Introduction

Uterine, or endometrial, cancer is the most common malignancy of the female genital tract [1]. With an estimated 43,470 new cases in the United States (US) in 2010, uterine cancer is the fourth most common cancer in American women [1]. Over the past 15 years, the incidence of uterine cancer in the US has increased by over 30% [1] and [2]. This upward trend is expected to continue due to an aging population and high levels of obesity [3] and [4]. Although most women diagnosed with uterine cancer have an excellent prognosis, an estimated 7950 women die from the disease each year in the US. Uterine cancer is classified as the second leading cause of mortality from gynecologic malignancies and the eighth leading cause of cancer-related death overall in American women [1].

For women with high-risk uterine cancers or recurrent disease, several chemotherapy regimens have shown activity [5] and [6]. However, no second-line chemotherapy regimen has produced a response rate of higher than 20%. If all women with advanced uterine cancer and 20% of women with stage I disease were treated with chemotherapy, approximately 14,000 women in the US would be receiving primary chemotherapy for uterine cancer every year [7] and [8]. Critically, for the 7000 women every year who develop recurrent metastatic disease, treatment options are extremely limited after primary chemotherapy. Therefore, there is a high unmet need for novel therapies.

An improved understanding of the biologic pathways leading to the development of uterine cancer provides an opportunity to develop targeted therapies. The phosphatidylinositol 3-kinase (PI3K) growth pathway is often altered in uterine cancer. Several drugs that inhibit different nodes of this pathway are in preclinical development or in clinical trials. They represent promising candidates for the future treatment of uterine cancer. This article will review the clinical development and mechanism of action of these drugs.

The PI3K pathway

The PI3K pathway transmits growth factor signals from transmembrane receptor tyrosine kinases (RTKs), including epidermal growth factor receptor (EGFR), human epidermal growth factor receptor 2 (HER2), vascular endothelial growth factor receptor, and insulin-like growth factor receptor, to activate cellular growth and survival mechanisms [9] and [10]. The oncogenic potential of the PI3K pathway was first demonstrated when abnormal cellular signaling caused by mutations in viral oncoproteins was associated with malignant growth [11]. Further evaluations indicated that this pathway was frequently mutated in human cancer, which has led to a drive for therapeutic intervention.

Members of the PI3K family of enzymes share the ability to phosphorylate the 3-hydroxyl group of phosphoinositides [12]. They are categorized into three major classes according to their structural characteristics and substrate specificity. Class IA PI3Ks have been associated with malignant transformation in many different cancer types. They are heterodimers comprising a regulatory and catalytic subunit. The PIK3R1, PIK3R2, and PIK3R3 genes encode a total of five splice variants collectively known as the regulatory subunit, p85 [13]. The catalytic subunit of class IA PI3K is referred to as p110, and the three isoforms are encoded by the PIK3CA, PIK3CB, and PIK3CD genes. The function of PI3K enzymes is primarily related to the p110 subunit. Within the class IA PI3K family, the p110-alpha subunit is involved in cellular growth signaling and is the isoform most often mutated or amplified in cancers. The p110-beta subunit is involved in insulin signaling [13]. The p110-delta subunit is primarily expressed in leukocytes and is likely to be related to immune function. Class IB PI3K enzymes have a p100-gamma subunit, which is also expressed in leukocytes. Class II PI3K enzymes are monomeric and involved in cell surface trafficking, and class III PI3K enzymes are implicated in the regulation of autophagy [13].

The PI3K heterodimer of the class IA enzyme complex is closely associated with phosphotyrosine residues on the intracellular membrane portion of RTKs. In an inactive state, p85 blocks the catalytic site on the p110 subunit. Once an appropriate ligand has bound the RTK, a conformational change occurs in p85 to activate the catalytic site of p110. Active p110 generates the second messenger phosphatidylinositol 3,4,5-triphosphate (PIP3) from phosphatidylinositol 4,5-diphosphate (PIP2) at the cell membrane (Fig. 1). This process is regulated by PTEN, a tumor suppressor oncogene, which downregulates the signal by converting PIP3 back to PIP2. Cells deficient in PTEN often have overactive PI3K pathway signaling due to lack of downregulation. Recently, a second tumor suppressor has been identified, inositol polyphosphate 4-phosphatase type II (INPP4B), which negatively regulates the PI3K pathway by converting PIP2 to phosphatidylinositol 3-phosphate (PIP) [14]. Dual knockdown of INPP4B and PTEN in human epithelial cells has been shown to result in cellular senescence [14]. The PI3K pathway can also be activated by RAS, which directly activates p110 [10]. Mutations in KRAS can therefore activate the PI3K pathway directly in a manner independent of extracellular signaling [10], [15] and [16].

