Efficacy of AKT Inhibitor ARQ 092 Compared with Sorafenib in a Cirrhotic Rat Model with Hepatocellular Carcinoma

Hepatocellular carcinoma (HCC) is the fifth most common cancer and the second cause of cancer-related death worldwide with 600,000 deaths per year (1). Liver cirrhosis, the latest stage of liver fibrosis, underlies HCC in approximately 90% of cases. The most frequent causes of liver cirrhosis are hepatitis B and C, chronic alcohol consumption, and nonalcoholic steato-hepatitis. Only 30% of the cases are accessible for curative treatment. In advanced stage, the only approved drug for HCC is sorafenib, a multikinase inhibitor targeting the vascular endothelial growth factor receptor, the platelet-derived growth factor receptor, and Raf. However, its efficacy is modest with a median overall survival of 10.7 versus 7.9 months with placebo in the pivotal phase III trial (2). Therefore, new treatment options with improved therapeutic efficacy are urgently needed.

Immunohistochemical and genomic studies indicate that the PI3K/AKT/mTOR signaling pathway is activated in approximately 50% of patients with HCC and cirrhosis of any cause (3–5). This pathway is divided into two unique complexes with distinct regulations and activities: mTOR complex 1 (mTORC1) and mTOR complex 2 (mTORC2). The serine/threonine kinase AKT is an important upregulator of mTORC1, which is involved in diverse cellular functions such as lipogenesis, energy metabolism, and lysosome biogenesis, and is a key actor in the control of protein synthesis (3, 4, 6).

This underlines that AKT is an essential player in liver tumorigenesis and progression, therefore making it a potential target in the management of HCC. Thus, we postulate that therapy with an AKT inhibitor capable of inhibiting the PI(3)K/AKT/mTOR pathway will be effective in treating fully developed HCC.

However, in order to identify specific adverse effects that could be related to the background of cirrhosis, inhibition of AKT should be preclinically tested in an appropriate animal model. Indeed, sorafenib antitumor efficacy is tested in xenograft mice models in more than 90% of cases which are immunocompromised animal with a normal liver function (7). As HCC develops on a cirrhotic liver with a modified vascularization, a severe fibrosis, and a liver deficiency which can influence drug metabolism, xenograft mice model does not reproduce the most frequent human HCC scenario. One of the models that most faithfully reproduces human HCC physiopathology is the diethylnitrosamine (DEN)-injured–induced rats model which develops an extensive fibrosis, leading to a compensated cirrhosis with a multifocal HCC after 14 weeks of induction (8).

Therefore, in this study, we tested safety and efficacy of a new allosteric AKT inhibitor, ARQ 092 (9), in a DEN-induced cirrhotic rat model with HCC and compared it with sorafenib-treated rats and untreated rats. In addition, we tested the effect of ARQ 092 on four different human cell lines.

Three different human HCC cell lines [Hep3B, Huh7, and phospholipase C (PLC)/PRF/5], one human hepatoblastoma cell line [HepG2; provided by Snorri S. Thorgeirsson (NCI, NHI, Bethesda, MDA) without authentication by the authors], and one rat HCC cell line (HR4) were used in this study [provided by Istvan Blazsek (INSERM U1193, Villejuif, France) without authentication by the authors]. HepG2 is expressing normal p53, while Hep3B is p53-depleted and PLC/PRF/5 and HuH-7 present p53 mutations. Based on COSMIC database, no mutations in AKT were detected in mentioned cell lines. Expression of p-AKT was reported to be normal in Hep3B, whereas it was decreased in HepG2, PLC/PRF/5, and HuH-7 cell lines (5). Rat HR4 cell line was obtained from DEN-induced rat model of HCC (10). All cell lines were tested for mycoplasma infection every 2 weeks by using the MycoAlert Mycoplasma Detection Kit (Lonza). Culture conditions are described in the Supplementary Information.

