Pereira, Tamara N.*; Lewindon, Peter J.*; Greer, Ristan M.†; Hoskins, Anita C.*; Williamson, Richard M.‡; Shepherd, Ross W.*; Ramm, Grant A.*
See “Cystic Fibrosis–associated Liver Disease: When Will the Future Be Now?” by Rudnick on page 312.
Liver disease has emerged as a significant contributor to the morbidity and mortality of patients with cystic fibrosis (CF) with onset in childhood. Not all patients with CF develop clinically significant liver disease despite the universal expression of the defective CF transmembrane conductance regulator (CFTR) protein in cholangiocytes (1–3), although minor pathology (bile duct plugs, inflammation, steatosis) is common. Significant focal biliary fibrosis is detected in 72% of patients at autopsy. Cirrhosis is prevalent in up to 10% of children (4) but <2% of adults (5) with CF, which suggests a survival disadvantage. Complications of CF liver disease (CFLD), including nutritional growth failure, portal hypertension, and, in some, synthetic liver failure, affect the quality of life in up to 20% of patients (6,7). Although CFLD originates early in life (6,7), the onset of fibrosis and progression to cirrhosis and portal hypertension are unpredictable and early diagnosis is elusive (4,6,7) because of the lack of sensitive noninvasive diagnostic tests (8). Liver biopsy is emerging as an important contributor to clinical management, although it is not widely used (8).
The pathogenesis of CFLD is now better characterized. The CFTR protein is expressed on the apical membrane (biliary aspect) of cholangiocytes (2,9) and biliary transport is altered (10), with retention of toxic bile acids, including taurocholate (11), which induces expression of a key fibrogenic chemokine, monocyte chemotaxis protein-1 (MCP-1), in hepatocytes and cholangiocytes (11). MCP-1 and other factors induce hepatic stellate cell (HSC) chemotaxis (11), leading to peribiliary fibrogenesis (11,12), the pathognomonic focal biliary fibrosis of CFLD (12), and some patients progress to multilobular biliary cirrhosis. Active fibrogenesis with collagen deposition from both activated HSC and myofibroblasts is seen along the expanding scar interface (11,12), reflected by the appearance of potential biomarkers in the serum (10,11,13,14). The occurrence of modifying genetic, environmental, immunological, and other factors such as compensatory membrane channels most likely explains the variability in onset, progression, and severity of liver fibrosis. Although modifying genes such as the SERPINA1 gene Z allele have been associated with CFLD (15), the molecular basis for various clinical phenotypes of CFLD remains undefined.
To further the search for molecular mechanisms underlying phenotypic differences in CFLD, we aimed to determine whether CFLD and/or the severity of liver fibrosis in CFLD could be differentiated at the transcriptional level. Such analyses have proved valuable in the study of other chronic liver diseases such as primary biliary cirrhosis (16), viral hepatitis (17,18), and biliary atresia (19–21). Using our unique CFLD liver biopsy repository, we initially targeted specific families of gene expression related to the potential pathways and candidate molecules implicated in our earlier mechanistic studies (10,11,13,14). Differential expression was analyzed comparing liver tissue from patients with CFLD to both normal pediatric controls and a non-CF cholestatic liver disease control cohort of patients with biliary atresia. The present study is the first to use cDNA array analysis coupled with both real-time reverse transcription-polymerase chain reaction (RT-PCR) and immunohistochemical validation to demonstrate differential hepatic gene expression in human CF liver tissue.
