Targeted approach to metastatic colorectal cancer: what comes beyond epidermal growth factor receptor antibodies and bevacizumab?

Insulin-like growth factor receptor inhibitors

The insulin-like growth factor (IGF) family consists of two ligands (IGF-I and IGF-II), two cell membrane receptors (IGF-IR and IGF-IIR), six IGF binding proteins (IGFBP1–6) and at least five adaptor proteins (insulin receptor substrate IRS1–4 and Shc), and other associated proteins including the insulin-receptor-related-receptor IRR, IGFBP-related proteins and IGFBP proteases [Guo et al. 1992]. IGF-I and -II are potent growth factors with a close structural similarity with insulin. IGF-I mediates growth in the adult, whereas IGF-II is expressed in the developing foetus. Both peptides are produced in the liver and in the gastrointestinal tract and act via the IGF-IR, by a number of mechanisms [Rouyer-Fessard et al. 1990]. IGF-IR is widely expressed throughout the gastrointestinal tract with the highest expression detected in the proliferating cells at the bases of the colonic crypts, whereas the type II IGF receptor has an unknown physiological role. IGF-I is a potent mitotic, antiapoptotic peptide and, in addition, has been demonstrated to promote angiogenesis by increasing the production of VEGF [Akagi et al. 1998; Bustin et al. 2002; Fukuda et al. 2002; Warren et al. 1996; Wu et al. 2002]. IGF-I can also promote cell invasion and motility by a variety of mechanisms and has a significant influence on the interleukin system, inhibiting production of the growth-inhibitory tumour necrosis factor α (TNFα) and increasing production of the antiapoptotic interleukin 8 (IL-8) [Andre et al. 2004; Remacle-Bonnet et al. 2000; Vallee et al. 2003].

The role of the IGF pathway in the pathogenesis of CRC has been demonstrated in different preclinical studies. Wu and colleagues conducted a study involving liver-specific IGF-I-deficient (LID) mice in whose serum the IGF-I level was 25% of that in control mice. In LID mice, the growth of orthotopically transplanted colon adenocarcinoma was significantly less than in controls and the appearance of a palpable caecal tumour was not only slower but also smaller than that in control mice. The frequency of hepatic metastasis was also significantly lower in LID mice. These results support the hypothesis that circulating IGF-I levels play an important role in colon cancer development and metastasis. CRCs often express 10–50 times higher levels of IGF-I and IGF-II than adjacent uninvolved colonic mucosa [Reinmuth et al. 2002]. Even more, CRC cells frequently overexpress IGF-IR and when activated by IGF-I they inhibit apoptosis and allow progression through the cell cycle [Ouban et al. 2003].

Significant cross talk has been observed between the IGF pathway and several other receptors. There is a crucial interaction between the IGF and EGF pathways. EGF is able to stimulate IGF-II and vice versa. In addition, EGF can suppress the expression of IGFBP3 and increase the ability of free IGFs. This interaction provides a rationale for combined therapy against these different pathways with the goal to improve antitumour activity [Burtrum et al. 2003]. Several MoAbs and small inhibitors of IGF-R have recently entered into clinical development (Table 1). IMC-A12, a fully human MoAb currently in clinical development, binds to IGF-IR with high affinity and inhibits ligand binding with a half maximal inhibitory concentration (IC50) of 0.6–1 nM [Burtrum et al. 2003]. In xenograft tumour models, IMC-A12 results in significant growth inhibitions of breast, pancreatic and colon tumours [Burtrum et al. 2003]. The safety and efficacy of IMC-A12 has been evaluated in a phase II randomized open-label study with or without cetuximab in patients with metastatic refractory anti-EGFR MoAb CRC [Reidy et al. 2010]. In this trial, IMC-A12 alone or in combination with cetuximab did not demonstrate meaningful antitumour activity and of 64 patients treated, only one patient achieved a durable partial response to the combination treatment. The limited clinical efficacy of IMC-A12 in this study suggests that additional preclinical work will be required to identify predictors of IGF-IR dependence in CRCs. Another fully humanized MoAb directed against IGF-IR, CP-751,871, is in advanced clinical development. In experimental models, it inhibits IGF-I binding to cells, inhibits IGF-I-induced receptor phosphorylation, and results in downregulation of IGF-IR expression at the plasma membrane through internalization of the receptor.

OPEN IN VIEWER

Table 1.

Membrane receptor targets and drugs.

Target	Category	Drug	Development stage in CRC	Population	Treatment
IGFR	MoAb	IMCA12	Phase II	Refractory	IMCA12 ± cetuximab
		Ganitumab (AMG479)	Phase II	Refractory KRAS wild type	Panitumumab ± AMG479
		Dalotuzumab (MK-0646)	Phase II	Second-line KRAS mutant	FOLFIRI ± ganitumab
			Phase II/III	Refractory	Irinoteccano + cetuximab ± dalotuzumab
	TKI	NVP-AEW541	Preclinical	–	–
c-MET	MoAb	Rilotumumab (AMG102)	Phase Ib/II	Refractory KRAS wild type	Panitumumab ± rilotumumab
	TKI	Tivantinib	Phase I/II	Refractory KRAS wild type	Irinotecan + cetuximab ± tivantinib
TRAIL	MoAb	LBY135	Preclinical	–	–
		CS-1008	Phase II	Second line	Irinotecan ± CS-1008
		Mapatumumab	Phase II	Refractory	Single agent
		Lexatumumab	Phase I	Refractory	FOLFIRI + lexatumumab
		AMG 655	Phase I/II	First line	FOLFOX + bevacizumab ± AMG655
			Phase II	Refractory KRAS wild type	Panitumumab ± AMG655
		HGS-TR2J	Preclinical	–	–
		Apomab	Preclinical	–	–

c-MET, mesenchymal epithelial transition factor; CRC, colorectal cancer; IGFR, insulin-like growth factor receptor; KRAS, Kirsten rat sarcoma; MoAb, monoclonal antibody; TKI, tyrosine kinase inhibitor; TRAIL, tumour necrosis factor-related apoptosis-inducing ligand.