PIP3 binds to the pleckstrin homology domain in the N-terminal region of AKT, anchoring it to the inner cell membrane. This action induces a conformational change in AKT revealing two amino acid residues that are essential for full activation. Phosphorylation at one of these residues occurs at threonine 308, by phosphoinositide-dependent kinase 1 (PDK1). PDK1 is constitutively active but is upregulated by activation of PI3K. Activated AKT promotes cellular growth, proliferation, angiogenesis, and prevents apoptosis through a number of downstream effectors. One such effector is tuberous sclerosis protein 2 (TSC2), which forms a heterodimer with TSC1. AKT phosphorylates and inhibits TSC2, which represses rheb (RAS homolog enriched in brain), leading to activation of the mammalian target of rapamycin (mTOR) pathway [17]. mTOR is present in two distinct protein complexes: mTORC1 and mTORC2. The mTORC1 consists of mTOR and raptor (regulator associated protein of mTOR), and activation increases production of two major downstream targets: eukaryotic initiation factor 4E binding protein 1 (4EBP1) and S6 kinase (S6K) [18]. Activated S6K serves to suppress PI3K signaling through a negative feedback loop (Fig. 1) [18]. The mTORC2 consists of mTOR and rictor (rapamycin insensitive companion of mTOR). Active mTORC2 phosphorylates AKT at serine 473 [18]. This, in combination with the phosphorylation event at threonine 308, fully activates AKT and results in a positive feedback loop between AKT and mTOR [18]. AKT signaling is suppressed by active S6K, which inactivates mTORC2. Drugs that solely inhibit mTORC1 can release the negative feedback and actually increase expression of AKT due to this effect. This represents a major limitation for treatment with mTORC1 inhibitors.

Cross-talk between PI3K and RAS/RAF/MEK pathways

There is considerable ‘cross-talk’ between the PI3K pathway and other key signaling cascades, such as the RAS/RAF/MEK (Fig. 1). In addition to its role in the RAS/RAF/MEK pathway, RAS also directly activates the p110 subunit of PI3K [19]. Several studies indicate that activation of the PI3K pathway is required for RAS-induced tumor initiation [20] and [21]. Cross-talk between the PI3K and RAS/RAF/MEK pathways also occurs at other nodes. For example, ERK, the downstream target of the RAS/RAF/MEK pathway, activates mTORC1 through inhibition of TSC2, or activation of its effector protein, ribosomal S6 kinase. AKT has also been demonstrated to inactivate BRAF through phosphorylation [22].

The PI3K pathway in uterine cancer

Risk factors for uterine cancer and the PI3K pathway

Some major risk factors for uterine cancer may lead to increased PI3K pathway signaling. Obesity is the best known independent risk factor for uterine adenocarcinoma and other hormone-dependent malignancies [23]. In post-menopausal women, this association is thought to be mediated through increased estrogen levels, which occur in response to aromatase activity in adipose tissue [24]. While the majority of Estrogen Receptors (ERs) are located in the cytoplasm and the nucleus, a small fraction are located on the cell membrane, and may be responsible for some rapid cellular effects of estrogen [25]. Activation of cell-surface ERs initiates signaling via the PI3K pathway, by direct interaction with the p85 subunit of PI3K, or via other pathways, such as RAS/RAF/MEK [26]. In addition, estrogen-mediated activation of the G-protein coupled receptor GPR30 can activate PI3K signaling [27].

Diabetes and insulin resistance, leading to increased insulin and blood glucose levels, are established independent risk factors for uterine and other cancers [23]. Elevated insulin levels stimulate PI3K signaling via activation of the insulin RTK. Demographic studies suggest that the anti-diabetes drug metformin may lower the risk of malignancy in diabetic patients through activation of the 5′ adenosine monophosphate-activated protein kinase (AMPK) pathway, which regulates glucose metabolism. Activation of AMPK phosphorylates TSC2 on residues distinct from those phosphorylated by AKT, thereby activating it, leading to mTOR inhibition and a reduction in translation initiation [28]. Metformin has also recently been shown to decrease AKT activation [29].

