Hepatocyte growth factor in patients with three different stages of chronic liver disease including hepatocellular carcinoma, cirrhosis and chronic hepatitis: An immunohistochemical study

Baskent University Faculty of Medicine, Department of Pathology, Bahçelievler, Ankara, Turkey Correspondence and reprints: Dr Banu Bilezikçi, Baskent Üniversitesi Tip Fakültesi Patoloji Anabilim Dali 12, Sokak No 7/1 Bahçelievler, Ankara, Turkey. Telephone +90-312-212-65-91, fax +90-312-212-75-72, e-mail banubil@yahoo.com Received for publication July 27, 2000. Accepted December 19, 2000 B Bilezikçi, AN Haberal, B Demirhan. Hepatocyte growth factor in patients with three different stages of chronic liver disease including hepatocellular carcinoma, cirrhosis and chronic hepatitis: An immunohistochemical study. Can J Gastroenterol 2001;15(3):159-165.

H epatocyte growth factor (HGF), originally described as a hepatocyte-specific mitogen, has been shown to be a potent stimulator of DNA synthesis in a variety of cell types (1)(2)(3)(4)(5).This growth factor also acts as a motogen, morphogen and tumour inhibitor (6).HGF-producing cells in the liver are nonparenchymal, presumably Kupffer, Ito and sinusoidal endothelial cells (7)(8)(9).This factor is also present in many organs other than the liver, and HGF messenger RNA expression is enhanced in the lung, kidney and spleen after liver damage.These findings have led to the characterization of HGF as an endocrine or paracrine factor that acts on hepatocytes and other organs (10).Levels of HGF may vary due to enhanced production or decreased hepatic clearance, or through both of these routes (10).The HGF receptor is a tyrosine kinase type encoded by the c-met proto-oncogene.The relationship between c-met and HGF expression may play an important role in hepatocellular carcinogenesis, metastasis and prognosis (11)(12)(13).
The aim of the present study was to determine the density of HGF-positive cells in three different stages of liver disease using immunohistochemical technique.Also examined were the relationship between histological activity index (HAI) and HGF staining density in chronic hepatitis (CH) specimens, and the link between HGF staining density and histological grade in hepatocellular carcinoma (HCC) specimens.

PATIENTS AND METHODS
Patients: Seventy patients with liver disease, 20 with CH (group I), 20 with cirrhosis (group II) and 30 with HCC (group III) were determined.Twenty normal liver biopsies were obtained as controls.Serological markers, abdominal ultrasonography, computed tomography and determination of alpha-fetoprotein levels were used to diagnose the three types of liver diseases, and all diagnoses were confirmed histologically.The clinical findings of patients with HCC are summarized in Table 1.The causes of the patients' liver dis-ease are presented in Table 2.At the time of biopsy, patients with HCC and patients with cirrhosis were not undergoing any specific therapy.In the CH group, one patient had been receiving steroid treatment for eight months, and three had been receiving interferon therapy for seven months at the time of biopsy.All but four specimens (resection samples in HCC cases) were obtained by needle biopsy.Each case of HCC was graded according to the Edmondson and Steiner scheme (14).The histological activity of CH was determined according to the method of Knodell et al (15).Histochemistry: All specimens were fixed in 10% formalin and were embedded in paraffin blocks.The blocks were then cut in serial sections of 4 to 5 µm thickness, and deparaffinized sections were prepared with hematoxylin and eosin (H&E), Masson trichrome, Manuel's reticulum stain and Perls' iron stain (16).Immunohistochemistry: All samples were immunohistochemically analyzed using the avidin-biotin complex method.First, deparaffinized sections were incubated with primary antibodies (HGF/Scatter Factor Ab-2, Clone SBF5, monoclonal [NeoMarkers, USA]) in a 1:10 dilution at room temperature for 16 h.The antibody used was a mouse monoclonal antibody of immunoglobulin G1 isotype that was raised against a recombinant human HGF protein.

Statistical analysis:
The Wilcoxon test and the t test were used to analyze the statistical relationships among the groups.P<0.05 was considered significant.

RESULTS
Immunohistochemical staining revealed diffusely scattered HGF-positive cells in all groups.While the staining was limited to nonparenchymal cells (Kupffer, Ito and sinusoidal endothelial cells) in groups I and II (Figure 1), in group III (HCC) the hepatocytes also stained (Figure 2 top, bottom).It was only possible to assess nonmalignant hepatocytes adjacent to the HCC in the four resection specimens, and these cases showed no positive staining for HGF in this area.The density of nonmalignant nonparenchymal positive cells in all groups was significantly higher than that in the control specimens.Also, the density of these cells in group I was significantly higher than that in groups II (P<0.05) and III (P<0.05)(Table 2).No association between the density of HGF-positive cells and grade of HCC (P>0.05)(Table 3), or between the density of HGFpositive cells and HAI in CH specimens (P>0.05) was found (Table 4).
There was cytoplasmic HGF positivity in the hepatocytes of 10 cases (33.3%) from group III.The distribution of HCC grading in these 10 patients was as follows: no grade I cases, two grade II cases, five grade III cases and three grade   IV cases (Table 5).Marked cytoplasmic staining was observed in the hepatocytes of four cirrhosis and four CH specimens, but this was identified as iron deposition (Figure 3 top, bottom).In contrast, no iron deposition was found in the 10 specimens from group III that had HGF-positive hepatoma cells (Figure 4 top, bottom).In addition, a diffuse pattern of oval-shaped noncytoplasmic globules that stained positive for HGF was noted in two of the cirrhosis specimens.These two cases exhibited severe cholestasis on H&E staining (Figure 5 top, bottom).Two patients in the cirrhosis group showed cholestasis as well, but the degree of cholestasis was minimal, and there was no staining for HGF in these two patients.

