BAY 80-6946 Is a Highly Selective Intravenous PI3K Inhibitor with Potent p110α and p110δ Activities in Tumor Cell Lines and Xenograft Models

The phosphoinositide 3-kinase (PI3K) family generates 3′-phosphoinositides that activates a variety of cellular targets important for cell proliferation, survival, differentiation, and migration (1, 2). The class I PI3Ks transmit signals from receptor tyrosine kinases (RTK) and from the oncogenic protein Ras via the PI3Kα, β, or δ isoforms and from G protein–coupled receptors (GPCR) via the PI3Kβ or γ isoforms (3–5) Activation of PI3K has been directly linked to cancer through somatic mutations or amplification of PIK3CA (6–8), and loss of function of tumor suppressor PTEN (9). The loss or inactivation of PTEN leads to increased PI3K signaling and tumor growth that seems to be mediated via PI3Kβ isoform (10–12). In addition, overexpression or mutational activation of RTKs occurs in multiple tumor types, leading to increased PI3K signaling (13, 14).

Aberrant activation of class I PI3Ks has been associated with both intrinsic and acquired resistance to targeted agents, such as imatinib and trastuzumab (15–17), as well as to traditional chemo- and radiotherapy (18–21). The prototype PI3K inhibitors (e.g., LY294002) did not enter clinical development due to their lack of specificity and therapeutic window (22). A number of second generation selective PI3K or PI3K/mTOR inhibitors have entered early clinical development. However, clinical benefit has not yet been clearly shown in solid tumors. This is largely due to the complexity derived from the existence of various PI3K isoforms and their differential roles in cancer pathology and compensatory feedback activation of oncogenic pathways (23). As such, development of PI3K inhibitors with differential pharmacologic and pharmacokinetic profiles would allow best exploration in different indications, combinations, and dosing regimens to maximize the tumor-killing effects of PI3K inhibitors.

The present report explores the pharmacologic and pharmacokinetic profile of the intravenous PI3K inhibitor BAY 80-6946, which preferentially inhibits PI3Kα and PI3Kδ and is highly potent in in vitro and in vivo preclinical tumor models with activated PI3K.

BAY 80-6946 [2-amino-N-[7-methoxy-8-(3-morpholinopropoxy)-2,3-dihydroimidazo[1,2-c]quinazolin]; Fig. 1A] was identified and synthesized at Bayer HealthCare Pharmaceuticals. The information about the reference compounds was described in Supplementary Fig. S1. For in vitro studies, 5 mmol/L stock solution of BAY 80-6946 (in dimethyl sulfoxide with 10 mmol/L trifluoroacetic acid) was used. All tumor cell lines, except KPL4 cells, were obtained from the American Type Culture Collection; KPL4 cells were obtained from the laboratory of Dr. Junichi Kurebayashi (Kawasaki Medical School, Okayama, Japan). All cell lines were authenticated by fingerprint techniques at German Collection of Microorganisms and Cell Cultures (DSMZ) after two to four passages of the original stock from the providers. An aliquot of the substock was used for the studies described here. The cell lines were maintained in RPMI-1640 medium containing 10% heat-inactivated fetal calf serum.

The effect of BAY 80-6946 on PI3Kα, PI3Kβ, and PI3Kγ activity was measured by the inhibition of ³³P incorporation into phosphatidylinositol as described in Supplementary Materials and Methods.

The effect of BAY 80-6946 on PI3Kδ was determined by Upstate/Chemicon using their proprietary PIProfiler assay system that uses homogeneous time-resolved fluorescence to measure inhibition of kinase activity. The effect of BAY 80-6946 on PI4K-II, PIP₄-5K, and PIP₅-4K was measured using assay conditions nearly identical to those for measuring PI3K activity, except that the p110 protein was replaced with either recombinant PI4K-II protein or PIP₄-5K protein, respectively. In addition, PIP₄ (Echelon) and PIP₅ (Matreya) were used as the lipid substrate in place of phosphatidylinositol in the PIP₄-5K and PIP₅-4K assays, respectively.