Genetic alterations of the PI3K pathway in uterine cancer

Genetic alterations in the PI3K pathway occur frequently in uterine cancers, leading to hyperactive pathway signaling. Type I and Type II uterine cancers demonstrate a different profile in terms of their underlying genetic alterations. The most frequent genetic alterations in Type I cancers occur in PTEN (Table 1) [30], [31], [32], [33] and [34]. Mutations in PIK3CA, KRAS, and AKT have also been reported (Table 1). Type II cancers are characterized by p53 mutation, inactivation of p16 and overexpression of HER2. Alterations in PTEN, PIK3CA, and KRAS have also been reported in Type II cancers (Table 1). Although most PI3K pathway alterations occur at lower rates in Type II cancers, PIK3CA mRNA overexpression and exon 20 PIK3CA mutation rates were found to be higher than in Type I cancers [33]. Genetic alterations in the PI3K pathway are widespread in uterine cancers, and thus represent a valid therapeutic target in both Type I and Type II tumors.

Sporadic loss of function mutations in PTEN occurs in up to 83% of Type I uterine cancers. Homozygous and hemizygous deletions of PTEN have also been observed in other tumor types [35] and [36], together with PTEN inactivation through epigenetic silencing due to promoter hypermethylation [10]. Glands devoid of PTEN exist in the normal endometrium and persist between menstrual cycles [37]. These glands are phenotypically normal, and so are not observed with normal microscopy. PTEN-null glands are more common in proliferative and hyperplastic endometrium, and may be more prone to stimulation of the PI3K pathway by excess estrogen or insulin-like growth factor-1 signaling, leading to malignant transformation [32] and [38].

Activating mutations in PIK3CA lead to gain of function of the p110 catalytic subunit of PI3K. There are two mutational hotspots where approximately 80% of mutations occur [39]. The first is in exon 9 which codes for the protein domain that interacts with p85. It is theorized that mutations in this area of the protein prevent the p110 catalytic subunit from being controlled by the p85 regulatory subunit. The second hotspot is in exon 20, which codes for the catalytic domain and probably creates a protein that is unaffected by binding with the p85 subunit. Mutations in both areas can lead to increased PIP3 and thus activation of the PI3K pathway. Mutations in PIK3R1, which encodes the p85 regulatory PI3K subunit, also lead to increased pathway signaling due to a disabled regulatory function of p85 over the p110 catalytic subunit. Mutations in KRAS can directly activate p110 and have been detected in approximately 15% of uterine cancers [16]. Although gain of function mutations in AKT–PDK enzymes can cause over-expression of important downstream targets, they are uncommon in uterine cancer (< 3% of cases) [40]. Loss of heterozygosity of the newly identified putative tumor suppressor INPP4B has been identified in a variety of other epithelial cancers, but currently a role in uterine cancer is unconfirmed [14].

Effects of PI3K pathway activation on prognosis in uterine cancer

Risk factors for uterine cancer and genetic alterations are both associated with increased PI3K signaling, which is implicated in disease development, progression, and tumor grade [41] and [42]. Several studies have also reported a link between activation of the PI3K pathway and prognosis. Among 99 patients with advanced endometrial cancer, those classified as having PTEN-positive and p-AKT-negative expression demonstrated a higher survival rate than other patients [41]. This was confirmed as an independent prognostic factor of survival by a subsequent multivariate analysis [41]. Furthermore, amplifications in PIK3CA were associated with poor recurrence-free survival compared with uterine carcinomas without the amplification (P = 0.002) [42]. In contrast, deregulation of the PI3K–AKT signaling pathway was only shown to influence overall survival when coexisting with p53 alterations. In these patients, survival was shorter (P < 0.001) than in patients with p53 alterations alone [33].

Targeting the PI3K pathway in uterine cancer

The PI3K pathway is unique in that dysregulation can occur as a consequence of multiple alterations at different nodes. The high frequency of these alterations in uterine cancer makes the PI3K pathway an attractive therapeutic target. There are currently several PI3K types of pathway inhibitors in Phase I trials in advanced solid tumors, as well as Phase II trials in uterine cancer (Table 2).

