This paper is a revision of the paper published in the Japanese Journal of Portal Hypertension 1999; 5: 247–56.
Dr F Kondo, Funabashi Central Hospital, Department of Pathology, 6-13-10 Kaijin, Funabashi, Chiba 273-8556, Japan. Email: firstname.lastname@example.org
Abstract Problems in definitive diagnosis and etiology of various benign nodular hepatocellular lesions were evaluated. Of these lesions, focal nodular hyperplasia (FNH), nodular regenerative hyperplasia (NRH), nodular lesions associated with idiopathic portal hypertension (IPH), non-cirrhotic large regenerative nodules (LRN), hepatocellular adenoma (HA)-like hyperplastic nodules, and partial nodular transformation (PNT) have been suggested to be related to abnormal hepatic circulation. However, the following points are considered to need further clarification: (i) is the abnormal circulation caused by thrombosis, vasculitis, or congenital anomaly?; (ii) is thrombosis a cause or a result of congestion?; (iii) are impaired blood vessels primarily the portal veins or arteries?; (iv) how are these disorders related to various syndromes, immunological abnormalities and abnormal blood flow of other organs, which are reported to coexist with these lesions often?; and (v) how should non-typical cases, which differ from typical cases, be interpreted? In addition, a concept that may lead to solving these problems (anomalous portal tract syndrome; a hypothesis that congenital vascular anomaly is the origin of these benign nodular hepatocellular lesions) was introduced.
The frequency of hepatic nodular lesion detection has increased because of recent advances in imaging diagnostic techniques. Benign hepatocellular lesions such as focal nodular hyperplasia (FNH), nodular regenerative hyperplasia (NRH), and hepatocellular adenoma (HA), as well as small hepatocellular carcinoma, are detected more frequently today. An increasing proportion of such lesions do not fit the concepts of known diseases, and are classified as non-typical cases.1 Also, some of these diseases are discovered in combinations.2,3 For these reasons, the etiology of these diseases is often difficult to explain by conventional theories.
This review clarified problems in the present diagnosis of benign hepatocellular lesions, re-evaluated their conventional classification and existing etiological theories, and proposed solutions by introducing a new viewpoint. The problems discussed here have been issues for several years in the medical societies of Japan.3 This paper incorporated new findings and the recent literature that have been derived from the discussion.
CONVENTIONAL CLASSIFICATION OF BENIGN NODULAR HEPATOCELLULAR LESIONS AND ITS PROBLEMS
Classification of nodules
Following is a summary of the definitions and knowledge concerning various benign nodular hepatocellular lesions.
Hepatocellular adenoma (adenomatosis) or liver cell adenoma
Hepatocellular adenoma is a benign neoplastic lesion. It often occurs as isolated lesions in the non-cirrhotic liver. However, as the term ‘adenomatosis’ suggests, it may occur as multiple lesions. Many of the reported lesions are several centimeters in diameter.
The clinical background is important for the diagnosis of HA. It is considered to occur in the liver under proliferative stimulation of hepatocytes, for example, a past history of the use of high-dose oral contraceptives,4 anabolic steroid, and the presence of a complication such as glycogen storage disease.
Focal nodular hyperplasia
Focal nodular hyperplasia is a hyperplastic nodule that contains scar-like tissue. In typical cases, the nodule contains stellate fibrous tissue called central scar. The nodule is considered to be a hyperplastic lesion reactive to abnormal arterial vessels in the central scar. Because the nodule may present with a profile grossly resembling cirrhosis because of multiple scar-like tissue in the nodule, it used to be called focal cirrhosis.5 It is reported to occur in the non-cirrhotic normal or nearly normal liver.6 Although the lesion is mostly solitary, multiple lesions have also been reported. The size of the lesion is often several centimeters in diameter.
Nodular regenerative hyperplasia
In this disease, small multiple nodules of hepatocytes occur diffusely in the liver. Unlike liver cirrhosis, there are no fibrous septa. Nodular regenerative hyperplasia is often complicated by portal hypertension and occurs with various syndromes, blood diseases, and immunological abnormalities in many cases,7–12 but not in others.
