Assessment of Liver Histology

Clinical-Pathologic Correlations

In current practices of hepatology, hepatobiliary surgery, and liver transplantation, pathologic studies remain integral in diagnosis and clinical decision making. Although many noninvasive tests continue to be developed (including combinations of serologic markers, imaging findings, and patient demographics), histo­pathologic studies remain the standard against which all other test results are judged. Stereotypic histologic features are commonly recognized and can broadly categorize disease processes as hepatitic, cholestatic, vascular, or neoplastic, but rarely do the histologic features alone suffice for the determination of a final diagnosis. Rather, in most situations, a differential diagnosis exists for identified lesions. An example is the duct lesion characterized by infiltration into the interlobular duct by lymphocytes. This lesion can occur in hepatitis C virus (HCV) infection (known as the Christoffersen-Poulsen lesion described in 1972), the florid duct lesion of primary biliary cholangitis, which is a necessary component of acute allograft rejection, a common finding in drug-induced liver injury (DILI), and an occurrence in some cases of autoimmune hepatitis (AIH). Thus adequate clinical information is necessary for, and often integrated into, a final pathology diagnosis. It is a personal choice of the pathologist as to when in the course of liver biopsy interpretation the clinical indications for the biopsy and accompanying pertinent clinical information are known.

Tissue Preparation and Stains

Rarely is there a reason not to place a liver biopsy into formalin immediately after procurement; the exception is for the rapid, frozen section diagnosis at the time of surgery. Routine histochemical studies and an ever-increasing array of immunohistochemical, in situ hybridization, and nucleic acid retrieval–based studies have been optimized for use with formalin-fixed, paraffin-embedded (FFPE) tissues. Copper and iron quantitation studies are routinely performed from the FFPE tissue remaining in the block. The exceptions to this include the need for fresh tissue (e.g., flow cytometry processing of a hematopoietic process, pediatric liver biopsies for enzymatic assays) and glutaraldehyde fixation for ultrastructural evaluation. On the other hand, the consideration of acute fatty liver of pregnancy can be successfully addressed if the specimen has not been processed into paraffin. Formalin-fixed biopsies can be frozen and cut onto charged or albumin-coated glass slides, with the application of oil red O (ORO) stain used to identify microvesicular steatosis. The same core can then be processed through paraffin embedding for routine evaluation and workup for the differential diagnoses, as discussed hereafter.

During submission for routine evaluation of any liver biopsy, embedding sponges and ink are not only unnecessary but often result in detrimental artifacts, and thus are both to be strongly discouraged. The former leads to distortion artifacts that create nonanatomic angled holes throughout the biopsy core ; the latter results in adherent pigment that may obscure cells and structures or may even be confused for bile or other pigments.

Abundant information can be obtained from a careful, systematic review of sectioned levels of hematoxylin- and eosin-stained (H&E) tissue; preferably three levels are available, one each from the beginning, middle, and end of the sectioned ribbons. Additional special stains utilized for evaluation can serve useful diagnostic purposes ( Table 12-1 ). Hepatopathologists have their own preferences, but most employ stains for connective tissue and architectural assessment at the very least. Arguments in favor of periodic acid–Schiff after diastase (PASd) and iron stains include the many uses of PASd (e.g., ease of observation of periportal globules of α ₁ -antitrypsin [A1AT] PiZ mutations; evaluation of bile duct basement membranes; visualization of enlarged, pigmented Kupffer cells; and the dark magenta pigment in zone 1 hepatocytes that typically correlates with copper) and iron (e.g., iron-free foci in cirrhosis with iron overloading can be a marker of dysplasia).

Approach to Microscopy

Systematic review of liver biopsies assists in the identification of diagnostic patterns of injury. Pathologists often interchangeably use anatomic terminology based either on Rappaport's acinar concepts of vascular supplies or on Kiernan's hexagonal lobular architecture. In the former, the hepatocytes most proximal to blood supply (i.e., portal tracts with portal venous and hepatic artery inflow) are referred to as zone 1 and those most distal from it are zone 3 . Zones 1 and 3 are pear shaped and starfish shaped, respectively. According to the lobular architecture described by Kiernan, the hepatocytes around the portal tract are referred to as periportal and those around the outflow vein as centrilobular or perivenular . Although both of these concepts are oversimplifications of the microarchitecture of the liver, they serve as aids in interpretation. It is well recognized that hepatocytes along the gradient from inflow to outflow have different functions and susceptibilities to pathologic insults. Thus recognition of zonal location of injury, pigment deposition, and/or fibrosis is often extremely useful in creating a differential diagnosis based on morphology. Prototypic zonal injuries are noted in Table 12-2 .