Levels of pAKT (T308), pAKT(S473), and p-4E-BP1(T70) in the presence or absence of PI3K inhibitors were quantitatively determined using AlphaScreen SureFire assays (PerkinElmer) or by Western blot analyses. Briefly, the effects of PI3K inhibitors on the basal levels of pAKT and p-4E-BP1 in KPL4 cells were measured after 2 hours of treatment with PI3K inhibitors. To determine the inhibition of lysophosphatidic acid (LPA)–stimulated phosphorylation of AKT and 4E-BP1, PC3 cells were cultured in serum-free medium overnight, then incubated with PI3K inhibitors for 1 hour before being stimulated with 50 μmol/L LPA for 10 minutes. Determination of pAKT and p-4E-BP1 levels was conducted according to the manufacturer's instructions.

Cell proliferation was determined using the CellTiter-Glo Luminescent Cell Viability Kit (Promega) as described in the Supplementary Data.

BT-20 or BT474 breast tumor cells were treated with test compounds for 24 hours. Activation of caspase-9 was determined using the Caspase-Glo9 Assay Kit (Promega). Phosphorylation of p53 at Ser15 and PARP cleavage was measured using the MSD Apoptosis Kit (Cat # K15102D-2) at 48 hours after treatment.

The in vivo antitumor efficacy of BAY 80-6946 was assessed in athymic nude rats or nude mice using xenograft models of human tumors. Animals were housed according to institutional guidelines with access to food (pelleted diet) and water ad libitum. Studies with patient-derived tumors were conducted in EPO GmbH or in Oncotest GmbH with written formed consent from each patient and the approval from local ethical committees. Tumor xenografts were generated by harvesting cells from mid-log phase cultures and injecting subcutaneously into the right flank of each rat and mouse. BAY 80-6946 was dissolved in a PEG400/acidified water (0.1 N HCl, pH 3.5; 20/80, v/v) or 5% mannitol vehicle. Treatment was initiated when all animals in an experiment had established tumors. BAY 80-6946 was administered every second day, every third day, or weekly via an intravenous bolus at the indicated doses. Tumor dimensions and body weights were recorded twice weekly starting on the first day of treatment. Tumor volumes were calculated using the equation |$(l \times x \times w^2)/2$|⁠, where l and w refer to the larger and smaller dimensions collected at each measurement. Antitumor efficacy was determined as a function of tumor growth inhibition (% TGI). TGI was calculated by the equation |$[1 - (T/C)]\, \times \,100$|⁠, where T and C represent the mean size of tumors in the treated (T) and control (C) groups, respectively. In addition, treatment responses were evaluated by means of clinically used Response Evaluation Criteria in Solid Tumors (RECIST) criteria; response rates (RR) were calculated as the percentage of animals with a complete or partial response (PR). The assessment of the pharmacodynamic effects of BAY 80-6946 in tumor specimens is described in the Supplementary Data.

To determine the pharmacokinetic properties of BAY 80-6946, a solution of BAY 80-6946 was prepared in PEG400/pH 3.5 water (20/80) and administered as an intravenous bolus dose in the tail vein at 2 mL/kg for rats and 10 mL/kg for mice. Each time point has a group of 3 animals. Blood was collected at 5, 15, and 30 minutes and at 1, 4, 7, 24, and 30 hours after administration for pharmacokinetic analysis. Samples for tumor versus plasma exposure and pharmacodynamics biomarker analysis were taken at 24, 48, 72, and 96 hours.

BAY 80-6946 represents a novel proprietary imidazolinoquinazoline class of PI3K inhibitors identified through high-throughput screening of Bayer's compound library and further optimized with respect to drug-like characteristics (Fig. 1A). BAY 80-6946 potently inhibited the catalytic activity of the class I PI3Kα, β, γ, and δ isoforms with IC₅₀ values of 0.5, 3.7, 6.4, and 0.7 nmol/L, respectively. It showed significantly weaker activity against mTOR with an IC₅₀ of 45 nmol/L. In contrast, 1 μmol/L BAY 80-6946 did not inhibit PI4K-II, PIP₄-5K, PIP₅-4K, or an additional 220 kinases in the Millipore kinase panel (inhibition <30%), indicating that BAY 80-6946 is a specific PI3K inhibitor with more than 2,000-fold selectivity against other lipid and protein kinases, except for mTOR. To further evaluate the selectivity of BAY 80-6946 against PI3K versus mTOR kinase, rat ELT3 cells, which exhibit a PI3K-independent activation of mTORC1 due to TCS2 deficiency, were used. Complete inhibition of PI3K-mediated AKT phosphorylation was clearly shown at a concentration of 5 nmol/L (Fig. 1B), whereas BAY 80-6946 showed only a minor reduction of PI3K-independent mTORC1-mediated phosphorylation of S70S6K levels and no effect at all on p-4E-BP1 at a concentration of 500 nmol/L (Fig. 1C). In contrast, rapamycin at 100 nmol/L completely inhibited p-p70S6K and the dual PI3K/mTOR inhibitor BEZ-235 at 500 nmol/L completely blocked the phosphorylation of both p70S6K and 4E-BP1, indicating potent activities directly against mTOR kinase. These results suggest that BAY 80-6946 has a 100- to 1,000-fold cellular selectivity of PI3K versus mTOR signaling and is, therefore, not a dual PI3K/mTOR inhibitor in cells.

