Correspondence to: Olaf van Tellingen, Department of Clinical Chemistry, The Netherlands Cancer Institute/Antoni van Leeuwenhoek Hospital, Plesmanlaan 121, 1066 CX Amsterdam, The Netherlands, Tel.: +31-20-512–2792, Fax: +31-20-512–2799, E-mail: email@example.com
Palomid 529, a novel dual mTORC1/2 inhibitor has displayed interesting activities in experimental models and is a candidate for clinical evaluation. We have assessed the interaction of Palomid 529 with ATP-binding cassette (ABC) drug efflux transporters ABCB1 (P-gp/P-glycoprotein) and ABCG2 (BCRP/Breast Cancer Resistant Protein) by in vitro transwell assays, and their effects on the brain penetration using drug disposition analysis of i.v. and oral Palomid 529 in wild-type (WT) and Abcb1 and/or Abcg2 knockout (KO) mice. Palomid 529 lacked affinity for these transporters in vitro, in contrast to GDC-0941, a small molecule PI3K inhibitor, which we used as control substance for in vitro transport. The plasma AUCi.v. of micronized and DMSO formulated Palomid 529 was similar in WT and KO mice. Importantly, the brain and brain tumor concentration of Palomid 529 at a high dose (54 mg/kg) was also similar in both strains, whereas a less than 1.4-fold difference (p < 0.05) was found at the low (5.4 mg/kg) dose. Because of poor solubility, the oral bioavailability of micronized Palomid 529 was only 5%. Olive oil or spray-dried formulation greatly improved the bioavailability up to 50%. Finally, Palomid 529 effectively inhibits the orthotopic U87 glioblastoma growth. In summary, Palomid 529 is the first mTOR targeting drug lacking affinity for ABCB1/ABCG2 and having good brain penetration. This warrants further evaluation of Palomid 529 for treatment of high-grade gliomas and other intracranial malignancies.
The oncogenic activation of the phosphatidylinositol 3-kinase (PI3K)/AKT/mammalian target of rapamycin (mTOR) pathway is considered to be crucial for tumorigenesis and tumor development in most cancer types.[1-3] This is also the case for glioblastoma multiforme (GBM), the most common and lethal primary central nervous tumor. In GBM, activation of this pathway occurs by amplification and/or mutation of EGFR gene, mutation and/or deletion of the PTEN gene and mutations of the PIK3CA gene (rare cases). Taken together, activation of the PI3K-mTOR pathway takes place in up to 82% of patients. Recently, several major components in this pathway have shown to be promising targets for small molecular inhibitors. In particular, mTORC2 may have a special role in promoting GBM growth or mediating chemotherapy resistance. Palomid 529 (8-(1-Hydroxy-ethyl)-2-methoxy-3-(4-methoxy-benzyloxy)-benzo[c]chromen-6-one;) markedly reduces the phosphorylation of Akt (S473-Akt) signaling through inhibition of both mTOR-raptor and mTOR-rictor association and accordingly inhibits the mTORC1 and mTORC2 activity. In vivo studies showed that Palomid 529 reduced angiogenesis, vascular permeability and tumor growth. Moreover, it has been reported that Palomid 529 enhances the anti-proliferative effect of radiotherapy in glioblastoma in an orthotopic model as well as in prostate tumor models, and more recently, this dual mTORC1 and mTORC2 inhibitor has exhibited its effect to sensitize the chemotherapy in hormone refractory prostate tumors.
