Fibrosis correlates with a ductular reaction in hepatitis C: Roles of impaired replication, progenitor cells and steatosis


  • Potential conflict of interest: Nothing to report.


The mechanisms for progressive fibrosis and exacerbation by steatosis in patients with chronic hepatitis C (HCV) are still unknown. We hypothesized that proliferative blockade in HCV-infected and steatotic hepatocytes results in the default activation of hepatic progenitor cells (HPC), capable of differentiating into both biliary and hepatocyte lineages, and that the resultant ductular reaction promotes portal fibrosis. To study this concept, 115 liver biopsy specimens from subjects with HCV were scored for steatosis, inflammation, and fibrosis. Biliary epithelium and HPC were decorated by cytokeratin 7 immunoperoxidase, and the replicative state of hepatocytes was assessed by p21 and Ki-67 immunohistochemistry. A ductular reaction at the portal interface was common. There was a highly significant correlation between the area of ductular reaction and fibrosis stage (r = 0.453, P < .0001), which remained independently associated after multivariate analysis. HPC numbers also correlated with fibrosis (r = 0.544, P < .0001) and the ductular area (r = 0.624, P < .0001). Moreover, steatosis correlated with greater HPC proliferation (r = 0.372, P = .0004) and ductular reaction (r = 0.374, P < .0001) but was not an obligate feature. Impaired hepatocyte replication by p21 expression was independently associated with HPC expansion (P = .002) and increased with the body mass index (P < .001) and lobular inflammation (P = .005). In conclusion, the strong correlation between portal fibrosis and a periportal ductular reaction with HPC expansion, the exacerbation by steatosis, and the associations with impaired hepatocyte replication suggest that an altered regeneration pathway drives the ductular reaction. We believe this triggers fibrosis at the portal tract interface. This may be a stereotyped response of importance in other chronic liver diseases. (HEPATOLOGY 2005;41:809–818.)

Why progressive portal fibrosis occurs in chronic hepatitis C (HCV) remains unknown.1 In particular, it is paradoxical that viral infection of hepatocytes within lobules induces fibrosis predominantly in portal areas. Factors contributing to increased fibrosis are well documented and include disease activity, age at infection, gender, duration of infection, and co-existent steatosis.

In earlier studies examining the relationship of fibrosis with steatosis in HCV, we found a significant correlation between steatosis, body mass index (BMI), and subsinusoidal fibrosis in lobules, as well as a correlation between steatosis and portal fibrosis.2 Additional support for the role of steatosis in HCV-related fibrosis has been provided by studies examining the effect of a reduction in steatosis. A weight loss and exercise intervention in overweight patients with HCV resulted in an improvement in the grade of steatosis and stage of fibrosis, and a reduction in the mean numbers of activated stellate cells and portal myofibroblasts.3 This effect was evident in portal areas (mean portal myofibroblasts pre- and post-weight loss: 333 mm-2 and 149 mm-2, respectively) as well as the lobules (mean activated stellate cells pre- and post-weight loss: 28 mm-2 and 3 mm-2, respectively), suggesting that steatosis directly or indirectly impacts on portal fibrosis. Currently there is no model to explain how this could be mediated.

Although not well described, a ductular reaction often occurs at the portal tract interface in patients with HCV and fibrosis.4, 5 These proliferating ductules are postulated to arise from hepatic progenitor cells (HPC),6–8 small periportal cells that are bipotential and capable of proliferation and differentiation into both hepatocytes and bile ductular epithelium. They are synonymous with oval cells in the rodent. In the normal liver, replacement of hepatocytes occurs through replication of other adjacent hepatocytes within the lobules. However, impairment of this primary pathway leads to proliferation of HPC, which by default become the source of regenerating hepatocytes.9 Causes of hepatocyte replicative arrest include experimental toxins,9 viral infection,10 and steatosis.11, 12

In cholestatic liver diseases, portal fibrosis occurs and appears to be induced by the ductular reaction.13 It has been shown that ductal epithelium can express proteins such as transforming growth factor-β (TGF-β), monocyte chemoattractant protein-1 (MCP-1), and platelet-derived growth factor (PDGF) that attract and activate stellate cells, leading to collagen deposition.14, 15 We hypothesized that replicative arrest of hepatocytes, induced by viral infection or inflammation and exacerbated by excess weight and steatosis, results in HPC expansion. We also hypothesized that the proliferation of HPC results not only in regeneration of hepatocytes, but also in a ductular reaction that contributes to periportal fibrogenesis. To test this, we quantified the number of HPC and the extent of the ductular reaction in a cohort of patients with HCV and then correlated these factors with the severity of fibrosis and steatosis as well as other host variables. In addition, the degree of replicative arrest in hepatocytes was evaluated to determine its role in this model.


