The PI3K/Akt/mTOR axis in head and neck cancer: functions, aberrations, cross-talk, and therapies


  • R Vander Broek,

    1. Tumor Biology Section, Head and Neck Surgery Branch, National Institute on Deafness and Other Communication Disorders, National Institutes of Health, Bethesda, MD, USA
    2. Medical Research Scholars Program, National Institutes of Health, Bethesda, MD, USA
    3. School of Dentistry, University of Michigan, Ann Arbor, MI, USA
    Search for more papers by this author
  • S Mohan,

    1. Tumor Biology Section, Head and Neck Surgery Branch, National Institute on Deafness and Other Communication Disorders, National Institutes of Health, Bethesda, MD, USA
    2. Medical Research Scholars Program, National Institutes of Health, Bethesda, MD, USA
    Search for more papers by this author
  • DF Eytan,

    1. Tumor Biology Section, Head and Neck Surgery Branch, National Institute on Deafness and Other Communication Disorders, National Institutes of Health, Bethesda, MD, USA
    2. Medical Research Scholars Program, National Institutes of Health, Bethesda, MD, USA
    Search for more papers by this author
  • Z Chen,

    Corresponding author
    1. Tumor Biology Section, Head and Neck Surgery Branch, National Institute on Deafness and Other Communication Disorders, National Institutes of Health, Bethesda, MD, USA
    • Correspondence: Carter Van Waes, MD, PhD or Zhong Chen, MD, PhD, National Institute on Deafness and Other Communication Disorders, National Institutes of Health, Building 10/CRC Rm 4-2732 or Rm 5D55, 10 Center Dr., Bethesda, MD 20892, USA. Tel: 301 402 4216, Fax: 301 402 1140, E-mails: or

    Search for more papers by this author
  • C Van Waes

    Corresponding author
    1. Tumor Biology Section, Head and Neck Surgery Branch, National Institute on Deafness and Other Communication Disorders, National Institutes of Health, Bethesda, MD, USA
    • Correspondence: Carter Van Waes, MD, PhD or Zhong Chen, MD, PhD, National Institute on Deafness and Other Communication Disorders, National Institutes of Health, Building 10/CRC Rm 4-2732 or Rm 5D55, 10 Center Dr., Bethesda, MD 20892, USA. Tel: 301 402 4216, Fax: 301 402 1140, E-mails: or

    Search for more papers by this author


Head and neck squamous cell carcinoma (HNSCC) is one of the most morbid, mortal, and genetically diverse malignancies. Although HNSCC is heterogeneous in nature, alterations in major components of the PI3K/Akt/mTOR pathway are consistently observed throughout the majority of HNSCC cases. These alterations include genetic aberrations, such as mutations or DNA copy number variations, and dysregulation of mRNA or protein expression. In normal physiology, the PI3K/Akt/mTOR axis regulates cell survival, growth, and metabolism. However, alterations in this pathway lead to the malignant phenotype which characterizes HNSCC, among many other cancers. For this reason, both pharmaceutical companies and academic institutions are actively developing and investigating inhibitors of PI3K, Akt, and mTOR in preclinical and clinical studies of HNSCC. Many of these inhibitors have shown promise, while the effects of others are tempered by the mechanisms through which HNSCC can evade therapy. As such, current research aimed at elucidating the interactions between PI3K/Akt/mTOR and other important signaling pathways which may drive resistance in HNSCC, such as p53, NF-κB, and MAPK, has become a prominent focus toward better understanding how to most effectively treat HNSCC.


Head and neck cancer is the sixth most common cancer worldwide and has a 5-year morbidity rate of ~50% (Siegel et al, 2013). The histological diagnosis of head and neck squamous cell carcinoma (HNSCC) accounts for over 95% of all clinical cases of head and neck cancer and is limited anatomically to one of five primary sites: oral cavity, oropharynx, nasopharynx, hypopharynx, or larynx (Leemans et al, 2011). In terms of genetic alterations, HNSCC is one of the most heterogeneous of all cancers (Vogelstein et al, 2013). However, The Cancer Genome Atlas (TCGA; and other studies recently revealed that the majority of HNSCC possess alterations in the PI3K/Akt/mTOR pathway (Grandis et al, 2012; Hayes et al, 2013), a critical regulatory axis for cell growth, survival, motility, and metabolism in both normal physiology and cancer.