The assessment of a medically indicated biopsy includes parenchymal architecture as assessed by the presence and spacing of vascular structures (terminal hepatic venules and portal tracts) and cord architecture (atrophy, hypertrophy, nodularity); the presence, location, and types of cellular infiltrates (focal, panacinar, portal predominant; predominant cell types); hepatocellular injury patterns (single cell apoptosis, spotty necroses, confluent necroses, zonal or nonzonal necroses, bridging necroses, regenerative hepatocellular changes and/or ductular reaction); and interlobular bile duct evaluation (presence, inflammation, features of cholestasis). The periportal space is often quite informative and supportive of the type of injury ( Table 12-3 ). Stereotypical patterns can provide diagnostic clues as the diagnostician considers injury patterns: hepatitic, cholestatic, and vascular diseases ( Table 12-4 ).

For directed lesional biopsies, the initial determination is the presence or absence of tissue consistent with a distinct mass lesion. Changes recognized as mass effect, such as sinusoidal dilatation with or without ductular reaction and portal chronic inflammation, suggest proximity to a lesion. If present, these findings should not be misinterpreted as necessarily evidence of underlying liver disease. Table 12-4 summarizes the morphologic features and associated diagnostic possibilities of hepatic injury, and Table 12-5 highlights unique histopathologic details for specific forms of acute and chronic liver diseases.

Acute Hepatitis

Acute hepatitis is characterized by intraacinar-predominant findings of hepatocyte injury, inflammation, and hepatocellular regeneration. The hepatocytes may be diffusely or zonally involved with swelling, cytoplasmic clearing, or lytic necrosis. Most types of acute hepatitis affect primarily zone 3, but hepatitides A and E are notable exceptions that may show zone 1 accentuation, and herpes hepatitis is characterized by nonzonal regions of necrosis. Clusters of ceroid-pigmented Kupffer cells and/or mononuclear cells may be sparse initially. With resolution, PASd-pigmented Kupffer cells may be present in zone 3. Eosinophilic, rounded, apoptotic hepatocytes may be noted singly outside the cords. Hepatocellular necrosis is detected by foci of spotty necrosis, which is recognized as scattered, intraparenchymal collections of Kupffer cells surrounding fragments of hepatocellular cytoplasm, zone 3 or zone 1 confluent necrosis, submassive (bridging) necrosis, or massive hepatic necrosis. The types and acinar extent of necrosis may be clues to underlying etiology. Necrosis of hepatocytes results in loss of plate architecture and reticulin collapse. Inflammatory cells infiltrate diffusely throughout the lobules, and sinusoidal lining cells become prominent as they are activated; an overall “busy” appearance is apparent. An interesting example is the relative lack of hepatocellular injury that accompanies the “beads on a string” sinusoidal lymphocytic infiltrates of Epstein-Barr virus (EBV) hepatitis. Inflammation in most forms of acute hepatitis is predominantly mononuclear cells. Polymorphonuclear leukocytes are uncommon in acute hepatitis. Microabscesses, however, can be indicative of cytomegalovirus (CMV) infection in the allograft liver, and polymorphonuclear leukocytes surrounding individual hepatocytes (satellitosis) or accompanying ductular reaction in zone 1 are characteristic of alcoholic hepatitis (discussed later in this chapter). Subendothelial or transmural inflammation (phlebitis) of outflow veins can be noted in cases of severe acute hepatitis. Chronic portal inflammation may be present in varying degrees. In acute hepatitis A or E, portal inflammation may be equal or predominant to the intraacinar findings and may be dominated by plasma cells. Spillover of inflammatory cells from portal tracts may be difficult to distinguish from piecemeal necrosis or interface activity. Evidence of hepatocellular regeneration is common in all forms of acute hepatitis except fulminant cases and is characterized by anisonucleosis of hepatocytes, binucleated or trinucleated hepatocytes, broad cords on reticulin stain, and, rarely, mitotic figures in hepatocytes. Ductular profiles in a seemingly empty stroma may represent the only form of regeneration in fulminant liver failure.