HCV, chronic hepatitis C; BMI, body mass index; HPC, hepatic progenitor cell; TGF-β, transforming growth factor beta; MCP-1, monocyte chemoattractant protein-1; PDGF, platelet-derived growth factor; SMA, smooth muscle actin; NASH, nonalcoholic steatohepatitis; ns, not significant.

Patients and Methods

Patients and Clinical Data.

The study assessed 115 Caucasian patients with chronic HCV who had undergone a liver biopsy at the Princess Alexandra Hospital, Brisbane, Australia. Informed consent was obtained from each patient, and the protocol was approved by the University of Queensland and Princess Alexandra Hospital Research Ethics Committees. Diagnosis of chronic HCV was based on standard serological assays and abnormal serum aminotransferase levels for at least 6 months. All patients were positive for HCV antibody by the third-generation ELISA (Abbott Laboratories, North Chicago, IL) with infection confirmed by detection of circulating HCV RNA by polymerase chain reaction using the Amplicor HCV assay (Roche, Nutley, NJ). Viral genotyping was performed using the Inno-Lipa HCV II assay (Innogenetics, Zwijnaarde, Belgium). Patients with other forms of chronic liver disease or antibodies to human immunodeficiency virus were not considered for the analysis. An additional 15 normal liver donor biopsy specimens taken before grafting were assessed to obtain baseline values for cytokeratin 7–positive biliary epithelial area and HPC numbers in normal liver.

Details about weight and height and average alcohol intake (g/day) during the preceding 12 months were obtained from all patients at the time of liver biopsy. Information regarding average alcohol intake (g/day) before the last 12 months was also obtained. Alcohol consumption was assessed retrospectively by interview on at least three occasions. The number and types of alcoholic drinks consumed each day were recorded, and the alcohol content of each drink was calculated. The alcohol intake over a weekly period was averaged and recorded in g/day.

Histopathological Analysis.

The sections were analyzed by an experienced hepatopathologist (A.C.) who was blinded to the laboratory parameters and clinical data. The degree of inflammation was graded according to the method of Ishak et al.,16 and a score of 0 to 4 was assigned for each of portal inflammation, interface hepatitis, and lobular inflammation, and the sum gave the histologic activity index. Fibrosis was staged (0 to 4) according to the method of Scheuer.17 Steatosis was graded as follows: 0 (<5% hepatocytes affected); 1 (5%-29% of hepatocytes affected); 2 (30%-70% of hepatocytes affected); or 3 (>70% of hepatocytes affected). Perls' stain was available for all patients and was graded 0 to 4. For statistical analyses, hepatic iron was scored as present or absent.

Immunoperoxidase Staining and Quantification.

Formalin-fixed, paraffin-embedded liver biopsy specimens (n = 115) were used for immunohistochemical studies as previously described.3, 18 The primary antibodies used in the study were: anti-cytokeratin 7 (DakoCytomation, Botany, New South Wales, Australia; dilution 1/200) (bile duct and ductular epithelium and HPC), Ki-67 (DakoCytomation, dilution 1/400) (proliferation marker), p21waf1 (Oncogene Research, Merck Pty Ltd, Kilsyth, Victoria, Australia, dilution 1/50) (replicative arrest), anti-α-smooth muscle actin (SMA) (DakoCytomation, dilution 1/400) (activated stellate cells and myofibroblasts), and anti–Bcl-2 (DakoCytomation, dilution 1/50) (inhibitor of apoptosis).