Phosphoinositide 3-kinase (PI3K), is composed of two subunits – an 85-kDa regulatory subunit and a 110-kDa catalytic subunit. It is activated in response to somatic mutation, stimulated receptor tyrosine kinases (RTKs) such as epidermal growth factor receptor (EGFR) and insulin-like growth factor receptor 1 (IGF-1R), or G-protein-coupled receptors (GPCRs). PI3K catalytic subunit alpha isoform (PIK3CA), the gene that programs for the p110α isoform of PI3K, is one of the most frequently mutated and amplified oncogenes in human cancers (Yuan and Cantley, 2008), including HNSCC. In normal physiology, PI3K is regulated by the tumor suppressor, Phosphatase and tensin homolog (PTEN). However, inactivating mutation or loss of PTEN is a frequent alteration in cancer and leads to hyperactivity of the PI3K pathway (Courtney et al, 2010). Protein kinase B (Akt) and mammalian target of rapamycin (mTOR) are effector proteins downstream of PI3K and also play important roles in carcinogenesis. Together, the members of the PI3K/Akt/mTOR axis interact with and contribute to the regulation of several other signaling molecules in HNSCC, including tumor suppressor, tumor protein p53 [p53 (TP53)], nuclear factor-kappa B (NF-κB), and mitogen-activated protein kinase (MAPK)/extracellular signal-regulated kinase (ERK) (Figure 1). For these reasons, there has been a surge in the development and use of pharmacologic inhibitors targeting PI3K, Akt, and mTOR for cancers such as HNSCC (Table 1). Initial preclinical and clinical studies indicate promise for PI3K and mTOR inhibitors to benefit patients with HNSCC.

Table 1. Currently ongoing clinical trials of PI3K/Akt/mTOR inhibitors targeting HNSCC and solid tumors. All ongoing clinical trials of PI3K/Akt/mTOR inhibitors that include any patients with HNSCC are listed, unless noted otherwise. The three most common adverse events associated with each agent are presented in order of frequency reported, to date. Clinical trial IDs are indexed by (National Institutes of Health)
Drug (company)CombinationStatusClinical trial IDConditionAdverse events
  1. HNSCC, Head and neck squamous cell carcinoma.

  2. a

    Open to all solid tumors including non-small-cell lung carcinoma, but might not include Head and neck squamous cell carcinoma HNSCC.

PI3K Inhibitors


(Novartis, East Hanover, NJ, USA)

CetuximabPhase I/IINCT01816984Recurrent or metastatic HNSCC

Rash, hyperglycemia,


Single agentPhase IINCT01737450Metastatic, recurrent, or progressive HNSCC


(Oncothyreon, Seattle, WA, USA)

CetuximabPhase I/IINCT01252628Incurable, progressive, recurrent or metastatic HNSCC

Diarrhea, nausea,


DocetaxelPhase I/IINCT01204099Locally advanced, recurrent or metastatic HNSCC



CetuximabPhase Ib/IINCT01602315Recurrent or metastatic HNSCC

Nausea, hyperglycemia,


Single agentPhase IaNCT01219699Advanced solid malignancies, including HNSCC, with an alteration of PIK3CA
mTOR Inhibitors



Single agentPhase IINCT01051791HNSCCRash, mucositis, fatigue
Single agentPhase IINCT01133678HNSCC
Carboplatin, cetuximabPhase I/IINCT01283334Advanced HNSCC
Carboplatin, paclitaxelPhase I/IINCT01333085Unresectable or inoperable locally advanced HNSCC
CetuximabPhase INCT01637194Metastatic or recurrent HNSCC
ErlotinibPhase IINCT00942734HNSCC


(Pfizer, New York,


Single agentPhase I/IINCT01195922Previously untreated HNSCC

Hyperglycemia, nausea,




Single agentPhase IINCT01172769HNSCC

Rash, mucositis, mood


CetuximabPhase IINCT01256385Recurrent or metastatic HNSCC
Carboplatin, paclitaxelPhase I/IINCT01016769HNSCC
PI3K/mT0R Dual Inhibitors



EverolimusPhase IbNCT01508104Metastatic or unresectable advanced solid tumorsaFatigue, nausea, diarrhea



Single agentPhase INCT00927823Advanced solid tumors including HNSCCFatigue, nausea, vomiting



Docetaxel, cisplatin, dacomitinibPhase INCT01920061Advanced solid tumorsaMucositis, nausea, fatigue
PD-901, irinotecanPhase INCT01347866Advanced solid tumorsa


(Sanofi, Bridgewater,


PimasertibPhase INCT01390818Advanced solid tumorsaRash, diarrhea, fatigue
Akt Inhibitors


(Merck, Whitehouse

Station, NJ, USA)

Single AgentPhase IINCT01349933Recurrent or metastatic HNSCCRash, nausea, pruritis
Figure 1.