Preparation of ARQ 092 (ArQule Inc.) and sorafenib (in vitro study: Bay 43-9006, Sigma-Aldrich; in vivo study: Nexavar, Bayer HealthCare) solutions for in vitro and in vivo experiments is described in the Supplementary Information.

Cell viability assay was performed by (3-(4,5-Dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide) (MTT), apoptosis analysis was assessed by flow cytometry analysis, and cell migration was studied by wound-healing assay. All in vitro experiments were realized in 4 different human liver cancer cell lines and in one rat cell line, and protocols are detailed in the Supplementary Information.

Twenty-six 8-week-old Fischer 344 male rats (Charles River Laboratories) were housed in the animal facility of the Grenoble Institute of Neuroscience (INSERM, University of Grenoble-Alpes, France). They were treated weekly with intraperitoneal injections of 50 mg/kg DEN (Sigma-Aldrich, Germany), diluted in pure olive oil in order to obtain a fully developed HCC on a cirrhotic liver after 14 weeks as previously described (8, 11).

After 14 weeks, rats were randomized in 3 different groups as follows: 10 in the ARQ 092 group, 10 in the sorafenib group, and 6 in the control group. Both treatments were dispensed by daily oral gavage during 6 weeks. ARQ 092 was administered for 7 days every other week (for a total of 3 weeks of treatment), at a dose of 15 mg/kg/day as recommended by ArQule Inc., whereas sorafenib was administered every day at a dose of 10 mg/kg/day.

Nutritional state was monitored by daily weighing of rats, and treatment doses were adapted accordingly. Protein-rich nutrition was added to the standard food in every cage where a loss of weight was observed.

All animals received humane care in accordance with Guidelines on the Humane Treatment of Laboratory Animals, and experiments were approved by the animal Ethic Committee.

Imaging study was conducted on a 4.7 Tesla MR Imaging system (BioSpec 47/40 USR, Bruker Corporation) in the Grenoble MRI facility IRMaGE.

All rats were subjected to 3 MRI scans: MRI1 was performed before randomization, MRI2 and MRI3 were respectively done after 3 and 6 weeks of treatment. Morphological analyses were performed on all MRIs, and perfusion study was done on MRI1 and MRI3 scans, both with blinded outcome assessment. Protocols for image acquisition and analysis are detailed in the Supplementary Information.

After the third MRI scan, all rats were euthanized with intracardiac blood sampling for hematologic and biochemical analyses. Each liver was weighed, the number of tumors larger than 1 mm on the surface of the liver was counted, and the diameter of the five largest tumors was measured in a blinded manner. The sum of these five diameters was calculated in order to obtain a histopathologic estimation of the tumor volume.

Histologic analysis of fibrosis and steatosis was performed with collagen staining by sirius red, and lipid staining with Oil Red O staining, respectively.

Tumor proliferation, apoptosis, and tissue vascularization were studied by using anti-Ki67 antibody, TUNEL marker, and anti-CD34 antibody. Protocols are described in the Supplementary Information.

Serum and plasma were taken in order to test biological safety and efficacy parameters as detailed in the Supplementary Information.

Frozen liver fragments (∼50 mg) were digested in 0.15 mL of 3 mol/L alcoholic potassium hydroxide (70°C, 2 hours), diluted 7 times in distilled water. Amount of liver triglycerides was measured by a Triglycerides kit (Erba Mannheim, Czech Republic), and samples' absorbance was measured by spectroscopy at 505 nm.

Western blot analysis of pAKT^(Ser473)/AKT and pERK/ERK, and the real-time qPCR analysis of Ras and AKT pathways downstream actors were performed on tumor and nontumor tissues of each group. The Phospho-Kinase Array Kit (Proteome Profiler Antibody Array; R&D Systems) was used according to the manufacturer's instructions. Experimental protocols are described in the Supplementary Information.

All comparisons of means were calculated by using ANOVA tests with Tukey HSD correction for multiple means comparisons, and independent t tests only when two means were compared. A P value of <0.05 was regarded as statistically significant, and data are presented as mean values ± SEM.