Patients With CFLD and Controls
The present study was approved by the ethics committees of The Queensland Institute of Medical Research and Royal Children's Hospital, Brisbane. Liver tissue was obtained from a systematic biopsy-based study of patients with suspected CFLD, and either control pediatric liver tissue from transplant donors or normal tissue adjacent to excised benign liver cysts. In addition, liver tissue was obtained from children with biliary atresia and served as a comparative cohort of non-CF cholestatic liver disease controls in real-time RT-PCR and immunohistochemical validation studies. As previously described (8,12), after written informed consent from parents or guardians, ultrasound-guided dual-pass percutaneous liver biopsies were carried out under general anesthesia in patients with suspected CFLD (defined by at least 2 of the following clinical features: hepatomegaly ± splenomegaly; persistent elevation of serum alanine aminotransferase [alanine aminotransferase >1 × upper limit of normal] >6 months; abnormal ultrasound scan with abnormal echogenicity or nodular edge suggestive of cirrhosis). Dual passes of liver biopsy were performed for the present study, as previously reported (8). The core with the highest liver fibrosis F-stage was used in all cDNA arrays, and real-time RT-PCR and immunohistochemical validation.
cDNA Array Study
Liver tissue from children initially enrolled between 1997 and 2001 was subject to clinical histopathological and research histochemical analyses. Where possible, a segment of liver biopsy tissue (3 mm) was stored at −70°C for RNA extraction. Tissue was available from 14 patients with CFLD and 2 control subjects. Two micrograms of RNA from each individual patient was pooled for cDNA array analysis (to compensate for the small amount of tissue available), according to the following histological groupings: subjects with CF with no fibrosis, F0 (n = 4; 3 F:1 M; 9.6 ± 3.6 years), stages F1–F2 fibrosis (n = 5; 2 F:3 M; 10.7 ± 3.5 years), stages F3–F4 fibrosis (n = 5; 3 F:2 M; 11.1 ± 2.9 years), and 5 μg of RNA was pooled from each of the pediatric control subjects (1 F:1 M; 9.9 ± 0.2 years). Table 1 shows the clinical characteristics of these patients.
Real-time RT-PCR Validation
With subsequent patient enrollment between 2001 and 2007, an expanded collection of liver samples was available for study. In addition to the specimens used for cDNA array analysis, tissue from a further 18 patients with CFLD and 4 pediatric controls was available for RT-PCR analysis. RNA from each patient was analyzed individually. Thus, tissue was available from 32 children with CFLD, stratified as fibrosis stages F0 (n = 9; 6 F:3 M; 11.9 ± 3.6 years), F1–F2 (n = 14; 9 F:5 M; 11.1 ± 3.8 years), F3–F4 (n = 9; 5 F:4 M; 10.9 ± 2.8 years), and 6 control subjects (3 F:3 M; 5.7 ± 4.6 years). Table 2 shows the clinical characteristics of these patients. Liver tissue was available from children with biliary atresia (n = 11; 3 F:8 M; 2.0 ± 2.3 years; range 0.2–7.6 years) and was used in real-time RT-PCR studies as a non-CF cholestatic liver disease control cohort. All of the patients with biliary atresia had stage 4 or severe fibrosis.
Histology and Immunohistochemistry
Liver was fixed in 10% buffered formalin and embedded in paraffin. A minimum of 10 levels of each biopsy were evaluated histologically following staining with hematoxylin/eosin, hematoxylin/van Gieson, orcin, periodic Schiff-diastase, Perls Prussian blue, and a silver-impregnated method for reticulin. A minimum of 5 portal tracts (range 5–13) were evaluated by a pathologist (R.M.W.) blinded to the patient's clinical characteristics, and fibrosis was scored according to Scheuer et al (22,23).
Immunohistochemistry was performed on liver tissue from CFLD, biliary atresia, or pediatric controls using heat retrieval in citrate buffer (pH 6) and either anti-plasminogen activator inhibitor-1 (PAI-1) goat polyclonal (AF1786, R&D Systems, Minneapolis, MN) or anti-tissue inhibitor of metalloproteinase-1 (TIMP-1) mouse monoclonal (M7293, DAKO, Glostrup, Denmark) primary antibodies. The goat or MACH 3 Mouse, Probe/HRP-Polymer (both Biocare Medical, Walnut Creek, CA) detection systems were used with 3,3′-diaminobenzidine as the chromagen. Nonimmune antihuman antisera (DAKO) were used in place of primary antibody in the negative control. Image analysis software was used to quantitate the immunohistochemical expression of PAI-1 and TIMP-1 proteins. Digital slides were created using the ScanScope System (Aperio Technologies, Vista, CA), and the Positive Pixel Count version 9 algorithm was used for quantitation.