Biochemically, BAY 80-6946 showed differential potency against the class I PI3K isoforms. To show its cellular activity against each individual isoform, the activity of BAY 80-6946 in inhibition of PI3Kα- and PI3Kβ-mediated activation of downstream effectors of the pathway was examined. Previous studies have shown that p110α is essential for the signaling driven by PIK3CA mutations and/or oncogenic RTKs, whereas p110β is the major isoform in mediating PTEN-deficient and GPCR signaling (24). KPL4, a breast carcinoma cell line that carries a PIK3CA-activating mutation and overexpresses HER2, was selected to assess the cellular PI3Kα activity. PI3Kβ activity was measured using LPA-stimulated AKT phosphorylation in PTEN-null PC3 prostate tumor cells. To validate these assays, TGX-221, a PI3Kβ-specific inhibitor, was tested in KPL4 and PC3 cells. TGX-221 potently inhibited LPA-stimulated AKT phosphorylation in PC3 cells with single-digit nmol/L IC₅₀ values, whereas it was completely inactive at 1 μmol/L in blocking AKT phosphorylation in KPL4 cells (Table 1). These data confirmed previous reports and validated the assays by showing that the phosphorylation of AKT in KPL4 cells is not driven by PI3Kβ, whereas LPA-stimulated AKT phosphorylation in PC3 cells is a PI3Kβ-dependent and PI3Kα-independent process (25). In KPL4 cells, BAY 80-6946 reduced basal levels of AKT phosphorylation at both Thr308 and Ser473 with IC₅₀ values of 0.4 and 0.6 nmol/L, respectively (Table 1). BAY 80-6946 also reduced pAKT levels in PC3 cells with activity similar to that of TGX-221, albeit at seven to 20 times higher concentrations than those required in KPL4 cells. This is consistent with biochemical data showing that BAY 80-6946 inhibits both α and β PI3K isoforms, but is more potent against the α isoform. In line with this observation, BAY 80-6946 was also approximately 10 times more potent in blocking AKT phosphorylation in breast cancer cell lines with PIK3CA mutations (e.g., BT-20, BT-474, and KPL4) than in those with loss of PTEN (e.g., ZR-75-1, MDA-MB-468) assessed by Western blot analysis in Fig. 1D.

BAY 80-6946 potently inhibited cell proliferation in a panel of human tumor cell lines, many of which exhibited constitutively activated PI3K signaling resulting from, for example, somatic mutations in PIK3CA and PTEN (Fig. 2A). Antiproliferative activity was evident across multiple tumor types. Of note, several breast cancer, endometrial cancer, and hematologic tumor cell lines were particularly sensitive to BAY 80-6946, with IC₅₀ values less than 10 nmol/L. This profile may correlate with the strong activity of BAY 80-6946 against PI3Kα and PI3Kδ. To further understand the correlation between sensitivity and molecular features, BAY 80-6946 was profiled in a broader panel of breast cancer cell lines. Indeed, BAY 80-6946 had mean IC₅₀ values of 19 nmol/L against cell lines with PIK3CA-activating mutations (n = 9) and 17 nmol/L against HER2-positive cell lines (n = 7; Fig. 2B and C), whereas the activity in PIK3CA wild-type and HER2-negative cells was about 40-fold less potent (average IC₅₀ = 774 nmol/L; n = 11). This result is in line with the potent biochemical and cellular mechanistic activity against PI3Kα and further indicates that PIK3CA mutation and HER2 overexpression are potential biomarkers to predict the sensitivity of breast cancer cells to BAY 80-6946. In contrast, there is no clear correlation between the sensitivity and loss of PTEN. In addition, BAY 80-6946 was active in inhibiting proliferation of tumor cell lines resistant to the anti-HER2 agents trastuzumab and lapatinib, such as T47D (PIK3CA-mut), ZR-75-1 (PTEN-null), and MCF7 (PIK3CA-mut) with IC₅₀ values of 6, 24, and 27 nmol/L, respectively. The potent activity in hematologic tumor cell lines (PIK3CA wild-type) may associate with the potent PI3Kδ activity of BAY 80-6946. Taken altogether, these observations show that BAY 80-6946 has antiproliferative activity against a subset of human cancer cell lines, particularly those tumor cells with PI3K and/or RTK alterations, such as PIK3CA mutation and/or overexpression of HER2 and may offer the potential to overcome resistance to anti-HER2 therapies.