The tissue sections were photographed using a PixeLink Colour Digital Camera (Total Turnkey Solutions, Mona Vale, New South Wales, Australia). Image analysis was used to quantify the immunoreactivity of cytokeratin 7 and α-SMA. For quantification of the total area of bile duct and ductular epithelium, non-overlapping fields of the entire biopsy after cytokeratin 7 staining were photographed at 100× magnification. For α-SMA, 10 non-overlapping fields at a magnification of 100× were photographed. Image analysis software (Image Pro Plus 4.5, SciTech Pty Ltd, Preston, Victoria) was used to assess the mean immunoreactive area per biopsy. The percent positive area was defined as the ratio of pixels set above the segmentation threshold to total number of pixels within a defined area of interest, multiplied by 100.

The number of HPC was calculated by counting isolated, cytokeratin 7–positive cells in the periportal area of the lobule and expressing these as number per portal tract. Counted cells were smaller than normal hepatocytes, and rare, morphologically typical hepatocytes with positive cytoplasmic staining were not included.

Hepatocellular proliferation and replicative arrest were assessed by counting Ki67-positive and p21-positive hepatocyte nuclei, respectively, in 10 random lobular fields at 400× magnification, and expressing this as a percentage of total hepatocyte nuclei to correct for cellular enlargement caused by steatosis. The replicative arrest ratio was determined by calculating the ratio (p21-positive nuclei) ÷ (Ki67-positive nuclei).

Statistical Methods.

Continuous normally distributed variables are summarized as mean ± SD and represented graphically as mean ± standard error of the mean (SEM). The grade of steatosis, stage of fibrosis, and alcohol consumption are summarized by median. Chi-square goodness of fit was used to determine the distribution of steatosis among the viral genotypes. To compare the means between groups, analysis of variance (ANOVA) or Student t test was performed. To determine differences between groups not normally distributed, medians were tested by Mann-Whitney U tests. Pearson's correlation coefficient was used to determine correlations between continuous normally distributed variables. The degree of association between nonparametric or ordinal variables was assessed by using Spearman nonparametric correlation.

Multivariate analysis was performed, correcting for age at biopsy, gender, viral genotype, stage of fibrosis, BMI, alcohol consumption, presence or absence of hepatic iron, grade of inflammation, and grade of steatosis. Independent effects of normally distributed variables were assessed by analysis of covariance (ANCOVA). A backward elimination approach was used to remove nonsignificant variables and determine the most parsimonious model including both fixed factors and covariates. Including a multiplicative term in the model tested interactions between variables. Ordinal linear regression was used to assess relative influence of variables on categorical data. All analysis was carried out using SPSS software version 12.0 (SPSS Inc. Chicago, IL) and a value of P < .05 was considered significant.


Clinical, Histological, and Laboratory Data.

Among the 115 patients, 75 (65.2%) were male and all were HCV RNA positive. The mean age and BMI, median alcohol consumption, and the distribution of the viral genotypes for the population are shown in Table 1. This table also details the distribution of subjects within each stage of fibrosis as well as the extent of hepatocyte steatosis. Fibrosis was stage 0 in all of the 15 donor biopsy specimens. The mean BMI was significantly higher in patients with steatosis (P = .001), and the prevalence of steatosis was higher in patients infected with viral genotype 3 (P < .001). There was no significant difference in current or past alcohol consumption between those with or those without steatosis (P = .205 and P = .967, respectively). Patients with steatosis had a higher necroinflammatory score (P < .0001) and more severe fibrosis (P < .0001) than those without steatosis.

Table 1. Demographic and Clinical Features of Patients With Chronic HCV
  • NOTE. Age and body mass index expressed as mean ± SD. Alcohol consumption expressed as median.

  • *

    Genotype unavailable for 5 patients, 4 patients had viral genotypes other than 1 and 3.