Total RNA was extracted and RNA integrity was assessed as described previously (12).
cDNA Array Analysis
We focused on genes potentially involved in regulating cytokine expression and hepatic cellular interaction based on our concurrent demonstration of a clear role for HSC, matrix remodeling, cytokine expression, and chemokine involvement in the pathogenesis of CFLD (11–13). Atlas Cytokine/Receptor (268 genes) and Cell Interaction (265 genes) Nylon arrays (Clontech Laboratories, Mountain View, CA) were chosen over other platforms as the most sensitive method for analyzing the needle biopsy tissue available at the time of initial patient enrollment (1998–2001). cDNA probes were synthesized; total RNA (5 μg) was reverse transcribed using 1-μL Superscript II (200 U/μL; Life Technologies, Gaithersburg, MD), 35 μCi of [α-32P] 2′-deoxyadenosine triphosphate (10 μCi/μL; Amersham, Arlington Heights, IL), and array-specific primers. Probes were purified and quantitated (Tri-Carb 2100TR, Packard Biosciences, Meriden, CT). Labeled probes (2–6 × 106 cpm) were hybridized (overnight, 68°C) to the arrays, exposed to a PhosphorImager screen for 1 to 3 nights, and scanned on a PhosphorImager using ImageQuaNT (Molecular Dynamics, Sunnyvale, CA). Atlas Image 1.0 (Clontech) was used for quantitation. The median count for each gene was corrected by subtracting its local background. The hybridization signal intensity of each gene of interest was standardized to the housekeeping gene β-actin. Pairwise comparisons were made between the intensity for each gene in the CFLD arrays with the control array.
Semiquantitative Real-time RT-PCR
The differential expression of selected genes was confirmed by semiquantitative real-time RT-PCR, as previously described (12). Five micrograms of total RNA was reverse transcribed, and real-time PCR was conducted with primers designed with Primer Express 1.0 software (Applied Biosystems, Foster City, CA). The sequence of primers is given in Supplementary Table 1 (http://links.lww.com/MPG/A80). The expression of the gene of interest was standardized to the expression of the housekeeping gene β-actin. The mean standardized gene expression for each CFLD patient was then expressed relative to the control subjects.
Statistical analysis was performed using GraphPad Prism 5.02 (San Diego, CA). Quantitative data were analyzed using the Kruskal-Wallis 1-way analysis of variance (KW-ANOVA), and posthoc comparisons were performed using the Mann-Whitney U test; P < 0.05 was considered statistically significant.
The cDNA array analysis results for all of the 533 genes in the Atlas Cytokine/Receptor and Cell Interaction arrays are listed in Supplementary Table 2 (http://links.lww.com/MPG/A80).
Genes Associated With HSC-mediated Fibrogenesis and Fibrolysis
As expected, the altered expression of many genes associated with hepatic fibrogenesis was confirmed in patients with CFLD by cDNA array analysis (Supplementary Table 2, http://links.lww.com/MPG/A80). MCP-1 was upregulated in F1–F4 fibrosis (as described earlier (11,24)) (Supplementary Table 2, http://links.lww.com/MPG/A80). The receptor for MCP-1, chemokine (C-C motif) receptor 2 (CCR-2), was upregulated in F1–F2 fibrosis. Macrophage inflammatory protein 1 beta (MIP1β/CCL4) was upregulated in early (F0 and F1–F2) fibrogenesis, and several endothelins and their receptors were also upregulated. Several fibrillar collagen and extracellular matrix components were elevated by up to 25-fold in all of the fibrosis stages, whereas the expression matrix metalloproteinases (MMPs)-3 and -14 were decreased in early fibrosis but were elevated, along with EMMPRIN, in severe F3–F4 fibrosis. Some MMPs, such as MMP-2, -10, and -11, were increased in F0 fibrosis, suggesting a potential role in early matrix remodeling and fibrogenesis.
