PI3 Kinase Inhibitors in the Clinic: An Update

Structure and Functions of the Phosphoinositide 3 Kinases

The phosphoinositide-3 kinases (PI3Ks) catalyze the production of lipid second-messengers from phosphoinositide-2 phosphate into phosphoinositide triphosphate in cellular membranes (1). There are eight classes of phosphoinositide kinase in mammals, among which only class I products are involved in second-messenger signaling with phosphoinositide triphosphate (2). Apart from being activated by tyrosine kinase receptors (RTKs), class IA PI3Ks are activated by G-protein-coupled receptors (GPCR) and oncoproteins, whereas class IB products are only regulated by GPCRs (3). The enzyme consists of the association of two subunits, p85 and p110, harboring regulatory and catalytic properties, respectively. There are three different isoforms of p110: p110α, β, or Δ, that are encoded by PIK3CA, PIK3CB and PIK3CD, respectively. All p110 isoforms have a similar structure, with well-identified domains: an amino-terminal adaptor-binding domain (ABD) interacting with p85, a rat sarcoma (RAS)-binding domain (RBD), a C2 (protein kinase-C homology-2) domain with affinity for lipid membranes, a helical domain, and a carboxyl-terminal kinase domain (4). The p85 regulatory subunit is encoded by three genes: PIK3R1, PIK3R2 and PIK3R3 (5). Among the four regions of p85, the nSH2 domain is mostly involved in p110 regulation, and inhibits its kinase activity, unless p110 is activated from an upstream signal, leading to the removal of the nSH2 inhibition (6). Besides the regulation of p110, p85 is also involved in a number of cellular functions, such as endocytosis, cytoskeleton organization, receptors trafficking, etc.. [reviewed in (5)]. There are four isoforms for the catalytic subunit, α, β, γ, Δ, respectively. In class I PI3Ks, either p110α, β, or Δ form a heterodimer with one of five regulatory proteins p85α, p85β, p85γ, p50α, and p55α out of which the two smaller proteins are splicing variants of p85α (7). Interestingly, only PIK3CA, encoding p110α is linked to a number of types of cancer, whereas PIK3CB has oncogenic properties only in cultured cell models (3, 8).

Activation of PI3K comes from an upstream signal [e.g. receptor tyrosine kinase (RTK) dimerization and activation through ligand binding], leading to i) the direct binding of p85 to the RTK and the subsequent p110 activation; or ii) the binding of growth factor receptor-bound protein 2 (GRB2) to the RTK, followed by GRB2-associated-binding (GAB) protein recruitment and further p85 activation; or iii) from RAS-mediated p110 activation through the recruitment of GRB2 and son of sevenless (SOS) to the receptor (9). A wide number of transmembrane receptors have been identified as PI3K activators.

The PI3K/AKT/mTOR Cascade and its Relevancy to Cancer

Signaling through PI3K drives a number of cell functions that are linked to cell growth, apoptosis regulation, cell adhesion, transformation, survival, and motility through the activation of the downstream AKT protein (10, 11). Importantly this pathway activation is regulated through the PI3K antagonistic phosphatase and tensin homolog (PTEN) that is deleted or inactivated in a wide range of malignancies (12). The downstream effector AKT phosphorylates a number of proteins, among which is mTOR. mTOR acts as the catalytic subunit in the protein complexes mTORC1 and mTORC2, in which the former regulates S6 kinase and 4E binding protein 1 (4EBP1), leading to translation initiation and protein synthesis, and the latter is involved in a feedback activation of the cascade through the phosphorylation of AKT on the Ser⁴⁷³ residue (13). Inhibition of mTOR has proven efficacious in numerous solid tumors, such as breast and kidney cancer, well-differentiated endocrine tumors and sarcomas. Hence, finding other targets within the PI3K/AKT/mTOR pathway has a high relevance to developing innovative compounds. Indeed, inhibition of mTORC1 has the caveat of a feedback activation of PI3K that might be responsible for rapamycin analogs (rapalogs) resistance, providing a strong rationale for combining mTOR and PI3K inhibition (14).

Oncogenic Mutations of PI3K

The PI3KCA gene is mutated in a wide range of cancer types, including those of breast, colon, prostate, endometrium, and of the ovary (7). Although mutations in the PI3K pathway seem to occur in a mutually exclusive manner, coexistence of PTEN and PI3KCA mutations have been reported in endometrial cancers (7, 15). About 80% of PI3KCA mutations occur within three hot spots, two of them mapping to the helical domain of PI3KCA and one in the kinase domain, with distinct mechanisms of protein activation (16). These mutations induce a gain of function in p110α, with increased and detectable levels of downstream effectors that can be surrogate markers of PI3KCA mutations (reviewed in 17). Mutations in the helical domain, such as E542K and E545K result in the lack of p110α regulation by the regulatory subunit, whereas H1047R in the kinase domain induces a conformational change in the activation loop, thus mimicking RAS-mediated activation (16, 18). This conformational change increases the protein interaction with the lipid membrane, thus making phosphoinositide diphosphate more accessible to the enzyme (19). Indeed, having identified such mutations, and having established their frequency among carcinomas provides a strong rationale for the development of anti-PI3K drugs, targeting the kinase domain through the inhibition of ATP binding. Mutations in the p85 regulatory subunit have also been identified and consist of truncations or deletions, resulting in the disruption of the interaction of nSH2 and iSH2 domains with the C2 domain of p110α, thus abolishing the kinase domain inhibition by p85 (19).

