1 MRNs may take a variety of forms and may be diffuse throughout the nodule or focal. Diffuse changes most often fall into the catcgory of cellular atypia. The definition of cellular atypia, as noted above, has usually included large LCD, although this should be discontinued.I2 On the other hand, small LCD-small, crowded hepatocytes with basophilic cytoplasm and an increased nuclear-cytoplasmic ratioI8-appears to be consistently related to the development of HCC in a variety of studies and should remain a criterion for "atypical." Small LCD is reported more frequently in studies from Japan than in those from other this discrepancy remains unexplained. Pseudoacinar structures resembling those seen in well-differentiated HCC are a form of architectural atypia in type I1 MRNs and may be either focal or diffuse. Focal atypia may merge with the surrounding MRN

parenchyma, but it more often occurs instcad as a "nodule-in-nodulc" lesion (Fig. 3)." Such subnodules often appear to compress the adjacent MRN parenchyma, and studies of proliferative rates of the cells making up these lesions indicate that they are proliferating more rapidly than the surrounding tissue.I9 These subnodules may display small LCD'''.'9 but may also show changes that are not classically "atypical," including fatty change,'" clear cell ~hange,l','~ clusters of hepatocytes with Mallory hyalin,>' increased iron uptake within the MRN,22iron resistance in an otherwise siderotic and accumulation of copper binding protein.'" Some expansile subnodules do not display any distinctive cytological features, although, architecturally, they may display a scirrhous or a pseudoacinar growth pattern. We have argued that all subnodules, with or without distinctive cellular changes, are appropriately defined as architectural atypia on the basis of the expansile growth and should warrant classification of the entire MRN as a type I1 lesion.'? HCCs may be identified in type 11 MRNs (Fig. 4).~.6.8-16,2627 These microfoci of HCC may display any of the features seen in larger HCCs, but they are usually well differentiated. Typical growth patterns include pseudoacinus formation, thickened trabeculae, or a scirrhous growth. Common cytological features include intracytoplasmic Mallory hyalin, fatty change, clear cell change, iron resistance, and multinucleation. Multiple foci of HCC may also be found in a single MRN, and the foci may differ from each other in their histologic feature^.^^^^^ The MRN parenchyma surrounding a microfocus of HCC usually contains portal tracts and may consist of normalappearing hepatocytes, suggesting a background of a type I MRN, or may show atypia, indicating a type I1 back-

Downloaded by: Albert Einstein College of Medicine. Copyrighted material.

SEMINARS IN LIVER DISEASE-VOL.

FIG. 3. This MRN in a liver with alcoholrelated cirrhosis is formalin fixed and paraffin embedded and measures 1.0 cm in greatest diameter. The dark color of the nodule is due to marked hemosiderosis, which is absent in the surrounding cirrhotic nodules. A iron-resistant subnodule (arrows) is grossly evident within the MRN. This nodule-innodule pattern classifies the lesion as a type II MRN (or "high-grade dysplastic nodule").

ground. Either way, MRNs containing foci of HCC are classified as type 11. Table 2 summarizes the features of the types I and I1 MRNs.

The Premalignant Nature of MRNs The association of MRNs with HCC is demonstrated in two ways. First, as mentioned above, MRNs sometimes contain one or multiple microscopic foci of HCC. Second, MRNs are sometimes found in livers that also contain

grossly apparent HCC elsewhere (Fig. 5). Both of these relationships have been found to achieve statistical significance.'&I2Beyond this statistical correlation, the atypical features often found in MRNs include many that have previously been thought to bc prcmalignant. Clustering of hepatocytes containing Mallory h ~ a l i n ~and ' . ~ foci ~ of iron resistance in siderotic nodule^^^^^^ have been independently described as premalignant changes. Small LCD, considered premalignant on the basis of morphometric analysis, is also seen and, in some series, is very ~ o m m o n . ~ ~ ' @ ~ " . ' ~ Other features that are commonly identified in mature HCC are found in type I1 MRNs. For example, immu-

FIG. 4. Multiple microscopic HCCs (arrowheads) arising in a macroregenerative ("dysplastic") nodule. The carcinomas are all moderately to well differentiated, displaying trabecular and focal pseudoacinus growth patterns. The nodule of carcinoma in the middle is also encapsulated with a focus of capsular invasion (arrows).The hepatocytes of the background nodule separating the carcinomas are devoid of atypia. (Hematoxylin & eosin, x35.)