The mining of cDNA array data initially focused on genes with differential expression at ±2-fold or greater in CFLD versus controls, as well as expected function via known linked pathways associated with hepatic fibrogenesis, and it is these differences that were further assessed by RT-PCR in the first instance. Because we previously showed a role for markers of matrix remodeling in CFLD (13), genes associated with this function were specifically selected for confirmation by RT-PCR, and comparison with disease control liver. It should be noted that although the array data also suggest that such differentiation may be possible based on differences in relative expression between fibrosis staging in CF liver, none of the reported differences (between different fibrosis stages of CFLD) could be validated by RT-PCR or immunohistochemistry. One of the most striking findings of the cDNA array analysis comparing CFLD to control liver was the decreased expression of endothelial PAI-1 by up to 25-fold in patients with CFLD (Supplementary Table 2, http://links.lww.com/MPG/A80), an observation confirmed by real-time RT-PCR analysis, (KW-ANOVA; P = 0.0043), in which the decrease was also seen in all of the patients with CF with F0 (P = 0.0047), F1–F2 (P = 0.0015), and F3–F4 (P = 0.0007) fibrosis compared with controls (Fig. 1A). Although there was a decrease in the expression of liver PAI-1 in biliary atresia versus control liver (Fig. 1B), this was not statistically significant. Immunohistochemical staining showed weak liver PAI-1 protein expression in CFLD liver even around fibrous tissue (Fig. 2A). In comparison, there was a marked expression of PAI-1 in both hepatocytes and bile duct cells in control subjects (Fig. 2B), as well as in biliary atresia, the comparative non-CF cholestatic liver disease control cohort (Fig. 2C). Image analysis was used to quantitate the immunohistochemical expression of PAI-1 protein in the liver (Fig. 3A). There was a marked decrease in patients with CFLD (KW-ANOVA; P = 0.012), with F0 (P = 0.0381), F1–F2 (P = 0.0381), and F3–F4 (P = 0.0061) fibrosis compared with controls. There was no significant difference in the expression of liver PAI-1 protein in patients with biliary atresia compared with controls.
Similarly, the expression of TIMP-1 mRNA was markedly decreased in patients with CFLD by cDNA array analysis (Supplementary Table 2, http://links.lww.com/MPG/A80). RT-PCR analysis confirmed this observation (KW-ANOVA; P = 0.0195) in patients with CFLD with F0 (P = 0.0076), F1–F2 (P = 0.0075), and F3–F4 (P = 0.02) fibrosis compared with controls (Fig. 1C). Conversely, there was no significant difference in the expression of liver TIMP-1 mRNA in patients with biliary atresia compared with controls (Fig. 1D). Immunohistochemical staining showed no detectable TIMP-1 protein in patients with CFLD, despite the presence of fibrosis and steatosis (Fig. 2E). TIMP-1 protein was seen in hepatocytes and bile duct cells in controls (Fig. 2F). In biliary atresia, TIMP-1 protein expression was comparable with that of pediatric controls and was observed in hepatocytes and HSCs (Fig. 2G) and in infiltrating immune cells (data not shown). Quantitation of the immunohistochemical expression of liver TIMP-1 protein (Fig. 3B) showed a marked decrease in patients with CFLD (KW-ANOVA; P = 0.012), with F0 (P = 0.0152), F1–F2 (P = 0.0317), and F3–F4 (P = 0.0326) fibrosis compared with controls. There was no significant difference in the expression of liver TIMP-1 protein in patients with biliary atresia compared with controls.