Statistical analyses were performed using IBM SPSS Statistics software, version 20.0 (IBM Corp.), and Prism 6 (GraphPad Software Inc.).

MTT assays showed a drastic decrease in proliferation rate for Hep3B (Fig. 1A), HepG2, Huh-7, PLC/PRF, and HR4 cell lines after ARQ 092 treatment. IC₅₀ were 2 to 6 times lower when compared with sorafenib, suggesting that ARQ 092 is more potent than sorafenib. The calculated IC₂₀ and IC₅₀ values are summarized in Supplementary Table S1.

Next, we examined whether growth arrest due to ARQ 092 treatment was associated with enhanced apoptosis. Compared with control cells, we observed significant dose-dependent decrease of cell viability in all tested HCC cell lines treated with ARQ 092 or sorafenib (Fig. 1B and Supplementary Fig. S1).

We next investigated whether ARQ 092 affected migratory behavior of human HCC cell lines by wound-healing assay. After 24 hours, both IC₂₀ and IC₅₀ of ARQ 092 strongly reduced migration of Hep3B, while sorafenib had significant effect at IC₅₀ only (Fig. 1C). In other cell lines, ARQ 092 decreased migration similarly to sorafenib (Supplementary Fig. S2A). Although control Hep3B cells almost recovered the wound by 72 hours, ARQ 092–treated cells had their wound area unhealed (Supplementary Fig. S2B). Similarly, ARQ 092 treatment decreased cell velocity (Supplementary Fig. S3A) and strongly reduced cell invasion (Supplementary Fig. S3B).

These results demonstrate that ARQ 092 suppresses proliferation and migration and promotes apoptosis in all tested cell lines.

At the end of the study, the mean weight loss was 5.8% ± 5.5% in the sorafenib group and 0.8% ± 0.6% in the ARQ 092 group compared with a gain of 5.9% ± 3.1% in the control group (P = 0.164; Table 1). Blood sample analyses (Table 1) revealed better liver function in the ARQ 092 and sorafenib groups compared with control, with a significantly lower total bilirubin level (ARQ 092: P = 0.0007, sorafenib: P = 0.0002). Albumin level was significantly higher in ARQ 092 compared with nontreated rats (P = 0.0170) and sorafenib group (P = 0.0098). There was no statistical difference in transaminases, alkaline phosphatase and gamma-glutamyl transpeptidase levels, but serum levels of AFP were significantly decreased by ARQ 092 treatment compared with control (P = 0.0328). Glucose, cholesterol, and triglyceride blood concentrations were similar to the control group. Assessment of triglycerides in liver and Oil Red O staining did not show any significant difference between groups (P = 0.467 and 0.355; Table 1; Supplementary Fig. S4A). Therefore, our results showed that ARQ 092 treatment does not interfere with lipid metabolism and improves liver function.

On MRI1 (n = 26), tumor sizes were comparable between groups with 21.3 ± 1.7 mm, 18.0 ± 1.2 mm, and 20.6 ± 2.0 mm in the control, sorafenib, and ARQ 092 groups (P = 0.424), respectively. As illustrated by Fig. 2A, on MRI2 (n = 24), tumor progression was significantly reduced in the sorafenib (+28.5% ± 3.0%; P < 0.0001) and ARQ 092 (+20.9% ± 3.8%; P < 0.00001) groups compared with control (+69.6% ± 9.0%). No statistical difference was found between sorafenib and ARQ 092 groups (P = 0.45). On MRI3 (n = 22), tumor progression rate was +57.0% ± 8.1% in the ARQ 092 group compared with +80.2% ± 9.3% in sorafenib group (P = 0.273) and +155.3% ± 16.0% in the control group (P < 0.0001; Fig. 2A).

These observations were further confirmed by macroscopic examination of the liver (Fig. 2B), which revealed a tumor size of 28.8 ± 1.8 mm in the ARQ 092 group compared with 37.9 ± 3.1 mm in the sorafenib group (P = 0.092) and 62.7 ± 4.4 mm in the control group (P < 0.0001).

