Taselisib (GDC-0032), a Potent β-sparing Small Molecule Inhibitor of PI3K, Radiosensitizes Head and Neck Squamous Carcinomas Containing Activating PIK3CA Alterations

The recent identification of the landscape of genomic alterations occurring in head and neck tumors has resulted in new insights and the identification of potential new targets for therapy. Among them, PIK3CA, the gene encoding for the catalytic p110α subunit of PI3K, is one of the most frequently altered. Given that increased PI3K signaling is associated with radioresistance, there is a strong rationale to inhibit PI3K signaling as a strategy to enhance the therapeutic effects of radiation therapy. We, therefore, have investigated the efficacy of combining radiotherapy and GDC-0032, a potent inhibitor of PI3K-α currently in clinical development. We show the GDC-0032 radiosensitizes PIK3CA activated HNSCC cell lines in vitro and impairs DNA damage repair. Further, combined GDC-0032 and radiation is more effective that either therapy alone in PIK3CA mutated HNSCC xenograft models. This data suggests combining GDC-0032 and radiation warrants further clinical investigation in HNSCC.

Purpose: Activating PIK3CA genomic alterations are frequent in head and neck squamous cell carcinoma (HNSCC) and there is an association between phosphoinositide 3-kinase (PI3K) signaling and radioresistance. Hence, we investigated the therapeutic efficacy of inhibiting PI3K with GDC-0032, a PI3K inhibitor with potent activity against p110α, with radiation in HNSCC. Experimental Design: The efficacy of GDC-0032 was assessed in vitro in 26 HNSCC cell lines with crystal violet proliferation assays, and changes in PI3K signaling were measured by western blot. Cytotoxicity and radiosensitization were assessed with Annexin V staining via flow cytometry and clonogenic survival assays, respectively. DNA damage repair was assessed with immunofluorescence for γH2AX foci, and cell cycle analysis was performed with flow cytometry. In vivo efficacy of GDC-0032 and radiation were assessed in xenografts implanted into nude mice.Results: GDC-0032 inhibited potently PI3K signaling and displayed greater antiproliferative activity in HNSCC cell lines with PIK3CA mutations or amplification, whereas cell lines with PTEN alterations were relatively resistant to its effects. Pretreatment with GDC-0032 radiosensitized PIK3CA-mutant HNSCC cells, enhanced radiation-induced apoptosis, impaired DNA damage repair, and prolonged G2/M arrest following irradiation. Furthermore, combined GDC-0032 and radiation was more effective than either treatment alone in vivo in subcutaneous xenograft models.Conclusions: GDC-0032 has increased potency in HNSCC cell lines harboring PIK3CA activating aberrations. Further, combined GDC-0032 and radiotherapy was more efficacious the either treatment alone in PIK3CA altered HNSCC in vitro and in vivo. This strategy warrants further clinical investigation.

Radiotherapy is the primary local treatment employed for head and neck squamous cell carcinomas (HNSCC) arising from many anatomic sites, including the oropharynx, hypopharynx, nasopharynx, and larynx. Concurrent administration of cisplatin-based chemotherapy has been shown to improve overall survival in patients with locally advanced HNSCC undergoing definitive radiation, primarily by decreasing locoregional relapse (1). However, concurrent chemotherapy also significantly increases toxicity during, and possibly after, radiotherapy (2-5). In addition to cisplatin, the anti-epidermal growth factor receptor (EGFR) monoclonal antibody cetuximab, has also been shown to improve overall survival and locoregional progression free survival, without increasing acute or late radiation toxicity, when administered concurrently with radiation (6). Despite this success, EGFR inhibitor-based radiation regimens are less effective than platinum-based chemoradiation (7). Therefore, novel agents that radiosensitize HNSCC in a tumor specific manner continue to be sought.The most common oncogenic alteration in HNSCC is the aberrant activation of PI3Kvia mutations or amplification of PIK3CA, the gene encoding the α-isoform of catalytic subunit of PI3K (p110α) (8-12). According to the TCGA (13), activating PIK3CA alterations are present in 56% and 34% of HPV(+) and HPV(-) HNSCC’s, respectively. Activation of PI3K leads to synthesis of phosphatidylinositol 3,4,5-trisphosphate (PIP3) at the plasma membrane that, in turn, leads to the recruitment of the pleckstrin-homology domain-containing proteins phosphoinositide dependent protein kinase-1 (PDK1) and AKT. PDK1 phosphorylates AKT at threonine 308 and activates AKT and downstream signaling elements, including mammalian target of rapamycin (mTOR) complex 1 (mTORC1), thereby promoting cell growth, proliferation, survival, angiogenesis, and regulating glucose metabolism (14).