Genes Associated With Cell Death Caused by Toxic Effect Bile
Tumor necrosis factor (TNF) -related apoptosis-inducing ligand (TRAIL) was upregulated by cDNA array analysis (Supplementary Table 2, http://links.lww.com/MPG/A80), but although RT-PCR showed an increase in expression, this did not reach statistical significance (Fig. 4A). Fas soluble protein/Apo1 ligand was elevated in early fibrosis, and its receptor (fasL receptor/CD95/Apo1) was elevated in all stages of fibrosis. The WSL protein/TRAMP/Apo-3/death domain receptor 3 and lymphotoxin-β/TNF-C were upregulated and apoptosis-related protein TFAR15 was downregulated in early fibrosis, whereas TNF-α was elevated in F1–F4 fibrosis.
Rho Family of Genes
Several of the Rho family proteins including p21-rac1/Ras-related C3 botulinum toxin substrate 1/ras-like protein TC25 and Ras-related C3 botulinum toxin substrate 2/p21-rac2/small G protein were upregulated in various stages of fibrosis. Other Rho family proteins such as transforming protein rhoB/ARHB/ARH6 and RND3/RhoE/Rho8/ARHE were downregulated (Supplementary Table 2, http://links.lww.com/MPG/A80). The decreased expression of RND3/RhoE/Rho8/ARH in the liver of patients with CFLD was confirmed by RT-PCR analysis (KW-ANOVA; P = 0.0028) and the decrease was seen in all of the groups, F0 (P = 0.0067), F1–F2 (P = 0.0012), and F3–F4 (P = 0.0016), compared with controls when assessed by real-time RT-PCR (Fig. 4B).
Liver disease develops in a subset of patients with CF. The present study provides the first evidence for a transcriptional basis for the pathogenesis of CFLD. Expression profiling of liver biopsies from children with CFLD demonstrates dramatic differences in CF versus age-matched controls. Coordinated behavior of gene expression may reflect key differences in pathogenic mechanisms associated with development of disease and fibrosis. The comparison with biliary atresia showing the opposite regulation of 2 key regulators of matrix remodeling, PAI-1 and TIMP-1, is novel and interesting. Our previous mechanistic studies demonstrated a role for HSC activation, matrix remodeling, and cytokine and chemokine expression in the pathogenesis of CFLD (11–13). The present study focused on gene sets potentially involved in regulating cytokine expression and hepatic cellular interaction to assess the differential hepatic gene expression in CFLD. The data indicate that there is a complex interaction of fibrogenic and inflammatory processes occurring in the liver of children with CF, as evidenced by the altered expression of genes known to be associated with hepatic fibrosis, tissue remodeling, cell death pathways and apoptosis, cholestasis and oxidative stress responses, and immune function, many of which have not been associated with CFLD. One of the aims of the present study was to determine whether expression of these genes varied with increasing severity of fibrosis; however, this was not seen in the genes analyzed further by RT-PCR, namely PAI-1, TIMP-1, TRAIL, and RND3. The array data suggest that there may be differential expression of additional genes based on differences in relative expression between fibrosis stages, a finding that remains to be validated in future studies. These data do not demonstrate causality, but they do provide insight into potential mechanisms and pathways that may be involved in CFLD, which require further investigation.
One of the most significant validated observations of the present study was the decreased hepatic expression of PAI-1 and TIMP-1 in cDNA array analysis, confirmed by RT-PCR and immunohistochemistry. This decrease was observed in all of the patients with CFLD regardless of fibrosis severity. PAI-1 is an inhibitor of tissue-type plasminogen activator and urokinase-type plasminogen activator, which in turn converts plasminogen to plasmin (25), a protease required for the activation of MMPs and other growth factors such as hepatocyte growth factor and transforming growth factor-β. TIMP-1 is an inhibitor of MMPs. Thus, the decrease in PAI-1 and TIMP-1 is proposed to increase activation of MMPs and ultimately to increased matrix turnover. Increased hepatic expression of several MMPs was indeed demonstrated in the cDNA array analysis. Our results suggest that excess matrix remodeling occurs in the liver of all of the patients with CFLD, even in patients with stage 0 fibrosis before any fibrotic tissue is laid down or is visible histologically.