No. of patients, n115
Age (yr)39.1 ± 8.2
Viral genotype*, n (%) 
 159 (51.3)
 347 (40.9)
Stage of fibrosis (Scheuer, 0–4), n (%) 
 014 (12.2)
 155 (47.8)
 221 (18.3)
 3/425 (21.7)
Necroinflammatory score (Ishak), n (%) 
 1–223 (20)
 3–457 (49.6)
 5–935 (30.4)
 median4 (range, 1–9)
Steatosis grade, n (%) 
 069 (60)
 122 (19.9)
 211 (9.6)
 313 (11.3)
Body mass index (kg/m2)24.2 ± 4.2
 (range, 16.5–42)
Alcohol—current (g/day)4.0
—past (g/day)30

A Prominent Ductular Reaction Is Seen in Patients With HCV: Relationship With HPC.

A ductular reaction was commonly found in biopsy specimens after immunohistochemical staining for cytokeratin 7. The area occupied by cytokeratin 7–positive biliary epithelium, comprising anatomical bile ducts and the ductular reaction, ranged almost 20-fold, from 0.11% to 1.93%, with a mean of 0.56% ± 0.39 SD. For comparison, in normal liver donor biopsy specimens, the mean biliary area comprising anatomical ducts only without any ductular reaction was significantly less at 0.26% ± 0.26 (P = .003). Generally the ductular reaction was located as a wreath at the periphery of the portal areas and septa, quite distinct from the anatomical bile ducts (Fig. 1A). In more prominent cases, small branches extended into the adjacent periportal region. Portal tracts that were enlarged by fibrosis did not show concentric rings of ductules, suggesting that the reaction is dynamic and may revise as portal tracts expand. Importantly, in 6 of 14 cases with stage 0 fibrosis, a fine ductular reaction was present (Fig. 1B), indicating that the reaction can precede fibrosis and is not simply a secondary effect. The ductular reaction was not seen in normal liver (Fig. 1C). Immunoperoxidase staining for the anti-apoptosis protein bcl-2 showed faint but distinct staining in many of the small bile ductules (not shown).

Figure 1.

The ductular reaction in chronic HCV infection. (A) Small ductules, stained brown, are present at the edge of an enlarged and fibrotic portal tract. Ductules also extend into the periportal parenchyma. (B) Early and focal ductular reaction in HCV without fibrosis (stage 0). There is mild portal expansion due to inflammation only. (C) Normal liver, which does not have this periportal ductular reaction. (D) Conspicuous hepatic progenitor cells (HPC) as well as a ductular reaction. The HPC are single positive cells without the cord-like arrangement of the ductules. (Cytokeratin 7 immunoperoxidase. Original magnifications: A, ×100, B and C, ×200; D, ×400)

Hepatic progenitor cells, identified as isolated, cytokeratin-7–positive cells in the periportal lobule, were located in close proximity to the ductular reaction (Fig. 1D). In these liver biopsy specimens, the average number of HPC per portal tract ranged from 0.5 to 22.4, with a mean of 6.9 ± 4.98. This was significantly greater than in the normal donor biopsy specimens (mean, 0.65 ± 0.99; range, 0-4.0, P < .0001). As predicted, if HPC are a potential source of the ductular reaction, there was a strong correlation between these 2 variables (Fig. 2).

Figure 2.

Close correlation between the extent of the ductular reaction and the number of hepatic progenitor cells (HPC). This is consistent with the hypothesis that the small ductules arise from proliferating progenitor cells, which have the potential to differentiate into both hepatocytes and biliary epithelium. (r = 0.624, P < .0001)

The Ductular Reaction and Expansion of HPC Are Associated With the Stage of Fibrosis.

There were highly significant correlations between fibrosis and both the area of the ductular reaction and number of HPC (rs = 0.453 and rs = 0.394, respectively, both P < .0001) (Fig. 3). These were also seen when the ductular reaction and HPC expansion were compared with the extent of α-SMA staining (Table 2). The ductular reaction and HPC expansion correlated with increasing age but not with age at infection, duration of disease, or ethanol intake (Table 2). There was no significant difference between sexes.

Figure 3.

Correlation between fibrosis stage and the extent of the ductular reaction (A) and hepatic progenitor cell expansion (B). Data are represented as mean ± SEM. *P < .05, **P < .01. ***P < .001. (NDL, nondiseased liver controls)

Table 2. Correlation of Hepatic Progenitor Cell Numbers and Bile Ductular Area With Clinical and Histological Variables in 115 Patients With Chronic Hepatitis C
 Hepatic Progenitor CellsBile Ductular Reaction
CorrelationPAdjusted P*CorrelationPAdjusted P*
  • *

    Corrected for age at biopsy, gender, viral genotype, stage of fibrosis, body mass index, alcohol consumption, presence or absence of hepatic iron, grade of inflammation, and grade of steatosis.