Downloaded by: Albert Einstein College of Medicine. Copyrighted material.

MACROREGENERATIVE (DYSPLASTIC) NODULES-THEISE

TABLE 2. Features That M a y Be Found in Type I (Ordinary) and Type II (Atypical) MRNs fipe I MRNs

Type 11 MRNs

Nonnal architecture Normal cytology or large cell dysplasia Iron accumulation in otherwise nonsiderotic liver Iron resistance in otherwise sidcrotic liver Diffuse fatty change Copper accumulation

Focal pseudogland formation Diffuse small cell dysplasia Nodule-in-nodule lesions with: Small cell dysplasia Iron accumulation Iron resistance in siderotic nodule Fatty change Clear cell change Mallory body clustering Decreased reticulin fibers Scirrhous growth Features diagnostic of hepatocellular carcinoma

nohistochemical studies of MRN sinusoids reveal increasing degrees of "capillarization," i.e., loss of endothelial fenestration, deposition of basement membrane, and expression of antigens such as factor VIII and CD34, with the development of atypia and HCC.17.3'.32 Studies of ploidy indicate increased frequency of aneuploidy in type I1 M R N s . ~ ~ .Immun~hi~tochemical '~ staining for alpha-fetoprotein (AFP)1'.75,'h and p-ras13demonstrates expression of these proteins in atypical foci and microfoci of HCC. MRNs may be clonal lesions. Tsuda et al?' exploiting integration of the hepatitis B surface antigen gene into the host genome using restriction fragment length polymorphism (RFLP) analysis, demonstrated that two MRNs in one cirrhotic liver were of two distinct clonal lesions. Moreover, one of these lesions contained a microfocus of HCC that consisted of cells derived from the same clone as the surrounding MRN parenchyma.

15, NO. 4, 1995

Of course, all these findings represent static samplings, virtual "snapshots," of the process of malignant transformation, from which we are left to infer a dynamic process over time. From the association of MRNs with HCC elsewhere in the same liver, we deduce that MRNs are at least a marker of a liver with a tendency to generate malignancies. A recent study confirmed this relationship by closely following patients who had had MRNs re~ e c t e dPatients .~ with MRNs that were histologically classified as type I1 following resection were found to have an increased risk of new HCC arising elsewhere in the same liver. Furthermore, within the 3-year follow-up period, all those with MRNs containing microfoci of HCC (313) developed subsequent macroscopic HCCs, 36% of those with type I1 MRNs without microfoci of HCC (41 11) developed HCC, and none of those with type I lesions (0110) developed HCC. All the various histologic, immunohistochemical, and molecular findings also suggest a stepwise progression of HCC development through intermediate stages of type I and then type I1 MRNs. Another longitudinal study supports part of this inference.' MRNs detected by screening imaging studies were serially biopsied, resulting in documentation of the development of HCC arising in preexistent type I1 MRNs in as little as four months from the time of initial screening. In both of these longitudinal studies, it is interesting to note that type I MRNs are not as obviously premalignant or as indicative of neighboring malignant transformation as are type I1 MRNs. The data addressing this difference appear contradictory. Studies of the various aspects of MRNs menti~ned-proliferation,'~~~',~~,~~ p-ras expression,13 p l ~ i d y ,endothelial ~ ~ . ~ ~ and extracellular matrix alteration^"^^'^^^-find similarities between type I1

FIG. 5. A type I MRN (or "low-grade dysplastic nodule"; arrow) and multiple nodules of HCC (arrowheads) in a cirrhotic liver from a patient with homozygous a'-antitrypsin deficiency. (Bar = 1.0 cm.)