A decrease in hepatic PAI-1 expression has been reported in adult human liver diseases such as viral hepatitis (26), alcoholic liver disease, primary biliary cirrhosis, and primary and secondary liver cancer (27), despite an increase in circulating plasma PAI-1 levels (26,27). Fitch et al (27) proposed that this may be caused by either release of hepatic PAI-1 into the circulation or decreased clearance of PAI-1 from plasma; however, animal models of liver disease show the opposite effect. Increased hepatic PAI-1 is seen in the bile duct–ligated mouse model of cholestatic liver injury (28,29). This discrepancy may be caused by the acute nature of the animal studies, in which liver PAI-1 levels are increased during initiation of liver fibrosis. Human studies may reflect low liver PAI-1 levels because they are obtained later in the natural history of chronic cholestatic disease as seen in CFLD. The immunohistochemical data presented in the present study supports this hypothesis, in which there was marked hepatic PAI-1 expression in patients with the acute, more aggressive cholestatic liver disease biliary atresia. In contrast, PAI-1 levels were much lower in the more chronic liver disease patients with CFLD. Similarly, the decrease in hepatic TIMP-1 mRNA expression is an apparent contradiction to previous biological studies in which serum levels of TIMP-1 protein were elevated in patients with CFLD with F0 fibrosis, returning to control levels with increasing fibrosis severity (13).
Recently, PAI-1 was classed as SERPINE1 and is one of a family of serine protease inhibitors. A different member of this family, the SERPINA1, has been identified as a modifying gene for CFLD (15); thus, it would be important to look for polymorphisms in the PAI-1/SERPINE1 gene in patients with CFLD. Of note, both the PAI-1 and CFTR genes are located on the long arm of chromosome 7. PAI-1 is at position 7q21.3-q22, whereas CFTR is at 7q31.2, and future studies should also determine whether there is any genetic linkage between these 2 loci.
The present study also confirmed the differential expression of several known markers of activated HSC, hepatic fibrogenesis, and matrix remodeling. Collagen-I, collagen-III, and MMP-2 were upregulated, as would be expected during active hepatic fibrosis. Several glycoproteins (fibronectin, laminins, tenascins) and proteoglycans (biglycan, decorin, vesicant) required for strengthening the collagen fibrillar structure were also differentially expressed in CFLD compared with controls. The hepatic accumulation of toxic bile acids in CF has been suggested as a determinant of hepatic fibrogenesis (10,11). In the present study, we observed increased expression of death receptors and ligands of the TNF superfamily, suggesting cell death via apoptosis. The TRAIL ligand was elevated but TRAIL receptors 1 and 2 were not. In addition, the Fas receptor was elevated but Fas ligand was not. TRAIL-mediated apoptosis has been demonstrated in HuH-7 hepatocytes exposed to bile acids (30) and could be activated in other liver cells.
RND3/RhoE/Rho8/ARHE is a member of the Rho superfamily. The decreased expression of this gene is a novel finding, not previously reported in liver disease, and requires characterization. RhoE is a negative regulator of actin cytoskeleton remodeling that, we speculate, may be involved in cell motility associated with HSC or inflammatory infiltration. Other genes involved in liver immunity such as interleukins and their receptors were elevated in patients with CFLD.