Inhibition of tumour growth has been documented in multiple xenograft models, and its combination with standard chemotherapeutic agents enhances its antitumour efficacy. The combination of CP-751,871 with 5-fluorouracil (5-FU) in a Colo-205 xenograft model resulted in improved antitumour activity compared with either agent given alone [Cohen et al. 2005]. Ganitumab (AMG 479) is a fully human MoAb against IGF-IR and a phase II trial has been presented by Eng and colleagues comparing panitumumab with panitumumab plus ganitumab (AMG 479) and panitumumab plus rilotumumab (AMG 102), an antibody against hepatocyte growth factor (HGF). In preliminary data, ganitumab as a single agent and in pretreated patients failed to generate any signal in this randomized trial [Eng et al. 2011]. In addition, a phase II randomized study with leucovorin, 5-FU and irinotecan (FOLFIRI) plus ganitumab or AMG655 versus FOLFIRI in Kirsten rat sarcoma (KRAS)-mutant mCRC is currently in progress.

Dalotuzumab (MK-0646), another humanized immunoglobulin G1 antibody, has shown efficacy in a xenograft colon cancer model and a favourable toxicity profile in a phase I trial [Scartozzi et al. 2010]. A multicentre, double-blind, randomized phase II/III study of dalotuzumab in mCRC has been conducted [Watkins et al. 2011]. Eligible patients were those whose condition had previously failed to respond to both irinotecan and oxaliplatin and had progressed on or within 3 months of their last therapy. Patients were randomized to receive either dalotuzumab 10 mg/kg weekly or 15 mg/kg loading followed by 7.5 mg/kg every 2 weeks or placebo combined with cetuximab and irinotecan at standard dose/schedule. The trial was stopped at the first interim analysis with 345 wild-type (wt) KRAS patients enrolled as the two dalotuzumab-dosing regimens did not achieve proof of concept. Median progression-free survival (PFS) for all randomized wt KRAS status patients [per intention to treat (ITT)] was 3.3 months in the weekly dalotuzumab group and 5.4 in the 2-weekly group versus 5.6 in the placebo group. Median overall survival (OS) for wt KRAS patients (per ITT) was 10.8 and 11.6 months in the two dalotuzumab groups respectively versus 14.0 in the placebo arm. The addition of dalotuzumab to cetuximab and irinotecan worsened PFS and OS in patients with chemorefractory wt KRAS mCRC. Based on these results a comprehensive, retrospective analysis has been performed to identify possible biomarkers predictive to cetuximab resistance and eventually dalotuzumab responsiveness [Watkins et al. 2012]. In particular, high IGF-I expression was found to be associated with resistance to cetuximab and sensitivity to dalatuzumab. Moreover, dalatuzumab was not effective in tumours with high IGF-II expression. Based on these data dalatuzumab will be further evaluated in a molecularly selected population of mCRC. This phase II/III study with dalotuzumab was performed without selection of patients with mCRC on KRAS status. KRAS-mutated mCRC is inherently resistant to EGFR-targeted therapy and carries an inferior prognosis to wt disease. For this reason the activity of dalotuzumab has been investigated in KRAS-mutant preclinical colon cancer models [Sathyanarayanan et al. 2012]. Moreover the efficacy data relating to a subset of patients with KRAS-mutant mCRC enrolled in a randomized phase II/III study were reevaluated. In preclinical KRAS-mutant models, dalotuzumab was found to be effective in inhibiting IGF-I-mediated cellular growth. Furthermore, in CRC xenograft models high IGF-I was found to identify a subset of dalotuzumab-responsive tumours. Combination studies demonstrated that irinotecan exposure resulted in the activation of IGF-IR and phosphatidylinositol 3 kinase (PI3K) signalling pathways, representing a possible cellular survival mechanism. In xenograft models the combination of dalotuzumab with irinotecan produced lasting tumour growth inhibition that persisted even after treatment withdrawal. Interestingly, efficacy data of 69 patients with KRAS-mutant mCRC enrolled in the randomized study were evaluated based on this preclinical information. Efficacy data demonstrated differential activity between colon and rectal tumours with evidence of activity in the rectal subgroup. Molecular analysis suggests that this observation may be mediated by differential IGF-I expression. These data suggest that dalotuzumab in combination with irinotecan may have utility in the treatment of patients with KRAS-mutated CRC. Small molecules that selectively inhibit the tyrosine kinase domain of IGF-IR without significant effect on the insulin receptor are under development (Table 1). NVP-AEW541, a kinase inhibitor, has shown induction of apoptosis and cell cycle arrest in two CRC cell lines, HT29 and HCT116, resulting in dose-dependent inhibition of proliferation. Combining this agent with either 5-FU or cetuximab resulted in additive growth inhibition. NVP-AEW541 alone inhibited proliferation in primary cancer cell cultures of tumours of eight patients with primary CRC [Hopfner et al. 2006].