The PI3K signaling axis is an attractive target for inducing tumor specific radio sensitization for a variety of reasons. DNA damaging agents, including radiation, induce phosphorylation of AKT, both at threonine 308 and serine 473, and activate downstream signaling within minutes of treatment (15-18). We and others have demonstrated that inhibition of PI3K or knockdown of PIK3CA enhances DNA damage and sensitizes breast cancer cells to PARP inhibition (19, 20).Furthermore, inhibition of PI3K-AKT-mTOR signaling has been shown to sensitize cancer cells to radiation-induced cytotoxicity (15, 16, 21-24). A major caveat of these previous studies was their reliance on nonspecific PI3K inhibitors such as wortmannin (25), LY294002 (26), and NVP-BEZ-235 (27) which also have potent inhibitory activity against PI3K-like kinases (PIKK’s) such as DNA-PKcs, ATM, and ATR, enzymes that play a central role in the repair of DNA damage following radiation. Thus, it is difficult to ascertain from these studies the relative contributions of PI3K inhibition, as opposed to PIKK inhibition, on the radiation-enhancing effects reported.Recently, several isoform-specific PI3K inhibitors have been developed and have entered into early phase clinical trials (28, 29). One of them, GDC-0032, is a potent inhibitor of p110α, p110δ, and p110γ, but with 31 times less potency for the remaining class IA PI3K enzyme p110β. Additionally, GDC-0032 is over 1000 times more selective for p110α than any tested PIKK, including no significant inhibitory activity against DNA-PKcs (30). GDC-0032 has shown in early clinical trials, clinical activity in tumors harboring PIK3CA alterations, including in head and neck cancer (29). We thus decided to investigate the efficacy of GDC-0032 in HNSCC, both as a single agent and in combination with radiotherapy with the goal to determine whether further clinical development of this class of agents is warranted in this disease.

GDC-0032 was provided by Genentech. For in vitro assays, all drugs were dissolved in dimethyl sulfoxide. For in vivo experiments, GDC-0032 was dissolved in sterile water, 0.5% methylcellulose, and 0.2% Tween-80.All HPV negative cells were obtained directly from American Type Culture Collection (Cal-33, FaDu, Detroit 562, SCC-4, SCC-9, SCC-15, SCC-25), the European Collection of Cell Cultures via Sigma-Aldrich (BICR-16, BICR-18, BICR-22, BICR-31), the Japanese Collection of Research Bioresources (HSC-2, HSC-3, HSC-4 ), or the Korean Cell Line Bank (SNU-46, SNU- 1076, SNU-1214, YD-8), with the exception of LB-771, which was obtained from The Center for Molecular Therapeutics at Massachusetts General Hospital. The HPV positive cell lines UD- SCC-2, UM-SCC-47, UPCI-SCC-90, and 93-VU-147T cell lines were kind gifts from the Paul Harari lab at the University of Wisconsin. UM-SCC-104 was purchased from the lab of Thomas Carey at the University of Michigan, and UPCI-SCC-154 was purchased from the Leibniz Institute DSMZ-German Collection of Microorganisms and Cell Cultures. All cell lines were maintained in humidified incubators at 37o Celsius in Dulbecco’s modified Eagle’s medium/Ham’s F-12 1:1, with the exception of
SNU-1076, SNU-46, SNU-1214 and YD-8, which were grown in RPMI1640. Cell culture media was supplemented with 10% heat- inactivated fetal calf serum, 2mM L-glutamine, penicillin (20 U/mL), and streptomycin (20μg/mL).uPIK3CA mutation and amplification status information for each cell line was obtained from the Cancer Cancer Line Encyclopedia (www.broadinstitute.org/ccle),(31) except for LB-771 and thesix HPV positive cell lines, which were sequenced using the MSK-IMPACT next generation sequencing platform.(32) Amplification was defined as greater than or equal to 4 copies of the PIK3CA gene.