Targeting PI3K in Anticancer Therapy: From the Bench to Phase II Trials

Given the elucidation of the PI3K structure, the identification of mutation hot spots, and the understanding of the related structure/function changes, efforts towards innovative drugs design should now focus on the design of isoform-specific, mutant-specific inhibitors (19). Although such a personalized approach is not yet feasible in routine practice, several PI3K inhibitors have been evaluated, both pre-clinically and in the clinic, with phase I and phase II trials for the latter. Initial data were obtained with two compounds, wortmannin and LY294002, which served as a proof of concept in preclinical models, though suffering from unfavorable pharmaceutical and toxicity profiles and yielding more in-depth research for better drugs (20, 21). In this section we focus on the clinical data that are available for these compounds, as well as on ongoing trials and future perspectives.

NVP-BEZ235

NVP-BEZ235 was synthesized by Novartis Pharma, (Novartis Institute for Biomedical Research Oncology, Basel, Switzerland) as a potent p110α inhibitor, with a dual mTOR inhibition capacity. NVP-BEZ235 has activity against p110α, p110α-H1047R and p110α-E545K, with a half maximal inhibitory concentration (IC₅₀) in the low nanomolar range (22). NVP-BEZ235 inhibits PI3K by blocking the ATP-binding site, a common mechanism of other tyrosine kinase inhibitors and possesses strong antiproliferative activity towards several cell lines and against tumor xenografts (22-24). Phase I data with NVP-BEZ235 showed an acceptable toxicity profile, with mild or moderate, and manageable nausea, and vomiting, diarrhea, fatigue, anemia, and anorexia. Two patients out of 59 had partial response (PR), whereas 16 experienced minor responses. Interestingly, in 14 patients with stable disease ≥4 months, 6 had deregulated PI3K pathway (25). Another phase I study of BEZ235, with a different formulation (SDS sachet) has been reported showing a maximal tolerated dose of 1600 mg/d, and a similar toxicity profile. The dose-limiting toxicities were grade 3 thrombopenia and fatigue (26). Clinical studies are ongoing with the drug in several malignancies, with an emphasis on breast cancer (Table I). Most of the trials focus on single-agent activity of NVP-BEZ235, despite a few trials investigating combinations with chemotherapy (capecitabine, paclitaxel) or targeted therapies (trastuzumab, everolimus and other PI3K inibitors such as BKM120). Genotype indicators for the NVP-BEZ235 efficacy have been found, at least in endometrial cancer, such as the presence of PTEN mutations without KRAS alterations, which might be strong predictive factors for NVP-BEZ235 efficacy in this condition (27). Unexpectedly, with regard to molecular abnormalities that were evidenced in endometrial cancer, NVP-BEZ2345 study has been stopped in this condition and endometrial cancer clinical trials were withdrawn.

BKM120

BKM120 has been developed by Novartis Pharma (Novartis Institute for Biomedical Research Oncology, Basel, Switzerland) and is a selective pan-PI3K inhibitor, with an IC₅₀ towards p110α, p110α-H1047R and p110α-E545K in the nanomolar range, and no activity towards mTOR and other kinases (14). BKM120 binds to the ATP-binding site of the lipid kinase in a mixed competitive and non-competitive manner. Preclinical data showed activity toward different cell lines, including glioma and myeloma (14, 28, 29). Interestingly, cell lines harboring PI3KCA oncogenic mutations had a greater sensitivity to BKM120 as compared to those with wild-type PIK3CA (14). Moreover, the presence of a concomitant KRAS activating mutation was found to reduce sensitivity to BKM120, a finding that has been previously observed in the clinic with mTOR inhibitors, such as everolimus (30). Interestingly, preclinical data combining BKM120 with a signal transducer and activator of transcription 3 (STAT3) inhibitor showed that the combination was synergistic in gastric cells harboring KRAS mutations as opposed to those with wild-type KRAS (31). A phase I study of BKM120 in patients with advanced solid tumors has recently been reported, showing an acceptable toxicity profile, with expected side-effects for PI3K/AKT/mTOR pathway inhibition (diarrhea, mucositis, hyperglycemia and anorexia). Importantly, mood disorders were observed both at the maximal tolerated dose (MTD; 100 mg daily) and the lower dose level (80 mg daily) (32). Particular attention should therefore be paid to a medical history of depression in patients likely to receive BKM120. Of note, among the 35 enrolled patients, one experienced a confirmed PR (triple-negative breast cancer) and seven had disease control for at least eight months. Many trials with BKM120 are currently recruiting patients (Table II) with various malignancies including, as opposed to NVP-BEZ235, endometrial cancer, notably the phase II GINECO ENDOPIK trial that has recently started enrolment (Clinicaltrials.gov identifier: NCT01397877). A number of other trials are not yet recruiting patients, but are scheduled to open rapidly worldwide.