The IGF-IR signalling system is now recognized as an important component of cancer development, progression and response to treatment across a wide spectrum of tumour types. Early stage clinical trials suggest that the main anticipated toxicities can be managed effectively. Further work is required to ascertain the efficacy of IGF-IR-targeted therapy in the clinical setting and to elucidate biomarkers of response to treatment. The role of IGF-IR-targeted treatment in CRC, which appears to respond due to autocrine production of IGF-I, likewise needs to be elucidated. The optimal treatment combinations with chemotherapy and other targeted agents and sequence of treatment also need to be established.

c-MET

The mesenchymal epithelial transition factor (MET) gene encodes a high-affinity receptor for the HGF, also known as scatter factor, and consists of an extracellular α chain disulfide bonded to a membrane-spanning β chain [Giordano et al. 1989]. HGF binds to MET, induces receptor homodimerization, and leads to phosphorylation of the cytoplasmic tyrosine kinase domain at two specific sites (Y1234 and Y1235) and activation of MET-mediated signalling [Longati et al. 1994]. These events are essential during embryogenesis and also play an important role for development, homeostasis and tissue regeneration, such as mitogenesis and morphogenesis in a wide range of cell types, including epithelial, endothelial and haematopoietic cells, neurons, melanocytes and hepatocytes [Furge et al. 2000]. Deregulation of the MET pathway has been observed in many human malignancies, and the effects of sustained MET activation have been extensively characterized in preclinical models [Iyer et al. 1990]. In vivo studies have shown that activation of HGF/MET signalling promotes cell invasiveness and triggers metastases through direct involvement of angiogenic pathways [Zhang et al. 2003]. A variety of mechanisms that lead to aberrant MET signalling have been characterized and these include overexpression of HGF or MET, MET gene amplification, mutations or structural rearrangements. Importantly, MET expression has been found in 55–78% of primary colon tumours. Moreover, MET amplification has been detected in patients with CRC, whereas no mutations involving the tyrosine kinase domain have been described in those patients [Umeki et al. 1999]. The prognostic role of HGF and MET has also been examined in several studies. Whereas the majority of these studies show a poor outcome for individuals in whom tumour HGF or MET overexpression was detected [Sawada et al. 2007; Shattuck et al. 2008; Siegfried et al. 1997], a few studies failed to show a prognostic role for MET overexpression [Nakamura et al. 2007; Resnich et al. 2004].

The prevalence of HGF/MET pathway activation in human malignancies has driven rapid growth in drug development programmes. Agents currently under development as HGF/MET pathway inhibitors can be broadly divided into biologicals and low molecular weight synthetic compounds. Biologicals, or protein-based agents, act through a variety of mechanisms and possess target selectivity and pharmacokinetic properties that are predictable and often desirable. Nonetheless, their size typically restricts their action to extracellular events and their complexity impacts drug manufacture, routes of administration and shelf life. Thus, it is not surprising that synthetic, low molecular weight tyrosine kinase inhibitors (TKIs) presently outnumber every other class of HGF/MET therapeutic. Among HGF/MET-targeted biologicals, the most advanced drug candidates are MoAbs directed against either HGF or MET (Table 1). Neutralizing MoAbs against human HGF, such as rilotumumab (AMG102), are currently being developed in patients with mCRC. Rilotumumab in vitro has been shown to bind the HGF light chain with a Kd of 0.22 nM and to block HGF/MET binding with an IC50 of 2.1 nM [Burgess et al. 2006; Giordano, 2009]. In a phase I clinical trial, it was well tolerated up to the planned maximum dose of 20 mg/kg, and the maximum tolerated dose (MTD) was not reached. Treatment-related adverse events (AEs) were moderate in severity, with the most frequently occurring being fatigue, constipation, anorexia, nausea and vomiting [Gordon et al. 2010]. Results of a phase Ib/II study of rilotumumab in combination with panitumumab in patients with wt KRAS mCRC have recently been presented [Van Cutsem et al. 2010]. In the dose-finding portion of the study, the standard dose of panitumumab and 10 mg/kg of rilotumumab every 2 weeks was tested. There were no dose-limiting toxicities reported in the first six patients. That dose was therefore set for the randomized phase II portion of the study. A total of 142 patients with wt KRAS metastatic colorectal tumours previously pretreated with irinotecan or oxaliplatin chemotherapy were enrolled. With a median follow up of nearly 7 months, 31% of patients in the panitumumab–rilotumumab arm showed a partial response, compared with 21% in the panitumumab– placebo arm. The median duration of response with the experimental arm was 5.1 months compared with 3.7 months for panitumumab alone. Median PFS showed a trend for improvement in the combination treatment at 5.2 months compared with 3.7 months in the panitumumab–placebo arm. Adverse events increased in the combination arms, and were within an acceptable range [Van Cutsem et al. 2011].