Downloaded by: Albert Einstein College of Medicine. Copyrighted material.

SEMINARS IN LIVER DISEASE-VOL.

364

MRNs and HCC. In the absence of these similarities between type I MRNs and HCC, most of these researchers conclude that type I1 lesions are related to HCC, whereas type I lesions are related to regenerative nodules and are less clearly implicated in hepatocarcinogenesis. Data from the largest series of liver explants both support and go against this view.I2 On the one hand, a subset of livers was identified (N = 13) that contained so many MRNs as to be virtually uncountable. When large LCD was excluded as a criterion for atypia, in all but one of these livers all the sampled MRNs were type I. None of these livers contained HCC. All such livers were in patients with chronic hepatitis (hepatitis B, hepatitis C, or autoimmune hepatitis), and the mean age of these patients was significantly lower than that of the other patients with few or no MRNs. These findings suggest that these MRNs are actually large regenerative nodules that have not yet scarred down into smaller nodules, examples of the "adenomatous hyperplasia" of Edmondson rather than neoplastic lesions. They might also correlate with large cirrhotic nodules that became undetectable on long-term ultrasound surveillance in a study by Kondo et aL40 On the other hand, statistical associations of type I lesions with coexistent HCC in this, as in earlier, series are strong, even when type I1 lesions are excluded from analy~is.~''~'~ Thus, the neoplastic, premalignant nature of type I lesions seems more open to question than it does for type I1 lesions. An explanation for these discrepancies which possibly resolves the apparent contradictions has been suggested on the basis of animal models of hepatic progenitor celk4'

SPECULATIONS ON THE EARLY STAGES OF HUMAN HEPATOCARCINOGENESIS The Standard Hypothesis As already mentioned, it is tempting to arrange these different types of lesions into a sequential process of neoplastic transformation (with or without type I MRNs). It has further been tempting to suggest that MRNs themselves are a result of exuberant regeneration in a cirrhotic liver.7This hypothesis suggests that a small cirrhotic nodule proliferates more rapidly, thereby becoming larger than other nodules (i.e., an MRN) and, at the same time, becoming more susceptible to carcinogenic events. Thus a small cirrhotic nodule becomes a large cirrhotic nodule and, subsequently, a neoplastic process begins in the midst of florid hyperplasia. Critical assessment of this concept reveals several inconsistencies with the observed properties of MRNs. First, as has already been stated, most MRNs contain portal structures, sometimes in a virtually normal distribut i ~ n , ' ~ yet , ~ ' most cirrhotic nodules contain few, if any,

365

portal structures, as most of these structures are caught up in the scarring process of cirrhosis. Therefore, according to this hypothesis, a rapidly expanding cirrhotic nodule would have to regenerate new portal structures in becoming an MRN. However, there is scant evidence, if any, to suggest that such regeneration, either in animal or human models of hepatic regeneration, can occur. Most such models involve regeneration following partial hepatectomy or massive necrosis and would thus concern portal regeneration from normal portal structures. But, in the context of a rapidly expanding cirrhotic nodule becoming an MRN, regeneration from scarred and linked portal tracts would be required-an even less likely proposition. Furthermore, recent work examining regeneration in the setting of Budd-Chiari syndrome documents that portal regeneration, when it occurs, results in abnormally structured portal tracts.43Although some abnormal structures as described in this work may be found in MRNs, they by no means predominate. Second, this concept considers MRNs to arise from a hyperplastic process, which theoretically would result in a polyclonal expansion of cells, yet MRNs, in the one limited study of Tsuda et a1,37appear to be monoclonal proliferations. As noted above, in this study, two MRNs in one cirrhotic liver were found to be different clones, and one of the MRNs contained an HCC that appeared to be derived from the same clonal expansion as the surrounding MRN parenchyma. On the basis of this evidence, it seems that MRNs, even prior to the development of a clearly neoplastic lesion such as HCC, are neoplastic rather than hyperplastic expansions of cells. Finally, until recently the general opinion was that MRN development assumes a preexisting cirrhotic nodule from which the MRN arises; however, several MRNs have been documented in livers with chronic hepatitis but without cirrhosis. The first such MRN is included in the survey of autopsy livers by Furuya et a1.8 This lesion was not associated with HCC. Subsequently, an MRN arising in the context of chronic hepatitis C, with fibrous septum formation, but without transition to cirrhosis, has been rep~rted.'~ This MRN contained multiple foci of morphologically distinct HCCs. A third case, similar to the last, was included in a study of proliferative rates in nodule-in-nodule lesions and consisted of an MRN containing a focus of HCC in a noncirrhotic, hepatitic liver of unspecified etiology.44 Thus, MRNs may arise independently of, though parallel to, the development of cirrhosis.