DISCUSSION
HGF is known to be a multipotent growth factor.Previous work has shown that the expression of HGF is associated with tissue regeneration, wound healing, normal tissue growth, tumour progression, embryogenesis and tumour invasion (13,17,18).The level of expression is increased in some patients with chronic active hepatitis (19).Investigation has proved that serum HGF levels are linked to HAI in this group (10,19).Our findings concurred with some of these earlier results in that immunohistochemical staining for HGF was of significantly higher density in CH specimens than in those from cirrhosis and HCC patients (P<0.05 and P<0.05, respectively); however, we found no association between density of HGF-positive cells and HAI (P>0.05).This could be a reflection of the small number of CH patients studied.These findings are in line with previous results indicating that hepatocyte growth factor is produced and expressed by nonparenchymal cells in nonneoplastic liver disease.Peptide growth factors are thought to be involved in the pathogenesis of hepatic fibrosis, and overexpression of transforming growth factor beta-1 (TGF-β1) has been proposed as an important initial step.Although HGF is expressed in CH, elevated TGF-β1 levels suppress HGF expression and induce hepatic fibrosis (20).All but one of the CH cases exhibited minimal fibrosis; thus, we contend that TGF-β1 levels were low and did not suppress HGF in these patients.

TABLE 5 Relationship between histological activity index (HAI) and hepatocyte growth factor (HGF)-positive nonparenchymal cell density in patients with chronic hepatitis
A particularly interesting aspect of this study was the detection of HGF in HCC cells.As noted earlier, HGF is known to be expressed in nonparenchymal cells, such as Kupffer, Ito and endothelial cells, both in normal tissue and in various liver diseases (11,19,(21)(22)(23).Most articles on this subject have stated that HGF may be detected in nonparenchymal cells within HCC nodules, but that neoplastic hepatocytes do not produce this factor (11,19,21,(24)(25)(26). In contrast, other reports have shown that some hepatoma cells do stain with HGF (13,27).One of these was an experimental study in rats, but the other was an investigation of human liver tissue.The latter authors reported that five of their six HCC cases had hepatoma cells that stained HGFpositive, and concluded that these cells were expressing HGF (13).Similarly, previous reports have documented HGF production by some carcinoma cells (lung carcinoma and glioblastoma cells).In our investigation, hepatoma cells in 10 of the 30 HCC cases stained positive for HGF.To confirm the specificity of our result, we ruled out the presence of pigments known to cross-react with HGF, including iron, lipofucsin and bile.
Concerning other possible confounding findings, careful analysis of the four cirrhosis and four CH specimens that showed marked cytoplasmic staining for HGF in the hepatocytes revealed that this staining was due to iron deposi-  tion.In addition, we noted a diffuse pattern of oval-shaped noncytoplasmic globules that stained positive for HGF in two of the cirrhosis specimens.These two cases exhibited severe cholestasis on H&E staining, whereas this was not observed in the other slides.We concluded that, before hepatocyte staining with HGF can be confirmed as a valid finding, hemosiderosis and cholestasis must be ruled out.
On this basis, we examined the immunohistochemically stained and H&E-stained specimens simultaneously.Regarding a possible link between HGF positivity and treatment received at the time of biopsy, none of the patients in the HCC and cirrhosis groups were undergoing specific therapy, and only three patients in the CH group were receiving interferon.On this basis, we believe that the HGF expression that we observed was likely not influenced by any therapy.Concerning potential relationships between HCC grade and HGF positivity, we observed that none of the 10 HGFpositive HCC cases exhibited grade I differentiation.However, our analysis revealed no association between density of HGF-positive cells and grade of HCC (Table 6).We think that there may be a connection between the reasons why hepatoma cells in grade I disease do not stain for HGF and why normal and non-neoplastic liver tissues do not stain.
In summary, our results should be considered preliminary.While they concur with the findings of some previous reports, other studies have failed to detect HGF in malignant cells.Our results may be related to the antibody that we applied.To our knowledge, our laboratory is the first to have used this antibody.If further research proves that hepatoma cells do produce this growth factor, immunohistochemical staining for HGF may be a valuable method for diagnosing well differentiated HCC.More investigation is required to determine the possible mechanisms of HGF staining in HCC.
Hepatocyte growth factor in chronic liver diseases Can J Gastroenterol Vol 15 No 3 March 2001 161

TABLE 1 Clinical findings in patients with hepatocellular carcinoma
*Positive staining in malignant hepatocytes for hepatocyte growth factor; ∅ Unknown; + Positive.AFP Alpha-fetoprotein; ALT Alanine aminotransferase; AP Alkaline phosphatase; AST Aspartate aminotransferase; GGT Gamma glutamyl transferase; HBsAg Hepatitis B surface antigen; HCV Hepatitis C virus

TABLE 3 Comparison of hepatocyte growth factor (HGF)-positive nonparenchymal cell density in groups I, II and III HGF-positive cells/mm 2 Group Number of cases (mean ± SD)
*Count in group I was significantly higher than the count in the other groups (t test; P<0.05)

TABLE 4 Relationship between hepatocyte growth factor (HGF)-positive nonparenchymal cell density and grade in hepatocellular carcinoma (HCC) Density of staining for HGF antibody (HGF-positive Grade of HCC* Number of cases cells/mm 2 , mean ± SD)
*There was no association between the density of HGF-positive nonparenchymal cells and grade of HCC (Wilcoxon test, P>0.05)