Several selective and nonselective c-MET kinase inhibitors have been developed and tested in different tumour types (Table 1). Among these, the one that is in clinical development for patients with mCRC is tivantinib, an oral, highly selective, nonadenosine triphosphate (ATP)-competitive c-MET inhibitor. Results from a phase I/II study of tivantinib in combination with irinotecan and cetuximab in patients with wt KRAS mCRC have been presented [Bessudo et al. 2011]. This study is based on the hypothesis that adding tivantinib to irinotecan plus cetuximab may decrease resistance to cetuximab treatment and improve patient outcomes. Patients with locally advanced CRC or mCRC who received more than one prior line of chemotherapy, were wt KRAS and had Eastern Cooperative Oncology Group performance status less than 2 were included in the study. Patients were treated with irinotecan (180 mg/m²) and cetuximab (500 mg/m²) every 2 weeks along with escalating doses of tivantinib (120, 240, 360mg) twice daily. Preliminary toxicity and efficacy data for nine patients showed no dose-limiting toxicities (DLTs) and grade 3/4 AEs included neutropenia (grade 4 in one patient), fatigue (grade 3 in two patients) and one case each of grade 3 leukopenia, acneiform rash, vomiting, diarrhoea, anaemia and syncope. In nine patients with evaluable responses, best responses included one complete response (after four cycles), two partial responses (after two cycles), five stable disease and one progressive disease. The randomized phase II portion of the study continues to accrue data for the recommended phase II dose of 360 mg tivantinib twice daily.

In closing, the wealth of basic knowledge about HGF/MET biology has enabled an accurate assessment of the pathway’s oncogenic potential and provided the insight needed to develop potent and selective inhibitors and use them with relative safety in humans. Patient selection, which is of primary importance, will advance as more robust methods are developed to analyse the many known potential diagnostic biomarkers of pathway activity. Methods that rely on DNA or RNA (e.g. detecting MET gene amplification or mutation) are now faster and more sensitive than those available for quantitating MET protein content and phosphorylation state, but efforts to improve both are under way. Similarly, the need for pharmacodynamic markers that track drug effect and patient response is recognized and clinical pharmacodynamic marker studies currently under way reveal solid candidates. Finally, although the complexity of cancer and the risk of acquired resistance may limit the use of HGF/MET molecular therapeutics as single agents for subgroups of patients, much evidence suggests that pathway involvement is widespread and critical for metastasis. Thus for HGF/MET pathway inhibitors in particular, combinatorial phase II trials with small, carefully selected patient groups may be the most expedient path to more effective cancer treatment.

Tumour necrosis factor-related apoptosis-inducing ligand receptor

The evolution of a cancer cell is dependent on six essential alterations, including self sufficiency in growth signals, insensitivity to growth-inhibitory signals, limitless replicative potential, sustained angiogenesis, tissue invasion and evasion of apoptosis [Hanahan and Weinberg, 2000]. Similar to the cell division cycle, the pathways that lead to apoptosis are complex and consist of a fine homeostatic balance between cell death blockers and inducers [Reed, 1999]. Because apoptosis is a physiological death culminating in fragmentation of cells cleared by phagocytosis, inflammatory reaction or tissue scarring usually does not occur. Defects in apoptosis can prolong cellular lifespan and contribute to neoplastic cell expansion and can create a permissive environment for genetic instability that can contribute significantly to carcinogenesis. The ability to directly induce apoptosis in cancer cells is a novel approach to cancer treatment that has recently begun to be evaluated. One emerging area of research is the evaluation of agents which activate the TNF death receptors [TNF-related apoptosis-inducing ligand (TRAIL)-R1 or DR4 and TRAIL-R2 or DR5], members of the TNF receptor superfamily that, when activated, directly induce programmed cell death in cancer cells. TRAIL induces apoptosis and is a member of the TNF ligand superfamily [Wiley et al. 1995]. Of four receptors identified to date, TRAIL-R1 and TRAIL-R2, upon binding with TRAIL, mediate downstream signalling, leading to apoptosis [MacFarlane et al. 1997]. TRAIL-R3 and TRAIL-R4 have nonfunctional or absent death domains, do not transmit apoptotic signals and may function as decoy receptors [Pan et al. 1997]. Upon binding of TRAIL to functional receptors, TRAIL-R1 and TRAIL-R2 recruit apoptosis-inducing caspases that activate the proapoptotic proteins Bid and Bax, leading to cytochrome C release from mitochondria [Li et al. 1998]. Strater and colleagues evaluated the expression of TRAIL and its cognate receptors on normal colon mucosa, colon adenomas and resected stage II/III colon carcinomas. They found TRAIL-R1 expression to be an independent predictor of disease-free survival, with increased expression associated with improved clinical outcome. In a severe combined immunodeficient mouse xenograft model prepared from fresh surgical colon cancer specimens, administration of TRAIL resulted in growth inhibition [Naka et al. 2002]. Notable synergy was evident with concurrent administration of either 5-FU or irinotecan and TRAIL. Similar results have been noted in xenograft models with established cell lines [Gliniak and Le 1999].

