Epidermal growth factor receptor (EGFR) binds transforming growth factor α (TGF-α) which is mitogenic for hepatocytes. Diverse lines of evidence suggest that activation of the TGF-α /EGFR pathway contributes to hepatocellular carcinoma (HCC) formation. Herein, we developed an experimental model of cirrhosis giving rise to HCC and tested the antitumoral effect of gefitinib, a selective EGFR tyrosine kinase inhibitor, in this model. Rats received weekly intraperitoneal injections of diethylnitrosamine (DEN) followed by a 2-week wash-out period that caused cirrhosis in 14 weeks and multifocal HCC in 18 weeks. Hepatocyte proliferation was increased in diseased tissue at 14 weeks compared with control liver and at even higher levels in HCC nodules compared with surrounding diseased tissues at 18 weeks. Increased proliferation was paralleled by upregulation of TGF-α messenger RNA expression. A group of DEN-treated rats received daily intraperitoneal injections of gefitinib between weeks 12 and 18. In rats treated with gefitinib, the number of HCC nodules was significantly lower than in untreated rats (18.1 ± 2.4 vs. 3.7 ± 0.45; P < .05), while EGFR was activated to a lesser extent in the diseased and tumoral tissues of these animals compared with untreated rats. HCC nodules from both untreated and gefitinib-treated animals displayed insulin-like growth factor 2 overexpression that contributed to tumor formation in treated animals. In conclusion, the blockade of EGFR activity by gefitinib has an antitumoral effect on the development of HCC in DEN-exposed rats, suggesting that it may provide benefit for the chemoprevention of HCC. (HEPATOLOGY 2005,41:307–314.)
Hepatocellular carcinoma (HCC) accounts for more than 5% of all cancers and more than 500,000 deaths per year worldwide. The vast majority of patients have pre-existing cirrhosis at the time they develop HCC. Because current therapies are rarely able to achieve complete tumor ablation, chemoprevention in high-risk patients with established cirrhosis has been envisioned as a promising alternative option.1, 2 Yet, data indicating that chemoprevention may be effective in patients at risk for HCC, are scarce. A preventive effect of an acyclic retinoid on the development of a second tumor after ablation of the original tumor has been shown by one group.3, 4 Conflicting results regarding a potential preventive effect of interferon have also been reported.5–7 To validate and expand the concept of chemoprevention to other therapeutics, the molecular events that contribute to hepatocarcinogenesis in the liver with cirrhosis need to be identified and targeted.
Epidermal growth factor and transforming growth factor α (TGF-α) stimulate mitogenesis in hepatocytes through their binding to the epidermal growth factor receptor (EGFR).8 TGF-α is produced by hepatocytes and may act as an autocrine factor in the regenerating liver.9, 10 Diverse lines of evidence suggest that activation of the TGF-α /EGFR signaling pathway may contribute to hepatocarcinogenesis: TGF-α messenger RNA (mRNA) is overexpressed in cirrhosis and in HCC11–14; immunohistochemical studies of human HCC have shown localizations of TGF-α and EGFR in HCC cells, consistent with autocrine and paracrine mitogenic actions of TGF-α in HCC;15 and enhanced expression of TGF-α in hepatocytes is sufficient to induce tumor formation in transgenic mice16 and dramatically accelerates the appearance of HCC in hepatitis B surface antigen and TGF-α bitransgenic mice.17 However, in these different animals, as in virtually all previously reported models of liver carcinogenesis, tumors arise from liver tissue without cirrhosis.
In the present study, a rat model of diethylnitrosamine (DEN)-induced liver injury that reproduces the progression of cirrhosis toward HCC was established. The profiles of TGF-α and EGFR expressions at different stages of liver injury in this model have raised the possibility that the administration of an EGFR inhibitor at the stage of cirrhosis may prevent the subsequent development of HCC. Gefitinib (AstraZeneca, Macclesfield, UK), an adenosine triphosphate mimetic anilinoquinazoline that is currently under clinical evaluation in the treatment of patients with lung cancer and other tumors, is an orally active EGFR-tyrosine kinase inhibitor that reduces epidermal growth factor–stimulated tumor cell growth.18 Based on the assumption that gefitinib would disrupt activation of the TGF-α/EGFR pathway, we herein tested a potential antitumoral effect of this drug in rats that had cirrhosis.