The brain is a pharmacological sanctuary site due to the inability of many drugs to cross the blood–brain barrier (BBB) in a therapeutic meaningful dose. The BBB is formed by the brain micro-vascular endothelial cells that are closely linked by tight-junctions, lack fenestrations and exhibit low pinocytic activity. Consequently, the entry of substances into the brain requires trans-endothelial trafficking, which is tightly regulated by a set of influx transporters (e.g., Glut1 for glucose). BBB penetration of agents that lack such an influx carrier is contingent on passive diffusion, which depends on the lipophilicity of the compound. However, even lipophilic substances may not reach into the brain as a further obstacle to BBB penetration is mediated by a set of ATP-binding cassette (ABC) drug efflux transporters that are expressed at the luminal side of the brain micro-vascular endothelial cells. In particular, ABCB1 (P-glycoprotein, P-gp, Mdr1) and ABCG2 (Breast Cancer Resistance Protein, BCRP), two dominant drug efflux transporters highly expressed at the murine and human BBB, are known to limit brain accumulation of many lipophilic anti-cancer agents as well as small molecule inhibitors.[13-15] Although brain tumors tend to cause a disruption of the BBB, it remains a major obstacle as this disruption occurs only in a part of the tumor and in particular many cells that have invaded into the normal surrounding brain are far away from this leaky area. Palomid 529 is a lipophilic substance, but whether it is a substrate for the ABC-transporters and if these would impede the drug delivery of Palomid 529 to the brain is yet unknown. In this study, we have explored the pharmacokinetic properties of this compound. We have assessed the affinity of Palomid 529 for ABCB1 and ABCG2, which are implicated in drug–drug interactions and drug resistance and that are known to limit the oral bioavailability and brain penetration of substrate drugs. Our results from both in vitro and in vivo models clearly suggest that unlike most of anti-cancer drugs, the interactions of Palomid 529 and ABCB1 or ABCG2 are minimal and that they do not restrict the brain penetration of Palomid 529.
Material and Methods
Palomid 529 (MW = 406.4) was provided by Paloma Pharmaceuticals (Jamaica Plain, MA) and was received as a micronized formulation suspended in Alcon Balanced Salt Solution containing 0.2% tyloxapol (prepared by AMRI, Burlington, MA), as a spray dried formulation (prepared by Bend Research, Bend, OR) and as pure (>98% purity) powder. GDC-0941 (MW = 513.2) was purchased from Chemdea (Ridgewood, NJ). Modified Eagles medium (MEM), l-glutamine, nonessential amino acids, MEM vitamins penicillin-streptomycin, fetal calf serum, trypsin-EDTA and other reagents for cell culture were purchased from Invitrogen (Invitrogen Corporation, Carlsbad, CA). Blank human plasma was obtained from healthy donors from the Central Laboratory of the blood Transfusion Service (Sanquin, Amsterdam, the Netherlands). All other chemicals were purchased from Merck (Darmstadt, Germany).
Drug analytical method
A high-performance liquid chromatography assay was developed and validated for the quantification of Palomid 529 in human and mouse plasma and in mouse tissue samples for the in vivo pharmacokinetic studies. In short, sample pretreatment involved liquid–liquid extraction with tert-butyl methyl ether. Palomid 529 and the internal standard Palomid 545 were chromatographically separated using a GraceSmart 5 μm RP18 column (2.1 × 150 mm2) and a mobile phase comprised of 50% (v/v) acetonitrile and 50% (v/v) milliQ water delivered at a flow speed 0.2 ml/min and were detected by a UV detector set at a wavelength of 315 nm.
