Ductular reactions in human liver: Diversity at the interface

Authors

  • Annette S. H. Gouw,

    1. Department of Pathology and Medical Biology, Pathology Section, University Medical Center Groningen, University of Groningen, Groningen, the Netherlands
    Search for more papers by this author
  • Andrew D. Clouston,

    1. Centre for Liver Disease Research, School of Medicine, University of Queensland, Brisbane, Australia
    Search for more papers by this author
  • Neil D. Theise

    Corresponding author
    1. Departments of Pathology and of Medicine, Division of Digestive Diseases, Beth Israel Medical Center of Albert Einstein College of Medicine, New York, NY
    • Division of Digestive Diseases, Baird Hall 17-61, Beth Israel Medical Center, First Avenue at 16th Street, New York, NY 10003
    Search for more papers by this author
    • fax: 212-420-4373


  • Potential conflict of interest: Nothing to report.

  • This work was supported in part by grants from the National Health and Medical Research Council of Australia and Royal Brisbane Hospital Foundation (to A.D.C.).

Abstract

Interest in hepatic ductular reactions (DRs) has risen in recent years because of a greater appreciation of their potential roles in regeneration, fibrogenesis, and carcinogenesis. However, confusion exists because there is significant, but often unappreciated diversity at the tissue, cellular, and subcellular levels in DRs of different diseases and stages of disease. DRs are encountered in virtually all liver disorders in which there is organ-wide liver damage and cell loss, but are also present in focal lesions such as focal nodular hyperplasia and adenoma. Moreover, diverse DR phenotypes can be present within any single disease entity, and are shaped by the etiology and evolution of the disease. Although much remains to be clarified, recent studies suggest that the diversity of appearances of the DRs are likely to reflect the differing signals at the anatomic, cellular, and molecular levels driving the proliferative response. These appear to determine the relative proportions of transit-amplifying cells, the degree of hepatocytic or cholangiocytic differentiation, and their relationships with stromal, vascular, and inflammatory components. The molecular signaling pathways governing these regenerative fate decisions closely replicate those found in human and other vertebrate embryos and more generally in stem cell niches throughout the body. Like the latter, complex interactions with matrix as well as mesenchymal and inflammatory cells, vessels, and innervation are likely to be of fundamental importance. Embracing systems/tissue biological approaches to exploring DRs, in addition to more traditional cellular and molecular biological techniques, will further enhance our understanding and, thereby, we believe potentiate new therapeutic possibilities. (HEPATOLOGY 2011)

Abbreviations

AIH, autoimmune hepatitis; CoH, canals of Hering; DR, ductular reaction; ECM, extracellular matrix; EHBA, extrahepatic biliary atresia; EMT, epithelial-to-mesenchymal transition; EpCAM, epithelial cell adhesion molecule; HCV, hepatitis C virus; IFN, interferon; K, keratin; PBC, primary biliary cirrhosis; PSC, primary sclerosing cholangitis; TGF, transforming growth factor; TNF, tumor necrosis factor.

 

Explanations should be as simple as possible, but no simpler.

—Albert Einstein

In recent years, diagnostic pathologists and their clinical colleagues, through detailed clinicopathologic correlations, have refined our understandings of ductular reactions (DRs), which are the reactive processes that arise, in disease and injury, at the interface of the portal (or septal) and parenchymal compartments, in human livers.1 Despite these advances, DRs remain obscure to many clinicians and scientists, even though liver pathologists, on a daily basis, see for themselves the variety of DR patterns in different clinical settings, with their marked or subtle differences often of essential diagnostic significance (Fig. 1). However, for everyone else, the words “ductular reaction” remain an abstraction. Although evocative in a general way, “ductular reaction” fails to convey the heterogeneity underlying the development, nature, and outcome that is necessary to give clinical or scientific relevance.

Figure 1.