An Alternative Hypothesis of MRN Development To explain the presence of portal tracts in MRNs and the identification of MRNs in noncirrhotic livers, while taking into account the possible clonality of these

Downloaded by: Albert Einstein College of Medicine. Copyrighted material.

MACROREGENERATIVE (DYSPLASTIC) NODULES-THEISE

lesions, we have previously suggested an alternative hypothesis of MRN d e ~ e l o p m e n t . This ~ ~ , ~hypothesis ' begins with a clonal expansion as one of the earliest stages of neoplasia, taking place in the setting of chronic liver disease and the resultant generalized increase in hepatocyte turnover. If the clonally expanded cells had a growth advantage over adjacent, nontransformed hepatocytes, the cells would expand to replace adjacent hepatic parenchyma. Such a growth advantage could be obtained through a higher proliferative rate or increased survival (e.g., through suppression of apoptosis). We then postulated that the expanding population of cells might grow in an infiltrative fashion, surrounding adjacent portal tracts and terminal hepatic venules. If these cells are resistant to the disease or scarring process affecting the rest of the hepatic parenchyma, then, as the rest of the liver becomes cirrhotic around it, this island of genetically altered hepatocytes with preserved hepatic architecture would take on the appearance of a large cirrhotic nodule, i.e., an MRN. Having already undergone the earlier genetic events leading to the clonal growth, these cells then are the most likely targets for the further "hits" necessary to develop an overt malignancy. The standard hypothesis of MRN development depends on increased proliferation to explain the size and premalignant potential of the lesions, whereas this alternative hypothesis is compatible with either an increased or a low proliferative rate compared to adjacent liver parenchyma. There are several studies of proliferative rates in MRNs using proliferating cell nuclear antigen (PCNA),'9,39Ki-67,13.38and argyrophilic nucleolar organizing regionP as quantifiable markers. The majority of these studies were performed to establish comparisons between type I and type I1 MRNs and HCC and did not specifically address hepatocyte proliferation in MRNs compared to adjacent cirrhotic tissue. Nonetheless, some of these data suggest that MRNs are relatively low proliferative lesions.39More recently we carried out a study to specifically address this question.19 Comparing PCNA expression in type I and type I1 MRNs and within the largest adjacent cirrhotic nodules in the same tissue section, as well as comparing atypical subnodules and microfoci of HCC to the surrounding MRN parenchyma, two results were evident: (1) the proliferative rates of MRNs were indistinguishable from, and usually lower than, the mean of proliferative rates of the adjacent cirrhotic nodules; (2) an increase in proliferative activity coincides with the emergence of atypia and then malignancy. Thus, MRNs derive neither their large size nor their premalignant potential from a continuing high proliferative rate, and these data support our alternative hypothesis of MRN development. One advantage of this alternative hypothesis is that it can explain in a unified fashion the development of HCC in a variety of clinical and pathologic settings: HCC arising in MRNs, in smaller cirrhotic nodules, or

SEMINARS IN LIVER DISEASE-VOL.