Interactive signaling of PI3K/Akt/mTOR and other pathways. Relationships between well-characterized molecular signaling routes in HNSCC (EGFR/PI3K/PTEN/Akt/mTOR, RAS/RAF/MEK/ERK, IKK/NF-κB, and MDM2/p53) are shown by black arrows (activating interactions) and dashed red lines (negative regulation). PI3K/Akt/mTOR activation or inhibition ultimately regulates the activity of many kinases and transcription factors (purple) important in the nuclear control of cell survival and the malignant phenotype

EGFR/PI3K/Akt/mTOR signaling in normal physiology

The EGFR/PI3K/Akt/mTOR signaling cascade is critical to cell growth, development, survival, and differentiation. Epidermal growth factor receptor (EGFR), also known as ErbB-1 or HER1, contains an extracellular ligand-binding domain, a hydrophobic transmembrane portion, and a cytoplasmic domain containing tyrosine kinase activity and multiple phosphorylation sites (Kalyankrishna and Grandis, 2006; Normanno et al, 2006; Freudlsperger et al, 2011). At least 30 ligands of the four ErbB family members have been characterized (Rogers et al 2005). Ligand binding triggers receptor activation through dimerization and auto- or cross-phosphorylation of the tyrosine kinase domains (Lu et al, 2012), which initiates EGFR signal transduction through multiple intracellular pathways, including the PI3K/Akt/mTOR pathway.

There are three classes of PI3Ks (I, II, and III), classified by structure and substrate preference. Class I PI3Ks, the subject of this review, are divided into subclasses IA and IB (Vanhaesebroeck et al, 2012). While initially postulated that IA enzymes are activated by RTKs, and IB enzymes are stimulated through GPCRs, recent work shows crossover in these stimuli (Katso et al, 2001). PI3K IA catalytic subunits include p110α (PIK3CA), p110β, and p110δ – all of which contain an N-terminal p85 regulatory subunit-binding domain and a Ras-binding domain (Domin and Waterfield, 1997; Fruman et al, 1998). p110α and p110β are ubiquitously expressed, while p110δ is mainly found in leukocytes, as it is a key player in lymphoid differentiation and development (Vanhaesebroeck et al, 2010). PI3K class IB includes the catalytic p110γ subunit, which is similar to the p110α-δ subunits, but does not contain a p85-binding site (Katso et al, 2001). Activated EGFR recruits and activates class IA PI3K at the membrane through direct binding of the Src homology 2 domains of p85 with the phosphotyrosine residues of the activated receptor (Engelman, 2009). Upon activation, the catalytic p110 subunit phosphorylates phosphatidylinositol-4,5-bisphosphate (PIP2), converting it into phosphatidylinositol-3,4,5-trisphosphate (PIP3). Tumor suppressor PTEN antagonizes PI3K function by dephosphorylating PIP3 to regenerate PIP2. PIP3 functions as a second messenger with binding sites for signaling proteins with pleckstrin homology (PH) domains including 3-phosphoinositide-dependent protein kinase-1 (PDK1) and Akt (protein kinase B) (Liu et al, 2009).

Akt is a serine/threonine protein kinase with three isoforms (Akt1, Akt2, and Akt3) (Bellacosa et al, 1991), which require phosphorylation for activation (Scheid and Woodgett, 2001). Akt's N-terminal region contains a PH domain that interacts with PI3K products and initiates its recruitment to the plasma membrane, conformational change, and subsequent phosphorylation. Upon binding with PIP3, Akt is phosphorylated at three different sites: threonine 308 stabilizing the activation loop of the catalytic kinase domain by Phosphoinositide dependent kinase (PDK1) (Vivanco and Sawyers, 2002), serine 473 in the hydrophobic COOH-terminal domain by mTORC2 (mTOR complex 2) (Sarbassov et al, 2005), and serine 129 between the PH and catalytic domains by casein kinase 2 (CK2) (Di Maira et al, 2005). Activated Akt has multiple downstream targets involved in cell survival, growth, proliferation, angiogenesis, metabolism, and migration (Manning and Cantley, 2007).