Histologic diversity of ductular reactions (DRs). Although there are commonalities to all DRs, each disease that incites a DR has its own characteristic constellation of features. These diverse appearances are, in fact, central to diagnostic histopathologic assessment of diagnostic liver biopsy specimens. (Top row, left) Normal human portal tract with bile duct and ductule/canal of Hering (CoH) profile (open arrow) extending beyond limiting plate to interface with periportal hepatocytes. (Top row, right) Similar normal portal tract immunostained for keratin 19, highlighting bile duct and ductule/CoH crossing the limiting plate. (Second row, left) In sepsis, the DR has little stroma or inflammation, but dilated ductule/CoH structures are prominent and often contain inspissated bile concretions. (Second row, right) Acute large bile duct obstruction with DRs, with abundant and edematous stroma and prominent neutrophilia. Hepatocytes adjacent to these DRs may show cholate stasis. (Third row, left) In nonalcoholic fatty liver disease (NAFLD), the DRs occur in later stage disease and are associated with dense scar and relatively scant mononuclear infiltrates. Intermediate hepatobiliary cells interface directly with often steatotic hepatocytes (open arrows). (Third row, right) As in NAFLD, HCV-associated DRs predominantly found in late-stage disease comprise intermediate hepatobiliary cells interfacing directly with hepatocytes and surrounded by densely compacted scar. Necroinflammation typical for chronic viral hepatitis is present (black arrow indicates acidophil body; white arrows show mononuclear cells in portal tracts and in DR). (Bottom row, left) Massive hepatic necrosis (MHN) from acetaminophen toxicity shows intermediate hepatobiliary cells with more prominent hepatocyte-like features (larger, lower nuclear:cytoplasmic ratio, hepatocyte-like nuclei) and surrounded by stroma that is less compacted than that seen in later stage NAFLD or chronic hepatitis. (Bottom row, right) Focal nodular hyperplasia (FNH) has intermediate cells in the DR that are more cholangiocyte-like and variably contain well-defined ductular lumens. The stroma is variable in FNH, being sometimes densely compacted, sometimes looser. (All images: original magnification, 20×. Top right panel is immunostained with DAB brown stain and hematoxylin counterstain; All other images are hematoxylin and eosin stained.)

That such nuances are important is clear from reviewing any contemporary literature regarding some of the key questions about hepatic physiology. DRs are now recognized to occur ubiquitously in many acute and chronic liver diseases, not just in biliary disorders, and are increasingly central to our understanding of hepatic stem and progenitor cells in liver regeneration, mechanisms underlying hepatic fibrosis, and hepatobiliary carcinogenesis.

This review will focus specifically on changes and concepts derived from studies of humans, not animal models, for concision and because much about human DRs is quite unlike their animal correlates. Although such models remain exceptionally useful, particularly for studies of hepatic regeneration, as far as fibrosis and neoplasia are concerned, the rodent models display very different processes from those seen in human livers. Where we include data from animal models, it is because they are clearly relevant to humans or they provide insights for which no human data are available.

Our key emphasis will be on the diversity of DRs, the word “diverse” applying in several ways. DRs show strikingly diverse patterns that are often diagnostically specific, varying markedly, for example, between predominantly biliary and hepatocellular injuries and acute or chronic processes (Fig. 1). DRs also contain a profound diversity of cellular and tissue elements, not just the hepatobiliary epithelial cells that are most prominent on quick glance (the “ductular” component of the name), but all the other elements of the tissue “reaction” (Fig. 2). The epithelial cells themselves show a range of differentiation states, particularly when studied by immunohistochemical expression (Fig. 3). There is also diversity of cell origin, with, in the most studied example, “intermediate hepatobiliary cells” of DRs shown to derive, variously, from activation of canals of Hering (CoH) and ductules (Fig. 4), circulating, marrow-derived precursors, biliary metaplasia of hepatocytes, and perhaps from mesenchymal-to-epithelial transition. Diverse molecular signaling pathways are also known to mediate human DRs and are the aspects of DRs perhaps best revealed by animal model analysis.5

Figure 2.

Immunohistochemical demonstration of cellular diversity within DRs, differing by disease. Three examples are given (from left to right): chronic biliary obstruction, massive hepatic necrosis (MHN) from acetaminophen toxicity, and nonalcoholic fatty liver disease (NAFLD). All DRs comprise a diversity of cell types including: hepatobiliary epithelial cells that may be specifically highlighted with stain for keratin 19 (K19) or in combination with hepatocytes when stained for K8/18; inflammatory cells, here highlighting the T cell components of the infiltrates with stain for CD3; endothelial cells highlighted with staining for CD34; stellate cells highlighted by stain for alpha-smooth muscle actin (αSMA) (which therefore also stains some of the matrix proteins already deposited by the stellate cells). Note that all components are present in all DRs, but their relative proportions and the cell:cell and cell:matrix relationships may differ significantly from one disease to another.

Figure 3.

Immunophenotypic diversity within DR hepatobiliary cells. Four-color immunofluorescence of HCV-associated DRs with K19 appearing as pseudo-colored blue, K7 appearing as green, CD56 (NCAM) appearing as red, and K18 (primarily staining hepatocytes) appearing as purple. (Original magnification, 20×; this figure courtesy of E. Prakoso, N. Shackel, and G. McCaughan, University of Sydney, Sydney, Australia.)

Figure 4.