  • ns = not significant.

Duration of infection0.159nsns0.172nsns
Age at infection0.003nsns0.164nsns
Current alcohol use0.068nsns0.029nsns
Previous alcohol use0.056nsns0.000nsns
Fibrosis stage0.453<.0001ns0.544<.0001.001
α-Smooth muscle actin area0.366<.0001ns0.472<.0001ns
Steatosis grade0.372<.0001.0070.374<.0001.043
Body mass index0.136nsns0.208.027ns
Total necroinflammatory score0.343<.0001ns0.463<.0001ns
Interface hepatitis grade0.407<.0001.0010.535<.0001.025
Lobular inflammation grade0.205.030ns0.228.015ns
Portal inflammation grade0.323.001ns0.417<.0001ns
Perls' grade (iron)0.188.048ns0.148nsns
Replication-arrested hepatocyte nuclei (p21+)
Proliferating hepatocyte nuclei (Ki67+)0.027nsns0.166nsns
Replicative arrest proliferation ratio0.145nsns0.021nsns

After multivariate analysis, fibrosis was independently associated with the area of the ductular reaction, total necroinflammatory score, steatosis, and age at biopsy (Table 3). The interdependence of the ductular reaction and HPC was confirmed in the multivariate analysis, where an increasing ductular reaction remained independently associated with fibrosis stage and the association with HPC lost statistical significance.

Table 3. Multivariate Analysis of Independent Associations With Fibrosis
Variable*P (Adjusted)Odds Ratio95% Confidence Interval
  • *

    Table includes only those variables reaching statistical significance.

  • Corrected for age at biopsy, gender, viral genotype, body mass index, alcohol consumption, presence or absence of hepatic iron, grade of inflammation, and grade of steatosis.

Ductular reaction<.000114.363.54–58.21
Total necroinflammatory score<.00017.222.88–18.05
Age (at biopsy).0181.061.01–1.12

Hepatic Steatosis and Inflammatory Activity Are Associated With Expansion of HPC and the Ductular Reaction.

Steatosis, assessed as both grade and percentage (data not shown) of steatotic hepatocytes, showed a highly significant correlation with increasing HPC numbers and the bile ductular reaction (Table 2; Fig. 4). The ductular reaction, but not number of HPC, also showed a correlation with increasing BMI. Steatosis was not an obligate feature, and patients without steatosis but with bridging fibrosis generally had an obvious ductular reaction. The relationships between steatosis and both HPC expansion and bile ductular area remained significant after multivariate analysis (Table 2).

Figure 4.

Steatosis and the ductular reaction. (A) Increasing steatosis is associated with a greater ductular reaction. (B) Higher grades of steatosis are also associated with increased numbers of HPC. Steatosis is not an obligate feature, and a prominent ductular reaction could be seen without it. Data are represented as mean ± SEM. *P < .05, **P < .01. NDL, nondiseased liver controls.

The hepatic inflammatory activity also correlated closely with the ductular reaction and HPC expansion (Fig. 5). The close relationship was seen for aggregate necroinflammatory activity (by Ishak scoring) and also for the individual inflammatory components of interface hepatitis, lobular inflammation, and portal inflammation. After multivariate analysis, interface hepatitis remained independently associated with both HPC and the bile ductular reaction (Table 2). Age was an additional association (Table 2).

Figure 5.

The relationship between inflammatory grade, the ductular reaction and HPC expansion. (A) Increasing necroinflammatory activity, the histologic activity index (HAI), is strongly correlated with the ductular reaction. (B) The number of HPC shows a similar association with the inflammatory score. Data are represented as mean ± SEM. ***P < .001.

Hepatocyte Replicative Arrest Is Associated With Expansion of HPC and the Ductular Reaction.