The PI3K/AKT pathway promotes tumor cell survival and prevents apoptosis by phosphorylation and inhibition of proapoptotic mediators such as caspase-9 and FOXO. BAY 80-6946 potently induced nuclear localization of FOXO1A with an EC₅₀ less than 1 nmol/L (Supplementary Fig. S2) and apoptosis in a subset of tumor cell lines, including BT20, BT474, and ZR-75-1 breast cancer cells (26). To investigate the apoptosis pathways activated by BAY 80-6946, a set of studies were carried out with anti-HER2–resistant PIK3CA-mutated BT20 breast cancer cells. BAY 80-6946 significantly increased caspase-9 activities following 24-hour incubation with 2- and 3-fold inductions in caspase-9 activity at 20 and 62 nmol/L, respectively (Fig. 3A). These values were comparable with the IC₅₀ of 6 nmol/L for antiproliferative activity in the same cell line. The effect of BAY 80-6946 on caspase-9 activation was even stronger than that of paclitaxel, whereas rapamycin was inactive at 1 μmol/L (Supplementary Fig. S3). Caspase-9 activation leads to cleavage of several procaspases, which in turn cleave cellular targets including PARP, thereby impairing DNA repair. DNA damage is reflected by the level of p53 phosphorylation at Ser15. Consistent with the induction in caspase-9, BAY 80-6946 at 200 nmol/L also increased levels of phosphorylated p53 at Ser15 (Fig. 3B) and cleaved PARP (Fig. 3C). Caspase-9 activation by BAY 80-6946, lapatinib or the combination of the two agents was further investigated in the anti-HER2–sensitive breast cancer cell line BT474. Interestingly, although lapatinib effectively inhibited tumor cell proliferation with an IC₅₀ of 21 nmol/L (data not shown), it was incapable of inducing caspase-9 activation at a concentration up to 10 μmol/L (Supplementary Table S1). In contrast, BAY 80-6946 induced caspase-9 activation with an EC₅₀ of 340 nmol/L. Interestingly, when BAY 80-6946 was combined with lapatinib, the same degree of caspase-9 activation could be reached with significantly less levels of BAY 80-6946 (e.g., 61 nmol/L) and lapatinib (e.g., 184 nmol/L) with combination indices of 0.20 to 0.29, across all concentrations tested (Supplementary Table S1). Taken together, these findings showed that BAY 80-6946, as a single agent, could induce cell death in both anti-HER2–sensitive and -resistant, PI3K-mutant breast tumor cells at clinically achievable concentrations (27). Combination with lapatinib further synergistically sensitized tumor cells to apoptosis.

BAY 80-6946 has a high plasma-free fraction across all species tested (35% in rats, 14% in mice, 33% in dogs, and 16% in humans). The pharmacokinetic profile of BAY 80-6946 was evaluated following administration of single or multiple intravenous doses in nude rats (Supplementary Fig. S4). BAY 80-6946 exhibited very large volume of distribution (V_ss = 32 L/kg), high plasma clearance (3.95 L/kg·h) and long T_1/2 (6.0 hours). Similar exposure and other pharmacokinetic parameters were observed following multiple administration (every second day × 5), suggesting that there is no accumulation of compound or changes in metabolic processes. In comparison with the pharmacokinetic in rats, BAY 80-6946 had a higher plasma clearance (16 L/kg·h) and a shorter T_1/2 of 0.7 hours. The V_ss was large (12.9 L/kg) in mice (Supplementary Table S2).