Relationship of DR hepatobiliary cells to the the ductule/canal of Hering (CoH) stem cell niche. DR hepatobiliary cells are largely the “transit amplifying progeny” of intrabiliary stem cells. Their primary source is therefore the stem cell niche located in the most proximal branch of the biliary tree, the ductule/CoH unit. This relationship is best seen in three-dimensional representations of DRs. (A) Schematic diagram of normal ductule and CoH (red arrow) structures and their relationship to bile duct (BD), portal tract stroma, limiting plate, hepatocyte canalicular system (narrow, branching lines), and terminal hepatic venule (THV). (Reprinted with permission from Saxena et al.2). (B) In primary biliary cirrhosis, computer-generated three-dimensional reconstruction shows a granulomatous duct destructive lesion (large white area with verticle dashed black lines) and DR (green brackets) arising from preexisting ductule/CoH structures. (Reprinted with permission from Yamada et al.3). (C) Serial, 4 μm sections of HCV-related cirrhosis. A small intraseptal hepatocyte nodule links to an interlobular bile duct via a single intermediate, K19-positive, CoH-like structure. The complete link can only be appreciated with examination of the serial sections (red arrow). (Immunostained with DAB, hematoxylin counterstain, original magnification, 20×; reprinted with permission from Falkowski et al.4). (D) Three sample tracings of K19-positive ductular reactions in sequential levels around a single portal tract. Colors are assigned to indicate contiguity of structures when analyzed in three dimensions. On each level, one bile duct is marked by an arrowhead. Note in levels 11 and 12 where the red arborizing structure connects via a single ductule/CoH branch to this bile duct (indicated by arrows). (Reprinted with permission from Theise et al.5).

Figure 5.

Elucidating the pathobiology of ductular reactions requires an interdisciplinary approach. Scientific techniques and methodologies are scale-dependent (circles). Each point of view provides a different perspective that sheds light on the others; however, no single point of view (tissue, cell, or molecular level) yields a complete picture. Transgenic models can be studied with any of these scale-related approaches and are therefore listed in all three, encircled classes of methodology. The relationships between data gathered at these different levels of scale can be used to build more complete models by creatively using different systems biology approaches (middle box).

We thus present a view of DRs from our own dual perspectives as research scientists and as diagnosticians who analyze DRs in daily clinical practice. We hope these combined perspectives will be of value for those investigators and clinicians who do not have the privilege of such intimate, daily contact with this increasingly fascinating realm, this “diversity at the interface.”

Histologic Diversity of Ductular Reactions

Normal Features.

Normal livers usually do not contain DRs, but the biliary tree and hepatic parenchyma interdigitate across the limiting plate where the hepatocyte canalicular system links to the CoH; these, in turn, link to bile ductules. Thus, the CoH/ductular unit, with somewhat variable anatomy (Fig. 4A) begins in the periportal parenchymal region as the first twig of the biliary tree, crosses the limiting plate, becoming a somewhat larger interlobular bile duct. The numbers, lengths, and shapes of the CoH/ductule units depend on the source and hence normalcy of the specimens and on the application of immunohistological markers, e.g., keratin 7 (K7), K19, or epithelial cell adhesion molecule (EpCAM).1, 5,6,7 Without immunostaining, there is an average of 0.4 ductules per portal tract (range 0-4) in normal human liver,8 whereas median values of 2.5-5 ductules per portal tract were observed after application of the K7 immunostain.9,10

Phenotypes of DRs in Liver Diseases.

Historically, human DRs have been grouped on the basis of morphology into “typical” and “atypical”, terms originally applied in and based on rodent studies.11 This terminology is discouraged, because a classification of DRs based on a quite limited set of experimental conditions in rodents cannot readily accommodate the range of patterns seen clinically; DRs are diverse, covering a spectrum of features rather than clear subphenotypes, which will now be described (Fig. 1).

DR morphologies may range from well-formed ductules with recognizable lumina to irregular counterparts without obvious lumina, sometimes merely consisting of a string of cells. Variable phenotypes between both ends of this spectrum can be present concomitantly in a single specimen, depending on the etiology and evolution of the disease. Even greater complexity emerges when there are concomitant disease processes such as primary sclerosing cholangitis (PSC) with both obstructive and regenerative DRs.12 Classification schemes suggested by Desmet12 and Turanyi et al.13 have attempted to integrate the histologic features, inciting disease, and/or immunophenotyping of DR, but these have not been subjected to a consensus-building process of review and are not (yet) recommended for DR subclassification.

Biliary obstruction produces the most well-known example of DR, featuring a multiplication of small ductules at the periphery of edematous portal stroma. The ductules show variable nuclear size and contain no bile. Obstructive-type DRs may be extraordinarily prominent, seemingly replacing parenchyma with markedly expanded portal tracts, but these can rapidly resolve. When bile concretions are present in dilated ductular lumina (“cholangitis lenta” or ductular cholestasis) a superimposed septicemia should be considered, either from cholangitis or a distant source.