15, NO. 4, 1995

in noncirrhotic livers. For example, if the neoplastic transformation and clonal expansion happens parallel to the cirrhotic process, an MRN develops as described. If the clonal expansion happens after cirrhosis is established, then malignant transformation would appear to occur within average-size cirrhotic nodules. If the clonal expansion occurs far in advance of the cirrhotic process, then HCC would emerge in a noncirrhotic liver. With the development of appropriate molecular studies to demonstrate hepatocyte clonality, this clonality would then serve to define the actual field of neoplastic growth rather than an arbitrary size criterion relying on nodular growth. Indeed, a recent publication concerning clonality of cirrhotic nodules4s has assessed the clonality of HCC and cirrhotic nodules in HCC-bearing livers-not MRNs, however-in hepatitis C based on RFLP of the X chromosome-linked phosphoglycerkinase (PGK) gene and on random inactivation of the gene by methylation. RFLP analysis demonstrated that seven HCCs in seven cirrhotic livers were all clonal. Of 76 nodules from these seven livers, 33 were clonal, demonstrating inactivation of one allele of the PGK gene or the other. Significantly, when clonal cirrhotic nodules were identified, they usually were identified in clusters, with all nodules in a cluster exhibiting the same PGK gene inactivation. Since the gene inactivation should be random, this result would be unlikely unless all these nodules were derived from the same clone. The authors therefore concluded that the clonal expansion must have preceded the development of fibrous septa and, thus, preceded the development of cirrhosis, as we have predicted concerning MRNs. An important difference, then, between these clusters of cirrhotic nodules derived from a single clone and clonal MRNs would apparently be the degree to which these clonally expanded hepatocytes are able to participate in the scarring process; perhaps there are differences in expression of whatever cytokines are required for recruitment and/or activation of myofibroblasts in the sinusoids. If such sinusoidal cells could be recruited and activated, then the clonal population would divide into individual cirrhotic nodules. If such recruitment or activation were retarded or suppressed, then the clone would appear as an MRN. In fact, using immunohistochemical staining for anti-smooth muscle actin to identify perisinusoidal myofibroblasts, we have found that both type I and type I1 MRNs do have a significantly reduced number of such cells compared to surrounding cirrhotic nodule^.'^

CLINICAL IMPLICATIONS Issues of Nomenclature and Classification The problems of the current standard terminology of MRNs are fairly obvious. The term "adenomatous hyperplasia" suggests that these lesions are adenomalike,

Downloaded by: Albert Einstein College of Medicine. Copyrighted material.

366

which they are not, and hyperplastic, which, as described above is questionable. Moreover, it misapplies a term taken from Edmondson, who used it to refer to lesions that are not premalignant. The term "macroregenerative nodule," however, implies that the lesions are simply large regenerative nodules and hyperplastic in nature, a conclusion that, again, appears problematic. Any attempt at a new nomenclature must be clinically useful, indicating, where possible, which lesions are benign, which are premalignant, and which malignant. In all these instances, diagnostic criteria must be set forth clearly and be reproducible. We must also recognize that as molecular investigative techniques are brought to bear that may distinguish neoplasia from hyperplasia, we should have a system of nomenclature that can accommodate expected developments without requiring additional major revisions. Two new systems of nomenclature have been recently proposed. One system of nomenclature, proposed by Ferrell et al,4h was developed from findings of an interobserver diagnostic study. The subsequent system was developed by the International Working Party on Hepatocellular Nodules, under the leadership of Ian Wanless, convened by the World Congress of Gastroenterology, a panel that included most of the authors of the Ferrell classification and therefore supercedes that classif i ~ a t i o n . ~However, ' the Ferrell study and classification highlighted an important issue, which was incorporated into the Working Party scheme. There was considerable disagreement among pathologists regarding a diagnosis of malignancy in a group of nodules that displayed increased nuclear density (implying increased nuclear-to-cytoplasmic ratio), cytoplasmic basophilia, nuclear hyperchromasia and membrane irregularities, and variable loss of reticulin fibers. These nodules were separated out as "borderline," a term intended to convey light microscopic findings suggestive of, but not definitive for, HCC. This "borderline" concept is an important advance over the previous nomenclature. It makes concrete the notion that there are some lesions about which we are unable to make definitive assessments of malignancy and yet which are related to the process of HCC development. Thus, the concept of a gradient of progression from premalignant to malignant without a distinct definable border is recognized and included in the terminology. This system also establishes reproducible criteria for such ambiguous lesions. The Working Party system eliminates both macroregenerative nodule and adenomatous hyperplasia as too problematic and turns to another earlier term: "dysplastic nodule." Similar to the Ferrell terminology, it recognizes that to be clinically useful, the nomenclature should include categories indicative of greater risk for impending development of HCC, even when a strict line cannot be drawn between premalignant and malignant. This concept is reflected by dividing nodules without definite HCC into "low-grade" and "high-grade" categories. High-