Receptor tyrosine kinase inhibitors

The first drugs targeting the PI3K pathway were monoclonal antibodies against RTKs, such as cetuximab and panitumumab against EGFR, and trastuzumab against HER2. More recently, novel small molecule inhibitors of the intracellular portions of RTKs, such as lapatinib (anti-EGFR and anti-HER2) and erlotinib (anti-EGFR) have been approved by the US Food and Drug Administration (FDA) in several oncology indications. However, given the relatively high incidence of PI3K pathway alterations downstream of the RTK, this may not be an optimal strategy in uterine cancer. In clinical trials of tyrosine kinase inhibitors, erlotinib demonstrated a partial response rate of 13%, and cetuximab demonstrated a total clinical response rate of only 15% (1 partial response and 2 stable disease) in women with recurrent and metastatic uterine cancer [43] and [44].

mTORC1 inhibitors

Rapamycin analogs, also known as rapalogs, inhibit mTORC1 and have been approved by the FDA in some indications. Temsirolimus, everolimus (RAD001), and ridaforolimus are being evaluated in uterine cancer as single agents or in combination with other agents (Table 2). In two Phase II studies of single-agent temsirolimus, rates of partial response were 26% [45] and 7% [46] in chemotherapy-naïve and previously treated patients with recurrent/metastatic uterine tumors, respectively. Grade 3/4 adverse events included pneumonitis, mucositis, fatigue, gastrointestinal pain, hypokalemia, hyperglycemia, hypoalbuminemia, and hypophosphatemia [46]. In combination, full doses of temsirolimus and topotecan were not tolerated in women with advanced and/or recurrent gynecologic malignancies including uterine cancer [47]. Similarly, in combination with hormone therapy (megestrol acetate alternating with tamoxifen), temsirolimus resulted in an unacceptable rate of venous thrombosis, and only modest activity in patients with advanced uterine cancer [48].

In a Phase II study of everolimus monotherapy in uterine cancer, 6 of 28 evaluable patients (21%) had a confirmed clinical benefit response (CBR) defined as a complete or partial response or prolonged stable disease (> 8 weeks) by RECIST at 20 weeks of therapy [49]. The most common drug-related toxicities were fatigue, anemia, pain, lymphopenia, and nausea. In combination with letrozole, everolimus demonstrated a CBR in 42% of patients. The confirmed objective response rate was 21%, with 1 patient demonstrating a complete response, and 3 patients with a partial response [50]. The most common drug-related toxicities were fatigue, nausea, stomatitis, hypertriglyceridemia, and hyperglycemia [50].

Adverse events were frequent in a Phase II study of ridaforolimus monotherapy in patients with metastatic/advanced recurrent uterine cancer, with 39% of patients discontinuing due to toxicity [51]. Of 26 evaluable patients, 2 had a confirmed partial response (8%), and 15 experienced stable disease (58%) [51]. In a Phase II study, PFS was 3.6 months in the ridaforolimus arm compared with 1.9 months for progestin (hazard ratio [HR] = 0.53; one-sided P = 0.008) in patients with advanced uterine cancer (N = 114) [52]. Rates of partial response were 4% in the control arm as assessed by independent radiology review; no partial responses were reported in the experimental arm. Mean survival was not statistically different between arms (8.9 months for control vs 10.0 months for ridaforolimus; HR 0.93, 0.55–1.58; P = 0.4). Toxicity was also greater in the experimental arm with hyperglycemia, anemia, back pain and asthenia the most commonly reported adverse events [52].

mTORC1/2 inhibitors

A limitation of mTORC1 inhibition relates to the negative feedback loops that regulate the PI3K pathway controlled by mTORC1. Inhibition of mTORC1 alone can lead to a paradoxical increase mTORC2 activity, thus perpetuating AKT signaling [53]. Dual mTORC1 and mTORC2 inhibition may provide more effective pathway shutdown [54] and [55], and small molecule ATP-competitive inhibitors of mTORC1 and mTORC2 have been developed [18]. AZD2014, AZD8055, and OSI-027 are currently in early-stage clinical trials in solid tumors, including uterine carcinoma (Table 2).

PI3K inhibitors

Several inhibitors that target all three class IA, and the class IB PI3K p110 enzymes are in Phase I trials (Table 2). BKM120 and XL147 are currently being evaluated as monotherapy in Phase II studies in uterine cancer. PI3K inhibitors specific for individual class I p110 isoforms have the theoretical advantage of an improved safety profile compared with pan-PI3K inhibitors, which inhibit all four class I p110 isoforms [13]. The majority of PI3K-activating mutations are observed in the p110-alpha isoform of the class IA enzyme (coded for by PIK3CA). An inhibitor against the p110-alpha isoform (BYL719) is currently being investigated in a Phase I study in patients with advanced solid tumors and PIK3CA activating mutations (NCT01219699). Although activating mutations have not been seen in the p110-beta isoform, it has been noted that p110-beta is required for PI3K signaling in a PTEN-deficient mouse tumor model, whereas p110-alpha is not [56]. In addition, non-p110-alpha isoforms can induce malignant transformation without mutation [57]. Inhibitors selective for p110-alpha and p110-beta are in preclinical development.