The TRAIL-related pathway has been targeted by agonist MoAbs directed to TRAIL-R1 or TRAIL-R2 or by recombinant variants of the ligand TRAIL itself. Currently one chimeric (LBY135) and six human (CS-1008, mapatumumab, lexatumumab, AMG 655, HGS-TR2J and apomab) monoclonal agonistic antibodies against TRAIL-R1 and TRAIL-R2 are in phase I/II clinical trials. LBY135, lexatumumab, AMG 665, HGS-TR2J and apomab are TRAIL-R2 agonists (Table 1). Generally, results to date indicate some antitumour activity, good safety profile with no hepatotoxicity, linear pharmacokinetics and relatively long half-life times (10–20 days) with no generation of antihuman antibodies [Oldenhuis et al. 2008]. In preclinical studies, lexatumumab has been demonstrated to induce apoptosis in a variety of tumour models, including the xenograft Colo205 model [de Bono et al. 2004]. In phase I studies, lexatumumab was safely administered up to a 10 mg/kg dose in a range of solid malignancies with stable disease observed in 12 patients (32%) that lasted a median of 4.5 months [Plummer et al. 2007]. The safety of lexatumumab in combination with gemcitabine, pemetrexed, doxorubicin or FOLFIRI is also being examined in a phase Ib trial [Wakelee et al. 2010]. Tumour shrinkage has been observed, including confirmed partial responses in the FOLFIRI and doxorubicin arms. More detailed results on patient responses are awaited. AMG655 is a fully human monoclonal agonist antibody that binds human TRAIL-R2. A phase Ib study of AMG655 in combination with FOLFOX6 (Oxaliplatin 85 mg/m2 (IV), Leucovorin 200 mg/m2 (IV), fluorouracil 400 mg2 (IV push), and fluorouracil 2400 mg2 (cont. infusion over 46 hours)) and bevacizumab for the first-line treatment of patients with mCRC has recently been presented [Saltz et al. 2009] and a randomized phase II trial of FOLFOX6 plus bevacizumab with or without AMG655 is currently in progress. Mapatumumab (HGS-ETR1, TRM-1 MoAb) is an agonistic fully human MoAb against TRAIL-R1. Preclinical data demonstrated growth inhibition of human colon cancer xenograft models in nude mice [Tolcher et al. 2007]. Phase I trials in patients with solid malignancies refractory to standard therapy concluded that it can be safely administered up to 10 mg/kg every 14 days. No adverse reactions were observed and out of the 49 patients enrolled, 19 had stable disease with two lasting 9 months. A phase II study of mapatumumab in advanced CRC has recently been published [Trarbach et al. 2010]. This trial was designed to evaluate the efficacy and safety of mapatumumab in patients with mCRC whose condition had failed to respond, who were intolerant, or not candidates for fluoropyrimidine, oxaliplatin and irinotecan-based regimens. No clinical activity of single-agent mapatumumab was observed in patients with advanced refractory CRC; in fact 32% of patients achieved stable disease for a median of 2.6 months with a median PFS of 1.2 months. However, on the basis of its favourable safety profile and preclinical evidence of potential synergy in combination with agents commonly used in the treatment of CRC, further evaluation of mapatumumab in combination with chemotherapy is warranted.

TRAIL is now in phase II clinical trials. Whether used as a single agent or in combination therapy, it is hoped that sooner rather than later, it will contribute to improving patient survival. Many questions remain unanswered. Top of the list is why some cells remain resistant to TRAIL. Could the combination of TRAIL receptor agonists with other therapeutics restore cancer cell sensitivity or inadvertently result in the death of normal cells? Why is it that certain tumour cells preferentially transduce apoptosis through one death-inducing TRAIL receptor but not the other; or preferentially induce the expression of only one TRAIL receptor in response to a given anticancer agent? Unfortunately, we are still unable to predict which TRAIL receptors are functional in which tumours and so, as yet, cannot foresee which TRAIL receptor-targeting treatment would be the most appropriate. Another essential question to ask is whether the differences in the pharmacokinetics and pharmacodynamics of the recombinant humanized TRAIL variants are problematic in vivo. Will a ‘safe TRAIL variant be found with potent antitumour effects? Given time, through translational research and more extensive clinical trials, we envisage that many of these questions will be answered. In the meantime, however, it appears that we are certainly on the right TRAIL.

PI3K/AKT/mammalian target of rapamycin inhibitors

The PI3K/AKT/mammalian target of rapamycin (mTOR) pathway provides an alternative signal transduction cascade with pleiotropic biological effects that are important for the formation and progression of cancer, including apoptosis, transcription, translation, metabolism, angiogenesis and cell cycle progression [Testa and Bellacosa 2001]. PI3K, a heterodimeric protein composed of a catalytic subunit (p110a) and a regulatory subunit (p85), is activated by growth factor receptor tyrosine kinases. The survival mechanism initiated by these proteins is executed through the downstream effectors of this kinase: Akt, mTOR and p70S6 kinase. Upon binding of ligands to various membrane growth factor receptors, the corresponding receptor tyrosine kinases activate PI3K, which results in recruitment of PI3K to the cell membrane and production of phosphatitidylinositol (3,4,5)-triphosphate (PIP3) [Raffioni and Bradshaw, 1992; Thompson and Thompson, 2004]. PIP3 subsequently recruits AKT to the plasma membrane where it is phosphorylated and activated. Phosphatase and tensin homologue dephosphorylates PIP3, thus acting as a negative regulator of PI3K signalling. Activated AKT exerts antiapoptotic activity by preventing release of cytochrome C from mitochondria, inactivating forkhead transcription factors known to induce expression of proapoptotic factors which results in transcription of antiapoptotic genes [Cross et al. 1995]. AKT activation also affects cell cycle progression, through regulation of cyclin D stability, inhibition of p27Kip1 protein levels and regulation of mRNA translation via control of phosphorylation of 4E-BP1 and its dissociation from the mRNA cap binding protein elF4E [Muise-Helmericks et al. 1998; Collado et al. 2000; Sonenberg and Gingras, 1998]. Finally, AKT mediates the activation of endothelial nitric oxide synthase (an important modulator of angiogenesis), promotes tumour invasiveness and enhances telomerase activity [Fulton et al. 1999; Kim et al. 2001; Kang et al. 1999; Vivanco and Sawyers, 2002].

