Targeting insulin-like growth factor axis in hepatocellular carcinoma

Both IGF-1 and IGF-2 bind to IGF-1 receptor 1(IGF-1R), a tyrosine kinase that is structurally similar to insulin receptor (IR) (Figure 1). After IGF ligand binding, the β subunit of IGF-1R undergoes conformational change which causes autophosphorylation of its own tyrosine kinase domain, which leads to the full activation of IGF-1R. IGF-1R induces anti-apoptosis and increases tumor cell mobility. The anti-apoptotic property of IGF-1R was shown in its response to p53, the tumor suppressor gene that promotes apoptosis. Wild type p53 expression inhibited the gene expression of IGF-1R, while mutant p53 increased the gene expression of IGF-1R [16, 29]. Oncogenes such as Src kinase and Akt kinase both stimulated the gene expression of IGF-1R, providing more evidence that IGF-1R is vital in carcinogenesis [16, 29‐31]. In addition, IGF-1R also stimulates cell mobility, as demonstrated by its activity in melanoma cell lines [32].

Another important role of IGF-1R in carcinogenesis is its ability to transform and maintain the transformed phenotype [33]. Mouse embryo fibroblasts possess an extremely strong tendency to spontaneously transform in culture without any additional factors [33], which was no surprise given IGF-1R overexpressed in mouse embryo fibroblasts led to transformation [33‐36]. However, when IGF-1R gene in mouse embryo fibroblasts was disrupted, these fibroblasts failed to transform, even in the presence of the most potent oncogenes such as SV40 T antigen, Ha-ras oncogene and activated c-Src [33, 37‐41]. An even more noteworthy observation was that when IGF-1R was reintroduced, these mouse embryo fibroblasts again restored their ability to rapidly transform.

IGF-1R is required for anchorage independent growth, and inhibition of IGF-1R causes apoptosis without toxicities in vivo. Human prostate cancer cells usually form anchorage independent growth, however; when IGF-1R was abolished, these cells failed to grow in culture, and the same model showed no tumor formation in mice [33, 42‐45]. These observations indicate that IGF-1R is an essential requirement for anchorage independent growth, a pattern common in cancer cell proliferation. In animal models with transformed tumors where IGF-1R was overexpressed, strategies that caused IGF-1R downregulation such as antisense against IGF-1R produced profound tumor apoptosis and massive reductions of metastases [46, 47]. Interestingly, IGF-1R is not required for normal cell growth, as its absence provided no growth inhibition on monolayer cell culture [45], eluting to the possibility that anti-IGF-1R strategies could produce minimal side effects on normal tissues.

One of the key regulators of IGF expression is the family of IGF Binding Proteins (IGFBPs) [6]. The predominant form of IGFBPs is IGFBP-3, which comprises of 90% of all IGFBPs in serum [15, 16], and it binds to the majority of circulating IGF-1 and IGF-2. IGFBPs that include IGFBPs 1, 3, 4 and 6 usually limit IGF access to IGF-1 receptor, therefore decrease the availability of IGFs and diminish their effects on cancer progression [6].

Other IGFBPs such as IGFBP-2 and 5 seem to increase the bioavailability of IGF ligands, therefore play an opposite role of IGFBP-3 [6]. Both in vitro and in vivo evidence support the observation that antisense strategy targeting IGFBP-2 or 5 decreases neoplastic growth [6, 52].

In human HCC tissues, IGF-1 mRNAs were expressed at lower levels than the surrounding normal liver tissues [18, 53]. This could be related to the observation that growth hormone receptor level was low in HCC tissues [18, 53], and growth hormone stimulation thus was low, and the downstream signals such as IGF-1 level would be accordingly low.

IGF-2 has been reported to be overexpressed in animal models of hepatocarcinogenesis and in human HCC [50, 54‐60]. IGF-2 has been linked to carcinogenesis by providing a stimulatory effect on cell proliferation and angiogenesis, both critical in HCC development. In a study using 2 human HCC cell lines, high levels of IGF-2 were demonstrated, and anti-sense oligonucleotides used to target IGF-2 mRNA showed reduction of IGF-2 mRNA and protein levels, which corresponded to a remarkable decrease in cell proliferation [18, 61]. In a study of molecular profiling of human HCC samples, overexpression of IGF-2 was related to a cluster of gene signature that downregulates apoptosis [62], indicating a potent anti-apoptotic effect of IGF-2. The relationship between IGF-2 and angiogenesis was demonstrated in human HCC cell cultures. Under hypoxia environment, IGF-2 mRNA levels in human HCC tissue increased, and IGF-2 overexpression directly increased vascular endothelial growth factor (VEGF) mRNA and protein levels [63]. It suggested a pro-angiogenic effect of IGF-2, an important pathway in HCC development and metastasis.