All animal care and experimentation conformed to the Guide for the Care and Use of Laboratory Animals from the National Academy of Sciences. Male Wistar rats weighing 200 g received intraperitoneal injections of DEN (Sigma-Aldrich, St. Louis, MO) at 50 mg/kg body weight once a week. Rats received 12 (n = 8) or 16 (n = 16) weekly injections of DEN, and were killed 2 weeks after the last injection (to allow recovery from acute necrosis). Three age-matched normal rats were used as controls. The 16 rats that were submitted to the 16-week DEN regimen received daily intraperitoneal injections of either gefitinib (n = 8), an EGFR-tyrosine kinase inhibitor (AstraZeneca), at 2 mg/kg body weight or of vehicle (n = 8) between weeks 12 and 18 (Fig. 1). At the time of sacrifice, animals were anesthesized with an intraperitoneal injection of thiopental. Blood was collected for analyses of serum aminotransferase activities and bilirubinemia. The count of malignant nodules was performed at macroscopic examination of the liver by two independent investigators based on the following criteria: nodules with a diameter of 3 mm or more and a dysmorphic or dyschromic aspect. Samples of both tumoral and nontumoral liver tissue were frozen and stored at −80°C or fixed in 10% buffered formalin and embedded in paraffin. Tissue samples were also homogenized in TRIzol lysis solution (Invitrogen SARL, Cergy Pontoise, France) and in protein lysis buffer as described later for subsequent analyses.
Histology and Immunohistochemistry.
Four-micrometer-thick tissue sections of formalin-fixed, paraffin-embedded liver samples were stained with hematoxylin-phloxin-safran for standard histology. Hepatocyte proliferation was assessed by Ki67 immunolabeling using an immunoperoxidase method. In brief, tissue sections were incubated sequentially with an anti-Ki67 antibody (Novocastra Laboratories, Newcastle, UK) at 1:100 for 30 minutes, with peroxidase-conjugated rabbit anti-mouse immunoglobulins (Dako, Glostrup, Denmark) at 1:40 for 40 minutes and with peroxidase-conjugated swine anti-rabbit immunoglobulins (Dako) at 1:20. Peroxydase activity was revealed by 3-amino-9-ethyl carbazole and counterstaining was performed using Mayer's hematoxylin. An eyepiece with a net micrometer (Carl Zeiss, Jena, Germany) at high magnification (×400, 0.96 mm2) was used to count Ki67-positive hepatocyte nuclei in 10 successive fields.
Reverse-Transcriptase and Real-Time Polymerase Chain Reaction.
Total RNA was extracted from liver tissue homogenates in TRIzol solution. The first-strand complementary DNA (cDNA) was generated from 5 μg of total RNA using the Moloney Murine Leukemia Virus Reverse transcriptase (Invitrogen SARL) and pd(N)6 primers (Amersham Biosciences, Orsay, France). Different amounts of calibrated mRNAs (e.g., 1.25 pg to 1.25 × 105 pg) (Applied Biosystems, Applera France S.A., Courtaboeuf, France) were submitted to reverse transcription with the Moloney Murine Leukemia Virus Reverse transcriptase and pd(N)6 primers and, subsequently, used in real-time polymerase chain reaction (PCR) for standardization of 18S transcripts. For standardization of target gene transcripts (TGF-α, EGFR, and vascular endothelial growth factor [VEGF]), a PCR product was amplified, sequenced, purified, and calibrated as a number of copies. The procedure used to generate VEGF-calibrated cDNA was reported in a previous study.19
To generate TGF-α and EGFR-calibrated cDNA, the primers were designated according to published rat cDNA sequences in the GenBank database using Primer Express software version 1.5 (PE Applied Biosystems): TGF-α forward 5′-TATGTATTAGGTGGATGACG-3′, reverse 5′-GGGAAACAAAACAAAACAAG-3-′, generating a 500-bp fragment; EGFR forward 5′-GACACCTGCCCACCACTCAT-3′; Reverse 5′-CTCCCTGCCTCTGCTCACAT-3′, generating a 799-bp fragment. The cDNA prepared from 200 ng RNA was added to 50 μL PCR buffer containing 150 μmol/L of deoxyribonucleoside triphosphate and 20 pmol of each forward and reverse primers. For real-time PCR, the primers, which were designed according to published rat cDNA sequences in the GenBank database within the respective calibrated PCR products, were previously defined for VEGF.19 For TGF-α, EGFR, and insulin-like growth factor 2 (IGF-2), the primers were designed according to published rat cDNA sequences in the GenBank database, using Primers Express software version 1.5 (Applied Biosystems): TGF-α-T forward 5′-TCAGTATCGGGCATCCATGTT-3′, reverse 5′-CCATCCCCACAGCCTTACTTT-3′; EGFR-T forward 5′-GATCACGGCTCGTGTGTCC-3′, reverse 5′-ATGCCTATGCCATTGCAAACT-3′; IGF2-T forward 5′-CGATGTTGGTGCTTCTCATCTC-3′, reverse 5′-AAGCGTGTCAACAAGCTCCC-3′. Real-time PCR was performed using the Sybr Green PCR Core Reagents Kit on a 7700 Sequence Detector (Applied Biosystems). In all reactions, 200 nmol/L of each target forward and reverse primer and 50 nmol/L of each 18S forward and reverse primer were used. Data were collected and analyzed with sequence Detector version 1.7 software (Applied Biosystems). The average efficiency of all PCR reactions was 90% as indicated by the calibration curves. Data were expressed as the number of target mRNA copies per microgram of 18S RNA.