In colon cancer, AKT phosphorylation and PI3K activation directly correlate with tumourigenic potential. PI3K activity is elevated in the vast majority of CRCs compared with normal mucosa, and mutations in the p110 α subunit of PI3K are found in approximately one-third of CRCs [Khaleghpour et al. 2004; Samuels et al. 2004]. Inhibition of PI3K and AKT confers antitumour activity in preclinical models [Meuillet et al. 2004; Ihle et al. 2004]. Further support for the development of AKT inhibitors in CRC is provided by a clinical study in patients treated with an EGFR inhibitor, in which responses were restricted to those tumours that showed complete abrogation of AKT phosphorylation after treatment. Despite the appeal of AKT as a target, however, development of specific AKT inhibitors has been challenging and candidate drugs for clinical trials have only recently become available [van Ummersen et al. 2004]. Several PI3K inhibitors are in clinical development, such as BEZ235 [Maira et al. 2008], SF1126 [Garlich et al. 2008], BGT226, XL147, BKM120, GDC-0941 and XL765 (Table 2). Other PI3K inhibitors are still in preclinical development, such as ZSTK474 [Maira et al. 2008; Garlich et al. 2008].

OPEN IN VIEWER

Table 2.

Downstream intracellular signalling targets and drugs.

Target	Drug	Development stage in CRC	Population	Treatment
PI3K	BEZ235	Phase I	Refractory	MEK162 + BEZ235
	SF1126	Phase I	Refractory	Single agent
	BGT226	Phase I	Refractory	Single agent
	XL147	Phase I	Refractory	XL147 + MSC1936369B
	BKM120	Phase I	Refractory	BKM120 + irinotecan
	GDC-0941	Phase I	Refractory	Single agent
	XL765	Phase I	Refractory	Single agent
mTOR	Temsirolimus (CCI-779)	Phase I	Cetuximab -	Temsirolimus + cetuximab
	RAD001	Phase I	Refractory	RAD001 + OSI906
		Phase II	Refractory	RAD001 + irinotecan +
		Phase Ib/II	Refractory	cetuximab
		Phase I	Refractory	Irinotecan + cetuximab + RAD001
	AP23573	Phase I	Refractory	5-Fluoruracile + RAD001
			Refractory	Bevacizumab + AP23573
Src	Bosutinib (SKI-606)	Phase I	Refractory	Capecitabina + bosutinib
	SU6656	Preclinical	–	–
	AP23464	Preclinical	–	–
	Dasatinib (BMS-354825)	Phase I	Refractory wild type	Dasatinib + FOLFOX + cetuximab
	Saracatinib (AZD0530)	Phase II	Refractory	Single agent
Aurora kinase	MK0457	Phase I	Refractory	Single agent
	PHA-739358	Phase I	Refractory	Single agent
	MLN8054	Phase I	Refractory	Single agent
	MLN8237	Phase I	Refractory	Single agent
	MK0457	Phase I	Refractory	Single agent

CRC, colorectal cancer; FOLFIRI, leucovorin, 5-fluorouracil and irinotecan; FOLFOX, 5-fluorouracil, leucovorin, and oxaliplatin; mTOR, mammalian target of rapamycin; PI3K, phosphatidylinositol 3 kinase.

Another potential therapeutic target in the PI3K/AKT signalling cascade is mTOR, which lies downstream of AKT. mTOR is phosphorylated in response to stimuli that activate the PI3K/AKT pathway and regulates protein translation through phosphorylation of ribosomal S6 kinase, eukaryote initiation factor 4E-binding protein 1 and p70 kinase [Hidalgo and Rowinsky, 2000; Hara et al. 1997; Kumar et al. 2000]. Inhibition of mTOR by the naturally produced macrolide antibiotic sirolimus (rapamycin) results in inhibition of the oncogenic effects of AKT and PI3K [Aoki et al. 2001; Sehgal et al. 1975]. Several mTOR inhibitors with superior pharmaceutical properties to rapamycin are currently undergoing clinical investigation [Clackson et al. 2003; O’Reilly et al. 2002; O’Donnell et al. 2008, Atkins et al. 2004; Punt et al. 2003]. There have been several early studies of the mTOR inhibitors suggesting some benefit in patients with mCRC (Table 2). In one phase I trial of RAD-001, another mTOR inhibitor, one partial response was seen in a patient with CRC (lasting 5.3 months with a disease control period of 9 months) [Tabernero et al. 2008]. In the subsequent phase II study, a disease control rate (defined as response rate plus rate of stable disease) of 25%, with an OS of 5.9 months was achieved. The combination was somewhat toxic, though, with fatigue, cytopenias and nausea/vomiting/dehydration being the main AEs. Recently a phase Ib/II study of RAD-001 in combination with tivozanib, a highly potent, selective oral inhibitor of VEGFR-1, -2 and -3, in refractory mCRC was presented [Wolpin et al. 2012]. The combination of RAD-001 plus tivozanib was well tolerated with MTD of RAD-001 of 10 mg/day and tivozanib 1 mg/day. The phase II trial has completed enrolment using these doses. The mTOR inhibitors will likely be most effective when combined with traditional chemotherapy, and in one phase I clinical trial, the combination of RAD-001 with 5-FU in patients with refractory CRC resulted in one partial response lasting 7.4 months [Punt et al. 2003]. Several phase I and II studies of the mTOR inhibitors in combination with chemotherapy are currently under way.