In rodents, diethylnitrosamine (DEN) induced 100% development of glycogen rich hepatic lesions, which are precursors to HCC, and up to 98% of such lesions expressed IGF-2 mRNA [54, 64]. These results highlight a vital role of IGF-2 early in hepatocarcinogenesis. The expression of IGF-2 has been shown to be a common pathway leading to hepatocarcinogenesis regardless of the species or the process of HCC development [65]. In transgenic mice where IGF-2 levels were persistently 20 times higher than normal control mice, a diverse spectrum of tumors were seen at a much higher frequency than the controls, and HCC was the most common malignancy by 18 months of age [66].

The expression of IGF-2 is very unique in fetal development, as it is maternally imprinted; therefore it is monoallelic [6, 18]. In adults, IGF-2 becomes biallelic [18]. In fact, IGF-2 overexpression in HCC showed re-emergence of fetal IGF-2 by the identification of fetal promoter activation [67]. In all 15 samples of human HCC tested in a study, the overexpression of maternally imprinted fetal IGF-2 was demonstrated [51]. In a study from Hong Kong, 30 HCC samples from patients examined using northern blot analysis showed more than 93% of the adult promoter IGF-2 transcripts were repressed, while 93% of the adult type IGF-2 transcripts were detected in nontumourous tissues [68, 69].

The importance of IGF-2 in HCC development is further demonstrated in its relationship with risk factors of HCC such as hepatitis B and C [70]. In patients with chronic hepatitis C and cirrhosis, the overexpression of IGF-2 was clearly related to hepatitis C viral replication [71]. In patients with chronic hepatitis B, HBV X protein stimulated IGF-2 expression by binding to the fetal promoter of IGF-2, therefore directly stimulating fetal transcript expression of IGF-2 in HCC [72]. Furthermore, aflatoxin has been shown to be synergistic with hepatitis B in the carcinogenesis of HCC, and p53 gene mutation induced by aflatoxin increased the expression of IGF-2 in HCC patients with hepatitis B infection [73].

In a study where 10 HCC cell lines (including PLC HCC cell line) were tested, all of them showed elevated IGF-1R mRNA [50]. Furthermore, the addition of both IGF-1 and IGF-2 to the PLC HCC cell line induced increased cell proliferation in a dose dependent manner, showing that the major tumor promoting effects of IGF ligands on HCC are exerted through IGF-1R [46].

In a model utilizing pancreatic islet transplantation into the livers of diabetic rats, a well established series of events led to development of HCC from preneoplastic foci [74]. When HCC developed from preneoplastic foci in this animal model, the expression of IGF-1R significantly increased, which could explain the phenomenon that the increase in mitotic activity was more than the increase in the rate of apoptosis [18, 75]. IGF-1R is therefore crucial in both the development of and the growth of HCC, making IGF-1R an ideal target in the treatment of HCC.

IGF-2R is closely associated with transforming growth factor β (TGF-β), a very potent growth inhibitor [76]. For instance, in human HCC tissues, the levels of both TGF-β and IGF-2R protein were reduced compared to those in adjacent normal liver tissues [66]. The expression of IGF-2R was significantly lower in several HCC cell lines in vitro, in HCC animal models and in human HCC tissues [77, 78]. The role of IGF-2R in IGF axis appears to serve as a site for IGF-2 clearance, therefore reduces the availability of a potent ligand for IGF-1R, the major gateway for carcinogenesis, tumor growth and proliferation. IGF-2R therefore provides an indirect inhibitory effect on IGF-1R.

The overexpression of IRS-1 has been described in human HCC cell lines and tissues [79]. IRS-1 leads to activation of downstream mitogens such as PI3K and MAPK. In human HCC cell lines, IRS-1 developed acquired resistance to apoptosis, indicating a potent role of IRS-1 in the promotion of continued cell growth in HCC [79].