Dual PI3K/mTORC1/2 inhibitors

Dual inhibition of PI3K and mTORC1/2 demonstrated greater growth inhibition in uterine cancer cell lines than mTORC1 inhibition alone [58]. This suggests that inhibition of multiple nodes of the PI3K pathway may be important for maximal suppression of tumor growth [58] and [59]. Cells with PIK3CA and/or PTEN mutations, without KRAS mutations, were particularly sensitive to dual PI3K/mTORC1/2 and mTORC1 inhibition [58] and [59]. As mutations in both PIK3CA and PTEN occur with high frequency in uterine cancer, dual PI3K/mTOR inhibitors are a rational molecular-targeted therapy for this disease [58]. BEZ235 is currently being evaluated as monotherapy in a Phase II trial in uterine cancer (Table 2).

AKT inhibitors

Inhibitors of AKT are either competitive, and contend with ATP at the active site, or are allosteric, and bind distally to the catalytic site and induce a conformational change that prevents ligand binding. As has been observed with rapalogs, inhibition of AKT could abrogate negative feedback loops, leading to a paradoxical increase in activity among non-AKT-dependent effectors of the PI3K pathway. In preclinical models, perifosine (KRX-0401) inhibited the growth of experimental models of human uterine cancers by blockade of AKT phosphorylation, and these effects were increased with cisplatin combinations [60]. A Phase I trial of perifosine in soft tissue sarcoma, including uterine, did not show evidence of activity [61]. MK2206 and a dual mTOR/AKT inhibitor MKC-1 are being explored in trials of advanced uterine cancer (Table 2) [62].

Practicalities of PI3K pathway inhibitors in uterine cancer

Single agent vs combination activity

Emerging data with PI3K pathway inhibitors in uterine cancer and other solid tumors indicate modest single-agent activity. For example, with mTORC1 inhibitors, there were no complete responses, and rates of partial responses ranged from 0 to 26% in patients with advanced uterine cancer [45], [46], [49], [51] and [52]. Furthermore, a PFS advantage of only 1.7 months was achieved with single-agent ridaforolimus compared with progestin [52]. Novel PI3K pathway inhibitors were based on rational design to address the limitations of rapalogs but, so far, single-agent activity remains disappointing. Interestingly, there is no clear correlation with response and potency for target on clinical data presented thus far (Table 3 and Table 4) [55], [62], [63], [64], [65], [66], [67], [68], [69], [70], [71], [72], [73], [74], [75], [76], [77], [78], [79] and [80]. Therefore, whether these limited responses are a function of the highly complex nature of the pathway, leading to ‘escape’ routes of signaling, a function of poor patient selection, or inadequate dosing is currently unclear. It is valuable to remember that HER2 antagonists and EGFR inhibitors both showed limited efficacy when tested in a general patient population. Only upon appropriate patient selection did the treatment benefit become clear. There is strong preclinical evidence for increased activity for PI3K pathway inhibitors in patients who have gain of function mutations in PIK3CA or loss of PTEN [81] and [82]. Early clinical results in various advanced solid tumors have shown responses in patients with and without pathway alterations [67], [69] and [71]. Nevertheless, several ongoing Phase II clinical trials in uterine cancer are screening for PI3K pathway activation at enrollment to test this hypothesis, including trials involving BKM120, BEZ235, and MK2206 (Table 2).

Combining PI3K pathway inhibitors with compounds that target parallel pathways, such as RAS/RAF/MEK, may potentially improve single-agent activity of PI3K pathway inhibitors. KRAS mutations have been shown to predict resistance to PI3K and mTOR inhibitors in preclinical and early clinical studies of uterine cancer [59] and [83] In preclinical models of various tumor types, simultaneous inhibition of PI3K and RAS/RAF/MEK pathways had a synergistic effect on tumor regression [84], [85], [86] and [87]. Multiple clinical studies examining the combination of inhibitors of the PI3K and RAS/RAF/MEK pathways are ongoing (Table 2) [88].