If obstruction cannot be relieved and becomes chronic, edema wanes and the extracellular matrix (ECM) of the DR becomes denser, with more mature forms of collagen that resist remodeling; eventually, obstructive biliary cirrhosis may develop. In other chronic biliary diseases such as extrahepatic biliary atresia (EHBA), PSC, and primary biliary cirrhosis (PBC), the morphology and severity of DRs depend on disease stage.12 All three will have foci of DRs with dense fibrous stroma similar to those in chronic obstruction. This is the only pattern seen in PBC, because only smaller bile ducts are involved. In EHBA and PSC, with involvement of larger ducts, there is a mix of this chronic form as well as superimposed obstructive-type DR. As disease progresses, the DR can become more variable, and may be sparse in the end stages of disease.

In liver diseases of nonbiliary origin, even more variable DR phenotypes are seen. The most profound DRs can be encountered in fulminant hepatic failure.14 The severe loss of hepatic parenchyma is accompanied by massive DRs with sparse fibrosis or inflammation.5 The hepatobiliary cells in these reactions are more “hepatocyte-like” than those that predominate in biliary tract disease. In fibrosing cholestatic variants of hepatitis B and C in the setting of a compromised immune system, DRs are still more dramatically expanded, floridly extending into the hepatic parenchyma in a “starburst” pattern and accompanied by more prominent stroma.15

In chronic viral hepatitis, DRs predominantly appear later in the disease process, years or even decades after infection, although they may be subtly present earlier. These DRs are more tightly compacted at the stromal–parenchymal interface.4,16 In autoimmune hepatitis (AIH), DRs may be variable: similar to fulminant hepatitis during severe flares or more like viral hepatitis when fibrosis is advanced or activity less marked. Hepatocytic rosettes are often a prominent feature of AIH-DRs containing a range of hepatocyte to cholangiocyte-like phenotypes, best highlighted by immunostains. The prominence of the DR varies with etiology. A greater magnitude of DRs is present in AIH compared with hepatitis C virus (HCV) infection, whereas the latter can show more DRs than prefibrotic alpha-1-antitrypsin deficiency.17

In fatty livers without steatohepatitis, DRs are inconspicuous, although increased periportal, K7-positive cells occur.18 The DRs become more prominent with steatohepatitis, particularly in those with portal or septal fibrosis.18 A slightly different DR occurs in ischemic diseases such as hepatic vein outflow obstruction, where the reaction is typically centrilobular.12 Focal liver lesions such as focal nodular hyperplasia and the inflammatory type hepatocellular adenoma also contain DRs.19

In summary, DR are encountered in virtually all liver diseases in which there is organ wide liver damage and cell loss, but are also present in focal lesions such as focal nodular hyperplasia and hepatocellular adenoma. Moreover, diverse DR phenotypes can be present within one disease entity, shaped by the etiology and the evolution of the disease.

Diversity: Cell Biology

Although the expansion of epithelial cells is the most obvious feature of the DR histologically, when the current nomenclature was being formulated in 2004, the term “reaction” was also included to recognize that the epithelial component is accompanied by a complex of ECM, inflammatory cells, vessels, mesenchymal cells, and other diverse components that are integral to its formation and sustenance.1 (Fig. 2)

The epithelial cells within DRs are not uniform in appearance and are called the intermediate hepatobiliary cells. They range from 6 μm (like the smallest cholangiocytes of the CoH) up to 40 μm (the normal hepatocyte diameter).1 These cells are a so-called transit-amplifying population of bipotent progeny of facultative hepatobiliary stem cells and range from nearly perfect cholangiocyte or hepatocyte morphologies to all imaginable intermediate morphologies, though different disease settings determine the range and distibution of forms. In general, hepatocyte-like cells are most prominent at the parenchymal border and cholangiocyte-like cells predominate at the portal/stromal border.1,20

Immunophenotypes of these cells have been evaluated to define their stem and/or progenitor cell function, as reviewed recently (Fig. 3).21 Immunophenotypes also reflect, however, the diversity of etiologies and relate to hepatic functions lost.12,13,22-23 For example, in obstructive biliary disease, positive staining with epithelial membrane antigen (EMA) suggests proliferation of mature cholangiocytes whereas hepatocellular differentiation is more prominent in fulminant hepatic failure and cirrhosis, and is associated with the simultaneous expression of neural cell adhesion molecule (NCAM, or CD56), EMA and CD10.13

The DR also contains other diverse, but essential components necessary for sustaining and modulating niche activity including mesenchymal, vascular, neural, and hematopoietic cells.24-26 The complex may even closely replicate the ductal plate structures in fetal liver, the source of parenchymal growth before birth.12 Niche components that appear critical for stem/progenitor cells include ECM, particularly laminin,27 and cellular components including endothelium26,28 and myofibroblasts/hepatic stellate cells.26 Infiltrating macrophages and other inflammatory cells are common and appear to have a role in the progenitor cell expansion.27

Sources of Hepatobiliary Cell Components of Ductular Reaction.

