Regression of Human Cirrhosis: Morphologic Features and the Genesis of Incomplete Septal Cirrhosis

Cirrhosis is widely regarded as being irreversible. However, several studies have demonstrated with serial liver biopsies that fibrosis may decrease with time in some cirrhotic livers.1 This partial regression has been documented in diseases in which progression can be arrested by therapy, as in alcoholism, hemochromatosis, Wilson disease, Indian childhood cirrhosis, autoimmune hepatitis, and chronic hepatitis C virus. Complete regression of cirrhosis has not been clearly documented.

We describe a patient with hepatitis B cirrhosis treated with lamivudine; serial biopsies demonstrated apparent regression of cirrhosis in this patient. Because of concern for sampling error, documentation of regression requires an ample biopsy followed by examination of the entire liver. In the present study, we examined a series of cirrhotic explants for histologic evidence of regression. Although we did not have the benefit of sequential specimens in the explant series, a number of observations and logical arguments support our belief that all cirrhotic livers show features of regression. We call these features the hepatic repair complex. This study allows one to understand For editorial comments, see pp 1585, 1587, 1589, 1591, and 1592.visually how septa and large fibrotic regions of parenchyma can be returned toward a normal appearance. Livers with cirrhosis and incomplete septal cirrhosis (ISC) demonstrate a continuum of regression changes, suggesting that some examples of ISC are actually regressed cirrhosis.

The clinical histories of patients undergoing liver transplantation at Toronto General Hospital (Toronto, Ontario, Canada) were reviewed for etiology of liver disease. Patients were selected at random for this study if cirrhosis (n = 39) or ISC (n = 13), as defined by Sciot et al,2 was present and no more than 1 etiology was documented. The cirrhotic livers were associated with alcoholism (n = 9), chronic hepatitis B virus (n = 8), chronic hepatitis C virus (n = 12), or primary biliary cirrhosis (n = 10). The apparent etiology of the ISC was alcohol (n = 3), hepatitis B (n = 5), hepatitis C (n = 2), α₁-antitrypsin deficiency (n = 1), and idiopathic (n = 2). The alcoholic patients were believed to have abstained for at least 6 months prior to transplant. The patients with hepatitis B had hepatitis B virus DNA in low or undetectable amounts, and had serum aspartate aminotransferase levels less than 2 times the upper limit of normal at the time of transplantation. Patients with hepatitis C had serum anti–hepatitis C virus by second- or third-generation test.

Slides from 1 block from each liver, chosen at random from those available, and stained with hematoxylin-eosin and elastic trichrome were reviewed. Blocks from 20 livers were stained for CD34 by the immunoperoxidase method (Novocastra, Newcastle upon Tyne, United Kingdom). Each of the selected parameters was evaluated on a separate day without knowledge of the etiologic diagnosis. The parameters were graded 0 through 2 unless otherwise stated, where 0 indicates absent; 1, occasionally present; and 2, present in numerous fields. General parenchymal injury was assessed as listed in Table 1. Sinusoidal fibrosis was said to be present if there were delicate collagen fibers in several adjacent spaces of Disse. Venoportal fibrous adhesions were recognized by the close approximation of portal tracts and hepatic veins with a short fibrous attachment between the structures. Large regions of parenchymal extinction were defined by the presence of scars at least 2 mm in least dimension, determined with an eyepiece graticule. The size of cirrhotic nodules was expressed as an approximate mean diameter.

Eight parameters likely representing hepatic repair were identified during a preliminary study of many specimens not included in the present review. We refer to these parameters as the hepatic repair complex. These parameters are defined as follows. Perforated delicate septa were fibrous septa that were incomplete but whose residual components formed a linear, usually curved, structure. Isolated thick collagen fibers were heavy bundles within the parenchyma not visibly attached to a portal structure, hepatic vein, or septum. Delicate periportal fibrous spikes were collagenous extensions from portal tracts but without visible connection to other portal tracts or hepatic veins. Hepatic vein remnants with prolapsed hepatocytes were hepatic veins with the lumen containing hepatocytes. Hepatocytes within portal tracts or split septa were clusters or cords of hepatocytes 2 or more cells in thickness, apparently located within portal tracts or wedged between layers of fibrous septa. Minute regenerative nodules (“buds”) were small clusters of hepatocytes (less than 10 cells in diameter) admixed with ductules. Aberrant parenchymal veins were veins close to portal tracts (within 5 hepatocyte diameters). Portal tract remnants had several appearances, including artery/duct pairs, unaccompanied arteries, or unaccompanied ducts, usually with absent portal vein. Portal tract collagen amount was less than found in normal portal tracts, so that hepatocyte plates closely approximated the portal tract elements.

The χ² or Fisher exact tests were used for statistical analysis.

A patient with cirrhosis and chronic hepatitis B virus who was successfully treated with lamivudine demonstrated apparent regression of cirrhosis on serial needle biopsies (Figure 1).