A major downstream effector of Akt is mTOR, also a serine/threonine protein kinase, and the catalytic subunit of mTOR complex 1 (mTORC1). mTORC1 regulates protein translation through phosphorylation of protein synthesis components based on growth factor and nutrient availability (Hennessy et al, 2005). mTORC1 and mTORC2 are structurally and functionally unique complexes; mTORC1 contains raptor (rapamycin-sensitive adaptor protein of mTOR), while mTORC2 binds rictor (rapamycin-insensitive companion protein of mTOR). Activation of mTORC1 occurs through Akt-mediated inactivation of the tuberous sclerosis complex (TSC). The TSC consists of a heterodimer of tumor suppressor proteins hamartin (TSC1) and tuberin (TSC2) and antagonizes mTOR signaling by inhibiting the Ras-like small GTPase, Ras-homolog enriched in brain (Rheb), a direct regulator of mTOR signaling (Li et al, 2004; Pan et al, 2004). Well-characterized targets of mTORC1 include ribosomal protein S6 kinase (RPS6K or S6K) and eukaryote initiation factor 4E-binding protein 1 (EIF4EBP1 or 4EBP1), both key regulators of protein translation. S6K phosphorylation of the S6 protein increases translation of ribosomal proteins and translation regulators (Meyuhas, 2000; Inoki et al, 2005). 4EBP1, a translation inhibitor, binds and inhibits eukaryotic translation initiation factor 4E (eIF-4E), which initiates translation (Gingras et al, 1999). mTORC2 regulates actin cytoskeletal reorganization and serves as a hydrophobic motif kinase for Akt at serine 473 (Bhaskar and Hay, 2007).

EGFR/PI3K/Akt/mTOR pathway aberrations in HNSCC

Recent developments in deep sequencing of tumor tissues have paved a powerful path for discovery of the most frequent genetic alterations in HNSCC. Unbiased high-throughput approaches have revealed that PIK3CA is one of the most commonly mutated and amplified genes in the protein-coding human HNSCC genome (Agrawal et al, 2011; Stransky et al, 2011), which validates earlier PIK3CA-targeted studies (Okami et al, 1998; Kozaki et al, 2006; Qiu et al, 2006; Murugan et al, 2008; Cohen et al, 2011; Morris et al, 2011). PIK3CA missense mutations are concentrated in three ‘hotspots’ on chromosome 3q26 (H1047R, E545K, and E542K) known to affect constitutive PI3K activity (Samuels et al, 2004). Recent mutation analysis of 279 patients with HNSCC by TCGA has provided evidence that PIK3CA mutations are among the most frequent mutations in HNSCC (18% in HPV-negative cancers and 37% in HPV-positive cancers) (Grandis et al, 2012; Hayes et al, 2013). Altogether, genetic changes (mutation or DNA copy number variation) in major components of the PI3K pathway are represented in 66% of the TCGA HNSCC cases studied (Figure 2). When including changes in mRNA expression, this statistic jumps to over 90% (, with alterations in one component of the pathway often being mutually exclusive with changes in a different component. PIK3CA DNA copy number is amplified in 20% of TCGA cases and mRNA expression is upregulated in 52% the TCGA tumors (Iglesias-Bartolome et al, 2013), with reduced PTEN expression in ~30% of another HNSCC cohort (Squarize et al, 2013). As a whole, the cumulative changes in PI3K/Akt/mTOR and molecules that communicate with this axis are represented at a significant rate in HNSCC. Consequently, it serves as an attractive target for molecular-oriented therapy.

Figure 2.

The frequency of genetic alterations of signaling molecules involved in PI3K pathway. The genetic data were obtained from cBioPortal for Cancer Genomics (, where the data from The Cancer Genome Atlas (TCGA) project were deposited. The data were collected from 279 HNSCC tumor samples through high-throughput sequencing and other array technologies. Samples with genomic amplification (red), homozygous deletion (blue), and/or mutation (green) are presented in the OncoPrint as individual bars or square dots. Only those samples with genetic alterations in the PI3K/Akt/mTOR pathway are presented (183/279 tumors = 66% of cases). Percentages reflect the frequency with which samples express any of the three alterations reported in a given gene. Copy number variations are putative. PI3K catalytic subunit alpha isoform (PIK3CA); epidermal growth factor receptor (EGFR); eIF-4E binding protein 1 (EIF4EBP1); ribosomal protein S6 kinase (RPS6KB); Harvey rat sacrcoma virus oncogene (HRAS); protein kinase B (AKT); 3-phosphoinositide dependent protein kinase 1 (PDK1); phosphatase and tensin homolog (PTEN); tuberous sclerosis (TSC); Ras-homolog enriched in brain (RHEB); mammalian target of rapamycin (mTOR)