Liver tissue samples were homogenized in a protein lysis buffer (50 mmol/L Tris-HCl [pH 7.5], 150 mmol/L NaCl, 1% Nonidet P-40, 0.5% deoxycholic acid, 0.1% sodium dodecyl sulfate) containing protease inhibitors (Roche Diagnostics, Meylan, France). Homogenates were centrifuged twice at 10,000g for 10 minutes at 4°C, and the were pellets discarded. Protein concentration was determined using the Bio-Rad protein assay (Bio-Rad Laboratories, Hercules, CA). Forty micrograms of proteins were separated via electrophoresis on 7.5% and 10% sodium dodecyl sulfate–polyacrylamide gels and were subsequently transferred to polyvinylidene fluoride (Bio-Rad Laboratories) or nitrocellulose membranes (Amersham Biosciences). Blots were incubated for 1 hour at room temperature in a blocking buffer and then overnight at 4°C with antibodies raised against extracellular signal-regulated kinase 1 (ERK-1), p-ERK-1/2 (Tyr204) (Santa Cruz Biotechnology, Santa Cruz, CA), EGFR, p-EGFR (Tyr992), Akt and p-Akt (Ser473) (Cell Signaling Technology Inc., Beverly, MA). Blots were then incubated with horseradish peroxidase–conjugated anti–immunoglobulin G (Cell Signaling Technology Inc.) diluted at 1:2,000. Immunoreactivity was revealed via enhanced chemiluminescence (Amersham Biosciences).
The Mann-Whitney U test was used to compare mean values between two groups; the Kruskall-Wallis test was used to compare mean values between more than two groups. The calculation of Spearman's rank correlation coefficient was used to assess the relationship between quantitative parameters. Data are expressed as the mean ± SEM. All reported P values are two-sided, and a P value less than .05 was considered statistically significant. Kendall's coefficient of concordance was calculated for evaluation of macroscopic tumor count between the two investigators.
Sequential Development of Cirrhosis and HCC in a Rat Model of DEN-Induced Liver Injury.
In all rats submitted to the present protocol of DEN administration (50 mg/kg/wk), cirrhosis developed after a short-term administration and HCC nodules arose from livers with cirrhosis after a long-term administration (Fig. 1). After the short-term administration of DEN (12 weeks plus a 2-week wash-out period), macronodular cirrhosis was evident at macroscopic examination of the liver and was confirmed by histological analyses (data not shown). No HCC nodule was detected either at macroscopic examination or at histological analysis at this stage. In contrast, after the long-term administration of DEN (16 weeks plus a 2-week wash-out period), malignant nodules that fulfilled the predefined criteria (a diameter ≥3 mm and a dysmorphic or dyschromic aspect) were detected on the surface of the liver (Fig. 2A). All malignant nodules were identified as HCC at histological examination (Fig. 2B-C). The surrounding peritumoral liver exhibited cirrhosis, and no additional tumor was detected in any case on full sections of the liver with cirrhosis either at macroscopic or histological examination. After the long-term administration of DEN, the mean number of malignant nodules detected at the surface of the liver was 18.1 ± 2.4.
Deregulation in Cell Proliferation and in the TGF-α/EGFR Signaling Pathway in Rats With Cirrhosis.