To investigate pharmacokinetic/pharmacodynamic relationship, tumor samples were collected from nude rats bearing H460 human non–small cell lung carcinoma (NSCLC) xenografts and treated with a single 6 mg/kg i.v. bolus dose. The concentrations of BAY 80-6946 and pharmacodynamic markers in tumors were assessed. As shown in Fig. 4A, BAY 80-6946 levels were approximately 100-fold higher in the tumor than in plasma at 48 hours and drug clearance from the tumor occurred more slowly than from plasma. This high tumor exposure might be explained, in part, by the basicity and large V_ss of the drug and acidic environment in tumors. The levels of pAKT-Ser473 in tumors showed more than 90% inhibition at 24 hours after intravenous dosing and remained suppressed for 48 to 72 hours (Fig. 4B). Strong reduction of pAKT levels in tumors was observed as early as 1 hour after intravenous dosing (data not shown). The potent inhibition of tumor AKT phosphorylation correlates well with the high level of tumor exposure of BAY 80-6946, suggesting a strong pharmacokinetic/pharmacodynamic relationship. The inhibition of AKT phosphorylation in response to BAY 80-6946 was further confirmed by immunohistochemical analysis with pAKT staining (Fig. 4C). Compared with vehicle control, tumors treated with 6 mg/kg of BAY 80-6946 for 24 hours showed 65% and 75% reduction in Ki67 and phosphohistone H3 staining, respectively, indicating that tumor cells were arrested at the G₀-phase of the cell cycle (Fig. 4D). Rats bearing H460 xenografts were also evaluated by 2[¹⁸F]fluoro-2-deoxy-d-glucose–positron emission tomography (FDG-PET). BAY 80-6946 inhibited tumor uptake of FDG in a dose-dependent manner, with sustained inhibition evident over 24 to 48 hours after a single intravenous administration of 6 mg/kg (Fig. 4E and F). Taken together, these results suggest that in vivo, BAY 80-6946 reached high and prolonged exposure in tumors, which resulted in potent inhibition of the PI3K pathway exhibited by marked reduction in AKT phosphorylation and FDG uptake, as well as cell-cycle arrest in G₀ in the H460 NSCLC xenograft model in rats.

BAY 80-6946 was well tolerated at all doses and schedules tested in the studies described here without producing any lethality. The maximum-tolerated dose (MTD) in rats was defined as 6 mg/kg. At the MTD, a maximum mean body weight loss of 6% to 10% occurred during the first few days of dosing and then consistently returned to the normal range by the end of the dosing period. The MTD in mice was more than 14 mg/kg with a every second day dosing schedule.

BAY 80-6946 was highly efficacious in a variety of human tumor xenograft models derived from different tumor indications that exhibit an activated PI3K pathway. The drug displayed robust antitumor activity in the rat KPL4 tumor xenograft model, which is an estrogen-independent HER2-positive breast carcinoma that carries a somatic PIK3CA mutation. BAY 80-6946 was administered at 0.5 to 6 mg/kg i.v. every second day for a total of five doses starting on day 14, following tumor cell implantation. On day 25, 3 days after the last dose, TGI rates of 77%, 84%, 99%, and 100% were observed with BAY 80-6946 at doses of 0.5, 1, 3, and 6 mg/kg, respectively (Fig. 5A). Complete tumor regression was shown in 10 of 10 rats in the 3 and 6 mg/kg groups, and all rats remained tumor free at the termination of the study on day 73. Tumor growth delays more than 25 days were observed in the 0.5 and 1 mg/kg dose groups.

Treatment of BAY 80-6946 led to tumor stasis in nude rats bearing HCT-116 human colon tumors, which carry PIK3CA and KRAS mutations. BAY 80-6946 was administered at 3 and 6 mg/kg i.v. every second day for a total of five doses, producing TGI of 75% and 88%, respectively, and tumor growth delays of 10 and 11 days, respectively (Fig. 5B, data not shown).

BAY 80-6946 was also effective in xenograft models established in nude mice with Lu7860 erlotinib-resistant, patient-derived NSCLC and MAXF1398 patient-derived luminal breast tumor models. BAY 80-6946 was administered at a dose of 14 mg/kg every second day for a total of 10 days, producing TGI rates of 88% in the NSCLC model (Fig. 5C) and 71% in the breast cancer model (Fig. 5D).