Rats from the group treated with ARQ 092 also displayed a significantly lower number of tumors (53.9 ± 7.0 tumors) when compared with sorafenib-treated animals (96.3 ± 13.5 tumors, P = 0.021) and controls (96.8 ± 9.4 tumors, P = 0.031).

Expert pathologic analysis with hematoxylin and eosin staining of liver section confirmed that tumors were HCC with several degree of tumor differentiation (Supplementary Fig. S3B).

Ki67 and TUNEL immunostaining (Fig. 2C; Supplementary Fig. S4C) showed that only ARQ 092 significantly decreased proliferation (41.1% ± 13.3% of control, P = 0.042) and induced apoptosis (148.6% ± 7.7% of control, P = 0.045), whereas sorafenib showed no statistical significance concerning these parameters (Ki67: 56.9% ± 19.6% of control, P = 0.160; TUNEL: 144.2% ± 16.5% of control, P = 0.072).

qPCR analyses of alpha fetoprotein (AFP) expression (Fig. 2D) revealed that only ARQ 092 significantly reduced expression by 96.6% ± 0.8% compared with control (P = 0.038), whereas sorafenib reduced AFP expression by 72.4% ± 20.5% without statistical difference (P = 0.163). Similarly, in ARQ 092–treated rats, serum levels of AFP were significantly decreased by 61% of control (P = 0.041; Table 1).

Therefore, ARQ 092 significantly reduces tumor progression and proliferation in DEN-induced HCC, and has a higher antitumor effect than sorafenib.

Antiangiogenic effect of treatments was assessed with dynamic contrast-enhanced (DCE) MRI as described in the Supplementary Information and illustrated in Fig. 3A. At baseline (MRI1), tumor enhancement was comparable between the groups (P = 0.732; Fig. 2B).

On MRI3, tumor enhancement was significantly different between groups (P = 0.013). ARQ 092 induced a lower tumor enhancement with 64.2% ± 8.5% of control (P = 0.114) and 54.4% ± 7.1% of sorafenib (P = 0.010; Fig. 3B). In each group of treatment, comparison between baseline and the end of the treatment (MRI1 and MRI3) revealed that only ARQ 092 treatment was associated with a significant decrease of tumor enhancement (P = 0.012; Fig. 3C).

Tumor vascularization was also studied by using a rat-specific anti-CD34 antibody to perform immunofluorescence staining of liver tissues. Although structural abnormalities of the tumor vasculature were numerous in control animals, normalization of vasculature was observed in both treated groups (Fig. 3D). Quantification of vascular density revealed that sorafenib decreased vascular density by 46% (P = 0.0008) and ARQ 092 by 68% (P < 0.0001) compared with nontreated rats (Fig. 3E). Thus, MRI results and CD34 staining proved that treatment by ARQ 092 leads to vascular normalization and inhibition of tumor angiogenesis.

As shown in Fig. 4A and B, collagen accumulation assessed by sirius red staining was significantly reduced in the ARQ 092 group compared with the control group (P = 0.001) and with the sorafenib group (P = 0.021). Difference between the control and the sorafenib groups was not significant (P = 0.348). No effects of treatment were observed concerning fibronectin levels (Supplementary Fig. S4D).

Improvement of liver fibrosis by ARQ 092 treatment was confirmed by qPCR analysis (Fig. 4C). The expression of fibrosis markers was downregulated in nontumor liver samples of the ARQ 092 group compared with the control group with significant differences for actin alpha (ACTA)1 (31.7% ± 10.9% of control, P = 0.029) and collagen 1 (9.9% ± 2.9% of control, P = 0.007). Accordingly, tissue inhibitor of metalloproteinases-1 (TIMP-1) was significantly decreased by ARQ 092 treatment compared with control, and matrix metalloproteinase MMP9 was upregulated.