To determine whether impaired hepatocyte proliferation could be contributing to the expansion of HPC, hepatocyte replicative arrest and proliferation were evaluated by immunohistochemistry for p21 and Ki67, respectively, and the percentage of reactive nuclei determined. The replicative arrest index ranged from 0 to 13.1 (mean, 1.1 ± 1.7). Increased replicative arrest was associated with higher HPC numbers and a ductular reaction (Table 2). Hepatocytes expressing p21 were increased in patients who were overweight (Fig. 6), although no significant relationship was seen with steatosis. There were also significant associations with the grade of lobular inflammation (r = 0.193, P = .041), portal inflammation (r = 0.214, P = .023), and total inflammation (r = 0.211, P = .025). Multivariate analysis with the replicative arrest index (p21-positive hepatocytes) as the dependent variable showed independent associations with BMI, viral genotype 1, and lobular inflammation (Table 4).

Figure 6.

Increasing hepatocyte replicative arrest with overweight and obesity. The p21 index, measuring the percentage of hepatocytes with a block in cell cycling, is increased in patients with HCV who have a body mass index of > 25 kg/m2. Data are represented as mean ± SEM. ***P < .001.

Table 4. Multivariate Analysis of Independent Associations With Hepatocyte Replicative Arrest (Measured by p21 Staining)
Variable*P (Adjusted)Odds Ratio95% Confidence Interval
  • *

    Table includes only those variables reaching statistical significance.

  • Corrected for age at biopsy, gender, viral genotype, body mass index, alcohol consumption, presence or absence of hepatic iron, grade of inflammation, grade of steatosis, and stage of fibrosis.

Body mass index.0014.351.83–10.37
Lobular inflammatory grade.0053.091.41–6.78
Viral genotype 1.0372.261.05–4.85

The hepatocyte proliferative index (Ki67-positive nuclei) showed a correlation with gender only (increased in female gender, P = .002). When the replicative arrest/proliferation ratio was assessed (determined by the ratio of p21-positive hepatocytes ÷ Ki67-positive hepatocytes), significant associations remained for gender (increased in male gender, P = .004) and BMI (r = 0.260, P = .006).

When analyzed according to the stage of fibrosis, hepatocyte proliferation peaked in patients with stage 3 disease and then decreased, which was the mirror image of the p21 expression in hepatocytes (Table 5). However, these trends did not reach statistical significance. The proliferative index in biliary ductal and ductular epithelium increased with stage of fibrosis (Table 5). It showed a significant correlation with increasing hepatocyte replicative arrest (rs = 0.308, P = .019).

Table 5. Cellular Proliferation and Replicative Arrest According to Stage of Hepatic Fibrosis
 Fibrosis Stage*CorrelP
  • Abbreviations: Correl, Correlation (rs); ns, not significant.

  • *

    Values expressed as means (±SEM).

Hepatocyte proliferative index0.220.390.490.550.45  
(Ki67-positive nuclei, %)(0.08)(0.14)(0.09)(0.18)(0.10)0.046ns
Hepatocyte replicative arrest0.240.890.761.142.04  
(p21-positive nuclei, %)(0.14)(0.26)(0.16)(0.31)(0.59)0.168ns
Bile ductular proliferative index0.  
(Ki67-positive nuclei, %)(0.06)(0.06)(0.07)(0.07)(0.16)0.308.008

Genotype Does Not Influence the Correlation Between the Ductular Reaction and Fibrosis.

Steatosis is multifactorial in HCV and is induced by both metabolic (predominant in genotype 1) and viral (predominant in genotype 3) mechanisms.19 The mean number of HPC did not differ significantly between the genotypes 1 and 3 (7.4 ± 5.1 and 6.8 ± 4.9, respectively, P = NS), and the area of the ductular reaction was similar (0.6% ± 0.04 and 0.5% ± 0.4, respectively, P = NS). This suggests that the associations are related to viral infection rather than a coincidental effect of metabolic fatty liver disease. Some differences were seen. As shown in Table 6, the correlations between the ductular reaction/HPC with fibrosis and steatosis were significant for both genotypes, and the ductular reaction correlated with the inflammatory activity. However, HPC were significantly associated with the inflammatory score only in the group infected with genotype 1. Similarly, the p21 index correlated with HPC and the ductular reaction for those infected with genotype 1 only.