A recent analysis showed that druggable events (copy gain with increased expression or activating mutation) associated with the PIK3CA or AKT1/2/3 loci could be targeted in 12/35 (34%) HNSCC tumors sequenced (Pickering et al, 2013). In a functional test of the druggability of PIK3CA mutations in HNSCC patient tumor grafts, mutated rather than wild-type PIK3CA conferred a greater sensitivity to the PI3K/mTOR inhibitor BEZ-235 (Lui et al, 2013). The same result held true for the PI3K inhibitor PX-866 in HNSCC patient tumor grafts (Keysar et al, 2013). Thus, the predilection of PI3K inhibitors activity for HNSCC with PIK3CA activating mutations provides one rationale for using PIK3CA mutational status as a predictor of treatment selection. When considering the importance of human papilloma virus (HPV) oncoproteins E6 and E7 in the etiology of a subset of HNSCC, and the association of HPV with PI3K pathway mutations, this rationale is strengthened. In a directed analysis of 15 genes in 64 HNSCC tumors, PIK3CA mutation was the most abundant mutation, found at a much higher frequency in the oropharynx than in any other primary site (McBride et al, 2013). This observation is consistent with a putative association between higher rates of HPV infection in oropharynx than in other head and neck cancer anatomic sites (Gillison et al, 2000), and higher rates of PIK3CA mutation in HPV+ HNSCC than in HPV- HNSCC (Lui et al, 2013; Nichols et al, 2013).

PI3K/Akt/mTOR aberrations promote the malignant phenotype

Functionally, aberrations in the PI3K/Akt/mTOR pathway are associated with many of the malignant characteristics of HNSCC, including immune suppression and inflammation, angiogenesis, survival, invasion, and metastasis. Inflammation, in specific, regulates many of these processes simultaneously (Grivennikov et al, 2010). In contrast to ubiquitously expressed PI3K p110-α and -β isoforms, p110-δ is mainly present in leukocytes and p110-γ in myeloid cells (Koyasu, 2003; Vanhaesebroeck et al, 2005). As such, both p110-δ and p110-γ are crucial for promoting inflammation in the cancer microenvironment. Although most of the PI3K pathway-directed anti-cancer therapies have aimed at p110-α suppression, isoform-specific inhibition of p110- δ or -γ also attenuates dysregulated inflammatory and immunosuppressive responses and may have an improved safety profile (Rommel et al, 2005; Fruman and Rommel, 2011; Schmid et al, 2011).

Dysregulated inflammatory signaling mediates angiogenesis, a critical process enabling tumor growth and spread. Angiogenesis is promoted by the secretion of cytokines such as vascular endothelial growth factor (VEGF), basic fibroblast growth factor (b-FGF), and interleukin-8 (IL-8) by tumor cells and associated myeloid lineage infiltrating cells. PI3K is activated downstream of the receptors for these signaling molecules and acts as a master regulator of angiogenesis and associated cytokine release (Yuan and Cantley, 2008). mTORC1 phosphorylation of eIF-4E facilitates tumor progression by increasing b-FGF and VEGF and increases risk for second primary in histologically cancer-free surgical margins (Nathan et al, 1999). While increased VEGF is associated with poor prognosis in HNSCC (Luangdilok et al, 2011), we have shown that pharmacologic inhibition of PI3K reduces phosphorylation of Akt in HNSCC cell lines, thereby blocking the secretion of VEGF and IL-8, even under epidermal growth factor (Bancroft et al, 2002) or hepatocyte growth factor stimulation (Dong et al, 2001a). In preclinical mouse studies, constitutive Akt activity causes abnormal vascular patterning (Hamada et al, 2005), and conditional class IA PI3K knockout compromises vascular integrity, limiting tumor size (Graupera et al, 2008; Yuan et al, 2008).

Although the PI3K/Akt/mTOR pathway enables vascular formation and growth factor delivery to tumors, it also confers a selective growth advantage and survival in conditions of nutrient deprivation. Normal cells live within 100–200 μm of a capillary bed, the diffusion limit of oxygen (Baish et al, 2011). Malignant cells with aberrations in PIK3CA, PTEN, or MTOR, however, are able to continue proliferating in hypoxic conditions outside of this range because of a modified cellular response to hypoxic stress, nutrient sensing, insulin signaling, and glucose uptake (Abraham, 2004; Kalaany and Sabatini, 2009; Kang et al, 2013; Vogelstein et al, 2013). Akt also promotes cell survival by attenuating pro-apoptotic factors, BAD, BAX, procaspase-9, and FOXO (Hennessy et al, 2005; Engelman et al, 2006).