Typical NRH is a diffuse condition and it is not common for a large nodule to develop that needs to be differentiated from the other nodular lesions. However, unusually large nodules have been recently detected within the liver with macro- and microscopic appearance of NRH (Figs 1,2). Some nodules have indistinct scar-like tissue. These nodules can be variously interpreted, for example an unusually large nodule of NRH, an HA in NRH-like liver, or a regenerative nodule resembling FNH. Therefore, NRH is included in this article.
Partial nodular transformation
A large nodule is formed at the hepatic hilus.13 The disease is often complicated by portal hypertension.
Idiopathic portal hypertension
Idiopathic portal hypertension (IPH) is basically a diffuse disease.14,15 However, nodular hyperplasia of hepatocytes is not infrequently seen in IPH. Such lesions are not surrounded by fibrous septa and show vague nodularity. They may resemble NRH, partial nodular transformation (PNT) or FNH.15 Many of these nodular lesions are benign, and reports of hepatocellular carcinoma are rare.
Large regenerative nodule
The meaning of the term, ‘large regenerative nodule (LRN)’, by the conventional usage16 is different from that by the definition of the Terminology of Nodular Hepatocellular Lesions by the International Working Party (IWP),6 and so caution is needed.
According to the conventional usage, LRN often meant regenerative nodules with very large size in the cirrhotic liver. Their nodule size is clearly larger than the surrounding regenerative nodules of ordinary size.16 Large regenerative nodule according to the Japanese General Rules for Clinical and Pathologic Studies on Liver Cancer17 means such lesions. Differentiation between LRN in cirrhosis and hepatocellular carcinomas (HCC) is very important in clinical situations. It is sometimes very difficult using only various images and clinical data. Detailed pathological examination is necessary for the differentiation.18,19
According to the IWP terminology, however, large regenerative nodules in non-cirrhotic patients are also called LRN. For example, large regenerative nodules after necrosis, large nodules of NRH, and nodular lesions concurrent with IPH are also LRN. Also, the IWP terminology distinguishes nodules that have characteristics of FNH, but are surrounded by abnormal liver tissues from typical FNH as LRN, because FNH is a result of reactive hyperplasia to large anomalous vessels,6,20 while such LRN are estimated to be formed by another mechanism, that is, repair and regeneration after tissue damage.
Thus, as LRN according to the IWP terminology includes nodules of diverse properties, the diagnosis of a lesion defined simply as LRN is never a definitive diagnosis. More detailed description of its characteristics is needed.
Problems with classification
Various non-typical and difficult cases
Many lesions encountered recently do not fit any of these typical cases and pose difficulty in definitive diagnosis. Therefore, a review of these typical lesions and evaluation of their problems are needed.
The following cases, which do not match the definitions of various lesions, have been regarded as difficult cases (to diagnose or classify): (i) the nodule itself matches the characteristics of FNH, but NRH-like or IPH-like lesions are observed outside the nodule, which contradicts with the definition that the lesion ‘occurs in the normal or nearly normal liver’6 (Fig. 1A,C); (ii) the nodule of NRH contains scar-like tissue and resembles FNH (Fig. 1B); (iii) the lesion is a benign hepatocellular nodule, but there are tissues that cannot be identified as a characteristic central scar of FNH, and distinction between FNH and HA is difficult (Fig. 1D,E). Conceptually, FNH is a hyperplastic lesion, and HA is a benign neoplastic lesion, but there is no decisive finding for their distinction in the hepatic parenchyma. It is sometimes very difficult to determine whether the lesion is neoplastic or non-neoplastic. The diagnosis is often made according to the principle that the lesion is FNH if the tissue is judged to be a characteristic central scar of FNH that contains a thick artery and radiating branches. The presence of a small bile ductule is believed to be more important for the diagnosis of FNH. However, this factor also has some problems, which will be discussed later; (iv) the nodule has the background of NRH, but it is much larger than a typical NRH nodule (Fig. 1F); (v) the distinction between NRH and IPH is difficult because of the presence of areas showing nodule formation and those not showing it in the liver with portal hypertension (Fig. 1A–C); and (vi) the distinction between FNH and PNT is difficult because of the presence of scar-like tissue in the nodule of the hilar region. Established diagnoses are difficult to assign to these lesions.