IRS-2 is a major downstream signal of insulin pathway in the liver, and its function in hepatocarcinogenesis is demonstrated in animal models. When SV40 large T antigen or DEN was applied in murine models, IRS-2 overexpression was detected in both preneoplastic foci and HCC lesions, with higher levels in HCC nodules [47]. A similar observation was reproduced in human HCC cell lines and tissue specimens, suppression of IRS-2 levels led to increased apoptosis. Together with IRS-1, IRS-2 also contributes to hepatocarcinogenesis, as demonstrated by its early emergence in preneoplastic lesions, and its anti-apoptotic property. IRS-1 and 2 therefore create an optimal environment for HCC growth.

In a study comparing IGFBP-3 levels in human normal liver, cirrhotic liver and HCC, the expression of IGFBP-3 mRNA levels was significantly reduced in HCC [80]. In a human HCC cell line, addition of exogenous IGFs stimulated mitosis, but this mitogenic effect was greatly reduced by IGFBP-3 [46]. Furthermore, addition of recombinant human IGFBP-3 induced growth inhibition of the human HCC cell lines HepG2 and PLC [81]. The role of IGFBP-3 on tumor growth inhibition can be further explained by IGFBP-3's induction by p53, a tumor suppressor gene essential in apoptosis and cell cycle arrest [15].

In a study examining radiation induced HCC mouse model, northern analysis showed decreased expressions of IGFBP-7 (a low affinity IGFBP) in HCC compared to normal liver tissues, which was inversely related to anchorage-independent growth in HCC cell lines [82]. A similar trend of reduced IGFBP-7 level was seen in human HCC tissues. When IGFBP-7 cDNA was injected to radiation induced HCC mouse model, the volume of HCC was greatly reduced. IGFBP-7, although has relatively low affinity toward IGF-1 and IGF-2, exerts a similar anti-tumor effect as its high affinity IGFBP counterpart IGFBP-3.

Metalloproteinase belongs to IGFBP proteases that degrade IGFBP-3. In a transgenic murine HCC model overexpressing the inhibitor of metalloproteinase (TIMP1), IGFBP-3 degradation was reduced, and serum level of IGFBP-3 was subsequently increased, which decreased the bioavailable IGF-2 ligand and its downstream signalling. This resulted in reduced liver hyperplasia, despite the activation of IGF-2 by a strong oncogene such as SV40 T antigen [83]. It provided evidence that IGFBP proteases and IGFBPs are equally important in the regulation of IGF ligand bioavailability and their downstream effects on IGF axis activation.

The majority of anti-IGF strategies focused on IGF-1R, the key component of IGF axis that provides mitogenic signal for tumor growth. The most common strategy utilized is the receptor-specific antibodies. For instance, pharmacodynamic studies of MK-0646 (Merck) on neoplastic tissues demonstrated reduction of phosphorylated AKT and phosphorylated S6 kinase, two downstream targets of IGF-1R. MK-0646 also decreased tumor proliferation as shown by reduction in the proliferation marker Ki67 [89, 90]. This observation provided a rationale to use this class of antibodies in the treatment of HCC, and it was supported by additional data generated using IMC-A12 (Imclone), a human monoclonal antibody that blocks IGF-1R, both in vitro and in vivo [91]. In hepatoma cell lines, 2 hour incubation with IMC-A12 completely blocked downstream signalling of IGF-1R as shown by the suppression of phosphorylated AKT and phosphorylated S6 kinase [91]. In addition, 10 day treatment with IMC-A12 in HCC xenografts led to 40% reduction of tumor volume and 40% prolongation of overall survival without additional toxicity compared to control animals [91]. In a phase I study of refractory solid tumors using IMC-A12 as a single agent, a patient with HCC had stable disease for up to 9 months [92].