Src kinase inhibitors

The nonreceptor tyrosine kinase Src was the first identified proto-oncogene, composed of a carboxy-terminal tail containing a negative-regulatory tyrosine residue, four Src homology domains and an amino terminal domain. Recent attention has been garnered by Src as an interesting and versatile target, with location at the nexus of a number of critical signalling pathways which acts as an intermediary between growth factor binding to the receptor and downstream signalling important for a number of cellular processes, including proliferation, differentiation and survival [Summy and Gallick 2000]. Src promotes tumour growth and invasion both through activation of downstream signal transducers and activators of transcription protein (STATs) and through modulation of the actin cytoskeleton focal adhesions and integrins [Brunton et al. 2005; Cao et al. 1996; Courter et al. 2005]. Enhanced Src activity is observed in 80% of CRCs [Talamonti et al. 1993]. The activity increases with the advanced stage of tumour development, for example, the high Src activity in dysplastic polyps with high malignant potential compared with more benign adenomas [Aligayer et al. 2002]. Src is also of particular interest in colon cancer because of its intimate connections with EGFR signalling. Upon binding of ligand to the EGFR, Src associates with EGFR and phosphorylates downstream targets [Muthuswamy and Muller, 1995]. Src can phosphorylate the EGFR itself, following activation by EGFR ligand binding, and through lateral activation via other growth factor receptor pathways [Biscardi et al. 2000]. Src activity has been correlated with shortened survival of patients with colon cancer [Aligayer et al. 2002]. Briefly, Src has an important role in oncogenic processes, which provides a rationale for Src as a therapeutic target.

Numerous Src inhibitors are entering phase I/II trials, including SKI-606 (bosutinib), SU 6656, AP 23464, BMS-354825 (dasatinib), saracatinib (table 2). Among these, the two that have been tested in CRC are saracatinib and dasatinib. Saracatinib has demonstrated preclinical activity against colon cancer metastases in vivo and is now undergoing phase II testing in patients with mCRC. In the phase I trial, 81 patients were tested, 28 of whom had CRC [Baselga et al. 2010]. Eleven patients were on study for over 12 weeks, five of whom had CRC, demonstrating some promise of activity, at least of disease stabilization in this patient population. Furthermore, a phase II study of the approved Src inhibitor dasatinib has been conducted in patients with unresectable, previously treated mCRC. The study was terminated after the first stage due to lack of efficacy. In fact, there were no objective responses and only one patient had stable disease for 7 months. Moreover, the median PFS rate was 1.6 months and median OS was 5.1 months [Sharma et al. 2011]. Src inhibitors are logical combination partners for EGFR inhibitors given the extent of cross talk between the two pathways. Building on this hypothesis, a combination study of dasatinib with FOLFOX and cetuximab was developed. A phase I, dose-finding trial in patients with previously treated mCRC has been conducted. The objectives of this study were to determine the MTD, toxicity and pharmacodynamics of the combination. Dasatinib at a dose of 100 mg daily in combination with FOLFOX and cetuximab was safe with promising early signs of clinical activity in previously treated patients [Kopetz et al. 2008].

Aurora kinase inhibitors

Within the complete cell-cycle process, mitosis constitutes one of the most critical steps. A copy of the duplicate genome is segregated by the microtubule spindle system into the two resulting cells. Errors in this process can lead to genomic instability, a condition associated with cancer. Contrary to its potential to induce cancer, mitosis has emerged as an important target for anticancer therapies. In fact, inducing aberrant mitosis in tumour cells leads to mitotic arrest and consequently, although not always, to cell death [Nigg, 2001]. Aurora is a name given to a family of seronine/threonine protein kinases that are key regulators of mitosis essential for accurate and equal segregation of genomic material from parent to daughter cells. Three AK family members have been identified in mammalian cells: A, B and C. These proteins are implicated in several vital events in mitosis and play a critical role as regulators of genomic stability [Keen and Taylor, 2004]. AKs are frequently overexpressed in human tumours. Deregulation of cell-cycle machinery can have an important impact on cellular proliferation. The first data that linked tumuorigenesis with the AK family came with the observation that AK-A DNA was amplified and its RNA was overexpressed in more than 50% of primary CRC specimens. Moreover this overexpression was associated with poor prognosis in patients with CRC [Bischoff et al. 1998; Katayama et al. 1999]. Recently, a study of the phenomena involved in the progression from colorectal adenoma to carcinoma identified that several oncogenes were potentially responsible for this malignant transformation [Carvalho et al. 2009]. These oncogenes were significantly amplified in carcinomas compared with adenomas, suggesting a potential role of the gene of AK-A in the development of colorectal carcinomas. Unlike AK-A, the role of AK-B in tumuorigenesis has been less studied. The reasons for this less extensive evaluation are not clear, although the pattern of expression of AK-B is not much different from that of AK-A. Increased levels of phosphorylation of histone H3 were correlated with overexpression of AK-B in colorectal tumour cell lines [Ota et al. 2002]. Like the other members of the AK family, AK-C has been shown to be overexpressed in multiple tumour cell lines; however, little is known about the role of its expression in human cancer specimens and prognosis [Fidalgo et al. 2009].

The evidence linking AK overexpression and cancer has supported the hypothesis that AK inhibition could be a target for cancer therapy. In fact, as AKs are only expressed and active as kinases during mitosis, nonproliferating cells would not be adversely affected by these drugs. In support of this hypothesis, proliferating tumour cells treated with ZM447439, a pan-AK inhibitor, were killed whereas nonproliferating cells were not affected by this agent [Ditchfield et al. 2003]. Given that most normal cells in the body do not proliferate rapidly, AK inhibitors could have a broader therapeutic index than nonspecific cytotoxic agents, such as some alkylating agents that act in a noncell-cycle specific manner. The mechanism of inhibition of the AKs could be to interfere with their ability to interact with their coactivating proteins or binding agents, such as by targeting the catalytic binding site of the ATP. This is the mechanism of action of a number of AK inhibitors.