Just as structure, functions, and immunophenotypes depend on each inciting disease it is probable that differences relate, in part, to differential stimulation of different niches of hepatic cellular repair.29 Several earlier studies using three dimensional reconstruction demonstrate that DRs are complex, arborizing networks of hepatobiliary cells branching from preexisting CoH (Fig. 4B-D).3-5 Proliferation studies in various liver diseases support these findings.4, 5,16-18 In cirrhosis from diverse causes, such links to CoH are present, and hepatocytes derive directly from the DRs,4,6 but “hepatocyte buds” also arise from some interlobular bile ducts themselves.4 In normal liver, clonal patches of hepatocytes apparently derived from the CoH have also been demonstrated.30

These kinds of studies, however, are not definitive. A functional assay for identification of the stem cell niche in living tissues is required. Such an approach, the label-retaining cell assay, depends conceptually on the following framework. Stem cells are defined as largely quiescent, rarely dividing multipotential cells.31 When they do divide, and in the liver this is usually in response to injury, they do so in an asymmetrical fashion, giving rise to a replacement stem cell on the one hand and a rapidly proliferative progenitor cell on the other. These rapidly proliferative progenitor cells, which form the majority of DR hepatobiliary cells, are analogous to the transit amplifying proliferative zone in the intestinal crypt, being a little larger and closer to final differentiation, but still bipotent. Even in a greatly expanded DR, true stem cells remain rare.

The label-retaining cell assay exploits these definitional rare and asymmetrical divisions of stem cells in their niches. Kuwahara et al. found that bromodeoxyuridine-label–retaining cells, marking true stem cells that divided asymmetrically and then became quiescent again, were observed in four different intrahepatic locations31: in CoH, within interlobular bile ducts, adjacent to ducts (“null cell” monocytes, negative for keratin or other differentiation markers), and peribiliary hepatocytes, where CoH link to hepatocytes. The last of these was considered to possibly represent a differentiated CoH cell rather than a true, resting stem cell. Others have identified and isolated multipotential cells from normal human liver that are 7-9 μm and express albumin (weak), biliary-type keratins such as K7 and K19, but not alpha-fetoprotein.32

Thus, the DR intermediate hepatobiliary cells are the transit amplifying progeny of hepatobiliary stem cells. Their immunophenotypes therefore combine antigens present on stem cells, hepatocytes, and cholangiocytes in varying combinations.1,7,33

The phenotypic diversity of DR during liver diseases has led to a concept that parallels development and regeneration. Zhang et al.7 demonstrated membranous EpCAM-positive cells with an intermediate hepatobiliary phenotype, adjacent or tethered to the CoH in adult livers and increasing in diseased livers. The immunophenotype and proliferation rates of these cells resemble fetal hepatoblasts, possibly suggesting common processes in regeneration and development. In fetal ductal plates, the fetal hepatoblasts represent the transit amplifying cell progeny of stem cells, and after development the intermediate hepatobiliary cells of postnatal DR are, likewise, the transit amplifying progeny of the CoH/ductules.7

Finally, three other origins for DR hepatobiliary cells have been hypothesized: (1) biliary metaplasia of hepatocytes may produce DRs,12 (2) marrow-derived circulating stem/progenitor cells contribute in some measure in rodent models34 and possibly in severe disease in humans,35 and (3) most controversially, it is hypothesized that there may be mesenchymal-to-epithelial transition, with myofibroblastic and/or angioblastic cells undergoing epithelial differentiation, although strongly supportive data are lacking.36

Diversity: Molecular Biology

Because DR cellular diversity is profound, unraveling DR signaling mechanisms is complex and remains incompletely understood. Factors such as interferon-γ (IFN-γ) and transforming growth factor-β (TGF-β) may have variable effects depending on cell type, location and stage of differentiation. By considering the DR as a combination of stem cell, transit-amplifying and differentiated cell compartments, the signaling mechanisms can be more easily understood as having several phases typical of stem cell niches of other organs: activation, proliferation, migration, and differentiation. Currently, in the liver, it is difficult to separate the factors and signaling pathways involved in DR activation and proliferation. These include, but are not restricted to cytokines signaling through the gp130 receptor (interleukin-6, oncostatin-M, and leukemia inhibitory factor),37 tumor necrosis factor (TNF) superfamily members including TNF-α and TWEAK,38,39 IFN-γ, hedgehog ligands,40 and growth factors such as epidermal growth factor, fibroblast growth factor-1 and hepatocyte growth factor.21,41 The presence of significant telomerase activity in DRs should also be noted.42