The antitumor activity of BAY 80-6946 was also evaluated using a weekly dosing schedule, in which two doses of BAY 80-6946 were administered on the first day of the week. In the rat HCT-116 colorectal xenograft model, the TGI achieved with BAY 80-6946 at a dose of 9 mg/kg twice a day once weekly was equivalent to that with the 6 mg/kg dose given every second day for 10 doses (Fig. 6A). BAY 80-6946 at a dose of 6 mg/kg twice a day once weekly was efficacious with a TGI of 64%, comparable with the activity of docetaxel at 5 mg/kg every second day given for three doses.

Activation of PI3K has been shown to cause intrinsic and/or acquired resistance to chemotherapy. BAY 80-6946 was tested in combination with paclitaxel in Lu7343, a patient-derived PIK3CA (E545K)–mutated squamous cell NSCLC model in nude mice (Fig. 6B). Paclitaxel was given at doses of 15 or 25 mg/kg on days 14, 21, and 28. BAY 80-6946 was administered weekly at 10 mg/kg twice a day on days 15, 22, and 29 (24 hours after each paclitaxel dose). Treatment of BAY 80-6946 at 10 mg/kg/twice a day/week and paclitaxel at 15 mg/kg (60% of the MTD) alone showed only moderate efficacy with T/C% (treatment/control) values of 49% and 38%, respectively. Paclitaxel at the MTD (25 mg/kg) was effective and produced T/C% of 4% and 70% RR at the end of treatment on day 33 (Supplementary Table S3); however, tumor regrowth was observed immediately after stopping the treatment. Thus, on day 55, the RR in paclitaxel MTD group had dropped to 30% and tumors had progressed in 60% of the animals. The combination of BAY 80-6946 with both high and low doses of paclitaxel produced tumor regression, which was not only superior to either agent alone, but the combination with paclitaxel at the MTD also produced 100% RR with complete response (CR) in 5 of 10 and (PR) in 5/10 animals at the end of treatment. Importantly, in contrast with paclitaxel single agent at the MTD, the combination group showed a long-lasting tumor regression with 60% CR and 40% PR on day 55 (22 days after stopping treatment). These findings suggest that BAY 80-6946, when combined with paclitaxel, synergistically improved tumor response rates and led to long lasting tumor regression even when dosed twice a day weekly over 3 weeks in mice.

Many selective small-molecule inhibitors of PI3K have been identified, including selective pan-class I PI3K inhibitors, selective PI3Kδ inhibitors, and dual PI3K/mTOR inhibitors. These new PI3K inhibitors are currently in the early stages of clinical development. Several pan-class I PI3K inhibitors, including BKM120 and XL147, have entered phase II trials for various advanced malignancies, such as castration-resistant prostate, endometrial, and breast cancer, and a p110δ-selective PI3K inhibitor has entered phase III trials for lymphoma (28, 29). Although the pan-class I inhibitors have activity against each p110 isoform, they may show preferential activity for specific members. It is likely that the various PI3K isoforms play different roles based on tumor type, somatic mutations, and/or previous therapy, as well as different toxicology profiles; therefore, understanding the pharmacologic profile of each PI3K inhibitor may be important for developing PI3K therapeutics for specific malignancies. Here, we present BAY 80-6946, a highly selective and potent pan-class I PI3K inhibitor, which has preferential activity against the p110α and p110δ isoforms, compared with p110β and p110γ, based on biochemical assays and PI3Kα and PI3Kβ cellular assays. As it is difficult to compare the biochemical activity of the PI3K inhibitors against each p110 isoforms under physiologic conditions, we explored cellular assays in which the phosphorylation of AKT is driven by p110α or p110β, the two dominant isoforms expressed in solid tumors. The selectivity of BAY 80-6946 for the PI3Ks versus mTOR was evident in both the biochemical and cellular assays: BAY 80-6946 was 90-fold more active against PI3Kα than against mTOR in biochemical assays and more than 250-fold more active in cellular assays using ELT3 cells with PI3K-independent constitutive activation of mTORC1 due to defective TSC2. This finding indicates that BAY 80-6946 is a highly selective and potent pan-class I PI3K inhibitor that does not exhibit significant cellular mTOR inhibition in contrast with BEZ-235 (30). In fact, based on the human pharmacokinetic data of BAY 80-6946, the plasma-free drug levels in human are significantly (>20-fold) below the concentration required for direct inhibition of mTOR kinase by BAY 80-6946 (27).