In vitro transport experiments
The parental LLC pig-kidney cell line (LLC-PK1) and sublines transduced with murine Abcb1a (LLC-Mdr1a) or human ABCB1 (LLC-MDR1) were used to determine whether Palomid 529 was a substrate of murine Abcb1a (Mdr1a) or human ABCB1 (MDR1). The parental Madine Darby Canine Kidney (MDCK) type II cell line (MDCKII-parental) and murine Abcg2 (MDCKII-Bcrp1) or human ABCG2 transduced sublines (MDCKII-BCRP) were used to determine whether Palomid 529 was a substrate of murine Abcg2 or human ABCG2. Cells were used for a maximum of 13 passages after thawing. Complete medium prepared from MEM medium containing l-glutamine, nonessential amino acids, MEM vitamins penicillin-streptomycin, and 10% (v/v) fetal calf serum was used throughout. For transport studies, cells were seeded on Transwell microporous polycarbonate membrane filters (3.0 µm pore size, 24-mm diameter; Costar Corning, NY) at a density of 2×106 cells per well in 2 ml complete medium. Cells were incubated at 37°C in 5% CO2 for 3 days. Both the conventional bidirectional transport assay[17, 18] and the more recently described concentration equilibrium transport assay were utilized. Briefly, 2 ml of complete MEM medium containing 5 µg/ml Palomid 529 was applied to either apical or basolateral compartments in the conventional bidirectional transport experiments, or to both the apical and basolateral compartments in the concentration equilibrium transport experiments. Zosuquidar (LY335979, 5 µM) was added to the medium to inhibit endogenous canine ABCB1 when doing all experiments with MDCK cell lines, and in one experiment of GDC-0941 with LLC-PK1 cell lines. [14C]-inulin (approximately 1.59 × 106 DPM/ml) was added to check the integrity of the membrane. Samples of 50 µl were taken at 60, 120, 180 and 240 min and used for drug analysis by HPLC, whereas at the same time points 20 µl samples were collected only from apical side for radioactivity counting. Wells showing leakiness in excess of 1.5% per hour were excluded. Control experiments with 0.5 µM of GDC-0941 were run in parallel. GDC-0941 is a small molecule PI3K inhibitor, which is a typical substrate for both ABCB1 and ABCG2.
Mice were housed and handled according to institutional guidelines complying with Dutch legislation. All experiments with animals were approved by the animal experiment committee of the institute. The animals used for pharmacokinetics studies in nontumor bearing animals were female wild-type (WT), Abcb1a/1b−/−, Abcg2−/− and Abcb1a/1b−/−;Abcg2−/− mice, all of a >99% FVB genetic background, between 9 and 14 weeks of age. Animals injected with tumor cells used for the efficacy study or for pharmacokinetics studies in tumor bearing animals were athymic (nude) mice of FVB background with WT or Abcb1a/1b−/−;Abcg2−/− genotype. These latter Abcb1a/1b−/−;Abcg2−/− nude mice have been obtained by intercrossing of FVB nude mice with Abcb1a/1b−/−;Abcg2−/− mice and will be described in detail elsewhere (manuscript in preparation). The animals were kept in a temperature-controlled environment with a 12-hr dark/12-hr light cycle and received a standard diet (AM-II, Hope Farm B.B., Woerden, the Netherlands) and acidified water ad libitum.
Plasma and brain pharmacokinetics of Palomid 529 in nontumor bearing mice
FVB WT and/or Abcb1a/b−/−;Abcg2−/− mice (n = 5 for each group) received Palomid 529 as a solution of 54 mg/ml in DMSO at a dose of 54 mg/kg (1 µl of DMSO per gram body weight) or as a micronized formulation of 18 or 1.8 mg/ml at a dose of 54 or 5.4 mg/kg by i.v. (intravenous) or i.p. (intraperitoneal) injection into the tail vein. Next, about 50 µl blood samples were collected from the tip of the tail at time points 0.5, 2, 4, 8, 12 and 24 hr after dosing using heparinized capillaries (Oxford labware, St. Louis). At 24 hr blood was also sampled by cardiac puncture and the mice were sacrificed by cervical dislocation. To investigate the brain distribution of Palomid 529, two studies have been performed with DMSO dissolved drug or micronized formulated Palomid 529, both administered i.v. at a dose of 54 mg/kg to WT and Abcb1a/b−/−;Abcg2−/− mice (n = 5 for each group). To further assess the role of Abcb1 and Abcg2 in limiting the brain distribution of Palomid 529, a dose of 5.4 mg/kg of micronized Palomid 529 was administrated i.v. to WT and Abcb1a/b−/−, Abcg2−/− and Abcb1a/b−/−;Abcg2−/− mice (n = 6, 6, 5 and 9 for above group, respectively). At 1 and 4 hr after administration animals were anesthetized with isoflurane and blood was collected by cardiac puncture. Next, the mice were immediately sacrificed by cervical dislocation and brain, liver, kidney, lung, spleen and heart tissues were dissected. Blood samples from all experiments were centrifuged (10 min, 5,000 rpm, 4°C) and the plasma fractions in supernatant were transferred into clean vials. Tissues collected were weighed and homogenized using a polytron (Polytron PT1200, Kinematica AG, Littau, Switzerland) in a solution of 1% (w/v) bovine serum albumin in water. We used 3 ml for brain, liver and kidneys and 2 ml for the other tissues. Both plasma and homogenized tissue samples were stored at −20°C until analysis.