Cells were seeded in replicates of 6 in 96 well plates with 500-5000 cells/well overnight and then treated with GDC-0032. After 4 days, the media was removed and the cells were fixed with 4% glutaraldehyde for 30 minutes. Fixed cells were stained with 0.1% crystal violet for two minutes, then washed and dissolved in 10% acetic acid. Absorption of light was quantified using a Biotek Synergy H1 plate reader.Cells were plated in 6 well plates in appropriate dilutions (37, 111, 333, 1000, 3000, or 9000 cells/well) and allowed to attach overnight. Cells were then treated with DMSO or drug for 24 hours and irradiated at various dosing using a cesium irradiator. Drugs were left in following irradiation. Ten to 14 days after irradiation, cells were fixed with 4% glutaraldehyde, and stained with 0.1% crystal violet. Colonies with greater than 50 cells were counted. Surviving fractions were calculated by normalizing to the plating efficiency at 0 Gy for either control or drug treated plates.Cells were washed twice in ice-cold phosphate-buffered saline, scraped from the plate, centrifuged, and then frozen at -20o after the supernatant was removed. The cells were then lysed in ice-cold radioimmunoprecipitation buffer supplemented with phosphatase inhibitor cocktails (Complete Mini and PhosphoStop, Roche), centrifuged, and the supernatant was removed for protein quantification (Pierce BCA Protein Assay Kit, Thermo Scientific).

Twenty-five to 50μg of protein was loaded into NuPAGE 4-12% bis-tris gels (Life Technologies) and resolved via electrophoresis, then transferred to Immobilon transfer membranes (Millipore). Membranes were blocked in 5% BSA in TBS-T for 1 hour prior to overnight incubation in primary antibody at 4o, and incubated in either mouse or rabbit horseradish peroxidase conjugated secondary antibodies (1:50,000, Amersham Biosciences) in 2% BSA in TBS-T for 1 hour. Membranes were imaged using SuperSignal West Femto Chemiluminescent Substrate (Thermo Scientific) and images were captured using a GBOX camera system. Antibodies used are listed in Supplementary Table 1.Cells were seeded into 8-chamber culture slides (BD Falcon), fixed at each time point with 4% formaldehyde at room temperature, and then rinsed with PBS and stored in PBS at 4oC. Cells were permeabilized with 0.8% Triton X-100 and 0.2% BSA in PBS for 10 minutes on ice and then blocked with 1% BSA in TBS-T for 1 hour at room temperature, followed by incubation with γH2AX (1:500, Millipore) and 53BP1 (1:200, Bethyl) antibodies in 1% BSA in TBS-T for 2 hours at room temperature. The cells were then washed with TBS-T and incubated with Alexa 555 anti-rabbit and Alexa 488 anti-mouse secondary antibodies (both 1:1000, Life Technologies)at room temperature. The cells were washed with TBS-T, rinsed in ddH20, and coverslipped with ProLong with DAPI (Life Technologies).

Slides were digitally scanned with Pannoramic Flash scanner (3DHistech, Hungary) using 20x/0.8NA objective. The images were then analyzed using Metamorph software (Molecular Devices, PA); briefly, nuclear regions were segmented using the DAPI channel, and the number of foci was counted using spot detection.Cells were treated with either GDC-0032 or DMSO for 24 hours, irradiated or mock-irradiated, then trypsinized and harvested along with cells in the media after an additional 72 hours. Cells were resuspended in Annexin V buffer, stained with Annexin V-FITC and propidium iodide (PI) according to the manufacturer’s instructions (BD Biosciences), and analyzed for fluorescence with a Fortessa flow cytometer. Resulting data was analyzed using FlowJo software.Cells were collected by trypsinization, washed in ice cold PBS, resuspended in 70% ethanol and stored at -20oC for at least 2 hours. Ethanol was then removed, cells were washed in wash buffer (1% bovine serum albumin, 0.25% Triton X-100, and 2.5mM EDTA in PBS), then incubated in wash buffer containing 40µg/mL propidium iodide (Sigma) and 100µg/mL RNase A (Sigma) for 20 minutes.