Fibrosis is not present in acute hepatitis. In cases with marked collapse, the two-cell-thick regenerative cords of surviving hepatocytes surrounded by dense reticulin condensation may be confused with cirrhotic remodeling. The trichrome stain may or may not distinguish the deep, vibrant color of the native collagen of portal tracts from the paler and less well-defined dye retention of the collapsed parenchyma. Reticulin stain highlights the collapse, as well as highlighting regenerative cords. Elastic fiber stains such as orcein can be quite helpful because elastic fibers are largely absent in recent passive collapse but are present in the true active septa of fibrosis.

Characteristic findings of particular entities that may result in lesions of acute hepatitis are presented in Table 12-5 .

Massive or submassive hepatic necrosis, characterized by panacinar or bridging parenchymal necrosis, respectively, may be histologic features in clinical cases of hepatitis. CD68-positive, PASd-pigmented Kupffer cells are easily noted in the residual stroma. Ductular reaction, characterized by increased numbers of ductular elements and variable amounts of stroma and inflammation, may be noted in acute hepatitis when extensive necrosis is present. In massive hepatic necrosis, the majority of viable epithelial cells remaining are the K7-positive and K19-positive interlobular bile ducts, as well as transit-amplifying cells derived from the hepatic progenitor cells that characterize the ductular reaction. Etiologic determination of the injury based on pathologic evaluation may not be possible in massive hepatic necrosis, but some patterns can be suggestive. Zone 3 coagulative necrosis with little inflammation characterizes both systemic shock and acetaminophen toxicity. Clusters of necrotic hepatocytes or massive necrosis may be seen with herpes simplex virus (HSV) or adenovirus hepatitis. Careful evaluation by light microscopy for the typical intranuclear inclusions of these viruses is suggested, particularly in cases of immunosuppression. The glassy nuclei, syncytial nuclear formation, and basophilic stippling of herpes simplex can be noted by routine stains. Immunohistochemical stains are available to confirm both herpes virus and the smudged nuclei of adenovirus.

Chronic Hepatitis

Chronic hepatitis is histologically characterized by the portal predominance of mononuclear or mixed chronic inflammatory infiltrates. In addition, PASd-pigmented portal macrophages are either more or equally prominent with intraacinar Kupffer cells. Lobular necroinflammatory activity ranges from that described for the previous lesions to near absence of activity. In most diseases that result in chronic hepatitis, fibrosis occurs initially in and around the periportal regions. The most notable exception to these concepts is fatty liver disease, which is discussed in the following section. Interface activity, a defining lesion of chronicity, is not specific to any disease entity and is characterized by chronic inflammatory cells breaching the limiting plate of the portal tract and surrounding adjacent hepatocytes in zone 1. The older term for this was piecemeal necrosis. Currently, actual necrosis (or apoptosis) is not required in the recognition of interface activity. The types and amounts of parenchymal necrosis and collapse depend on the specific disease and the individual host's immune response. If the underlying disease is unchecked, fibrosis in chronic hepatitis may progress and may involve portal-portal bridging septa, portal-central approximation from loss of intervening parenchyma, and ultimately cirrhosis, defined not only by loss of hepatic parenchyma and replacement by septal scar tissue but also by vascular remodeling that results in (or from) intraparenchymal shunts. The extinct parenchyma, or septa, commonly contains not only elastic fibers and collagens but also ductular profiles, thin-walled vascular and lymphatic channels, and varying degrees of inflammatory infiltrates. Neither the inflammation nor the fibrosis is uniform throughout the organ in many cases. This lack of homogeneity can result in sampling error in liver biopsies, particularly if sufficiently large samples are not obtained.

Fig. 12-1 is an example of AIH. AIH may manifest a variety of histopathologic findings, but rarely as an acute hepatitis. The most common lesions of AIH are marked overrepresentation of plasma cells in the otherwise mixed chronic portal infiltrates. Sheets of plasma cells are common at the portal-parenchymal interface. Plasma cells are also easily noted in foci of lobular necroses and/or infiltrates; swollen groups of hepatocytes form hepatitic rosettes; zone 3 confluent necrosis or bridging necrosis are commonly described. Eosinophils are a minor but common component of the portal inflammation. When cirrhosis occurs, there may be broad areas of extinction (septa). The untreated process is notable for more severe necroinflammatory activity and interface hepatitis than with hepatitis C. On the other hand, inactive burned-out cirrhosis may be due to AIH. Overlap of AIH and primary biliary cholangitis may have either clinical and/or histologic features of both disease processes. Treatment decisions will include consideration of the liver biopsy findings, as well as biochemical and serologic tests. It is important, however, to reiterate that no single finding distinguishes AIH and other forms of chronic hepatitis, thus careful clinical correlation (determination of total proteins and elevated globulins may be as informative as measurement of autoantibodies) is needed.