Oral drug formulation and administration
Oral formulations included an olive oil formulation (2 mg/ml), a micronized formulation (5.4 mg/ml) and a spray dried formulation (2 mg/ml). The olive oil formulation, micronized formulation and spray dried formulation (suspended in 0.5% (w/v) methyl cellulose) were orally administrated by gavage to FVB WT mice at dose of 20, 20 and 54 mg/kg, respectively (n = 5 for each group). The preparation of above formulations was described in Supporting Information. Sampling of tail blood and sample processing were performed as described above.
Brain and tumor distribution and in vivo efficacy in intracranial GBM model
We used FVB or Abcb1a/b−/−;Abcg2−/− nude mice with orthotopic grafted U87 GBM to investigate the Palomid 529 tumor distribution and its effect of GBM growth inhibition. To establish tumors, 100,000 U87-luc cells in 2 µl were injected stereotactically in the right forebrain (caudate nucleus) of mice as described previously. Mice with U87-luc tumors were used for the drug distribution study when the bioluminescence signal exceeded 108 photons per second by bioluminescence imaging using an IVIS 200 camera (Caliper Life Science, Alameda, CA). The micronized formulated of Palomid 529 was administered i.v. at a dose of 54 mg/kg (n = 4 or 5 for each group). At 1 and 4 hr after administration, whole blood was collected by cardiac puncture. The U87 tumors were collected together with the tumor-free left forebrains. Sample homogenization, pretreatment and analysis were performed the same as in the pharmacokinetic studies.
For in vivo efficacy study, the tumor load was established for each animal after seven days by bioluminescence imaging, and the animals were stratified, according to the bioluminescence signal, into a control group receiving no therapy and a treatment group receiving micronized Palomid 529 i.p. at a dose of 54 mg/kg per day (n = 8 or 9 for each group, respectively). Bioluminescence imaging was repeated twice every 3–4 days to establish the efficacy of the therapy. The amount of bioluminescence in each animal was calculated relative to the first measurement when therapy was initiated (arbitrarily set at 100%).
Pharmacokinetic calculations and statistical analysis
Pharmacokinetic parameters were calculated using an add-in program for Microsoft Excel PKSolver. The area under the plasma concentration–time curve (AUC) was calculated using the computing area under curve function of GraphPad Prism 5.01 (GraphPad Software, Inc., La Jolla, CA). The oral bioavailability (F) was calculated as the dose-corrected area under the plasma concentration–time curve after oral dosing (AUCp.o.) divided by the dose-corrected AUC after intravenous dosing AUCi.v. of the corresponding drug formulations using the following formula:
AUCp.o. of micronized formulation is compared with AUCi.v. of micronized formulation; AUCp.o. of olive oil formulation and spray dried formulation are compared with AUCi.v. of free drug dissolved in DMSO.
