The PI3K pathway is one of the most frequently altered pathways in human cancer (Fig. 1). Regardless of how it is activated, the PI3K pathway regulates numerous processes such as cell growth, metabolism, survival, and proliferation, underlining its critical role in human malignancy. It is, therefore, not surprising that tremendous time and resources have been committed to the clinical development of therapeutic agents targeting nearly every aspect of this pathway. In this article, we will review the PI3K signaling pathway, its role in development and progression of malignancy, and progress in the clinical development of targeted therapeutic agents.
PI3K Signaling Pathway
There are three classes of PI3Ks (I-III) with distinct structure, cellular distribution, mechanism of action, and substrate preference.1 The Class I PI3Ks are further subdivided into Class IA, which are activated by growth factor signaling through receptor tyrosine kinases (RTK), and Class IB, which are activated by G protein-coupled receptors (GPCR). The Class IA PI3Ks are the most relevant PI3Ks from an oncologic perspective and are heterodimeric kinases consisting of a p85 regulatory subunit and a p110 catalytic subunit. There are three Class IA p110 isoforms (alpha, beta, and delta) encoded by three genes (PIK3CA, PIK- 3CB, and PIK3CD, respectively) and one related Class IB p110 isoform (gamma). Of these, the alpha and beta isoforms are believed to be expressed ubiquitously, whereas the delta and gamma isoforms are expressed only in hematopoietic cells. There are also three isoforms of the Class IA p85 subunits (alpha, beta, and delta) encoded by three genes, PIK3R1, PIK3R2, and PIK3R3, respectively.
Like all PI3- Ks, the Class I PI3Ks are lipid kinases that phosphorylate the 3’-OH group of phosphoinositides. In particular, the Class IA PI3Ks phosphorylate PIP2 and generate PIP3. This activity is directly opposed by the tumor suppressor PTEN, which dephosphorylates PIP3 to PIP2. PIP3 induces the membrane localization and activation of an array of kinases through interaction with their respective pleckstrin homology (PH) domains. Although the best described of these is Akt (protein kinase B), there are several other PI3K-dependent pathways including those possibly relevant to cancer, such as serum and glucocorticoid kinases (SGKs) and Bruton tyrosine kinase (BTK).2,3 Nonetheless, Akt has canonically been regarded as the primary executer of PI3K and regulates the function of a broad array of proteins involved in cell growth, proliferation, motility, adhesion, neovascularization, and apoptosis. 4 One downstream effector of Akt, which is particularly relevant to cancer, is the kinase mTOR. The PI3K/Akt pathway is one of many inputs into the activity of mTOR complex 1 (TORC1), which serves as master regulator of cell growth responding to numerous environmental inputs including the availability of oxygen, nutrients, ATP, and amino acids. (Fig 2)5 As each of aforementioned kinases (PI3K, Akt, and mTOR), regulate critical signaling networks for cancer, there is ample rationale for their emergence as therapeutic targets for drug development.
Activation of the PI3-K Pathway in Cancer
Given the central role of PI3-K and its downstream pathways in many cellular functions related to growth and proliferation, it is perhaps not surprising that the pathway is constitutively activated in many cancers. Class IA PI3-Ks are primarily activated in cancer through two mechanisms: signaling through RTK or through mutations within the pathway itself. Activation of RTKs, most commonly through growth factor signaling, results in the phosphorylation of the Y-X-X-M motif present in the cytoplasmic tail of the RTK, which then binds to the Src-homology (SH2) domain of the p85 regulatory subunit. This results in a functional dissociation of the p85 subunit from the p110 subunit, augmenting kinase activity of the latter. In cancer, RTKs can be pathologically activated through amplification and mutation, perhaps the best known of which is EGFR-mutant non-small cell lung cancer. Other examples of frequently mutated or amplified RTKs in cancer include HER2, c-KIT, C-MET, and PDGFR-alpha, all of which result in downstream activation of PI3-K. In many cases, the ability of individual RTK inhibitors to have clinical efficacy in specific RTK-driven cancers is dependent on their ability to abrogate PI3-K signaling.6,7
In addition to RTKs, virtually every other element of the PI3-K pathway has been found to be genetically altered through mutation or amplification in an array of cancers. One of the most common and earliest described is the down-modulation or loss of PTEN.8 Class IA PI3-K can also be constitutively activated by mutations in the gene encoding the p110-alpha subunit, PIK3CA, and such alterations are among the most commonly found mutations in cancer specimens.9 The vast majority of PIK3CA mutations occur in either the helical or kinase domains of p110-alpha, both of which result in enhanced kinase activity. Mutations in the gene encoding the p85-alpha regulatory subunit, PIK3R1, have also been described, which result in enhanced kinase activity by abrogating the inhibitory interaction between p85-alpha and p110-alpha.10 These mutations, however, appear to be less common than those in PIK3CA. Class IA PI3-Ks can also be activated through association with Ras and through signaling through GPCR.11,12 However, although activating mutations in Ras are common in many cancers, it is clear that the growth and survival-promoting effects of such alterations do not signal entirely through the PI3-K pathway because these mutations can predict for resistance to PI3-K inhibitors.13
Many downstream effectors of PI3-K can also be altered in human cancer. Gain-of-function mutations in Akt have been described in all three Akt isoforms.14-16 Similar mutations in PDK-1, the kinase responsible for phosphorylating and partially activating Akt at the Thr308 residue, have also been described.15 Further downstream, knowledge of mutations in many genes—which would be predicted to constitutively activate mTOR, including LKB1, TSC2, FBXW7, and mTOR itself—is emerging. The diverse array of mechanisms, which tumors have undertaken to activate various elements of the PI3-K pathway, speaks to its critical importance in tumorigenesis, as well as tumor growth and proliferation.