Histologic characteristics of the series of explanted specimens are summarized in Tables 1 and 2. Most livers had little or no activity of the original disease. The elements of the hepatic repair complex were found in the majority of cirrhotic livers, independent of etiology. The cirrhotic livers having low grades of the repair features were those livers with significant activity, as repair was obscured by active inflammation or bridging necrosis.

The features of the hepatic repair complex are illustrated in Figures 2 through 10 and interpreted in Figures 11 and 12. Delicate perforated septa were found in almost all livers examined (Figures 2 through 4), but were found in only 50% of livers with primary biliary cirrhosis, likely because of continuing periportal and paraseptal necrosis in that disease. At sites of perforated septa, adjacent nodules appeared to merge. The degree of septation often varied within the same liver, so that the histologic appearances of both complete septal cirrhosis and ISC were present (Figure 5). Livers with incomplete septal cirrhosis also showed a spectrum both within the same liver and among livers, with nearly normal architecture in its mildest form (Figure 6).

Sinusoidal fibrosis was seen in 74% of livers, but was always focal and usually minimal in severity. The 7 livers with the most severe sinusoidal fibrosis (grade 2) had some fibrosis in nearly all nodules, but this involved only 10% to 30% of sinusoids. Parenchyma with sinusoidal fibrosis often appeared to be compressed by nodules of regenerative hepatocytes that lacked sinusoidal fibrosis (Figure 7).

Adhesions between medium-sized portal tracts and hepatic veins were found in all cirrhotic and ISC livers (Figure 8). Aberrant parenchymal veins were found in all ISC livers but in only 51% of cirrhotic livers, in which the number of patent hepatic veins was reduced markedly.

Large regions of extinction were present in 64% of cirrhotic and 8% of ISC livers. Minute regenerative nodules tended to occur within large regions of extinction. Such minute nodules (buds) were seen in 85% of livers having large regions of extinction (Figure 9). Other features of repaired parenchyma are illustrated in Figure 10.

The general clinical experience with chronic liver disease is that the histologic and clinical severity of fibrosis progresses with time.3–5 Regression of fibrosis was seen in many experimental models, including biliary obstruction,6–8 carbon tetrachloride toxicity,9–11 nutritional models,12 ethionine toxicity,13 and schistosomiasis,14,15 among others, as reviewed by Perez-Tamayo.1 The cellular mechanisms of collagen resorption have been studied in some of these models.11,16 

In humans, one study reported that 68 of 75 patients with micronodular cirrhosis on biopsy had macronodular cirrhosis at autopsy, with a median conversion time of approximately 2 years.17 At least partial regression of human cirrhosis has been documented in treated patients having biliary obstruction,18 hemochromatosis,19–22 Wilson disease,23,24 intestinal bypass–related cirrhosis,25 Indian childhood cirrhosis,26 autoimmune hepatitis,27 primary biliary cirrhosis,28 chronic viral hepatitis,29–32 and alcoholic liver disease.29 

In most studies, histologic changes have been reported as a decrease in fibrosis scores on needle biopsy specimens.28,31,33 One study showed decrease of fibrosis using a morphometric technique.32 One patient with hemochromatosis and cirrhosis on biopsy showed progressive loss of fibrosis during phlebotomy therapy and total absence of hepatic fibrosis at the time of autopsy.34 We present a patient with hepatitis B cirrhosis who underwent serial biopsies and in whom lamivudine therapy was followed by nearly total disappearance of fibrous septa.

A detailed histologic description of regression has not been reported for human cirrhosis. In the present study of explants with cirrhosis or ISC, 8 histologic parameters were found that we interpret as features of regression.

Cirrhosis occurs with the accumulation of many discrete lesions of parenchymal extinction (Figures 11 and 12).35 Extinction lesions form as contiguous liver cell plates collapse and collagen is deposited in the site of injury. The histologic appearance of the active injury varies with the particular disease. In viral and autoimmune hepatitis, the collapse is recognized as bridging necrosis. Regression involves 2 main processes, namely, removal of fibrosis and the repopulation of scarred regions with hepatocytes.

Hepatic fibrosis is present as a continuum that we divide into 4 categories for purposes of analysis: sinusoidal fibrosis, septal fibrosis, fibrous adhesions, and large regions of extinction. The histologic appearance suggests that all of these categories of fibrosis undergo regression, although with different ease.

Active alcoholic liver disease usually demonstrates fibrosis in the subendothelial spaces of sinusoids. In contrast, histologically visible fibrosis is absent in the majority of sinusoids in our livers with inactive alcoholic cirrhosis. This observation suggests that sinusoidal fibrosis is either removed in situ or displaced and accreted onto septa. We believe that both processes occur simultaneously. Resorption of sinusoidal collagen likely begins while the hepatocyte plate is still intact. If the hepatocytes die, the sinusoidal fibers may collapse into broader septa. However, this accretion is transient, since septa appear to be resorbable.