One of the most studied IGF-1R antibodies is CP-751871 (Pfizer) and it showed rather promising activity in a phase II study in patients with advanced non-small cell lung cancer. When it was added to carboplatin and paclitaxel as a first line regimen, the response rate increased from 32% to 46%. What was even more impressive was in patients with squamous histology, the response rate was as high as 71% [93]. The most common side effect in this phase II study was hyperglycemia. The subsequent ambitious Phase III study looked at patients with stage IIIB or IV non-small cell lung cancer, and randomized them to receive carboplatin and paclitaxel either with or without CP-751871. This study was halted in late 2009 due to unexpected increase in fatal events in the experimental arm [93], and it could be partially explained by the most common side effect of hyperglycemia. The consequence of IGF-1R inhibition leads to compensatory increase of growth hormone stimulation that promotes liver gluconeogenesis, resulting in hyperglycemia [94]. What we could learn from this surprising result is that there are subsets of patients who could potentially benefit from IGF-1R inhibitors such as CP-751871 [95]. For instance, in the experimental arm, patients with low IGF-1 levels (< 5 pg/ml) before treatment with CP-75871 were more likely to suffer fatal events within 60 days of treatment. The same group of patients also had much shorter median overall survival compared to the ones with higher pretreatment IGF-1 levels (7 months vs. 10.4 months). Conversely, for patients with higher pretreatment IGF-1 levels, those who received the experimental treatment that included CP-751871 had a trend toward higher median overall survival compared to those who received the standard chemotherapy (10.2 months vs. 7 months). Further analysis from the phase II study also showed that IGF-1R was present in the highest level in patients with squamous histology, which could explain the observed high response rate in squamous cell patients who received CP-75871 [96]. Such observation was consistent with a presentation at ASCO GI in 2011, in which data of 288 patients with HCC were analyzed. In this study, pretreatment lower plasma IGF-1 and higher plasma VEGF levels significantly correlated with advanced clinicopathologic parameters and poor overall survival, with an optimal cut off point of 26 pg/mL and 450 pg/mL, respectively. The combination of low IGF-1 and high VEGF predicted median overall survival of 2.7 months compared with 19 months for patients with high IGF-1 and low VEGF (p < 0.0001) [97]. Such information provided insights into the specific patient subsets in HCC where IGF-1 levels would offer additional prognostic significance. Whether baseline plasma IGF-1 levels could be used to predict response to IGF axis inhibition in HCC remains to be explored.

IMC-A12 was studied as a single agent in patients with advanced HCC as a front line systemic therapy. This study unfortunately was terminated due to futility. The pre-planned primary endpoint of progression free survival rate at 4 months was only 30% and median overall survival of 8 months [98]. Up to 46% of patients developed grade 3-4 hyperglycemia, similar to what was seen in the phase II NSCLC study of CP-751871 [93], thus raising the possibility that hyperglycemia could be the dose limiting toxicity of IGF-1R monoclonal antibodies. Hyperglycemia and its subsequent increase of growth hormone could also contribute to the disappointing activity of this class of drugs. BIIB022 (Biogen-Idec) is an anti-IGF-1R monoclonal antibody that blocks binding of both IGF-1 and IGF-2 to IGF-1R [92]. It does not contain Fc effector function, therefore can potentially minimize toxicities in healthy tissues expressing IGF-1R [92]. This agent does not appear to cause hyperglycemia, a common side effect of receptor specific antibodies [92]. Hyperglycemia has been attributed to insulin resistance secondary to high levels of growth hormone, a compensatory reaction to IGF-1R antibodies [94, 95]. The class of IGF-1R monoclonal antibodies share similar side effect profiles, including minimal dose limiting toxicities. These favorable safety profiles make them ideal candidates in the combination therapy with current available chemotherapy or biologic therapy [6]. BIIB022 showed inhibition of tumor growth in HCC cell line HepG2, and this inhibitory effect was enhanced by addition of sorafenib [92], the only FDA approved medication for patients with advanced HCC. A planned phase I/II study comparing sorafenib with or without BIIB022 in patients with advanced HCC was terminated due to a business decision of Biogen-Idec. AVE-1642 (Sanofi-Aventis) is another IGF-1R antibody that was initially studied in advanced HCC patients in a phase I study in combination with sorafenib [99], the study was terminated not related to either efficacy or toxicity concerns. Although IMC-A12 lacks single agent activity in HCC, its combination with sorafenib could potentially yield synergy. It is currently undergoing phase I study in combination with sorafenib in patients with HCC, the result of this clinical trial may help understand the clinical benefits of combining IGFR-1R monoclonal antibodies and sorafenib in HCC.