After short-term DEN treatment, hepatocyte proliferation was markedly increased (approximately 14-fold) in the liver tissue of rats with cirrhosis compared with normal liver, as assessed by Ki67 immunolabeling (Fig. 3A). In keeping with a stimulation of proliferation, the mitogen-activated protein kinase pathway was activated. The levels of phospho-ERKs were markedly increased, while the levels of total ERKs were unchanged (Fig. 3B). This stimulation of cell proliferation coincided with an increase in the expression of TGF-α mRNA (Fig. 3C), while the expression of EGFR mRNA was lower (Fig. 3D) than in normal liver tissue. Consistent with this latter result, total EGFR, at the protein level, was decreased in liver tissue exhibiting cirrhosis compared with normal liver (Fig. 3E). A decreased expression of EGFR in proliferating hepatocytes has been previously reported and attributed to ligand-mediated EGFR downregulation.20–22 TGF-α mRNA expression and hepatocyte proliferation were further increased in HCC nodules compared with the liver tissue exhibiting cirrhosis at 18 weeks, suggesting that the TGF-α/EGFR signaling pathway may contribute to increased proliferation of hepatocytes in the liver with cirrhosis and to the development of HCC in this setting (Fig. 3F).
Antitumoral Effect of Gefitinib Treatment in Rats With Cirrhosis.
The above results led us to examine whether disruption of the TGF-α/EGFR signaling pathway could have a prophylactic effect on carcinogenesis in animals with cirrhosis. To test this possibility, we used gefitinib, a specific EGFR tyrosine kinase inhibitor, in these animals. Some of the rats under long-term DEN administration received daily intraperitoneal injections of gefitinib for 6 weeks, starting at the stage of cirrhosis. In gefitinib-treated rats, the number of HCC nodules was significantly lower (3.7 ± 0.45) than in untreated animals (18.1 ± 2.4) (P < .05) (Fig. 4). Mean tumor size was not different between untreated and gefitinib-treated animals (3.7 ± 0.2 mm and 4.0 ± 0.6 mm, respectively). All tumors enumerated in Fig. 4 were unambiguously identified as hepatocellular carcinoma, with a histological pattern similar to that shown in Fig. 2C. No evidence of drug-induced toxicity was detected at histological examination of the liver in gefitinib-treated animals, while the serum levels of liver enzymes were similar in gefitinib-treated and untreated animals.
To verify that the antitumoral effect of gefitinib was the result of a disruption in the EGFR-dependent signaling cascade, we examined the phosphorylation status of EGFR and ERKs in rats treated with gefitinib. Even though TGF-α mRNA expression was similar in livers with cirrhosis from gefitinib-treated and untreated animals (Fig. 5A), the amounts of both phospho-EGFR (Fig. 5B) and phospho-ERKs (Fig. 5C) were dramatically decreased in the liver tissues exhibiting cirrhosis from rats treated with gefitinib compared with those from untreated rats. These results confirmed the strong inhibitory effect of gefitinib treatment on EGFR tyrosine kinase activity and downstream mitogenic signals.
In the residual tumors that developed in rats despite gefitinib treatment, the level of phospho-EGFR was also dramatically decreased compared with tumors from untreated rats (Fig. 6A). However, the activation of ERKs was maintained (Fig. 6B), suggesting that the dose of gefitinib sufficient to suppress EGFR kinase activity may not be sufficient to completely inhibit downstream EGFR-dependent signaling pathways or that alternative proliferation and/or cell survival pathways might be activated in these tumors. The serine/threonine kinase Akt, a major component of survival pathways,23 was also activated in residual tumors from gefitinib-treated animals as in tumors from untreated animals (Fig. 6C). Residual tumors from gefitinib-treated animals contained lower levels of TGF-α mRNA than tumors from untreated animals (32.1 ± 5.3 vs. 96.4 ± 22.5 transcripts × 103/μ g RNA) (Fig. 6D). This result suggested that gefitinib treatment might have an additional inhibitory effect on TGF-α mRNA expression in HCC nodules. Because EGFR activation also regulates the expression of VEGF,24 a major angiogenic factor that may contribute to the progression of HCC, we examined in parallel the effect of gefitinib treatment on VEGF expression. Similarly, we found that VEGF mRNA levels were lower in tumors from gefitinib-treated animals compared with untreated animals (102.1 ± 15.3 vs. 164.6 ± 30.6 transcripts × 103/μg RNA) (Fig. 6E). Such inhibition in VEGF expression might also have contributed to the antitumoral effect of gefitinib in the present model. Finally, because IGF-2—a potent growth and survival factor for hepatocytes—has been implicated in liver carcinogenesis,25 we next evaluated its expression level in the model using quantitative reverse-transcriptase PCR. We found that IGF-2 mRNA expression was increased in HCC nodules compared with livers exhibiting cirrhosis with or without gefitinib treatment (Fig. 6F). Altogether, these results suggested that IGF-2 overexpression together with persistent activation of ERKs and Akt signaling pathways may contribute to the formation of residual tumors in animals with cirrhosis under EGFR inhibitory treatment.