Tumor outgrowths as a result of aberrant PI3K/Akt/mTOR signaling can subsequently lead to invasion (Samuels et al, 2005). Mechanistically, PI3K oncogenic disruption through PTEN loss disturbs membrane polarity and leads to epithelial-to-mesenchymal transition (Martin-Belmonte and Mostov, 2008), a key event causing invasion. 3q26 (PIK3CA locus) copy number gain and amplification is much more frequent in invasive carcinoma and high-grade dysplasia than in low-grade dysplasia (Woenckhaus et al, 2002). Late-stage HNSCC lesions also express higher levels of PI3K p110α mRNA and protein (Woenckhaus et al, 2002), lower levels of PTEN (Squarize et al, 2002, 2013), and more simultaneous mutations in multiple PI3K pathway genes (Lui et al, 2013). In addition, increased phospho-Akt and phospho-S6 expressions have been correlated with HNSCC progression (Amornphimoltham et al, 2004; Molinolo et al, 2007). This suggests that disruption of any number of components in the PI3K pathway might act as an oncogenic switch leading to invasion. Once invasion has occurred, PI3K-PTEN interplay regulates chemotaxis and intravasation of cells into endothelial networks (Kolsch et al, 2008). Consequently, the capacity of the PI3K/Akt/mTOR pathway to concurrently regulate the processes of angiogenesis, cell motility, and invasion also gives it solid footing for contributions to metastasis.

Accordingly, lymph node metastases are directly associated with PIK3CA gain and mutation (Fenic et al, 2007). Because metastasis is the leading cause of death for patients with solid tumors, PI3K activation portends a worse prognosis. Indeed, two independent studies of HNSCC show that 3q26 gain (PIK3CA locus) can predict poor clinical outcome for patients with early disease tumors that have not yet metastasized (Redon et al, 2001; Suda et al, 2012). Reduced PTEN expression is a separate prognostic indicator of poor clinical outcome (Lee et al, 2001). Additionally, mTOR blockade is able to diminish lymphangiogenesis and cervical lymph node spread, thereby increasing survival (Patel et al, 2011). Because PI3K, Akt, and mTOR activation and PTEN inactivation are all independently associated with adverse outcome in HNSCC (Nathan et al, 2004; Massarelli et al, 2005; Molinolo et al, 2009), they serve as logical targets in clinical trials.

PI3K/Akt/mTOR interactions with other signaling molecules

TP53 is one of the best-studied and most frequently deleted or mutated tumor suppressor genes. Inactivation of p53 can occur through physical association with murine double minute 2 homolog (MDM2), which results in centrosome hyperamplification, chromosomal instability, and aberrant mitosis (Carroll et al, 1999). PI3K phosphorylation of Akt activates MDM2 by translocating it from the cytoplasm to the nucleus, thereby diminishing cellular levels of p53 (Mayo and Donner, 2001). Concurrent inactivating mutations in TP53 and activating abnormalities in PIK3CA are uncommon in cancer, implying that they are often mutually exclusive and independent promoters of malignancy. In HNSCC, PI3KCA activation confers resistance to p53-induced apoptosis, while p53-mediated apoptosis involves transcriptional inhibition of PIK3CA (Singh et al, 2002). We have shown that treatment with the PI3K/mTOR inhibitor, PF-04691502 (PF-502), induces apoptosis and expression of wild-type p53 in two separate in vivo models of HNSCC (Herzog et al, 2013). We have also provided evidence that Akt inhibition reverses the direct suppression of p53 by transcriptional and post-translational mechanisms (Friedman et al, 2013). In glioma and endothelial cells, PI3K inhibition has also been reported to increase p53 transactivation and transcription of its target genes (Su et al, 2003). Furthermore, p53 is able to induce PTEN, favoring cell death in malignancy (Mayo and Donner, 2002). Other studies have evinced that mTOR signaling contributes to p53 inactivation through IGF-1R-mediated upregulation of MDM2 (Du et al, 2013), and the use of the mTOR inhibitor, rapamycin, is able to prevent oral cancer progression in a K-ras/p53 double-knockout mouse model (Raimondi et al, 2009). Thus, therapies targeting PI3K/mTOR may be doubly effective because of their capacity to both limit PI3K/Akt/mTOR-mediated growth and survival and intensify p53-mediated apoptosis.

In contrast to loss of p53 as a mechanism for evading apoptosis, overactivity of the transcription factor, NF-κB, is another important anti-apoptotic and prosurvival event in HNSCC (Van Waes, 2007). We have compiled a large body of evidence showing that NF-κB regulates the gene expression and secretion of multiple cytokines and chemokines that favor tumor progression (Chen et al, 1998, 1999; Duffey et al, 1999; Loukinova et al, 2000; Dong et al, 2001b; Allen et al, 2007). Activation and nuclear translocation of NF-κB occurs primarily through phosphorylation of inhibitor-κB kinases (IKK) (Van Waes et al, 2007; Nottingham et al, 2013), which can be phosphorylated downstream of active PI3K/Akt (Ozes et al, 1999; Romashkova and Makarov, 1999; Yang et al, 2001; Hutti et al, 2012) in a cell-type-specific manner (Gustin et al, 2004). Accordingly, we have shown that the PI3K inhibitor, LY-294002, is able to reduce both constitutive and inducible NF-κB activity in HNSCC (Bancroft et al, 2002). Furthermore, rapamycin has also been shown to inhibit IKK activity, as the stimulation of IKK through Akt requires the interaction of the mTOR-associated protein, raptor, with IKK (Dan et al, 2008). While the mechanisms of interaction between the PI3K/Akt/mTOR and NF-κB pathways have been elucidated in many cancers, it remains to be studied further in HNSCC.