Cells were characterized for fluorescence with either a FACSCalibur or a Fortessa flow cytometer and the resulting distributions were analyzed using FlowJo software.All in vivo studies were conducted according to Memorial Sloan Kettering Research Animal Resource Center approved protocols and Johns Hopkins University Animal Care and Use Committee approved protocols. Six week old Nu/Nu mice were order from Harlan Laboratories. For cell line-derived xenograft studies, mice were injected bilaterally with 5×105 cells resuspended in 200uL of culture media and Matrigel (BD Biosciences) mixed in a 1:1 ratio.After tumors reached approximately 100-200cm3, mice were randomized into treatment arms with 8-10 tumors per group. GDC-0032 (5mg/kg) was dissolved in a vehicle containing 0.5% methylcellulose with 0.2% TWEEN-80 and was administered via daily oral gavage. Tumors were irradiated using an X-RAD 320 X-ray system with appropriately sized lead shields.For patient tumor-derived xenograft studies, mice were implanted with a tumor obtained from a patient with oropharynx squamous cell carcinoma. The tumor DNA was sequenced using CancerSelect-203 R platform (Personal Genome Diagnostics, Baltimore, MD) and found to have a PIK3CA E542K mutation. After tumors reached approximately 100-250cm3, mice were randomized into treatment arms with 5-6 tumors per group. GDC-0032 (5mg/kg) was administered via daily oral gavage for 14 days. Tumors were irradiated 2Gy daily on days 2-4 using a Small Animal Radiation Research Platform at the Johns Hopkins University. Tumor volumes were calculated as (π/6) x length x width2.

We tested the antiproliferative activity of GDC-0032 across a panel of 26 HNSCC cell lines, including the majority of commercially available HPV-positive HNSCC cell lines. As expected, there was a gradient of sensitivity in HNSCC cell lines. Cell lines harboring either mutations or amplification of PIK3CA tended to be sensitive to GDC-0032, with IC50’s in the nanomolar range (Fig. 1A). By contrast, 4 of the 6 most resistant cell lines to GDC-0032 had mutation or loss of PTEN, consistent with previous reports of PTEN aberrations leading to resistance to PI3Kα inhibitors through upregulation of PI3Kβ signaling (33-35). The observation of a preferential antiproliferative effect of GDC-0032 in cells with activated PI3KCA has been observed also with other isotype specific PI3K inhibitors in other tumor types, suggesting that this could selectivity could be important in the clinic (36, 37). By contrast, neither PIK3CA nor PTEN status correlated with sensitivity to GDC-0941, a pan-PI3K inhibitor with similar potency against all class IA PI3K isoforms (Fig. 1B). Thus, although PIK3CA and PTEN status may help identify tumors sensitive to GDC-0032, this is not a property shared amongst all classes of PI3K inhibitors.
Next, we investigated the effects of GDC-0032 on downstream PI3K signaling in several genetic contexts. Treatment with GDC-0032 in Cal-33 cells (harboring a PIK3CA H1047R mutation) prevented phosphorylation of AKT and inhibited downstream mTOR targets such as ribosomal protein S6 kinase (S6K), eukaryotic translation initiation factor 4E-binding protein 1 (4EBP-1)and S6 (Fig. 2A). This translated into an induction of apoptosis, as assessed by cleavage of poly- ADP ribose polymerase (PARP). Similar results were obtained in LB-771, a cell line containing amplification of PIK3CA (Supplementary Fig. S1A). In cell lines containing either PTEN homozygous deletion (UD-SCC-2) or mutation (UPCI-SCC-90), GDC-0032 was appreciably less effective at downregulating AKT/mTOR signaling and inducing cell death (Fig. 2A and Supplementary Fig. S1B). This supports the notion that downregulation of PI3K signaling is necessary for the pro-apoptotic effects of GDC-0032 in HNSCC.