The protocol of DEN-induced liver injury used in the present study caused the sequential formation of cirrhosis and HCC. In this model, we show that TGF-α mRNA expression increases in the liver with cirrhosis compared with normal liver and even further in HCC nodules compared with the surrounding liver with cirrhosis. This increase in TGF-α expression coincides with a parallel increase in hepatocellular proliferation. Consistent with a causal relationship between the activation of the TGF-α/EGFR pathway and the emergence of HCC, treatment of the animals at the stage of cirrhosis with the EGFR inhibitor gefitinib caused a dramatic reduction in the number of HCC nodules arising from the liver with cirrhosis. Of particular interest with respect to therapeutic perspectives, this prophylactic effect was obtained without liver toxicity.
Different models of DEN-induced liver carcinogenesis have been described previously.26–28 However, in contrast to previous models, the present model reproduces the sequence of cirrhosis and of HCC—as is most often the case in human liver disease—and thus provides a unique tool to test preventive treatments of HCC. The upregulation of TGF-α expression that occurs in this model may result both from the effect of cytokines produced by inflammatory cells and from the local regenerative response to cell loss in the cirrhotic liver.29 TGF-α, which is produced mainly by hepatocytes,10 acts both as a paracrine and autocrine factor, induces its own expression,30 stimulates hepatocyte proliferation, and may cause the onset of liver tumors at least partly through activation of the ERK signaling pathway. In keeping with these data, we observed a decrease in the phosphorylation levels of both EGFR and ERKs in liver tissues with cirrhosis from animals under gefitinib. This inhibitory effect was associated with a decrease in HCC occurrence in gefitinib-treated animals. Other mechanisms might also be involved in the prophylactic effect of gefitinib such as an inhibition of angiogenesis. Indeed, TGF-α is also a mitogen for endothelial cells24, 31 and the blockade of the EGFR signaling pathway with gefitinib leads to apoptosis in endothelial cells32, 33 and almost completely inhibits angiogenesis.34 Therefore, the antitumoral effect of gefitinib might be partly mediated by an inhibition of angiogenesis. Consistent with this possibility, we observed that gefitinib treatment induced a decrease in VEGF mRNA expression in tumors.
Gefitinib-induced EGFR inhibition, although effective in HCC nodules and surrounding tissues exhibiting cirrhosis, did not completely prevent the onset of HCC in our experimental conditions. In residual tumors from gefitinib-treated animals, ERKs and Akt were phosphorylated similar to tumors from untreated animals. Possible resistance of ERKs and Akt signaling pathways to EGFR kinase inhibitors has been previously reported.35 Further activation of these pathways by other growth factors such as IGF-2 could explain the emergence of residual tumors. Deregulation in insulin and IGF signaling pathways including re-expression of fetal IGF-2 mRNA, which may contribute to hepatocarcinogenesis, have been reported in human and murine HCC.25, 36 The possibility that IGF-2 overexpression contributed to the formation of gefitinib-resistant tumors is supported by recent in vitro findings showing that IGF-dependent signaling may antagonize the growth inhibitory effect of trastuzumab (Herceptin), an anti-c-erbB2 receptor monoclonal antibody, in human breast cancer cell lines.37–39 Other mechanisms might also be implicated in gefitinib-acquired resistance. For example, the loss of function of the phosphatase and tensin homolog phosphatase has been recently reported to counteract the antitumor action of EGFR tyrosine inhibitors in vitro.40 Other members of the erb-B receptor family, such as c-erbB3 and c-erbB2—which are expressed in 84% and 21% of human HCC, respectively—might also contribute to liver carcinogenesis.41 The effect of gefitinib on these receptors is poorly known. However, c-erbB2 does not appear to be involved in gefitinib resistance, because its association with Herceptin did not improve the antitumoral effect of gefitinib in our model (data not shown).
In conclusion, the present study provides a demonstration that a specific inhibitor of EGFR reduces the onset of HCC tumors in animals with cirrhosis. Further studies are now required to determine the optimal conditions for gefitinib alone or in combination with other drugs to completely overcome resistance and inhibit liver carcinogenesis upon cirrhosis. Altogether, our results suggest that inhibition of the TGF-α/EGFR loop with gefitinib could prevent or at least delay the emergence of HCC in patients with cirrhosis.