These lesions may be described collectively as ‘LRN unaccompanied by liver cirrhosis’ following the IWP terminology, but the similarities and differences in their pathological profiles and etiology compared with those of the typical cases of known diseases (HA, FNH, NRH, IPH) must still be discussed.
Even in the microscopic examination, the definitive diagnosis is difficult in these lesions. The parenchymal features are similarly recognized as benign regenerative or hyperplastic liver tissue in the nodular areas. The nodular hyperplastic areas gradually transform into the peripheral atrophic (or non-hyperplastic) areas (Fig. 2). Interstitial features are also similar to one another. They show various abnormal blood vessels and bile ducts in the portal tracts (Fig. 2).
Figure 3 shows the possibility of various diagnoses for non-typical cases depending on the interpretations of the findings.
Part (A) is a non-typical case. Although there is tissue resembling central scar in the nodule, the findings are not typical. Nodule formation, although obscure, is observed in the surrounding background hepatic tissue. Part (B) shows various interpretations of this case: (i) the diagnosis is HA as the background liver is interpreted to be in the normal range and central scar to be absent in the nodule; (ii) the diagnosis is FNH as the background liver is interpreted to be in the normal range and central scar to be present in the nodule; (iii) the diagnosis is NRH as nodule formation is interpreted to be present, and the lesion in question to be a similar nodule; (iv) the diagnosis is LRN (IWP terminology) as nodule formation is considered to be present in the background liver. Because the background liver is abnormal, the lesion is interpreted to be LRN rather than true FNH despite its similarity to FNH (such type of LRN is described as FNH-like LRN hereinafter in this paper).
Thus, non-typical cases may be diagnosed differently because of the diversity of subjective interpretation.
Evaluation of the etiology of these lesions has also been insufficient. There have been widely accepted etiological concepts for typical cases, and abnormal hepatic circulation has often been considered responsible. However, abnormal hepatic circulation alone is not a sufficient etiological explanation, and a solution to the following problems is needed:21 (i) how should non-typical cases that do not fit various typical lesions be interpreted? If FNH, NRH, IPH, and HA are regarded as different pathological entities resulting from different etiological mechanisms, it is difficult to offer the etiological explanation of the non-typical case; (ii) is the cause of abnormal hepatic circulation thrombosis, vasculitis, or congenital anomaly?; (iii) is thrombosis a cause or a result?; (iv) are the damaged vessels pri-marily in the portal system or in the arterial system?; (v) how are these diseases related to various syndromes, immunological abnormalities, and blood diseases, which are reported to complicate them often?; and (vi) can the etiology of these lesions be explained by the regeneration after tissue damage or compensatory increase in the blood flow?
For example, FNH is considered to be a hyperplastic nodule caused by the blood flow of an anomalous large artery.6,20 However, Kumagai et al.22 considered that FNH is derived from acquired thrombosis. Nodular regenerative hyperplasia is considered to be a result of obstruction or narrowing of portal branches caused by thrombosis and atrophy of areas with severely impaired blood flow, and hypertrophy of areas with relatively mild impairment of blood flow, leading to nodule formation.10 Mechanisms such as reactive hyperplasia after hepatocellular injury induced by vasculitis11 have also been suggested. However, etiological interpretation of the concurrence of FNH and NRH and the presence of intermediate types is difficult.
Nodular regenerative hyperplasia is known to occur with various syndromes,8,9 immunological abnormalities,12 thrombosis,10 and vasculitis,11 and these conditions have been considered to be involved in its etiology. However, their concurrence is unclear in some patients. Also, whether NRH or IPH is caused directly by thrombosis or thrombosis is made secondarily by portal congestion remains controversial.
Furthermore, HA is defined as a benign tumor, and the use of high-dose oral contraceptives is considered likely to be its etiologic factor.4 However, some patients diagnosed to have HA have no history of the use of oral contraceptives. Focal nodular hyperplasia-like lesions with unclear scar tissue, hyperplastic lesions based on abnormal circulation, and LRN according to the IWP terminology may have been diagnosed as HA. The presence or absence of proliferation of the bile ductules has been suggested as a point of differentiation, but this criterion also has some problems, which are as follows. Evaluation of this finding becomes difficult if the normal portal region is trapped in the nodule.6 Proliferation of bile ductules can be found even in a definite neoplastic lesion, namely hepatocellular carcinoma. Even in HA (a benign neoplastic lesion), the presence of tiny ductules may possibly be proved to be positive if immunohistochemical staining for biliary cytokeratins were used.