Many of these factors also stimulate replication of mature hepatocytes, but the preferential emergence of a DR in many liver diseases can be explained by the relative susceptibility of mitochondria-rich hepatocytes to oxidative stress leading to inhibited replication from up-regulation of the cell cycle inhibitor p21.43 Oxidative stress affects the cells in the DR far less, so that cell cycling is not inhibited. Aging of the liver also impairs normal hepatocyte regenerative capacity, explained in part through increased expression of cyclin-dependent kinase inhibitors such as p15INK4b,44 and also from age-related vascular changes in the sinusoids.45 This may provide a proliferative advantage for the progenitor cells in the DR as was shown recently for transplanted fetal progenitor cells44 and would also explain the increased DRs found in older patients with chronic hepatitis C.46 Of the factors listed above, TWEAK is the only known factor specific for progenitor cells, due to restricted expression of its receptor Fn14 on hepatic progenitors, but not mature hepatocytes.38 Additional stimulatory factors include neuroendocrine stimulation47 and increased bile salts.48 The intracellular signaling pathways activated in the proliferative phase include Wnt,22,49 hedgehog,40 nuclear factor-κB,50 TGF-β/bone morphogenic protein, and JAK/STAT pathways.

With respect to differentiation and the ultimate fate of bipotential progenitor cells, and again replicating developmental biology of the liver, biliary differentiation is associated with Notch signaling.22,51 Hepatocytic differentiation proceeds following exposure to fibronectin27 in the absence of Notch signaling.22

Diversity: Tissue Biology

Allen Cowley Jr, Chairman of Physiology at the University of Wisconsin and editor of Physiological Genomics, has written: “At its core, systems biology represents a renewed recognition that biology is best understood by taking a coordinated, integrative systems view. In practice, systems biology means the application of many currently defined disciplines with the goal of bringing together information from the smallest units of the biological system (genes) to help understand the function of the whole organism.”52 Certainly, studies of individual cell types, specific cell: cell interactions, molecular intracellular signaling pathways, and genomic analyses have elucidated many DR mechanisms and behaviors. However, the point of cell and molecular biology reductivism was to make things simple enough to study when our tools were insufficiently complex for modeling cell:cell and cell:matrix interactions at the tissue level. We should not lose sight of the actual goal, however, of reintegrating these data into what might be considered a sub-branch of systems biology as considered by Dr Cowley, namely, tissue biology, the level of scientific analysis most appropriate to considerations by academic pathologists who concern themselves, first and foremost, with considerations of the tissue level of scale.25,53,54

DRs, similar to all stem cell systems in all organs, fully reveal their physiologic behaviors only at the tissue level.47,48,55 Thus, a tissue biology approach is needed for full understanding. Complexity theory suggests that cells and matrix interact as members of complex adaptive systems.54,56,57 Such systems can change their community-level self-organization so that the community as a whole can adapt to a changing environment. This approach opens up the possibility of computerized, agent-based modeling of cell:cell and cell:matrix interactions to test hypotheses generated from molecular and cell biological data.54,58 This class of computational modeling simulates the actions and interactions of autonomous agents (such as cells and matrix elements) with a view to assessing their effects on the system as a whole. This methodology has been fruitfully and extensively applied in studying the participation of hematopoietic stem cells in homeostasis or in pathophysiology, in particular neoplasia (reviewed by Roeder59). A recent application to liver disease has been presented by Hoehme et al. in which they modeled data derived from immunohistochemistry and confocal microscopy of murine hepatic regeneration following CCl4 injury.58 The modeling of vascular and hepatocyte structures and interactions “unambiguously predicted a so-far unrecognized mechanism as essential for liver regeneration, whereby daughter hepatocytes align along the orientation of the closest sinusoid, a process which we named ‘hepatocyte sinusoid alignment.’”51 Commentary on this study has highlighted the novelty and predictive power of this model and the richness of the approach to guide future experiments and, perhaps, therapeutic applications from this single modeling effort.60 (To appreciate the aesthetic beauty of such modeling alone, the reader is directed to the Hoehme laboratory Web site at the University of Leipzig: http://www.bioinf.uni-leipzig.de/∼hoehme/)

Although full discussion is beyond the scope of this article,54,56,57 it is worth emphasizing that fundamental concepts apply to all complex systems independent of scale. Thus, microscopic self-organization of cells and matrix into tissues is similar to self-organization of biota and inorganic substrates biota and inorganic substrates into ecosystems at the macroscale.55 (Fig. 5) This point of view realigns thinking about DRs and opens up a host of possibly interesting perspectives and methodologies for studying liver pathophysiology, some of which we suggest here. For example, in the language of landscape ecology, subdomains of tissue compartments such as the normal structures of portal tracts and parenchyma and the sharp boundary between them at the limiting plate/interface, can be conceptualized as ecosystem mosaics.61 Landscape ecology indicates that these are often sites of increased biological diversity called the “edge effect”. Such edge effects, so-called ecotones, are an engine for greater adaptation to environmental pressures. Examples include the meeting of bodies of water with land or where forest meets prairie. These can be caused by or can parallel ecoclines, where physiochemical gradients occur, such as ecosystem thermoclines (gradients of temperature), chemoclines (chemical gradients), haloclines (salinity gradients), and so forth.