Several AK inhibition drugs are in preclinical and clinical development in the CRC setting: MK0457, PHA-739358, MLN8054, MLN8237. MK0457, initially developed by Harrington and colleagues with the name of VX-680, is a diaminopyrimidine that targets the ATP-binding site common to all AKs [Harrington et al. 2004] (Table 2). VX-680 is an inhibitor of the three AKs with a more than 100-fold selectivity for AKs over a panel of 55 other kinases, with the exception of fms-like tyrosine kinase receptor 3 [Tyler et al. 2007]. In its first phase I clinical trial, MK0457 was given as an intravenous continuous infusion over several days to patients with previously treated solid tumours. The principal DLT was grade 3 neutropenia, accompanied by some nonspecific side effects, including nausea and fatigue [Rubin et al. 2006]. Another AK inhibitor is PHA-739358 which is a pan-aurora inhibitor with documented antitumour activity in multiple tumour xenograft models, including colon cancer, which have shown sustained tumour growth inhibition after discontinuation of treatment [Carpinelli et al. 2007]. In a preliminary assessment, DLT was reported to be grade 3–4 neutropenia. No tumour responses were observed but 8 of 40 patients had stable disease for at least 4 months. Pretreatment and post-treatment skin biopsies showed downregulation of phosphorylated histone H3Ser10 levels in eight of nine patients tested. Recently, the final results of the pharmacokinetics and pharmacodynamics of phase I with PHA-739358 were communicated [Steeghs et al. 2009]. As in the previous study, the drug was administered intravenously on days 1, 8 and 15, but in 6 h or 3 h infusions. Dose levels from 45 mg/m² in 6 h and from 250 mg/m² in 3 h were studied. Fifty patients with advanced or metastatic solid tumours were treated. For the 6 h infusion the most frequently reported adverse effects were neutropenia (55%), nausea (25%) and anorexia (23%). In the 3 h infusion they were fatigue (70%), neutropenia (60%) and diarrhoea (50%). The MTD was 330 mg/m² for the 6 h infusion and not identified for the 3 h infusion. This systemic exposure to PHA-739358 increased linearly with dose. Biomarker analyses showed inhibition of histone H3 phosphorylation indicative of AK-B inhibition at doses of 190 mg/m² or greater. Stable disease was observed in 23.7% of patients and disease stabilization occurred in 6 or more months in five patients. A small-molecule inhibitor of AK with relative specificity for AK-A is MLN8054. It is an ATP competitor and reversible inhibitor. This compound induced G2/metaphase accumulation and spindle defects and inhibits proliferation in multiple tumour cell lines. Growth of colon tumour xenografts in nude mice was dramatically inhibited after oral administration of MLN8054 at well tolerated doses. Moreover, the tumour growth inhibition was maintained after discontinuing MLN8054 treatment. This molecule demonstrated activity against a broad spectrum of human xenografts, including colon cancer [Manfredi et al. 2007; Huck et al. 2006]. Toxicity in animals was reversible and AEs included myelosuppression, cataracts and gastrointestinal mucosa damage. In a recent phase I trial, 43 patients with advanced solid tumours were treated [Macarulla et al. 2010]. In the two initial cohorts treated with 10 or 20 mg of MLN8054 once daily on days 1–5 and 8–12 of 28 days, patients experienced benzodiazepine-like AEs, mainly somnolence, so subsequent cohorts in this trial were treated with 25, 35, 45, 55, 60, 70 and 80 mg/day in four divided doses on days 1–14 with the largest dose at night. To manage somnolence, methylphenidate was administered in some patients with daytime doses. Dose escalation was stopped at 80 mg due to DLTs, including G3 somnolence (one patient) and G3 transaminitis (one patient). Pharmacokinetics showed a rapid absorption with a long half life (30–40 h) and pharmacodynamic studies confirmed that higher doses of MLN8054 corresponded with an increased mitotic count in tumour and skin biopsies and reduced chromosome alignment and spindle bipolarity in the mitotic cells in these biopsies. Stable disease was observed in three patients with colorectal and non-small cell lung cancer, and melanoma. A second-generation oral inhibitor, more potent than MLN8054, is MLN8237, which is a competitive and reversible inhibitor of the ATP of AK-A. Like MLN8054, this new inhibitor has activity for the binding site of the receptor of γ-aminobutyric acid α1; however, this interaction occurs at a level of dose much higher than the DLT. These advantages over MLN8054 have led to stopping the development of MLN8054, focusing efforts on the MLN8237. In preclinical, in vivo studies, MLN8237 has demonstrated anticancer activity in xenografts with colon cancer with a maximal inhibition of the tumour growth greater than 90%. Doses and maximal efficacy schedules in nude mice models varied from 10 to 30 mg/kg. A phase I trial testing this compound included 27 patients with advanced solid tumours [Cervantes-Ruiperez et al. 2009]. MLN8237 was administered orally and once or twice daily for 7 days followed by a 14-day recovery period. Patients were treated in three once daily cohorts with 5, 80 or 150 mg, and four twice daily cohorts with 50, 60, 75 and 100 mg. DLTs observed at 60–100 mg twice daily included neutropenia, pancytopenia, stomatitis and somnolence. Somnolence (of any grade) was seen with once daily dosing but was reduced with divided doses each less than 100 mg. MLN8237 was tolerable with twice daily dosing for 7 days and exhibited favourable pharmacokinetics and pharmacodynamics.