The dose of 100nM GDC-0032, which inhibits AKT/mTOR signaling in PIK3CA mutant cell lines but not in cells with loss or mutation of PTEN, was chosen for subsequent time-course experiments. This dose of GDC-0032 inhibited phosphorylation of AKT and downstream signaling in Cal-33 (PIK3CA mutant) and LB-771 (PIK3CA amplified) cells, but had little effect on PI3K signaling in UD-SCC-2 (PTEN homozygous deletion) or UPCI-SCC-90 cells (PTEN mutant) (Fig. 2B), confirming that this is a concentration of GDC-0032 that inhibits PI3Kα-, but not PI3Kβ-dependent signaling.Given that inhibition of PI3K signaling has been purported to affect expression of DNA damage repair (DDR) proteins (19, 20, 38) and alter DDR signaling in response to radiation (15, 17), we next sought to study the effect of GDC-0032 on HNSCC cell lines treated with radiation. In Cal- 33 cells (PIK3CA H1047R), the combination of GDC-0032 and radiation resulted in both more apoptotic (Annexin-V positive, PI negative) and non-apoptotic (Annexin-V positive, PI negative) cell death than either treatment alone (Fig. 3A). GDC-0032 and radiation also slowed cell growth rates more than either treatment alone in PIK3CA mutant cell lines Cal-33 and HSC-2, but had
little effects in PTEN altered cell lines UPCI-SCC-90 and UD-SCC-2 (Fig 3B). The increased antitumor effects of combined radiation and GDC-0032 compared to either treatment alone were confirmed using Annexin staining in three additional cell lines bearing activating PIK3CA alterations (LB-771, SNU-1076, and HSC-2), whereas no significant additional activity was observed in cell lines with wild type PIK3CA and/or inactivating PTEN alterations (HSC-3, UD- SCC-2, HSC-4, and FaDu; Fig. 3C).

Similar results were seen with the structurally unrelated p110α inhibitors BYL719 (Supplementary Fig. S2) and A66 (Supplementary Fig. S3), suggesting that GDC-0032 induces cell death following radiation primarily through p110α inhibition, rather than inhibition of other PI3K isoforms or off target enzymes.Inhibition of other downstream PI3K components using an allosteric AKT inhibitor MK-2206 or the allosteric mTORC1 inhibitor RAD001 also increased radiation-induced apoptosis, albeit to a smaller degree than GDC-0032 (Fig. 3D). Similar results were observed with the PIK3CA- mutated cell line HSC-2, although RAD001 did not enhance radiation-induced apoptosis in this cell line (Supplementary Fig. S4).The gold standard for assessing radiosensitization is clonogenic survival. Using this assay, Cal- 33 cells pretreated with GDC-0032 had significantly decreased cell survival following radiation (Fig. 3E). Similar radiosensitization was seen when LB-771 cells were treated with GDC-0032 prior to radiation (Supplementary Fig. S5). However, GDC-0032 had no effect on the radiation response of HSC-3, a PIK3CA wt cell line resistant to single agent GDC-0032 (Fig. 3D).PI3K signaling is a key regulator of the DNA damage response (15, 17-21, 39). Therefore, we decided to study whether the GDC-0032-dependent radiosensitization was at least in part attributable to impaired DNA damage repair under a state of PI3Kα inhibition. We quantified the amount of DNA double strand breaks (DSBs), as assessed by γH2AX foci, with and without GDC-0032 pretreatment in Cal-33 cells. Cells pretreated with GDC-0032 had significantly more γH2AX foci at 24 and 48 hours after irradiation than control treated cells (Fig. 4A). This increase in DNA damage upon combination of radiation and GDC-0032 was accompanied by increased formation of p53-binding protein 1 (53BP1) foci, a mediator of the DSB repair downstream of γH2AX (Fig. 4B) (40), and induction of PARP cleavage (Supplementary Fig. S6A). Consistent results were observed also in LB-771 cells (Supplementary Fig. S6B), supporting the notion that GDC-0032 impairs DSB repair in these cells following radiation.