Thus, DRs are microscopic ecotones with indistinct boundaries, arising where ecoclines develop in response to liver injury. Chemoclines might develop as hepatitis produces viral-response cytokine and chemokine gradients spanning the mesenchymal/parenchymal interface; nutrient gradients change with alterations of portal venous or hepatic arterial flow into the liver and, microscopically, into the portal/sinusoidal interface. Haloclines, altered salt gradients, could occur where bile salt passage across the interface (through CoH) is altered by diminished production or obstructive accumulation. Might we also consider altered “ferroclines” in hemochromatosis or “cuproclines” in Wilson's disease?

Thus, the “species diversity” of DR cellular components is a function of the microarchitectural landscape ecology of hepatic mosaic domains and ecotones/ecoclines. It can lead to increasing opportunities for adaptive reorganization when successful. It is also noteworthy that the mesenchymal/parenchymal interface and DRs of end-stage livers (“burned-out cirrhosis”) is often marked by decidedly less cellular diversity within DR ecotones than in earlier disease stages.

Another approach for conceptualizing DR that we may similarly consider, in parallel with landscape ecology, is evolutionary biology.62 Data indicating that fields of EpCAM-positive hepatocytes are progeny of EpCAM-positive DRs suggest that mutational events may arise in the DR ecotones, paralleling species evolution geographically. Some may be evolutionarily adaptive at the cell population level. For example, mutations in rapidly proliferative and therefore mutationally susceptible hepatobiliary progenitors might lead to emergence of hepatocytes resistant to disease (e.g., resistant to copper accumulation in Wilson's disease or to lipotoxicity in fatty liver disease). Of course, it also opens the door to deeper, more complex understandings of hepatocarcinogenesis associated with emergence of DRs.

Outcomes of Ductular Reactions

Parenchymal Regeneration.

The prevailing concept of liver regeneration is that replenishment of cellular loss is by proliferation of mature cells and that activation of the stem/progenitor cell compartment(s) occur(s) when the proliferative capacity of mature cells is exhausted or inhibited.

The absence of DRs in normal livers supports this. However, a basal activity of the stem cell niche to generate hepatocytes as a regular process of cellular renewal is not excluded, particularly given the discovery of clonal patches of CoH-derived hepatocytes.30 This maintenance activity does not occur through overt DR, but possibly through “post-natal hepatoblasts” and “peribiliary hepatocytes.”7,23,31

The relative dynamics and contributions of hepatocyte replication versus DRs to parenchymal restitution in chronic liver disease appears to change with time, with increasing proliferation of DR hepatobiliary cells correlating with diminishing hepatocyte replicative potential and increasing senescence.4,16,18 A recent study of hepatic progenitor cells in HCV found a significant correlation between DRs and older age, which supports the role of senescence.46 That DRs are a source of hepatocellular restoration is most strongly supported by recent studies showing that EpCAM-positive hepatocytes had telomere lengths longer than those of EpCAM-negative hepatocytes (which presumably are older and/or derive from replication of earlier hepatocytes), but slightly shorter than those of the DR cells, which express telomerase.6

Thus, in diseases of all kinds, DRs mediate repair, at least in part, and may reflect activation of multiple stem/progenitor niches. From a tissue biological point of view, the basis of DR success as a prevalent reparative mechanism lies in the “geographic” uniqueness of the niche from which DR arises, the portal–parenchymal interface. Here, interactions between the hepatobiliary cells with portal and periportal mesenchymal cells are likely through both production of diffusible growth controlling factors and physical cell–cell contact. For example, hepatocyte:sinusoidal endothelium contact is instrumental in regeneration.58 Nerve and myofibroblast contact with progenitors during activation have been documented.21,47 Hepatocyte:cholangiocyte contact at the CoH may also play a role. Isolation of hepatocyte couplets for canalicular studies sometimes also captures hepatocyte:cholangiocyte couplets; when these are split, to isolate the very small cholangiocyte of the pair, the cholangiocyte undergoes hepatocellular differentiation (Ron A. Faris, personal communication April 2001).

Fibrogenesis.

Although long recognized as the likely driver of fibrosis in biliary obstruction, increasing DRs correlate closely with worsening stage in many chronic liver diseases.4,6,16,18 This suggests a model of portal fibrogenesis reliant on two key features. First, inhibition of normal hepatocyte replication due to replicative senescence or oxidative stress promotes stem/progenitor cell activation. Second, these cells need to be subject to increased proliferative drive due to hepatocytic injury and loss, expanding the DR.16 Profibrogenic factors from the cells of the DR, or other DR-dependent mechanisms, could then stimulate fibrosis. This model links lobular injury to portal fibrosis. It also explains why cofactors such as metabolic syndrome or alcohol exacerbate a range of other chronic parenchymal diseases such as hepatitis C or hemochromatosis,63 because any disorder affecting regeneration could promote portal fibrogenesis.