Systems for semiquantitation of lesions of chronic hepatitis, initially developed for comparative analysis of treatment effects, have evolved over recent decades into methods of evaluation in clinical practice. Although each has subtle differences, they share a thorough evaluation of lesions with the separation of necroinflammatory lesions (portal and lobular; grade) from evaluation of fibrosis amount and architectural alterations (stage). Fig. 12-2 is an example of the progression of fibrosis in chronic hepatitis due to HCV. Guidelines for application of scoring systems include knowledge of the clinical situation, but blind application of grade and stage without clinical evidence of the chronic hepatitis being evaluated is not recommended.

Fatty Liver Diseases

The two most common diseases that are characterized by intrahepatocellular steatosis in some phase of their evolution are alcoholic fatty liver disease (AFLD) and nonalcoholic fatty liver disease (NAFLD). Although alcoholic hepatitis, an entity commonly without steatosis, may exist within the spectrum of alcoholic liver disease (ALD), to date no equivalent lesion (or clinical equivalent) has been described in nonalcoholic fatty liver. Likewise, the entity of alcoholic foamy degeneration has no equivalent in NAFLD. Otherwise, there may be substantial overlap of histologic lesions in these two processes, including the findings of steatosis on one end and cirrhosis on the other. Further, patients may have clinical features of both metabolic syndrome and deleterious exposure to alcohol and thus liver biopsy evaluation may show features that could be attributed to either. Both are characterized in adults by mixed large and small droplet steatosis, zone 3 hepatocellular ballooning, presence of Mallory-Denk bodies , and, initially, zone 3 perisinusoidal fibrosis ( Fig. 12-3 ). Both may progress to fibrosis in and around portal tracts, including varying degrees of ductular reaction, bridging septa between portal structures or outflow veins or portal-central bridging ( Fig. 12-4 ), and both may result in cirrhosis with or without lesions of the active disease. Thus distinguishing the two may be impossible without adequate clinical information. On the other hand, canalicular cholestasis, large and abundant Mallory-Denk bodies, pericholangitis, lesions of alcoholic hepatitis (nonsteatotic), the various occlusive lesions of the outflow veins (including sclerosing hyaline necrosis), and large regions of parenchymal extinction strongly favor alcohol abuse.

Similar to the separation of necroinflammatory lesions (grade) from those of fibrosis and remodeling (stage) in chronic hepatitis, scoring systems for semiquantitative evaluation are commonly utilized for assessment and treatment trials of NAFLD. Unlike the recent system proposed by a consortium in Europe, the National Institute of Digestive Diseases and Kidney; Nonalcoholic Steatohepatitis Cooperative Research Network scoring system has shown that simple addition of lesional values for score and actual diagnostic classification of pattern of injury correlate with different processes and deserve to be assessed separately.

A multinational validated study resulted in a method of predicting 90-day prognosis for alcoholic hepatitis. The Alcoholic Hepatitis Histologic Score, an aggregate score of lesions for mild, moderate, and severe, was based on extensive fibrosis, canalicular and ductular cholestasis (i.e., bile plugs in the canal of Hering), severe polymorphonuclear infiltration, and lack of megamitochondria. Steatosis was not a factor included in the model.

Hepatic steatosis and/or steatohepatitis may occur concurrently with other forms of chronic liver disease. Steatosis is noted in up to 70% of HCV biopsies and this is particularly common with genotype 3. Trichrome stain is considered essential by some to aid in diagnosis or exclusion of concurrent steatohepatitis, as zone 3 perisinusoidal fibrosis, whether attributable to alcohol or obesity, is not associated with other forms of chronic hepatitis. Thus zone 3 perisinusoidal fibrosis is necessary for diagnosing both causes of fatty liver. Other investigators have used hepatocellular ballooning as the distinguishing feature of concurrent steatohepatitis.

Nonalcoholic fatty liver disease in the pediatric and adolescent age groups is known to have unique histologic manifestations from those in adults. First shown in a landmark study, the distinctions are most apparent by gender and age. Boys and younger children are more likely to have portal accentuation of the lesions, and less likely to have ballooning, Mallory-Denk bodies, and zone 3 perisinusoidal fibrosis. Transitions through adolescence result in altered histologic features of fatty liver disease, as noted in a cross-sectional study.