Role of PI3K Pathway Alterations in Tumor Development
It is clear that certain genetic alterations in the PI3K pathway are frankly oncogenic Mutation of p110-alpha, both in the kinase and helical domain, are sufficient to transform and immortalize fi broblasts.17 PTEN is an established tumor suppressor gene whose germline loss results in Cowden Syndrome, a hereditary hamartoma tumor syndrome. Furthermore, binding of Ras to p110- alpha has been shown to be necessary for Ras-mediated malignant transformation, highlighting the relevance of PI3K in tumorigenesis.18 It remains interesting, however, that various genetic alterations in the PI3K pathway can overlap in the same tumor, for example the coexistence of PIK3CA mutation and PTEN loss.19 Although each of these alterations would be expected to activate the PI3K pathway, their coexistence in the same tumor implies that the biologic consequences of each of these alterations are not necessarily overlapping. Such findings suggest that the acquisition of such mutations over time may confer a growth advantage to transformed tumor cells and be critical to tumor growth and proliferation. The likelihood that genetic alterations in PI3K do not completely overlap functionally also suggests that such alterations should not be dealt with equivalently from a therapeutic standpoint.
Progress in Therapeutic Targeting of PI3K Pathway
From a pharmaceutical standpoint, the PI3K pathway is perhaps one of the most therapeutically investigated targets in cancer. The first agents targeting this pathway to enter clinical assessment were the rapalogues, which are allosteric inhibitors of TORC1. Agents such as temsirolimus and everolimus have demonstrated single-agent clinical activity in renal cell carcinoma, mantle cell lymphoma, and some neuroendocrine tumors. Preclinical studies with these agents, however, suggest that the presence of regulatory feedback loops which activate PI3K/Akt following treatment with a rapalogue may undermine the efficacy of these agents.20,21 Following the development of the rapalogues, a broad array of agents has been developed and advanced into clinical assessment, including pan-isoform PI3K inhibitors, isoform-specific PI3K inhibitors, dual inhibitors of PI3K and mTOR, and inhibitors of Akt. The progress in clinical development of these agents is summarized further in this article.
Pan-Isoform PI3K Inhibitors
Emerging from compounds such as wortmannin, some the earliest agents to enter clinical assessment have been panisoform inhibitors of PI3K. These agents are direct inhibitors of p110 kinase activity that act as ATP mimetics, binding competitively and reversibly to the p110 ATP-binding pocket. For the most part, these drugs have moved through phase I testing and into phase II assessment in multiple disease types. Early clinical trials have suggested that, in general, the pan-isoform PI3K inhibitors are reasonably well tolerated. The most common toxicities that have been noted are hyperglycemia, skin toxicity (most commonly a maculopapular rash), stomatitis, and gastrointestinal side effects (nausea, vomiting, diarrhea, and anorexia). Of these, hyperglycemia is likely a mechanism-based toxicity given the well-described role of PI3K in insulin signaling. Of the pan-isoform PI3K inhibitors in clinical development, BKM120 might be the furthest along and is being assessed in a placebocontrolled randomized phase III clinical trial comparing BKM120 and placebo with BKM120 and fulvestrant in patients with hormone-positive HER2- negative breast cancer (NCT01610284).