It is not yet proven that the DR causes fibrosis.64 Indeed, as discussed earlier, the mesenchymal and matrix components are important in the stem cell niche and several groups have shown that early matrix deposition or remodeling occurs prior to or with the DR in rodent models.27,65 Conversely, increased DRs have clearly been shown to precede detectable fibrosis.18,66 It is possible that stroma is a necessary requirement for a regenerative response, but that sustained injury leads to an unregulated stromal deposition.

Signaling factors include platelet-derived growth factor-B, TGF-β, connective tissue growth factor and monocyte-chemoattractant protein-1/CCL2.66 Notch signaling, important in biliary differentiation, appears to have some role because impairment of this signaling is associated with attenuated fibrosis in humans (Alagille syndrome67) and rodents.68 An accessory role for inflammatory cells, including lymphocytes, natural killer cells, and macrophages, needs also to be considered.39

Epithelial-to-mesenchymal transition (EMT) is perhaps the most intriguing hypothesized mechanism for hepatic fibrosis with demonstration that DR epithelia can lose markers of epithelial differentiation and acquire those of mesenchyme.69,70 However, whether this change progresses to full myofibroblastic differentiation and collagen production has not, to our knowledge, been demonstrated.70,71 It remains possible that a “partial” EMT by cells in the DR could contribute less directly to fibrosis through an altered expression profile of profibrogenic mediators such as TGF-β.72

Neoplasia.

DRs are thought to play two roles in hepatocarcinogenesis: their stem/progenitor cells may undergo malignant transformation and they may contribute to the development of the tumor microenvironment.

Supportive of the role of DRs in hepatocarcinogenesis, Ziol et al. described an increased risk of malignancy in HCV-infected patients whose biopsies showed foci of intermediate hepatocytes expressing EpCAM and K19.73 In single hepatobiliary tumors, morphologic and antigenic heterogeneity spans the spectrum of hepatocellular and biliary features, sometimes suggesting a shared lineage deriving from progenitor cell origin. By virtue of the bipotentiality of progenitors, following oncogenic transformation they may give rise to progeny with a heterogenous maturational pattern, some retaining progenitor cell functions. The most dramatically suggestive of these are mixed hepatocellular-cholangiocarcinomas with stem cell features. Even in pure hepatocellular carcinoma, the presence of small, subpopulations of tumor cells expressing a progenitor cell profile has been confirmed.74-76 Whether these intratumoral progenitor cells represent true cancer stem cells depends on the demonstration of tumor initiating capacity,77 which has been identified in some subpopulations.74,78

Apart from the tumor parenchyma, the tumor microenvironment, comprising matrix and various stromal cell populations, has been acknowledged as important to tumor emergence, growth, and invasion.77 For example, cancer-associated fibroblasts, which are themselves heterogeneous, may play an important role and could derive from the stromal components of DRs.

Two reports reveal the diminishment of cirrhosis-associated DRs with the stepwise emergence of HCC. Ikeda et al. show that persistent portal tracts and portal structure–containing fibrous septa within dysplastic nodules and HCCs showed diminishing DRs at the stromal–parenchymal interface with a parallel increase in K7-positive small hepatocytes.79 This observation, however, does not have significant explanatory power for the novel finding of these K7-positive small hepatocytes. Lennerz et al., however, go a step further by considering diffuse cellular elements beyond the hepatobiliary cells themselves and investigate molecular signaling between these cellular elements.80 These authors suggest that changes in DRs might contribute to the generation of cancer-associated fibroblasts in the process of a developing tumor microenvironment, in parallel with the oncogenic transformation of hepatocytes themselves, with reciprocal signaling interactions between these hepatocytes and neighboring cells shaping the tumor microenvironment. These novel hypotheses arise because the authors, true tissue biologists, actively consider the DR as a dynamic interplay of diverse molecular, cellular, and tissue level effects and can therefore form hypotheses about the underlying mechanisms whereby DRs condition or respond to early events in hepatocarcinogenesis.

Conclusion

DRs are of increasing interest for scientists and physicians, having diagnostic and prognostic import and appearing to be central to pathophysiologic processes as diverse as hepatobiliary regeneration, fibrogenesis, and hepatocarcinogenesis. It should be clear to interested clinicians and investigators that there is no single “ductular reaction”; rather, DRs are a protean array of changes in liver tissue in response to acute or chronic injury, as diverse as the wide array of diseases and injuries that cause them, cellularly and geographically diverse within themselves, and diverse in their physiologic and pathologic outcomes. Embracing systems biological approaches to exploring DRs, in addition to the more traditional cell and molecular biological techniques, will further enhance our understanding and, thereby, advancement of therapeutic possibilities.

Ancillary