Moreover, there are problems also with the concept that nodule formation is a result of the compensatory tissue repair process. It has been confirmed that an area where the portal blood flow is blocked by embolism becomes atrophic.23 A compensatory increase in the arterial blood flow, if there is any, is considered to be insufficient to lead to nodule formation.
There is also the theory that FNH-like LRN develops by reactive regeneration after liver injury.6 In some cases, however, the liver tissue damage is not severe enough to lead to nodule formation. Massive necroinflammatory changes that may cause large regenerative nodules have not been observed. Also, the FNH structure is usually absent in true large regenerative nodules following fulminant hepatitis. In addition, abnormalities are often discovered in the portal branches and arteries of the portal area on close examination, even in true FNH occurring in the nearly normal liver (i.e. reactive hyperplastic lesion caused by vascular malformation),24 and the distinction between normal and abnormal liver is extremely difficult. Further discussion concerning the interpretation of compensatory nodule formation and background liver (extranodular liver tissue of FNH) of FNH is needed.
HYPOTHESIS FOR THE SOLUTION OF PROBLEMS
Under these circumstances, correct understanding of the etiology and characteristics of various nodular lesions is demanded. The author would like to present a viewpoint useful for finding solutions to these problems. This hypothesis needs further confirmation before it is established, but it provides reasonable explanation of these problems. It is described in detail below.
The presence of various non-typical, intermediate, and transitional types (namely FNH and NRH, or FNH and HA etc.) is explained better by ascribing their etiology to a single factor.
As suggested previously,3,24 many of the diseases mentioned above may be explained by supposing a common causation, that is, anomaly of the portal tract (Glisson's sheath) or anomaly of hepatic vessels including the hepatic vein. Both the arterial and portal blood flows may be impaired by vascular anomaly (or so slight an anomaly that it may be called a variation), and this disturbance of the blood flow causes hyperplasia and consequent nodule formation in areas with a richer blood flow.
Figure 4 describes a hypothesis of the relationship between the vascularity and nodule formation.
Owing to the recent advances of image diagnosis, vascularity of nodular hepatocellular lesions is more easily evaluated than before. By using CT arteriography (CTA) and CT arterioportography (CTAP), we can easily examine which blood vessels are dominant in nodular lesions.25 Clinical doctors are now trying to make a definitive diagnosis based on these vascular findings. Therefore, schematic explanation on the relationship between the nodule formation and its vasculature is needed.
Nodules may be formed in whichever of the arterial flow and the portal flow is dominant. An area becomes hyperplastic if it receives a higher blood flow with the arterial and portal flows combined, compared with the surrounding tissues. In fact, both types of nodules are present. However, it should be noted that an artery-dominant nodule detected by using imaging techniques must not be immediately regarded as a neoplastic nodule simply because of the arterial dominance. Also, even if the portal vein is widely patent, the influx of the portal blood may be prevented because of an increased sinusoidal pressure if there is a rich arterial blood flow. In such a case, the portal vein is not visualized by the use of portography. Such a lesion must not be interpreted as an artery-dominant hyperplastic nodule resulting from compensation of a reduced portal blood flow by the arterial blood flow. The fact is that the portal blood cannot enter the nodule in which the arterial circulation is originally dominant. Indeed, the liver is atrophied after portal branch embolization as mentioned above.23 Simple occlusion of portal vein and compensatory increase of arterial blood flow were proved to be insufficient for nodule formation. Many other conditions may be related to nodule formation.
Similar hyperplastic nodules are formed also when the portal blood flow is dominant.
The areas with a low blood flow and those with a blood flow comparable to that in the surrounding regions become atrophic areas and nearly normal areas with no nodule formation, respectively.
Interpretation of the terms ‘hyperplasia’ and ‘regeneration’ should be mentioned here. Theoretically, ‘regeneration’ occurs as a process of repair of damaged hepatic tissue, and ‘hyperplasia’ is simply a reaction to proliferative stimulation (e.g. an increase in the blood flow) without hepatocellular injury. In the actual situation, however, regeneration and hyperplasia may occur simultaneously, and they may not be differentiated in a rigid way (e.g. regenerative hyperplasia).