We applied the GLM (General linear model) repeated measures procedure to analyze the results of concentration equilibrium transport experiments using SPSS (v17.0; SPSS Inc, Chicago, IL). The differences of the percentage ratio of peak area of the measured samples to the reference (medium sampled at time = 0) between apical and basolateral compartments were considered as the observed values from repeated measurements. They were grouped by defining four sampling time points (1, 2, 3 and 4 hr) as a four-level within-subjects factor. Simple contrast was selected to compare the differences between the mean observed values of 2, 3 and 4 hr and 1 hr. Then, the multivariate significance tests were performed to determine whether the apical-basolateral differences of the Palomid 529 levels were significantly increased by the factor of time.
For in vivo pharmacokinetic experiments in which only WT and Abcb1a/b−/−;Abcg2−/− mice were involved, the two-tailed student's t-test was used to determine the significance between two groups. For the experiment in which more than two strains were included, one-way analysis of variance and post hoc Bonferroni was performed. Difference were considered statistically significant when p < 0.05. For in vivo efficacy study, the two-tailed student's t-test was used to determine the significance between tumor growth ratios of control and treated group.
In vitro transport of Palomid 529
We first determined whether Palomid 529 is a substrate of Abcb1 and Abcg2 using LLC-Mdr1a and MDCK-Bcrp1 cell lines in a conventional bidirectional setup, with apical-to-basolateral and basolateral-to-apical transport. After 4 hr, about 25–30% of Palomid 529 was recovered at the opposite side of the membrane, showing that this compound has appropriate cell membrane permeable properties (Figs. 1a and 1b). Overexpression of Abcb1 or Abcg2 did not cause vectorial basolateral-to-apical translocation of Palomid 529 in LLC-Mdr1a and MDCK-Bcrp1 cells. Since it has been reported that the conventional bidirectional transport assay may be less suitable to identify weak Abcb1 substrates, we have used the more recently described concentration equilibrium transport assay for all subsequent experiments. The concentration equilibrium transport system showed good sensitivity when we tested the translocation of GDC-0941, a PI3 kinase inhibitor that is a good substrate of Abcb1 and Abcg2 (Figs. 1c, 1f, 1i, 1j and 1l). With this setting, we were even able to detect translocation of GDC-0941 in the LLC-PK1 cell line that is most likely mediated by the low level of endogenously expressed porcine Abcb1 in these LLC-PK1 cells, because this translocation was attenuated by the specific P-gp inhibitor zosuquidar (Figs. 1f vs. 1g). Despite the sensitivity of this setup, there was still no basolateral-to-apical translocation of Palomid 529 by Abcb1a or ABCB1 (Figs. 1d and 1e), suggesting that this compound is indeed not a substrate of Abcb1a. We found a small albeit significant (p = 0.028, Figs. 1j and 1m) increase of the apical to basolateral translocation of Palomid 529 in murine Abcg2 overexpressing cells over time, but not in MDCK-parent cells, suggesting that this compound is a very weak Abcg2 substrate. Furthermore, we used the same setup to test translocation of Palomid 529 in corresponding cell lines that overexpress human ABCB1 and ABCG2, but we did not observe any basolateral-to-apical directed translocation of Palomid 529 in both cell lines, indicating that Palomid 529 is not a substrate of human ABCB1 or ABCG2 (Figs. 1e and 1k).
Plasma pharmacokinetics of Palomid 529 in mice
Next, we evaluated whether Abcb1a/b and/or Abcg2 have an impact on the plasma pharmacokinetics of Palomid 529 by using WT and Abcb1a/b−/−;Abcg2−/− (Abcb1a/b and Abcg2 deficient) mice. We measured the plasma concentration following i.v. administration of the micronized formulation and of Palomid 529 dissolved in DMSO at a dose of 54 mg/kg for up to 24 hr. There was no statistically significant difference in the AUC0–24 between WT and Abcb1a/b−/−;Abcg2−/− mice (Figs. 2a and 2b and Table 1). To check whether drug clearance was dose-independent, we also tested a 10-fold lower dose level of 5.4 mg/kg of Palomid 529 (Fig. 2c). All results demonstrate that Abcb1a/b and Abcg2 have no impact on the plasma clearance of Palomid 529. We also checked the plasma pharmacokinetics of Palomid 529 when it was administrated i.p. at a dose of 54 mg/kg and found that the pharmacokinetic parameters were similar to those of WT mice receiving i.v. Palomid 529 (Table 1 and Fig. 2d). Because of the convenience of i.p. administration, it provides a less stressful manner for tumor growth intervention study, which requires daily repeated drug administration.