No significant difference was observed for TGFβ1 (40.1% ± 15.9% of control, P = 0.115). For sorafenib group, collagen 1 and TIMP-1 were the only significantly downregulated fibrosis markers.

Overall, ARQ 092 significantly decreased hepatic collagen deposition and improved liver fibrosis in DEN-induced cirrhotic rat model of HCC.

Western blot analyses showed that ARQ 092 treatment completely blocks phosphorylation of AKT^(Ser473) in all human HCC cell lines at both IC₂₀ and IC₅₀ concentrations (Supplementary Fig. S5). Immunofluorescence staining of p-AKT on liver tissues confirmed these results (Supplementary Fig. S6).

Accordingly, ARQ 092 inhibited phosphorylation of AKT^(Ser473) in both tumor and nontumor liver tissues (Fig. 5A and B), with a pAKT/AKT ratio of 29.5% ± 2.27% of control (P = 0.002) in tumor samples and 17.2% ± 2.33% of control (P = 0.034) in surrounding liver samples. Interestingly, sorafenib treatment significantly increased the pAKT/AKT ratio in tumor samples (P < 0.0001) compared with the control group. By profiling kinase phosphorylation (Supplementary Table S2), we found that the levels of phosphorylated mTOR, proline-rich Akt/PKB substrate 40 kDa (PRAS 40), PLCγ1, and Ribosomal protein S6 kinase (S6K1) were significantly decreased in tumor tissues after the ARQ 092 treatment compared with the control (Fig. 5C). As expected, qPCR analyses did not show a significant difference in AKT gene expression, but confirmed that ARQ 092 downregulates AKT pathway downstream actors such as mTORC1 (44.2% ± 11.4% of control, P = 0.005) or S6K1 (54.6% ± 11.9% of control, P = 0.142), as shown in Fig. 5D.

Regarding the ERK pathway, Western blot analyses did not show significant differences in the pERK/ERK ratio between the groups. Accordingly, we observed no differences between the groups in gene expression of ERK in tumor samples. Interestingly, the gene expression of ERK was downregulated in nontumor tissues of both ARQ 092 and sorafenib-treated groups compared with the nontreated group (P = 0.029 and 0.039).

In this study, ARQ 092 showed antitumor, antiangiogenic, and antifibrotic effects with significantly better efficacy than sorafenib in terms of tumor number, as well as tumor contrast enhancement, and the level of liver fibrosis. In vitro, ARQ 092 was also highly efficient in HCC cell lines with a 2 to 6 times more potent effect on cell viability than sorafenib.

ARQ 092 was easily managed in rats with a mean weight loss of only 0.8% at the end of the study. The most frequent side effects of mTOR inhibitors are diabetes and hyperlipidemia. In our hands, with ARQ 092, there was no increase in glucose, cholesterol, and triglyceride blood levels as well as liver cholesterol and triglyceride levels compared with control- and sorafenib-treated rats. The dose strategy for ARQ 092 for in vivo study was based on a previous toxicity study (data provided by ArQule Inc.). The “one week on/one week off” schedule probably contributed to the good tolerability of the tested regimen.

Previous publications have demonstrated the effect of sorafenib on HCC in noncirrhotic rats with a good tolerability at doses between 10 mg/kg in association with another drug (12) and 30 mg/kg when given alone (13–15). In a previous pilot study, we tested a 20 mg/kg sorafenib dose in our cirrhotic rat model with HCC, but due to an important weight loss and other symptoms, after first days of sorafenib administration, we had to stop the study. Therefore, in this study, we have chosen 10 mg/kg for sorafenib. This underlines that HCC new drugs have to be tested in cirrhotic animal models to better assess their side effects that can be very different between noncirrhotic and cirrhotic patients.

Another particularity of this study is the demonstration of the kinetic of tumor progression through three sequential MRI scans per rat. The dramatic increase of tumor size after 6 weeks in control rats (+155.3% ± 16.0%) confirmed the high level of aggressiveness of the DEN model. Tumor progression between MRI1 and 3 was significantly reduced in both groups of treatment compared with the control. There was no statistical difference between ARQ 092 and sorafenib groups possibly because of a type 2 error.