Table 6. Correlation of Hepatic Progenitor Cell Numbers and Bile Ductular Area With Clinicopathological Variables, Analyzed According to Viral Genotype
 Hepatic Progenitor CellsBile Ductular Reaction
Genotype 1Genotype 3Genotype 1Genotype 3
  1. NOTE. Table includes only variables reaching statistical significance in univariate analysis.

  2. Abbreviation: ns, not significant.

Age at biopsy0.238.0690.292.0510.336.010.411.005
Fibrosis stage0.543<.00010.320.0320.653<.0010.430.003
Inflammation grade0.437.0010.178ns0.547<.0010.412.004
Hepatocyte p21 index0.351.0070.074ns0.292.028−0.087ns


This study demonstrates a significant independent relationship between a periportal bile ductular reaction and hepatic fibrosis in a large cohort of patients with chronic HCV. These proliferating ductules are likely to arise from HPC as evidenced by their location in close proximity to the ductular reaction, and the strong correlation between these 2 variables. Steatosis and interface hepatitis are both independently associated with an increase in the number of HPC and the extent of the ductular reaction. This finding provides one mechanism whereby steatosis contributes to the progression of portal fibrosis. The significant independent relationship between the number of HPC and the number of hepatocytes in replication arrest supports our hypothesis that the periportal ductular reaction in chronic HCV is a consequence of altered hepatocyte proliferation, exacerbated by steatosis. To our knowledge, this is the first study to describe this potential mechanism for the progression of fibrosis in HCV.

The replacement of hepatocytes lost from normal hepatic parenchyma is known to occur by replication of mature hepatocytes.9 Inhibition of this replication by drugs,9 alcohol,11 steatosis,12 or viral infection10 promotes the activation of a secondary replicative pathway through the bipotential HPC,9 capable of producing both hepatocytes and bile ductules. Reactive oxygen species such as superoxide anion and hydrogen peroxide are potential mediators of this effect.12 Steatosis as well as inflammation are potential sources. It was also suggested recently that the combination of interferon-γ and tumor necrosis factor alpha, cytokines overexpressed in HCV,20, 21 may inhibit primary hepatocyte replication and stimulate HPC expansion.22 A ductular reaction is a by-product of this HPC proliferation,4 but a clear understanding of the molecular pathways involved remains elusive.

HPC proliferation has been described in a variety of chronic liver diseases.4–6, 12, 23 In livers with cirrhosis, Falkowski et al.6 found that intraseptal hepatocyte buds were associated with and appeared to arise from ductular reactions, the latter usually draining into the biliary system. Roskams and colleagues12 found a significant association between fibrosis and HPC numbers in non-alcoholic and alcoholic fatty liver disease, and a similar association was found in small numbers of human biopsies of other chronic liver diseases,24 supporting the associations found in the current study. A periportal ductular reaction was noted in association with HPC expansion in fatty liver disease,12 and in alcoholic liver disease the ductular reaction was associated with increased fibrosis.23 Findings from the current study indicate that this change could also be of importance in HCV. Recently, it was shown that increased HPC in HCV-infected patients correlated with increasing fibrosis, although a mechanism was not suggested.25

Aspects of the role of the ductular reaction in fibrogenesis have been studied in animal models and humans, particularly in biliary disorders. In an experimental model, newly formed cholangioles were found to express MCP-1 and PDGF B chain,15, 26 cytokines capable of recruiting and activating stellate cells to produce collagen.14 In human liver diseases of various etiologies, “proliferating” ductular epithelium was shown to express similar cytokines, including TGF-β1 and PDGF.27 In chronic biliary diseases of childhood, the expression of proteins such as MCP-1, TGF-β, and PDGF by ductal epithelial cells has been postulated to underlie the portal fibrosis that develops.13, 28, 29 The ductular reaction occurring in submassive hepatic necrosis has also been shown to have increased expression of profibrogenic factors and to intimately localize with activated stellate cells or myofibroblasts.30