A second combination approach involves combining PI3K pathway inhibitors and standard therapy, with the aim of achieving clinically effective synergies and restoring sensitivity to the original agent. Over-activation of the PI3K pathway has been associated with resistance to a range of anticancer agents, including chemotherapy and hormone therapy [89], [90] and [91]. Chemoresistance is a consequence of multiple cellular events that can include inhibition of apoptosis, dysfunction of the tumor suppressor gene p53, and constitutive activation of PI3K/AKT signaling. The role of the PI3K pathway in resistance to chemotherapy is exemplified by the observation that PTEN transfection significantly enhanced doxorubicin chemosensitivity by induction of apoptosis in PTEN-null Ishikawa cells [91]. In preclinical cancer models, PI3K pathway inhibitors have been shown to enhance the efficacy of chemotherapy [92], [93], [94] and [95]. For example, mTORC1 inhibition with rapamycin resulted in synergistic effects when combined with paclitaxel and cisplatin in uterine cancer cell lines through inhibition of cell proliferation, induction of apoptosis, and potentially increased polymerization and acetylation of tubulin [94] and [95]. Progestin resistance is a significant problem in the successful hormonal treatment of uterine cancer. The PI3K inhibitor LY294002 was found to upregulate progesterone receptor expression, diminish cell growth in progestin-resistant uterine cancer cells, and reverse progestin resistance in mouse uterine cancer xenografts [90]. Additive effects on growth inhibition and apoptosis of uterine cancer cell lines were also reported when everolimus was combined with tamoxifen [96]. The combination of temsirolimus with hormone therapy was, however, associated with a significant increase in toxicity over temsirolimus alone in a Phase II study [48].

Adverse events

Given the critical role of the PI3K pathway in multiple cellular processes, it is not surprising that treatment with all of these compounds has been associated with class effects. Fatigue, nausea, and diarrhea are recognized as class-associated effects of PI3K pathway inhibition (Table 4). Hyperglycemia was predicted to be a significant problem with PI3K pathway inhibitors due to the role of the PI3K pathway in insulin signaling and regulation of cellular metabolism. Consequently, patients with diabetes mellitus are generally excluded from clinical trials with these compounds. Reassuringly, early clinical results do not suggest that hyperglycemia is a major limitation to therapy.

Rare adverse events have also been reported that may be a direct function of PI3K pathway inhibition. These include abnormal ECGs, and mood alterations (Table 4). Abnormal ECGs manifesting as T-wave inversions have been reported as dose-limiting toxicities with XL765 and GDC0941. Preclinical evidence suggests that PI3K pathway signaling is generally cardioprotective, and that inhibition could have detrimental effects [97]. However, cardiac effects have not been a major source of toxicity in clinical trials of these compounds (Table 4). Preclinical studies have implicated PI3K/AKT and downstream activation of GSK3-beta in the regulation of behavior, and the signaling pathway responds to changes of the monoamine transmitters dopamine and serotonin 5-HT in vivo [98], [99] and [100]. In a Phase I study of BKM120, mood changes were reported [77], including anxiety and depression, leading to exclusion of patients with a history of, or active, mood disorder in Phase II trials.

Conclusions

The PI3K pathway is an important promoter of cellular growth, metabolism, differentiation, proliferation, survival, and angiogenesis. This critical pathway is altered in multiple tumor types, including uterine cancer. The most frequently altered signaling component genes in uterine tumors are PTEN, PIK3CA, and KRAS, which can result in constitutive signaling in the absence of growth factor stimulation. In some tumor types, over-activation of the PI3K pathway has been associated with a reduced response to chemotherapy and hormone therapy, making it an attractive target for therapeutic intervention.

Inhibitors against different nodes of the PI3K pathway are currently in clinical development for advanced solid tumors and uterine cancer. Monotherapy has not produced dramatic decreases in tumor volume but has provided long-term disease stabilization in some patients. The assumption is that these agents may be more effective in tumors that are dependent on PI3K pathway signaling, and clinical trials are underway to investigate this hypothesis [101]. Preliminary evidence suggests that over-activation of the PI3K pathway may be associated with drug resistance, and thus further evaluation of the ability of PI3K pathway inhibitors to restore sensitivity to standard hormonal treatment or chemotherapy is underway. The PI3K pathway is complex due to feedback loops and cross-talk with parallel pathways. The potential for enhancing clinical outcomes with a combination of PI3K and RAF or MEK inhibitors warrants investigation, and early results indicate similar toxicities to single agents [102] and [103]. As the challenges are worked through in each tumor type, predictive biomarkers should facilitate selection of monotherapy and appropriate combinations to suit individual tumors.

Conflict of interest statement

The author is a speaker for GlaxoSmithKline and Merck, and conducts research supported by Genentech, GlaxoSmithKline, Novartis, and OSI Pharmaceuticals.

Acknowledgments