The greater activity of BAY 80-6946 against p110α versus p110β seen in the biochemical assays (IC₅₀, 0.5 vs. 3.7 nmol/L) was confirmed in tumor cells. First, BAY 80-6946 was 16- to 20-fold more potent in inhibiting AKT phosphorylation in PIK3CA-mutated and HER2-overexpressing KPL4 cells (PI3Kα-driven p-AKT) compared with AKT phosphorylation in PTEN-null, LPA-stimulated PC3 cells (PI3Kβ-driven p-AKT). Second, BAY 80-6946 was about 10-fold more potent in inhibition of basal-level pAKT and p-p70S6K in cells with activated PI3Kα (such as BT20 and BT474) than in cells with loss of PTEN where PI3Kβ is activated (ZR-75-1 and MDA-MB-468; Table 1 and Fig. 1D). Interestingly, GDC-0941 showed equal potency in inhibition of cellular PI3Kα and PI3Kβ activities (Table 1). This result for GDC-0941 is in line with previous reports by O'Brien and colleagues (31)).

Profiling of BAY 86-6946 indicated that PIK3CA mutations and HER2 overexpression might be effective predictive biomarkers for predicting the sensitivity of breast tumor cells to BAY 80-6946. Thus, BAY 80-6946 had approximately 40 to 45 times greater antiproliferative activity against breast tumor cells with PIK3CA-activating mutations and/or HER2 overexpression than against those with HER2-negative tumors with wild-type PIK3CA (Fig. 2C). In contrast, the mean IC₅₀ for GDC-0941 in PIK3CA-mutant and/or HER2-overexpressing cells compared with PIK3CA wild-type and HER2-negative cells showed only about a 2- to 3-fold difference in this study (data not shown) and in another report (31). The difference of the two pan-class I PI3K inhibitors in this aspect might be explained by the differential profiles in PI3K isoforms and kinase selectivity. BAY 80-6946 is a more potent PI3Kα inhibitor compared with GDC-0941, but GDC-0941 has equally potent activity against both PI3Kα and PI3Kβ. The higher sensitivity in cell lines harboring PIK3CA-activating mutations or amplification of HER2 was also reported recently with the highly selective allosteric AKT inhibitor, MK-2206 (32). Hence, sufficient selectivity against other kinases and potency in blocking PI3Kα are indispensable for using PIK3CA mutation and HER2 overexpression as potential predictive biomarkers for PI3K pathway inhibitors.

Taken together, the differential isoform selectivity observed with BAY 80-6946 and other investigational PI3K inhibitors results in a differential antitumor profile (as described above for BAY 80-6946 versus GDC-0941). These findings suggest that the efficacy of the various PI3K inhibitors may differ across tumors with different genetic makeups. Understanding the preclinical profile of each PI3K inhibitor may help to guide its clinical development to the proper tumor type/subtype and also aid in formulating an optimal biomarker strategy for identifying patients most likely to benefit from treatment with a PI3K pathway inhibitor with a certain isoform profile. On the basis of these preclinical data, BAY 80-6946 was particularly active in a subset of breast cancer, endometrial cancer, and multiple myeloma cell lines, particularly those with PIK3CA-activating mutations and/or HER2 overexpression (Fig. 2).

Of note, BAY 80-6946 also potently induced apoptosis in BT20 human breast cancer cells with PIK3CA mutation when it was tested as a single agent at a concentration similar to the IC₅₀ for blocking cell proliferation. A detailed analysis of the molecular features for apoptosis induction is described in another article (26).

BAY 80-6946 administered intravenously showed potent antitumor activity in multiple cell-line–derived and patient-derived xenograft tumor models. In line with the in vitro cellular data, the greatest efficacy was seen in xenograft models derived from tumor cells with PIK3CA mutations or HER2 overexpression. In the HER2-positive/PIK3CA^mut KPL4 xenografts, BAY 80-6946 monotherapy dosed intravenous at the MTD and 50% of MTD (6 and 3 mg/kg every second day) produced complete tumor regression that persisted after discontinuation of therapy. To our knowledge, CR in 100% animals in this model has not been achieved by any of to date investigational PI3K pathway inhibitors, including GDC-0941 (33), CH5132799 (34), and MK-2206 (35), when they were dosed at their respective MTD. BAY 80-6946 also exhibited potent antitumor activity in HT116 colon cancer (Fig. 5B) and H460 NSCLC (data not shown) xenografts with coexisting PIK3CA and KRAS mutations, with complete tumor stasis observed in these models. However, in these models with KRAS mutations, it seems necessary to administer BAY 80-6946 in combination with another active agent, such as a MAP–ERK kinase (MEK) inhibitor, to achieve tumor regression.