We have observed that most cirrhotic livers with low-grade activity have some delicate septa with varying degrees of perforation and fragmentation. The following indirect evidence indicates that these septa are resorbing rather than progressing.

1. Several livers in this study had identifiable populations of hepatocytes that differed in adjacent nodules (Figure 2). This appearance implies that septa have a barrier function isolating hepatocytes from free migration. Broad septa can be seen to have complete barrier function, while delicate and perforated septa allow mingling of adjacent populations. Assuming that distinct populations develop when barrier function is complete, mingling at a perforation site would imply that the septum is being resorbed.

2. Adjacent cirrhotic nodules have independent microcirculations. These independent microcirculations can be recognized in adjacent nodules separated by perforated septa (Figure 3), suggesting that perforation occurred after the development of the microcirculation.

3. Reduction in portal tract collagen to amounts less than normal, as found in portal remnants (Figure 10, D and E), suggests an active resorptive process.

4. Progressing sinusoidal fibrosis, as seen in active alcoholic hepatitis, is characterized by collagen fibers that are mostly delicate. However, in our study, large collagen bundles dominate (Figure 10, D), suggesting that most delicate fibers have been resorbed, leaving the heavier original architectural collagen.

Septa represent collagen bundles that have linked the heavier architectural elements, especially portal tracts and hepatic veins. As septa are resorbed, portal tracts and hepatic veins become separated from each other and are recognized as remnants with a number of abnormalities. Portal tracts lose much of their native collagen and small portal veins are obliterated, so that portal tract remnants are most often artery/duct pairs with scanty collagen. Hepatocytes often appear to have migrated into the stroma of portal tracts, as recognized by the finding of hepatocytes immediately adjacent to ducts or arteries (Figure 10, D and E) and rarely in the former lumen of a portal vein. Medium and large hepatic veins usually return nearly to normal. Most small hepatic veins are lost without a trace, but a few medium-sized veins remain as fibroelastic cords (“VOD” lesions) or rings of collagen bundles with the former lumen containing prolapsed hepatocytes (Figure 10, F). Such an appearance is important evidence that normal (ie, nonneoplastic) hepatocytes are able to migrate into regions previously devoid of hepatocytes. Similar hepatocellular prolapse has been reported in idiopathic portal hypertension36 and other noncirrhotic conditions.37 

Adhesions are broad and very short septa between portal tracts and hepatic veins. Adhesions appear to persist longer than narrower septa and thus become a prominent feature of late regression, as in ISC. As adhesions lyse, the hepatic veins remain close to the portal tract and are recognized as aberrant (Figure 11, F). Therefore, adhesions and aberrant veins, as indicators of previous extinction events, document the prior existence of a cirrhotogenic disease.

Adhesions and aberrant veins have both been reported in idiopathic portal hypertension, ISC, and cirrhosis.37,38 Several authors have discussed the origin of aberrant veins, variously called misplaced dilated central veins,39 aberrant vasculatures,40 and periportal angiomatosis.2 Our analysis provides a new perspective on the genesis of these veins.

Large regions of extinction41 are a major obstacle to the reversal of cirrhosis. These regions appear to repopulate from buds composed of proliferating ductules and hepatocytes surrounded by collagen and CD34-positive vascular channels (Figures 9 and 12). The hepatocytes eventually remodel into nearly normal plates with scanty or no visible sinusoidal collagen. However, large extinction lesions often remain visible as well-demarcated regions with sinusoids lined by CD34-positive endothelial cells (Figure 10, C). Because these CD34-positive regions often contain unaccompanied arteries and telangiectatic sinusoids (Figure 10, B), we believe these regions contain arteriovenous shunts. Arterialized sinusoids, lined by CD34-positive endothelial cells, may be equivalent to “capillarization,” as described by Schaffner and Popper and others.42,43 

Cirrhotic nodules appear to enlarge by expansion against septa as well as by lysis of septa (Figure 12). Nodules surrounded by thin septa coalesce first, giving rise to macronodules or large regenerative nodules surrounded by broad septa and large regions of extinction (Figures 3; 4, A; and 10, A).

Incomplete septal cirrhosis is a condition characterized histologically by delicate and incomplete fibrous septa.2,44,45 The livers with ISC in our study demonstrated all of the features of the hepatic repair complex, including venoportal adhesions and aberrant veins that would not be expected in early-stage disease. These results are consistent with the view that ISC may arise from complete cirrhosis. This hypothesis requires confirmation with examination of adequate serial specimens. It is likely that the cirrhosis must become inactive with little or no addition of new extinction lesions, so that at the time of examination all the lesions are delicate and highly regressed.

In conclusion, we propose that the anatomic state of the liver in chronic liver disease is a balance between the effects of 2 processes, those of injury and repair. While injury is often episodic, repair processes appear to be slow and continuous with similar histologic features (the hepatic repair complex) in all the diseases studied.