We acknowledge and recognize that there are limitations of the technologies used in the present pilot study. The use of cDNA array for discovery of differential gene expression has the potential to generate false-positive data; however, cDNA arrays are used to obtain a global understanding of differentially expressed genes/pathways. In the present pilot study in CFLD, >500 genes were represented, and to avoid “noise,” only genes that showed differential expression of 2-fold or greater were reported, with specific gene differences of interest further validated by both real-time RT-PCR analysis and immunohistochemistry performed on individual patient samples. Only a small number of patients with CFLD available at the start of the present study and only a portion (<3 mm) of the core biopsy were available for research. Thus, pooling the specimens was necessary for the cDNA array analysis, and it is possible that outliers may have influenced the cDNA array data analysis; however, real-time RT-PCR was used to validate the array data and because this technique required less starting material, analyses were conducted on individual patient tissue and not pooled tissue. Furthermore, as the study progressed and more patients were enrolled, tissue from an additional 18 patients with CFLD and 3 controls was available for RT-PCR analysis (in total 32 CFLD + 6 controls). Despite these limitations, several genes (PAI-1, TIMP-1, TRAIL, and RND3) showed concordance between the array and RT-PCR results. In addition, the 2 genes (PAI-1 and TIMP-1) that showed the most significant differential expression at the mRNA level also showed significant differential expression at the protein level as determined by immunohistochemical analysis. Because of the lack of tissue, we were unable to extend this validation to other genes in the present study.
The coordinated expression of these functionally related genes underscores a predominant fibrogenic footprint and provides a basis for identification of gene groups that play regulatory roles in the pathogenesis of duct injury and obstruction in CFLD. Validation of many of these genes identified by cDNA array analysis is still required because this was a pilot study with both limited biopsy material available and patient enrollment. Future analysis of these genes, and their transcriptional or posttranscriptional regulation, and studies exploring cellular localization and levels of protein expression will assist in elucidating the mechanisms associated with hepatic fibrogenesis in CFLD and may reveal markers for the early detection of liver injury in patients with CF.
1. Kerem B, Rommens JM, Buchanan JA, et al. Identification of the cystic fibrosis gene: genetic analysis. Science 1989; 245:1073–1080.
2. Riordan JR, Rommens JM, Kerem B, et al. Identification of the cystic fibrosis gene: cloning and characterization of complementary DNA. Science 1989; 245:1066–1073.
3. Rommens JM, Iannuzzi MC, Kerem B, et al. Identification of the cystic fibrosis gene: chromosome walking and jumping. Science 1989; 245:1059–1065.
4. Lindblad A, Glaumann H, Strandvik B. Natural history of liver disease in cystic fibrosis. Hepatology 1999; 30:1151–1158.
5. Cystic Fibrosis Foundation Annual Data Base Report. Bethesda, MD: Cystic Fibrosis Foundation; 2008.
6. Sokol RJ, Durie PR. Recommendations for management of liver and biliary tract disease in cystic fibrosis. Cystic Fibrosis Foundation Hepatobiliary Disease Consensus Group. J Pediatr Gastroenterol Nutr 1999; 28 (suppl 1):S1–13.
7. Colombo C, Battezzati PM, Crosignani A, et al. Liver disease in cystic fibrosis: a prospective study on incidence, risk factors, and outcome. Hepatology 2002; 36:1374–1382.
8. Lewindon PJ, Shepherd RW, Walsh MJ, et al. Importance of hepatic fibrosis in cystic fibrosis and the predictive value of liver biopsy. Hepatology 2011; 53:193–201.
9. Kinnman N, Lindblad A, Housset C, et al. Expression of cystic fibrosis transmembrane conductance regulator in liver tissue from patients with cystic fibrosis. Hepatology 2000; 32:334–340.
10. Smith JL, Lewindon PJ, Hoskins AC, et al. Endogenous ursodeoxycholic acid and cholic acid in liver disease due to cystic fibrosis. Hepatology 2004; 39:1673–1682.
11. Ramm GA, Shepherd RW, Hoskins AC, et al. Fibrogenesis in pediatric cholestatic liver disease: role of taurocholate and hepatocyte-derived monocyte chemotaxis protein-1 in hepatic stellate cell recruitment. Hepatology 2009; 49:533–544.