Ionizing radiation induces two molecularly distinct G2/M checkpoints (41). The first, known as the “early” G2/M checkpoint, consists of a transient, ATM-dependent mitotic block affecting cells in late G2 occurring within minutes of irradiation and can be assessed by the proportion of cells with phosphorylation of histone H3 (HH3) after irradiation (41). The second, more prolonged and known as the “late” G2/M checkpoint, is independent of ATM and results in an accumulation of cells with 4N DNA content (41). It is well accepted that the duration of the G2/M checkpoint reflects the number of unrepaired DSBs (42). Therefore, based on our results we hypothesized that GDC-0032 may alter the DNA damage-induced cell cycle arrest that occurs following irradiation.As single agent, treatment with GDC-0032 resulted in a mild increase in the proportion of cells in G1 over 72 hours, with a concomitant decrease in the G2 phase of the cell cycle (Fig. 5A). As expected, radiation induced both the early and late G2/M checkpoints, with a marked decrease in pHH3 positive cells (Supplementary Fig. S7) and an accumulation of cells with 4N DNA content (Fig. 5B). However, by 48 hours after irradiation, the cell cycle profile was nearly back to baseline. When we examined the impact of PI3Kα inhibition in this setting we observed that, while treatment with GDC-0032 for 24 hours prior to irradiation did not significantly alter the cell cycle profile of Cal-33 cells (Fig. 5B) or affect the early G2/M arrest as assessed by pHH3 positive cells (Supplementary Fig. S7), the proportion of cells with 4N DNA content at 24 hours and 48 hours after irradiation was considerably higher with than in control treated cells.This suggests that the late G2/M arrest induced by radiation is enhanced by pharmacological inhibition of PI3Kα (Fig. 5B).

Because the late G2/M checkpoint is known to be ATM-independent, we investigated whether the enhanced late G2/M checkpoint induced by GDC-0032 could be abrogated by either inhibition of other known regulators of the G2/M cell cycle progression, such as Wee1 and ATR. We found that although both Wee1 inhibition with AZD-1775 and ATR inhibition with VE-821 abrogated the early G2/M checkpoint following irradiation as assessed by pHH3 (Supplementary Fig. S8 and S9), only ATR inhibition with VE-821 reversed the late G2/M arrest induced by the combination of GDC-0032 and radiation (Fig. 5C and Supplementary Fig. S10). Taken together, these results indicate that GDC-0032 enhances the late G2/M checkpoint induced by irradiation in an ATR-dependent fashion, resulting in increased DNA damage overtime.In order to test the ability of GDC-0032 to inhibit PI3K signaling in vivo, we treated nude mice implanted with subcutaneous Cal-33 xenografts with 5mg/kg of GDC-0032 and harvested the tumors after 2, 6, and 24 hours of treatment . As expected, treatment with GDC-0032 resulted in nearly complete abrogation of AKT and PRAS40 phosphorylation, as well as decreased phosphorylation of 4EBP-1 and S6, at 2 hours after drug administration (Fig. 6A). However, by 6 hours after oral gavage, a rebound in phosphorylation of all of these PI3K targets was detected, probably a reflection of the short half-life of the compound.We next assessed the efficacy of combined PI3K inhibition and radiation. Mice received daily GDC-0032 throughout the experiment and 20 Gy in 5 daily fractions on days 2 to 6. Although treatment with either radiation or GDC-0032 alone resulted in tumor growth delay compared to vehicle treated mice, only the combination of GDC-0032 and radiation resulted in durable tumor regressions (Fig. 6B). In fact, at the experiment endpoint (90 days of treatment), none of the tumors in the combined radiation and GDC-0032 arm had progressed in comparison to the start of treatment. We also saw superior activity with transient administration of GDC-0032 during radiation in a HPV (+) PIK3CA-E545K patient derived xenograft model (Supplementary Figure 11). These data also suggest that both mutations (E542K in the helical domain and H1047R in the kinase domain) are functionally active in HNSCC.Moreover, our in vivo findings suggest that even a transient inhibition of the PI3K/AKT/mTOR pathway (Fig. 6A) is sufficient to sensitize tumors to radiation, an observation with potential clinical implications.