BAY 80-6946 exhibits unexpected higher and prolonged exposure in tumors compared with plasma concentrations after a single intravenous dose (Fig. 4A). It is plausible that the higher exposure in tumors led to marked and sustained reductions in tumor pAKT levels and FDG uptake (lasting for at least 48 hours). Our initial data comparing p-AKT inhibition in tumors versus in blood monocytes indicated that BAY 80-6946 indeed suppressed p-AKT in tumors with better potency and over a longer time period (N. Liu; unpublished data). This result was consistent with the antitumor activity seen in xenograft models using a every second day schedule despite shorter plasma T_1/2 in rat (6 hours) and in mice (1 hour). Interestingly, administration of BAY 80-6946 even on a twice a day once-weekly schedule in HCT116 xenograft model in rats resulted in a similar antitumor activity as the every second day schedule. Furthermore, when BAY 80-6946 was combined with other agents, such as paclitaxel (Fig. 6B), weekly dosing was sufficient to generate synergistic antitumor effects and to induce complete tumor regression in mice. This result indicated that maintaining constant inhibition of PI3K pathway might not be necessary especially when combining with paclitaxel. Hence, intermittent combination with targeted and chemo therapeutics could be a promising approach to synergistically enhance the antitumor efficacy and meanwhile, maintaining the safety profile in combination treatment. Currently, a series of pharmacokinetic–pharmacodynamic efficacy studies of tumor versus plasma and continuous versus intermittent dosing schedules are being conducted. These data will provide further insights to best explore PI3K inhibitor BAY 80-6946 in clinical development.

On the basis of its antitumor profile and unique physical–chemical and pharmacokinetic properties, BAY 80-6946 is currently being investigated in phase I trials dosed weekly intravenous in patients with advanced malignancies and in combination with paclitaxel as well as with an allosteric MEK inhibitor (36). Analysis of the upcoming clinical data will provide further insights on how the PI3K isoform profile together with pharmacokinetic properties impact on the antitumor efficacy of BAY 80-6946 in the clinic.

S.M. Wilhelm has ownership interest (including patents) in Bayer HealthCare Pharmaceuticals. No potential conflicts of interest were disclosed by the other authors.

Conception and design: N. Liu, B.R. Rowley

Development of methodology: N. Liu, B.R. Rowley, P.R. Fracasso, D.P. Wilkie

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): N. Liu, B.R. Rowley, C.O. Bull, C. Schneider, A. Haegebarth, C.A. Schatz, P.A. Fracasso, D.P. Wilkie, M. Hentemann

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): N. Liu, B.R. Rowley, C.O. Bull, A. Haegebarth, C.A. Schatz, P.R. Fracasso, W.J. Scott

Writing, review, and/or revision of the manuscript: N. Liu, C.A. Schatz, A. Haegebarth, M. Hentemann, S.M. Wilhelm, W.J. Scott, D. Mumberg, K. Ziegelbauer

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): B.R. Rowley, W.J. Scott, M. Hentemann, D. Mumberg

Study supervision: N. Liu, B.R. Rowley, K. Ziegelbauer

The authors thank all the colleagues and collaborators who have worked on developing the PI3K inhibitor BAY 80-6946. The authors also thank Susan Jenkins, Randall Jones, Stefan Prechtl, Enrico Stasik, Katja Haike, and Andrea Sturz for their excellent experimental support and are grateful to Drs. Neal Rosen and Wei Jiang for critical review. While in preparation of this manuscript, the key contributor of this work, Dr. Bruce Rowley passed away from NSCLC on July 16, 2013. Bruce was truly an intelligent, solid, and creative scientist. We will be missing him on our way towards developing effective therapeutics for cancer patients.

The work was supported by Bayer HealthCare Pharmaceuticals.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.