12. Lewindon PJ, Pereira TN, Hoskins AC, et al. The role of hepatic stellate cells and transforming growth factor-beta(1) in cystic fibrosis liver disease. Am J Pathol 2002; 160:1705–1715.
13. Pereira TN, Lewindon PJ, Smith JL, et al. Serum markers of hepatic fibrogenesis in cystic fibrosis liver disease. J Hepatol 2004; 41:576–583.
14. Leonardi S, Giambusso F, Sciuto C, et al. Are serum type III procollagen and prolyl hydroxylase useful as noninvasive markers of liver disease in patients with cystic fibrosis? J Pediatr Gastroenterol Nutr 1998; 27:603–605.
15. Bartlett JR, Friedman KJ, Ling SC, et al. Genetic modifiers of liver disease in cystic fibrosis. JAMA 2009; 302:1076–1083.
16. Shackel NA, McGuinness PH, Abbott CA, et al. Identification of novel molecules and pathogenic pathways in primary biliary cirrhosis: cDNA array analysis of intrahepatic differential gene expression. Gut 2001; 49:565–576.
17. Honda M, Kaneko S, Kawai H, et al. Differential gene expression between chronic hepatitis B and C hepatic lesion. Gastroenterology 2001; 120:955–966.
18. Shackel NA, McGuinness PH, Abbott CA, et al. Insights into the pathobiology of hepatitis C virus-associated cirrhosis: analysis of intrahepatic differential gene expression. Am J Pathol 2002; 160:641–654.
19. Bezerra JA, Tiao G, Ryckman FC, et al. Genetic induction of proinflammatory immunity in children with biliary atresia. Lancet 2002; 360:1653–1659.
20. Chen L, Goryachev A, Sun J, et al. Altered expression of genes involved in hepatic morphogenesis and fibrogenesis are identified by cDNA microarray analysis in biliary atresia. Hepatology 2003; 38:567–576.
21. Zhang DY, Sabla G, Shivakumar P, et al. Coordinate expression of regulatory genes differentiates embryonic and perinatal forms of biliary atresia. Hepatology 2004; 39:954–962.
22. Scheuer PJ. Liver biopsy size matters in chronic hepatitis: bigger is better. Hepatology 2003; 38:1356–1358.
23. Scheuer PJ, Standish RA, Dhillon AP. Scoring of chronic hepatitis. Clin Liver Dis 2002; 6:335–347.
24. Ramm GA. Chemokine (C-C motif) receptors in fibrogenesis and hepatic regeneration following acute and chronic liver disease. Hepatology 2009; 50:1664–1668.
25. Kruithof EK. Plasminogen activator inhibitors—a review. Enzyme 1988; 40:113–121.
26. Inuzuka S, Ueno T, Torimura T, et al. The significance of colocalization of plasminogen activator inhibitor-1 and vitronectin in hepatic fibrosis. Scand J Gastroenterol 1997; 32:1052–1060.
27. Fitch P, Bennett B, Booth NA, et al. Distribution of plasminogen activator inhibitor in normal liver, cirrhotic liver, and liver with metastases. J Clin Pathol 1994; 47:218–221.
28. Wang H, Vohra BP, Zhang Y, et al. Transcriptional profiling after bile duct ligation identifies PAI-1 as a contributor to cholestatic injury in mice. Hepatology 2005; 42:1099–1108.
29. Bergheim I, Guo L, Davis MA, et al. Critical role of plasminogen activator inhibitor-1 in cholestatic liver injury and fibrosis. J Pharmacol Exp Ther 2006; 316:592–600.
30. Higuchi H, Gores GJ. Bile acid regulation of hepatic physiology: IV. Bile acids and death receptors. Am J Physiol Gastrointest Liver Physiol 2003; 284:G734–G738.