In this study, we determined that HNSCC cell lines containing activating PIK3CA alterations are significantly more sensitive to GDC-0032, a novel, potent inhibitor of PI3Kα, than cell lines without these alterations. This was consistent with previous studies of other PI3Kα inhibitors studied in breast cancer (36, 37, 43). Additionally, we found that PTEN aberrations were associated with resistance to GDC-0032, with 4 of the 6 most resistant HSNCC cell lines to this agent containing PTEN alterations, also consistent with previous data from our group (33).Combining GDC-0032 and radiotherapy resulted in greater cell death than either treatment alone in cells with intrinsic sensitivity to GDC-0032. This translated in profound in vivo antitumor activity of the combination of GDC-0032 and radiation in PIKCA mutant tumors.The mechanisms underlying the strong antitumor efficacy of combined radiotherapy and PI3K inhibition in HNSCC are almost certainly multifactorial given the diverse phenotypic outputs induced by both radiation and PI3K signaling, including modulation of growth, survival, metabolic activity, angiogenesis, and immune response. However, we speculate that a major contributor to the efficacy of this combination is impaired DNA damage repair (DDR) as a result of PI3K inhibition. Supporting this, we found that GDC-0032 pretreatment delays the resolution of deleterious DNA lesions following radiotherapy and prolong the accumulation of cells in G2/M following irradiation in an ATR-dependent manner. It should also be noted that we observed strong in vivo effects of GDC-0032 and radiation despite the fact that inhibition of PI3K signaling was relatively transient. Although further studies may be needed with respect to the optimal timing and sequencing of GDC-0032 and radiation, our findings suggest that a strong pulsatile PI3K pathway inhibition would be sufficient to achieve remarkable antitumor efficacy.

Combining PI3K pathway inhibitors and radiation may be a particularly promising therapeutic strategy in HPV(+) HNSCC for both biologic and clinical reasons. PIK3CA mutations and amplifications are present in 22-37% and 20-25% of HPV(+) tumors, respectively, an approximately two to three fold higher incidence over what is observed in HPV(-) tumors (12, 44). Furthermore, although PIK3CA mutations are found in the kinase domain, the helical domain, and other non-hotspot locations through the gene in HPV(-) HNSCC, almost all PIK3CA mutations in HPV(+) HNSCC occur within the helical domain of p110α at either E542 or E545, suggesting that these specific helical domain codon changes convey a critical function in the pathogenesis of this disease (13). The limited PIK3CA mutation spectrum in HPV(+) HNSCC also will likely simplify patient selection for clinical trials, obviating the complexity introduced by multiple mutations of unknown biochemical and phenotypic significance.Importantly, HPV(+) tumors have increased radiosensitivity both preclinically and clinically in comparison to HPV(-) HNSCC (45, 46), and several clinical trials are currently underway to investigate strategies to de-escalate therapy without compromising cure rates. Thus, given the high prevalence of activating PIK3CA alterations in HPV(+) HNSCC and the need to improve the toxicity profile of radiotherapy (by eliminating concomitant administration of chemotherapy), combining PI3K inhibitors with radiotherapy in select HPV(+) patients may be a promising strategy for future clinical investigations. In fact, we are currently enrolling patients with HNSCC and PI3K mutations in a phase I clinical trial and our findings could catalyze the launch of a follow up study in combination with radiation therapy.

In summary, we have demonstrated that specific PI3Kα inhibition is an effective strategy for HNSCC tumors harboring activating PIK3CA alterations in combination with radiotherapy. Given the importance of PI3K signaling in both the pathogenesis of HNSCC and in the response of cells to radiation, these types of strategies may represent an important step to reducing the toxicity of treatment by personalizing therapy while maintaining the excellent outcomes provided by cytotoxic Taselisib chemoradiation.