Similarly, according to histologic examination, both sorafenib and ARQ 092 significantly reduced the tumor size compared with the control, but only ARQ 092–treated rats displayed a significantly lower number of tumors. This suggests that ARQ 092 inhibits tumor initiation. To be confirmed, this hypothesis needs further experiments with an earlier introduction of ARQ 092 during the DEN-induction phase like it was done for erlotinib (11).

Our in vivo and in vitro analyses confirmed that the ARQ 092 treatment strongly and selectively affects the AKT pathway. In fact, ARQ 092 is a highly selective allosteric inhibitor that suppresses pan-AKT activity by blocking its phosphorylation and by preventing the inactive form from localizing into plasma membrane which together leads to strong and specific downregulation of downstream targets of AKT (9). Such high specificity was missing in action of catalytic AKT inhibitors that have been previously developed (16). Besides, as sorafenib, ARQ 092 plasma protein binding is very high, around 99% in both rat and human (data provided by ArQule Inc.). Despite this fact, in vitro study showed that ARQ 092 IC₅₀ and IC₂₀ were lower than sorafenib's ones, suggesting a higher efficacy of this new drug.

In sorafenib-treated rats, the absence of downregulation of the ERK pathway on qPCR and Western blot analyses can be surprising, as it has been previously shown that sorafenib downregulates pERK in rat HCC (12). Nonetheless, as DEN induces a strongly aggressive type of HCC, multiple resistance mechanisms have probably already been developed in this model. The overexpression of pAKT in this group is a surrogate marker of such resistance.

Thus, despite difficult conditions with an aggressive model of cancer in cirrhotic rats, ARQ 092 showed its efficacy in controlling tumor progression and demonstrated a good safety profile that makes this experimental drug promising in the treatment of HCC in cirrhotic patients. The results presented here also confirm the importance of targeting AKT in HCC development and progression.

G. Abbadessa has ownership interest (including patents) in ArQule. No potential conflicts of interest were disclosed by the other authors.

Conception and design: G.S. Roth, Z.M. Jilkova, G. Abbadessa, Y. Yu, T. Decaens

Development of methodology: G.S. Roth, Z.M. Jilkova, T. Decaens

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): G.S. Roth, Z.M. Jilkova, A.Z. Kuyucu, K. Kurma, S.T. Ahmad Pour, B. Busser, T. Decaens

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): G.S. Roth, Z.M. Jilkova, A.Z. Kuyucu, K. Kurma, S.T. Ahmad Pour, G. Abbadessa, Y. Yu, B. Busser, V. Leroy, T. Decaens

Writing, review, and/or revision of the manuscript: G.S. Roth, Z.M. Jilkova, A.Z. Kuyucu, G. Abbadessa, Y. Yu, B. Busser, P.N. Marche, V. Leroy, T. Decaens

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): G.S. Roth, G. Abbadessa, P.N. Marche, T. Decaens

Study supervision: T. Decaens

We thank Sylvain Andrieu, Pauline Durrenbach, Manon Leportier, and Fabien Mehr (Animal platform, Grenoble Institute of Neurosciences, Grenoble, France) for the quality of their animal care and their crucial help in oral gavage, and Vasile Stupar (MRI platform, Grenoble Institute of Neurosciences, Grenoble, France) for his help in the realization of MRI acquisitions.

Grenoble MRI facility IRMaGe was partly funded by the French program “Investissement d'Avenir” run by the ‘Agence Nationale pour la Recherche' and the grant ‘Infrastructure d'avenir en Biologie Santé’ - ANR-11-INBS-0006.

Professor Nathalie Sturm, Head of the Pathology Department of Grenoble-Alpes University Hospital.

Dr. Istvan Blazsek for DEN rat cell lines.

T. Decaens was awarded an AFEF (Association Française pour l'Etude du Foie) research bursary in 2013.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.