BAY 80-6946

BAY80-6946 (Bayer Healthcare Pharmaceuticals, Berlin, Germany) is a potent and highly selective pan-class I PI3K inhibitor with antitumor activity described in preclinical models. A phase I trial of BAY 80-6946 administered weekly for three out of four weeks, has been reported in solid tumors, with an MTD of 0.8 mg/kg. Dose-limiting toxicities included grade 4 hyperglycemia and acute left ventricular dysfunction, other toxicities being mild (fatigue, nausea, diarrhea, mucositis, anemia and dysgeusia). Two patients experienced stable disease for six and eight months, with endometrial and oesophageal cancer, respectively (33). Four trials exploring BAY 80-6946 are currently recruiting patients (Table III).

GDC-0941

GDC-0941 (Genentech, CA, USA) is a potent, selective, orally bioavailable inhibitor of PI3K that has shown efficacy in various preclinical models (34-36). Moreover synergy with Map-Erk Kinase (MEK) inhibitors was observed with enhanced tumor cell apoptosis in preclinical models, including gefitinib-resistant lung cancer cell lines (37, 38). Phase I data showed that GDC-0941 has an acceptable toxicity profile, without observed dose-limiting toxicities, and with moderate side-effects, including diarrhea, nausea, dysgeusia, peripheral sensory neuropathy, dry mouth, thrombocytopenia, and increased aspartate aminotransferase (39). Similar results were obtained in another phase I study, where the MTD was exceeded at 450 mg daily, and responses were observed in melanoma, gastro-intestinal stromal tumors (GIST) and ovarian cancer patients (40). Trials investigating GDC-0941, currently recruiting patients are shown in Table IV.

GDC-0980

GDC-0980 (Genentech, CA, USA) is a novel, dual PI3K/mTORC inhibitor that targets all PI3K class I isoforms and mTORC at low nanomolar concentrations, which is active in several cancer cell lines of breast, prostate and lung cancer, including those with activated PI3K or PTEN loss (41, 42). A phase I study of GDC-0980 in patients with advanced solid tumors recommended a daily dosing <70 mg, on a 21/28 day schedule. Toxicity consisted of rash, hyperglycemia, mucositis and pneumonitis, which resolved with drug cessation. Antitumor effects were observed in 3/33 patients with mesothelioma. Fluorodeoxyglucose positon emiting tomography (PET-FDG) responses were also observed in GIST and adrenal gland cancer (43). A prior phase I study was also reported, at a daily dosing of 16 mg, again on the 21/28 day schedule, showing consistent results (44). Studies currently recruiting patients in GDC-0980-based therapies are shown in Table V.

Future Perspectives for Targeting the PI3 Kinase

Since the mTOR inhibitors have demonstrated that inhibiting the PI3K/AKT/mTOR pathway is a relevant goal for improving the outcome in several types of cancers (kidney, breast, endocrine tumors and sarcomas), this pathway has been extensively studied. Indeed, as modern chemistry and molecular biology offer the opportunity to decipher in detail the molecular mechanisms of protein activation though the kinase family, there is a continuously growing number of novel compounds that are being designed and evaluated. There is now a solid body of evidence suggesting that selective inhibition of the PI3K isoforms that are involved in carcinogenesis and the mechanisms of cancer progression is a relevant approach. Indeed, this review identified a large number of phase I studies, contrasting with a much lower number of phase II trials. Hence, more data are awaited from single-agent phase II studies to ascertain the actual antitumor properties of PI3K inhibitors, in terms of indicators that are more relevant to the patients and the physicians, such as survival. Moreover, it is so far uncertain whether transient responses observed in early trials, as well as surrogate markers of efficacy such as TEP-FDG standard uptake value (SUV) responses in heavily pretreated patients are actually meaningful. A more problematic issue arises from resistance to the inhibition of PI3K. Indeed, as evidenced upon mTOR inhibition, activating feedbacks may quickly jeopardize the benefits of the targeted therapy. Hence, future directions should focus on concomitant inhibition of pathways that are involved in the feedback activation loops, with obvious tolerance issues. Trials are ongoing for this purpose, as well as those combining PI3K inhibitors with cytotoxic drugs. Finally, much is awaited from the identification of tumor biomarkers associated with the efficacy of PI3K inhibitors, such as concomitant KRAS mutation, or PI3K-activating mutations. Tumor genotyping in patients receiving such drugs is not only mandatory in early phase trials, but will certainly be part of routine practice in the near future. Hence, we will be able to identify patients which are most likely to benefit from these new targeted therapies.