367

grade lesions include but are not limited to the "borderline" lesions of Ferrell et al,46following the criteria of that study. On the other hand, low-grade lesions are those which do not display such features of atypia or progression, but are convincingly thought to be neoplastic on the basis of clinical and pathologic clues by which neoplasia or clonality can be recognized. For example, when a very limited number of large nodules are identified in a liver with coexistent HCC elsewhere, whether these nodules are atypical or not, they may be considered "dysplastic." Or when morphologic features suggest that the whole nodule is itself clonal even in the absence of atypia, as with a siderotic nodule in an otherwise iron-free liver, the nodule may be classified as "dysplastic." Of course, many large nodules without atypia or such clonal features may be seen in cirrhosis; these nodules, under the Working Panel classification, may simply be diagnosed as "large hepatocellular nodules, not otherwise specified." Among the advantages of this system of nomenclature are that it avoids terms which prematurely define nodules as hyperplastic, some of which may actually be neoplastic, and that it incorporates criteria concerning the full range of morphologies reported in these nodules, particularly the concept of nodule-in-nodule "clonelike domains." Furthermore, it recognizes the indistinct nature not only of the border between premalignant and malignant, but of the border between hyperplastic and neoplastic. Of course, as newer techniques emerge to more definitively discriminate neoplasia from hyperplasia, the blurring of this "lower" border may resolve and the "not otherwise specified" category will fall increasingly out of use. Thus, it may be expected that although use of the terminology may be refined over time, it should not require drastic modification. As with the Ferrell terminology, the issue of rapidly proliferative subnodules has still not been directly engaged by this terminology; however, since most nodulein-nodule lesions have features suggesting that they represent a clonal domain, their presence ensures a diagnosis of dysplastic, rather than hyperplastic, regenerative nodule under this scheme. What remains is to determine whether it is a high-grade or a low-grade lesion. The demonstration of higher proliferation rates in these subnodules indicates that most, if not all, such nodule-in-nodule lesions represent a further step toward the development of HCC and thus should be considered high-grade.I9 A potentially controversial aspect of the Working Party system lies in the virtual elimination of the size criterion for defining "dysplastic nodules" by its being reduced to 0.1 cm. In fact, this system classifies not only dysplastic nodules of any size, seen predominantly in cirrhotic livers, but also dysplastic foci, sometimes seen in noncirrhotic livers. In our alternative hypothesis of MRN development, we have already suggested that the size of these nodules is merely an artifact of the timing

Downloaded by: Albert Einstein College of Medicine. Copyrighted material.

MACROREGENERATIVE (DYSPLASTIC) NODULES-THEISE

15, NO. 4, 1995

TABLE 3. Features Distinguishing Dysplastic Nodules from Regenerative Nodules

TABLE 4. Features Distinguishing High-Grade from Low-Grade Dysplastic Nodules

Dysplastic nodules Clonelike features in the entire nodule or clone-like domains of cells within the nodule Unpaired arteries (i.e., without accompanying bile ducts) Dysplasia, focal or diffuse, particularly of small cell type Compression of portal structures Dysplastic nodules are usually few in number (