The PI3K/Akt/mTOR signaling route is also activated in concert with the Ras/Raf (MAPK kinase kinase)/MAPK pathway downstream of EGFR. Although the contributions of these signaling pathways to malignancy are often independent of EGFR overactivity, recent evidence indicates that MAPK activity is differentially affected by the EGFR-dependent PI3K/Akt/mTOR axis. Many studies have shown that through various feedback loops or resistance mechanisms, inhibition of one pathway downstream of EGFR or IGF-1R causes compensatory signaling through other parallel pathways (Carracedo et al, 2008; Pernas et al, 2009; Ercan et al, 2012; Limesand et al, 2013). The pleiotropy and resistance to therapy of downstream EGFR signaling has led to preclinical and clinical trials of combinations of EGFR, PI3K, mTOR, MAPK kinase kinase, effector of Ras (RAF), and MAPK kinase (MEK)/ERK inhibitors. Although the interactions between the PI3K/Akt/mTOR axis and p53, NF-κB, and MAPK signaling pathways are perhaps the most studied, this is by no means an exhaustive list of its interactions. The EGFR/PI3K/Akt/mTOR axis has also been shown to be modified in HNSCC in response to the modulation of transforming growth factor beta (TGF-β) (Bian et al, 2012), signal transducer and activator of transcription 3 (STAT3) (Lee et al, 2008; Sen et al, 2012), and Wnt signaling (Kavitha et al, 2013), among others.

PI3K/Akt/mTOR inhibitors in clinical studies

The summary of currently ongoing clinical trials using PI3K pathway inhibitors for head and neck cancer is presented in Table 1. BKM120, a pan-isoform PI3K inhibitor which is furthest along in development, is currently in a phase III trial for breast cancer (NCT01633060). Several studies have been initiated to select for patients with HNSCC, including a phase II trial using BKM120 as a single agent and a phase I/II study combining BKM120 with cetuximab. The orally available, irreversible pan-isoform PI3K inhibitor PX-866 has also moved into phase II trials for HNSCC. Although showing limited effect as a single agent in solid tumors including HNSCC (Hong et al, 2012), phase I/II trials combining PX-866 with cetuximab and docetaxel have been initiated in patients with HNSCC and may provide further insight into the efficacy of this drug.

Several isoform-specific PI3K inhibitors are currently in clinical trials, offering the potential benefit of improved target selectivity with fewer off-target side effects. Of these, only BYL719, a PI3Kα-specific inhibitor, has been applied to patients with HNSCC. The phase I trial on this drug was the first to selectively enroll patients with advanced solid tumors with prescreened mutations or amplifications of PIK3CA. In the dose escalation portion, eight patients including those with oral cancers showed prolonged disease stabilization lasting ≥4 months (Juric et al, 2012), while the dose expansion study that followed reported a partial response in 7/39 patients, one of which had HNSCC (Gonzalez-Angulo, 2013). The safety profile for BYL719 was similar to pan-PI3K inhibitors studied, with the most common adverse effects including hyperglycemia, nausea, and GI toxicities. Based on the encouraging results from this initial trial, BYL719 is currently being evaluated in combination with cetuximab in a phase Ib/II trial in patients with HNSCC.

Previously, the Akt inhibitor perifosine showed promising results in preclinical studies, but when moved to phase II clinical trials in HNSCC, it failed to show any anti-tumor effect (Argiris et al, 2006; Hideshima et al, 2006). More recently, a newer Akt inhibitor MK-2206 has shown synergistic activity with paclitaxel in preclinical models and phase I trials of HNSCC (Ahmed et al, 2013). It is currently being evaluated in phase II trials as a single agent in recurrent or metastatic HNSCC and may prove to be more effective than previous agents.