Congenital anomaly theory
The grounds for our support of the theory that congenital anomaly rather than acquired thrombosis is the common cause of benign hepatocellular nodules are as follows.
The involvement of thrombosis or vasculitis has often been suggested in the etiological process of IPH and NRH, but the findings in the portal tract were similar in FNH, NRH, and IPH according to our past results.24 That is, narrowing of the portal vein and the bile duct and muscular thickening of the arterial wall (Fig. 2) were commonly observed. The changes in the arterial wall occasionally caused irregular protrusions into, (Fig. 2) or complete obstruction of the arterial lumen. Occasionally, also, abnormal arrangement of the portal vein, artery, and bile duct, irregularity of their external morphology or the width of their lumens and their numerical increases or decreases were observed. Abnormal arrangement of the hepatic vein was also noted. Of these changes, changes in the muscular element of the artery and bile duct and abnormal vascular arrangement, which cannot be explained by thrombosis or organization of thrombi, compensatory changes, or inflammation, suggest congenital anomaly.
The abnormalities in the background liver of FNH-like LRN can also be explained by congenital anomaly of the portal areas. Mild abnormalities observed in the background liver are anomalous vessels similar to those in the nodules. The development of various lesions shown in Figs 1–3 can be explained by considering that nodule formation occurs in areas with more severe abnormality, but not in areas with milder abnormality.
Such nodular lesions are observed also in children and infants.26,27 Reports of lesions accompanied by clear congenital vascular anomalies28–30 and concurrence of lesions with vascular abnormalities in other organs31 also support the congenital anomaly theory.
Detection of lesions or the appearance of symptoms in and after the middle age is not contradictory to congenital anomaly. Many of these nodules are asymptomatic, and younger individuals have fewer opportunities to receive screening of the vascular system. Also, symptoms may develop only after changes in the hemodynamics associated with the growth. In the case shown in Fig. 1C, atrial septal defect (congenital cardiac anomaly) became symptomatic after middle age.
Coexisting disease and secondary factors
The relationships of these lesions with various coexisting diseases must also be evaluated.
Figure 5 shows the process of nodule formation and relationship with various diseases that often concur with them. Hepatic nodular lesions are known to concur with thrombosis, immunological abnormalities, vasculitis, various syndromes, and vascular anomalies of the liver and other organs. According to the theories to date, these coexisting diseases were considered to play some etiological roles (Fig. 5A, serial theory). These coexisting diseases were considered to cause abnormal hepatic circulation or hepatocellular damage, and nodules are formed as a result of reactive regeneration. Some nodules may be formed in the acquired process of the serial theory. Some dramatic clinical event known to affect vessels initially occurs, and then regenerative nodules are formed. However, such dramatic events do not necessarily precede (Fig. 1). Moreover, no coexisting disease is observed in some patients. In fact, all of the five cases in Fig. 1 showed no symptoms caused by these nodules. The signs of portal hypertension were found in the two cases (A-B) and in case C. Only case E was a viral carrier (hepatitis B virus). Liver function tests were normal or nearly normal, and the clinical data did not show immunological abnormalities, coagulation disturbance or vasculitis in all these cases. Also, vasculitis, thrombosis, immunological abnormalities, and various syndromes do not always result in abnormal hepatic circulation or nodule formation. It is summarized that the nodules are not always caused by these diseases, and that these diseases do not always cause nodule formation.
Alternatively, these conditions may be in simple coexistence and not have direct etiological relations (Fig. 5B, parallel theory). According to this view, there are congenital factors (genetic abnormalities) that cause abnormal hepatic vasculogenesis, and the presence or absence of coexisting diseases or their types are determined by the severity or the pattern of original genetic abnormality or associated congenital factors (genetic abnormalities). The direct cause of nodule formation is abnormal hepatic circulation, and coexisting diseases are simply concurrent phenomena. Thrombosis and vasculitis, which intensify the unevenness of distribution of the blood flow, may be involved as secondary worsening factors in nodule formation or abnormal circulation. Similarly, drugs that promote regeneration or proliferation of hepatocytes may also serve as secondary worsening factors.