Table 1. Plasma pharmacokinetic parameters from FVB wild type and/or Abcb1a/b;Abcg2−/− mice after i.v. or i.p. administration of Palomid 529 using different dose levels and formulations
Data except p values are mean ± SD, n = 5. Students t-test was performed to evaluate the difference in AUC values between wild type and Abcb1a/b;Abcg2−/− mice. We also established the pharmacokinetic parameters after i.p dosing because this route was used for daily dosing during the efficacy study. Abbreviations: AUC, area under plasma concentration–time curve; Cmax, maximum plasma concentration; T½, el, elimination half-life, calculated from 1 to 2 hr; Cl, clearance.
AUC (0–24 hr) ng/ml × hr
15690 ± 2340
14600 ± 985
120980 ± 20410
132690 ± 25830
119330 ± 5250
146360 ± 41180
132002 ± 13362
3390 ± 920
3590 ± 300
15090 ± 1090
15270 ± 1300
37020 ± 3770
50890 ± 12670
15358 ± 2041
3.33 ± 0.25
3.30 ± 0.380
5.22 ± 0.40
4.80 ± 1.27
2.42 ± 0.08
2.63 ± 0.24
5.15 ± 0.49
0.36 ± 0.06
0.38 ± 0.03
0.45 ± 0.07
0.42 ± 0.11
0.49 ± 0.02
0.44 ± 0.19
0.39 ± 0.02
The formulation of Palomid 529, however, did have some effect on the plasma concentration time curve. At the first sampling point of 30 min after i.v. dosing the plasma concentration with the micronized formulation was about twofold to threefold lower than when Palomid 529 was dissolved in DMSO. On the other hand, the half-life of micronized Palomid 529 was longer, but overall the AUC (and plasma clearance) between the two formulations was similar. Most likely, a fraction of the relatively large micronized particles (0.8 µm) is removed from the circulation during their passage through tissues of the mononuclear phagocyte system and is retained there. Next, redistribution occurs as Palomid 529 is (slowly) released from the particles and re-enters the blood circulation. This hypothesis is supported by the finding of much higher Palomid 529 levels in liver and spleen at 24 hr, whereas the levels in brain and kidney follow the plasma level (Fig. 2e).
Role of Abcb1 and Abcg2 in brain penetration of Palomid 529
Because both ABCB1 and ABCG2 are very efficient in keeping substances out of the brain, we first used FVB WT versus Abcb1a/b−/−;Abcg2−/− mice lacking both Abcb1a/b and Abcg2 to evaluate the impact of Abcb1 and Abcg2 on brain penetration of Palomid 529. In line with the previous results from tail blood samples, the plasma concentration of Palomid 529 does not differ significantly between these two genotypes at both 1 and 4 hr after i.v. administration of 54 mg/kg micronized or DMSO solubilized Palomid 529 (Figs. 3a and 3b). Moreover, the brain penetration of Palomid 529 after administration of micronized Palomid 529 was not higher in Abcb1a/b−/−;Abcg2−/− mice at 1 and 4 hr. Administration of Palomid 529 in DMSO resulted in a very modest 1.33-fold higher brain–plasma ratio in Abcb1a/b−/−;Abcg2−/− mice compared with WT mice at 1 hr (p = 0.036) and a nonsignificant 1.22-fold higher ratio at 4 hr after drug administration (p = 0.059). Thus, it shows that Abcb1 and/or Abcg2 play only a very minor role in limiting the brain penetration of Palomid 529.