Rapamycin, the first mTOR inhibitor, targets mTOR exclusively and has shown anti-tumor activity in preclinical models (Amornphimoltham et al, 2005). The first clinical trial of rapamycin as a single agent in HNSCC is ongoing in phase I/II. Several rapamycin analogues have been developed, including temsirolimus (CCI-779) and everolimus (RAD001). Temsirolimus, after showing initial safety and strong inhibition of mTOR targets S6 and 4E-BP1 in HNSCC patient tumors (Ekshyyan et al, 2010), has moved to phase II trials in HNSCC as a single agent, in combination with carboplatin and paclitaxel or with cetuximab. Meanwhile, everolimus is also being evaluated extensively in HNSCC, with two phase I trials that have recently completed. In the first trial, combining everolimus with cisplatin and docetaxel as induction therapy for radiation, 2-year overall survival rate was 91% and progression-free survival rate at 2 years was 76.6% (Fury et al, 2013b). In the second trial, everolimus was given with cisplatin and concurrent radiation, showing a 2-year overall survival rate of 92% and 2-year progression-free survival rate of 85% (Fury et al, 2013a). The most impressive response, however, has come from the ongoing phase I-II trials of everolimus in combination with carboplatin and paclitaxel as induction therapy in patients with unresectable locally advanced HNSCC. Recently presented results showed a 2.6% complete response, 76.3% partial response, as well as 21% stable disease in the 38 evaluable patients (Raymond et al, 2013). Everolimus continues to be investigated alone and in combinations in patients with HNSCC in five additional phase I-II trials (Table 1).

An important drawback in selectively inhibiting mTOR is the negative feedback loop that exists, in which S6K phosphorylates and blocks insulin receptor substrate 1 (IRS1), leading to the activation of PI3K and subsequent tumor growth (O'Reilly et al, 2006; Huang and Manning, 2009). To overcome this drawback, there is growing interest in the development of dual PI3K-mTOR inhibitors that can target the pathway at two points simultaneously. Dual PI3K-mTOR inhibitors that have shown promising results in preclinical studies of HNSCC include the orally available PF-502 (Herzog et al, 2013) and intravenously administered PF-05212384 (PF-384). Preliminary results from a phase I trial of PF-502, which is currently open to HNSCC, but to date only includes other advanced solid tumors such as non-small-cell lung carcinoma (NSCLC) and colorectal carcinoma, have shown no objective tumor reduction, but stable disease ≥16 weeks in approximately 16% of patients (LoRusso et al, 2011). Similarly, approximately 25% of patients with solid tumors including NSCLC given weekly PF-384 as a single agent achieved stable disease ≥16 weeks, although no objective tumor reduction was observed (Tabernero et al, 2011). Additional phase I trials of PF-384 in combination with docetaxel, cisplatin, and dacomitinib (EGFR inhibitor), as well as irinotecan (topoisomerase 1 inhibitor) and PD-901 (MEK inhibitor), may prove to have greater anti-tumor effects than single agents alone in HNSCC.

Conclusion and future directions

Following these initial clinical trials of agents targeting the PI3K pathway, several key questions remain. Additional studies will necessitate decision making about the most biologically active dose for each drug, the ideal method of administration, and the acceptance of adverse events including on-target effects such as hyperglycemia. It remains to be determined which class of agents targeting PI3K, mTOR, or Akt will be most effective in HNSCC, and whether there are factors that would make one class superior to the other in specific cases. Several clinical trials have retroactively analyzed PIK3CA mutation status in patient samples, while few have limited enrollment to patients with known PI3K pathway mutations. There has been some evidence that preselected patients with mutations in PI3K have better response to treatment, revealing improved outcomes compared to non-selective trials. Prescreening for molecular alterations has proven beneficial in preclinical studies as well, as it has uncovered critical mechanistic information regarding sensitivity or resistance and effects on other pathways. PI3K pathway cross-talk with other signaling routes, such as p53, NF-κB, and MAPK, should be investigated more extensively when designing future clinical trials. As large-scale screening protocols for molecular alterations become more prevalent in drug development trials, the feasibility of translating genetic information about the PI3K/Akt/mTOR axis to the clinic increases. Ultimately, this will allow clinicians to tailor treatment to an individual molecular profile and improve patient care.


R.V., S.M., and D.E. were supported through the National Institutes of Health (NIH) Medical Research Scholars Program, a public–private partnership supported jointly by the NIH and generous contributions to the foundation for the NIH from Pfizer Inc, The Leona M. and Harry B. Helmsley Charitable Trust, and the Howard Hughes Medical Institute, as well as other private donors. For a complete list, please visit the foundation website at Z.C. and C.V. were supported by NIDCD intramural project ZIA-DC-000016.

Author contributions

R. Vander Broek, S. Mohan, D.F. Eytan, Z. Chen, and C. Van Waes contributed to the conception, design, drafting, and editing of the manuscript, tables, and figures.