Clinical course (shrinking, disappearance, and growth) of nodules
These nodular lesions may show enlargement or shrinking during a follow-up study.32 The author has also experienced lesions that became undetectable on imaging studies and lesions detected newly after resection of main nodules as well as enlargement and shrinking of nodules (Kondo, unpubl. data, 2000). Although causes of these changes cannot be identified, they may be related to the above secondary worsening factors. The hemodynamics may change with the growth, abdominal surgery, or various coexisting diseases. The nodules may enlarge (and become visible by using imaging techniques) if the perfusion increases. But they may shrink and become undetectable if the perfusion decreases. Actually, in our past case, many nodules were observed in autopsied livers, some of which were visible, but others were not on imaging studies. The nodules in Fig. 1A(a,b) and Fig. 1B(a–d) were not visualized. The nodules were clearly observed in macroscopic photographs. Microscopically, however, the contrast showed gentle transition in the margins of the nodules from the hyperplastic area to the non-hyperplastic area in the non-nodular part, probably because of insufficient contrast of uneven distribution of the blood flow. For these reasons, these lesions are considered to have been undetectable by imaging modalities.
Integration of disease concepts
Figure 6 shows a new concept developed by rearranging the conventional classification on the basis of the above integrated etiological theory.
There is abnormality of the portal tract (Glisson's sheath; congenital anomaly) as a common pathological condition, and individual lesions are classified according to localization and distribution of primary lesions: (i) some conditions of non-cirrhotic portal hypertension without nodule formation (like IPH) can be caused by congenital vascular anomaly; as etiology of IPH has been controversial, the term ‘non-cirrhotic portal hypertension’ was used in Fig. 6; (ii) nodular regenerative hyperplasia is a small multinodular (diffuse nodular) lesion; (iii) in FNH, a few nodules are often localized in peripheries of the liver and contain scar-like tissues; and (iv) large regenerative nodules (particularly non-cirrhotic LRN) show a small number of localized nodules without scar-like tissues. Some lesions classified as HA are not considered to be genuine tumors or to be similar in nature to LRN.5 Partial nodular transformation forms localized nodules in the hilar region.
Schematic representation of these lesions facilitates an understanding of the presence of incomplete or intermediate types of typical lesions (namely FNH, NRH, LRN, HA etc.) in addition to these basic types. Also, FNH-like LRN with abnormal surrounding liver tissues (IWP terminology), and typical FNH are regarded as nearly the same disease entity according to this classification. The author (Kondo) expects that these lesions can be lumped as a single syndrome once their etiology has been established. Such a syndrome may be called anomalous portal tract syndrome or anomalous vasculature syndrome.3 These lesions may be complicated by intrahepatic or extrahepatic vascular lesions and vascular lesions of other organs, but the coexistence of these lesions is easy to understand by considering that at least part of this syndrome is a subtype of generalized vascular anomaly.
Such integration of pathological concepts has been attempted to be included in the classification of bile duct lesions.33 Focal nodular hyperplasia has also been reported to frequently concur with biliary hamartoma,1 and the possible relationship of this syndrome with congenital anomaly of the bile duct also awaits clarification.
Abnormalities of hepatic circulation have been suggested for many years as possible etiological factors in benign nodular hepatocellular lesions. However, detailed evaluation of how abnormal circulation is related to the diseases is needed for explanation of various phenomena associated with the disease.
In this paper, the author reviewed conventional etiological theories, clarified their problems, and proposed a hypothesis that may lead to their solutions. This hypothesis was derived from our own experience3,16,24,30 and that of others and a review of the literature. Although this hypothesis needs further verification and modification, it is considered to be worth being developed further along with conventional theories. The author admits that some nodules are formed in a completely acquired process. The conventional theory should also be carefully evaluated. However, congenitally formed nodules also exist. An introduction of the congenital anomaly theory is necessary, at least for the interpretation of various difficult cases.
The author hopes that this hypothesis contributes to clarification of the pathology of nodular lesions of the liver and leads to further progress of terminology and advance of medicine.
The author wishes to express his sincere gratitude to Professor Emeritus Kunio Okuda (Chiba University) for offering this opportunity for presentation of work.