Dual PI3K/mTOR Inhbitors
Structural similarities between the ATP-binding domain of p110 and the catalytic domain of mTOR have led to the development of a class of agents that are both pan-isoform inhibitors of PI3K and mTOR. Unlike the currently available rapalogues, these agents are active site inhibitors of mTOR and have the theoretical advantage of inhibiting the kinase activity of mTOR regardless of whether it is in complex with raptor (TORC1) or rictor (TORC2). As noted before, mTOR responds to many inputs besides PI3K/Akt. Therefore, these dual inhibitors may have broader activity in cancers in which PI3K/Akt is not the primary driver of mTOR activity. Finally, unlike the rapalogues, these agents might be able to inhibit TORC1 activity while preventing the aforementioned feedback activation of PI3K. Thus far, early-phase clinical studies suggest that the toxicity of the dual inhibitors of PI3K/mTOR is similar to that of the pan-isoform PI3K inhibitors. Interestingly, toxicities characteristic of the rapalogues, such as hyperlipidemia and pneumonitis, have not been routinely observed with this class of agents. A few of these agents such as GDC-0980 and BEZ235 have entered phase II clinical trials in a variety of diseases. In general, although some objective tumor responses have been observed, the activity of these agents thus far has been modest, and no clinical trial with either a pan-isoform PI3K inhibitor or dual PI3K/mTOR inhibitor has reported robust clinical activity, even in tumors with known genetic alterations.22
Isoform-specific PI3K Inhibitors
The aforementioned modest efficacy of earlier-generation PI3K inhibitors is perhaps reason that the greatest enthusiasm remains for PI3K inhibitors that are isoform-specific. The primary theoretical advantage of the isoform-specific inhibitors is that they may have more focused toxicities compared with the pan-isoform inhibitors, allowing these agents to be tolerated at doses, which can result in more complete and reliable inhibition of kinase activity. Thus, although these new agents are not mutation specific, there is hope that greater inhibition of mutated PI3K might be achieved. This paradigm has proven true in diseases such as in V600E BRAF mutant melanoma, in which substantial antitumor responses were not observed until BRAF inhibitors with greater specificity for the mutant kinase were developed.23 Concurrent with the development of the isoform-specific PI3K inhibitors has been the realization that the pro-survival effects of various genetic alterations may signal preferentially through specific isoforms of p110. For example, ERBB-amplified breast carcinoma may depend primarily on the p110-alpha isoform.24 It has also recently been suggested that the transforming effects of PTEN loss largely depend on the p110-beta isoform in models of prostate intraepithelial neoplasia.25 Finally, p110-delta isoform is perhaps the dominant isoform in the lymphocytic lineage. At the same time, there is growing concern that the pan-isoform inhibitors of PI3K are not equally effective against every isoform at their deliverable doses. Not surprisingly, there is great interest in the activity of PI3K-alpha inhibitors such as BYL719 in cancers with PIK3CA mutations, PI3Kbeta inhibitors in tumors with PTEN loss, and PI3K-delta inhibitors in hematologic malignancies. PI3K-delta inhibitors have already shown very promising activity in patients with chronic lymphocytic leukemia and a randomized double-blind placebo-controlled phase III study of GS-1101 (idelalisib) in combination with rituximab has been launched (NCT01539512; Abstract 7003).26
Like the PI3K inhibitors, Akt inhibitors are slowly advancing through clinical assessment. Amongst the inhibitors in clinical development are both allosteric inhibitors such as MK2206 and ATPcompetitive catalytic inhibitors such as GSK690693 and GDC-0941. These agents thus far are all pan-isoform inhibitors of Akt. In early-phase clinical trials, both the allosteric and catalytic inhibitors of Akt have shown toxicities similar to those observed with the PI3K inhibitors, such as hyperglycemia, rash, stomatitis, and gastrointestinal side effects. Also similar to the PI3K inhibitors, the clinical activity of these agents has been modest thus far and questions regarding appropriate patient selection remain. 27 For example, it was recently suggested that certain PIK3CA mutations result in relatively low activation of Akt in comparison to PTEN loss.3 Thus, these agents may be more appropriately directed to those cancers with Akt alterations and PTEN loss.
Future Challenges in Development of PI3K Inhibitors
For most of the agents described herein, the phase I clinical trials have been completed and the toxicity profiles established. Although these agents have been well tolerated, their early-phase studies have also been highlighted by the relatively modest single-agent clinical activity observed thus far from this entire drug class. As these drugs move deeper into clinical assessment, there are many challenges to enhance their therapeutic index. First, it remains unclear if all of these agents are optimally or even effectively modulating their biologic targets. Efforts going forward have focused heavily on obtaining treatment biopsy specimens during treatment for pharmacodynamic studies concurrent with identification of more robust biomarkers of treatment response. Hopefully, these efforts will allow the optimization of dose and schedule to achieve either the maximal inhibition of biologic target or the greatest clinical benefit. Secondly, it is clear that each class of agent results in distinct biologic effects, perhaps not surprising given the complexity of the PI3K network. Therefore, efforts must be focused on identifying the most appropriate cancers and genetic alterations for each class and individual agent. Lastly, it remains rationale to believe that even greater clinical responses might be achieved through combination of the PI3K/Akt inhibitors with other molecularly targeted agents. Pre-clinical studies have identified numerous mechanisms of possible resistance to PI3K inhibitors, including feedback activation of RTK (e.g., HER2 and IGFR1), activation of parallel nodal pathways (e.g., MAPK), and alterations in hormone signaling. Not surprisingly, PI3K inhibitors are in active clinical assessment in combination with inhibitors of RTK signaling (e.g., trastuzumab in breast cancer), MEK inhibitors in many solid tumors, and hormone therapy (antiandrogen therapy in prostate cancer and antiestrogen therapy in breast cancer).
In conclusion, early-phase clinical trials of the next generation of PI3K inhibitors have mostly been completed and have shown that these agents are able to modulate the PI3K/Akt/mTOR signaling pathway with a reasonable toxicity profile. These same trials have identified several questions as these agents progress deeper into clinical development. Although the clinical benefits observed thus far have not completely justified the enormous investment of resources in this class of agents, optimism remains driven by the promise of the continued development of isoform-specific PI3K inhibitors, identification of more predictive biomarkers of response in conjunction with enhanced knowledge of the biology of specific genetic alterations, and the promise of combinational approaches.