This study does not prove that the ductular reaction directly promotes portal fibrosis. It is possible that a common or related stimulus drives the HPC/ductular reaction and fibrosis independently, or that the reaction is simply a by-product of portal fibrosis. This cannot be formally tested in the current study, but the presence of a fine ductular reaction in some patients without portal fibrosis (stage 0) suggests that it precedes fibrosis. Moreover, the intimate association between the bile ductules and the periportal and septal interface supports a role for it in the concentric deposition of fibrosis, which has been shown recently to develop as a web-like expansion between portal tract branch points.31 The close relationship between the ductular reaction and activated stellate cells and myofibroblasts is also compelling.32 A second consideration is that the distinction between HPC, canals of Hering, and the periportal ductular reaction can be problematic, and the clear delineation between the components is difficult and somewhat controversial.8 It is likely that some single cytokeratin 7–positive cells, counted here as HPC, represent canals of Hering cut in cross-section. Conversely, the HPC and canals of Hering are currently believed to be part of a continuum or compartment,32 with the former constituting a significant part of the canals.33, 34

The auxiliary role of increased BMI and steatosis in promoting replicative arrest in hepatocytes, HPC expansion, and a ductular reaction is of importance in HCV. Steatosis is promoted in a genotype-specific manner by metabolic (overweight) and viral effects and is associated with increased fibrosis.35 Multivariate analysis in this study shows a significant correlation between steatosis and the ductular reaction, providing a potential mechanism to explain the link between fatty hepatocytes and portal fibrogenesis. Although increased BMI and infection with genotype 1 appear to be important contributing factors in hepatocyte replicative arrest by multivariate analysis, we believe that steatosis induced as a viral effect36 is also likely to promote fibrosis through the ductular pathway, as shown in Table 6. However, the lack of a correlation between p21 expression and either HPC expansion or the ductular reaction in genotype 3–infected patients suggests that metabolic steatosis may be a more potent inhibitor of hepatocyte replication. This putative pathway of HPC proliferation implies that a progenitor cell compartment capable of promoting hepatocarcinogenesis could be expanded.12 The recent description of an increase in the prevalence of hepatocellular carcinoma in HCV-infected patients with steatosis supports this hypothesis.37 Expansion of HPC may not show a linear association with fibrosis stage. In one study, HPC proliferation, measured by Ki67 staining, was increased in livers with cirrhosis.6 This was postulated to occur because of senescence associated with long-term hepatocyte division. This has been confirmed in an animal model of decompensated cirrhosis, where p21-deficient animals developed less architectural distortion and had better compensation and hepatocyte mass.38 We found similar trends both for increasing hepatocyte senescence and ductular epithelial proliferation, as well as a relationship between the 2, although not all of these reached statistical significance.

Other diseases in which steatosis occurs, such as non-alcoholic steatohepatitis (NASH) and alcoholic liver disease, warrant further study to define the ductular reaction and its relationship with portal fibrosis. The classical lobular “chickenwire” fibrosis of NASH does not explain why progression is characterized in many cases by portal expansion and linkage.39 Brunt et al.39 particularly noted increasing portal fibrosis in their widely used staging system, but its cause has not been explained. Even more compelling, pediatric NASH often shows minimal sinusoidal fibrosis and progresses by deposition of portal fibrosis.40 A ductular reaction is common in these cases and can be seen in early fine, fibrous septa linking portal areas (A. Clouston, unpublished observation). Fibrosing cholestatic hepatitis occurring in immunosuppressed patients with chronic viral hepatitis also shows a close relationship between a florid ductular reaction and rapidly progressive fibrosis.41

In summary, this study demonstrates a striking relationship between increasing hepatic fibrosis in chronic HCV and a periportal ductular reaction. Steatosis is associated with an increase in both the number of HPC and the extent of the ductular reaction, providing a potential mechanism whereby steatosis contributes to the progression of portal fibrosis. The relationship between increased replicative arrest of hepatocytes and an increase in the number of HPC is consistent with the hypothesis that an altered regeneration pathway drives the ductular reaction, which we believe then triggers a profibrogenic reaction at the portal tract interface. If correct, it is possible that this is a common and stereotyped response that could be of importance in other chronic liver diseases. The potential mediators and modifiers of these reactions are of great interest and warrant further study.