To identify which transporter contributes mainly to the small increase of brain penetration of Palomid 529 in the compound knockout (KO) mice, and to check whether Abcb1 and Abcg2 may be more important at lower plasma concentrations, we have performed an additional experiment with Palomid 529 at a dose of 5.4 mg/kg. We have also included single Abcb1a/b−/− and Abcg2−/− animals, next to WT and Abcb1a/b−/−;Abcg2−/− mice in this analysis. In line with results from the high-dose study, a significant increase (1.45-fold, p = 0.001) of the brain–plasma ratio of Palomid 529 was observed in Abcb1a/b−/−;Abcg2−/− mice relative to WT mice at 1 hr after administration (Fig. 3c). We also observed a 1.38-fold increase (p = 0.015) of the brain–plasma ratio of Abcb1a/b−/− mice relative to WT mice whereas that of Abcg2−/− mice was 1.31-fold higher but this difference relative to WT mice did not reach statistical significance (p = 0.081). At 4 hr, the brain concentration and brain–plasma ratio of Palomid 529 was also 1.45-fold higher in Abcb1a/b−/−;Abcg2−/− mice relative to WT but not significantly higher in any of the single KO mice. Thus, overall only a very moderate restriction of the brain entry of Palomid 529 is noted, which appears to be caused by a concerted action of Abcb1 and Abcg2.
Bioavailability of Palomid 529 is improved by an optimized oral formulation
Chronic treatment period is usually required for targeted therapeutics such as mTOR inhibitors, thus a convenient p.o. application form with acceptable oral availability is certainly a preferred feature for such drugs. To investigate the bioavailability of Palomid 529, we first tested the micronized formulation of Palomid 529 in FVB WT mice at a dose of 54 mg/kg and monitored the plasma concentration of Palomid 529 for up to 24 hr after administration. Unfortunately the bioavailability was only 4.8% (Fig. 4 and Table 2). To assess whether the poor bioavailability might be due to a slow dissolution rate of Palomid 529 from the micronized form or due to poor penetration and uptake by the intestinal epithelium, we switched to a more solubilized drug formulation in olive oil. To minimize the possibility of drug precipitation in the gut as a possible source of low oral bioavailability, we started with a lower dose level of 20 mg/kg. As shown in Table 2, the oral bioavailability of mice receiving 20 mg/kg of Palomid 529 formulated in olive oil was 54.0%, indicating that the low oral bioavailability with the micronized form was probably due to a slow dissolution rate of Palomid 529 in vivo. When we increased the concentration of Palomid 529 in olive oil formulation to deliver a dose of 54 mg/kg, the plasma AUC did not increase (Table 2). As Palomid 529 is a very poorly water-soluble compound, this might be due to a more extensive precipitation of Palomid 529 occurring when the more concentrated drug solution in oil is being mixed with the aqueous gastric fluids of the animal. Still, the oral bioavailability of 54 mg/kg of Palomid 529 in the oily formulation was about 18% and therefore much better than using the micronized formulation. To explore a more clinically acceptable drug formulation, we performed a pilot study with administration of a spray-dried formulation of Palomid 529. At a dose of 20 mg/kg, we found a bioavailability of 32.2% (Fig. 4 and Table 2). With this formulation, a further increase of the dose may be feasible without compromising the bioavailability. The results obtained from the latter two formulations, particularly the spray-dried formulation, suggested that despite of a poor solubility of Palomid 529, a relatively acceptable bioavailability could be achieved with further optimization of the formulation.
Table 2. Plasma pharmacokinetic parameters from WT mice after oral administration of Palomid 529 at different doses and in different formulations
Micronized formulation (dose = 54 mg/kg)
Olive oil formulation (dose = 20 mg/kg)
Olive oil formulation (dose = 54 mg/kg)
Spray dried formulation (dose = 20mg/kg)
Data are mean ± SD, n = 5/6. Abbreviations: AUC, area under plasma concentration-time curve; Cmax, maximum plasma concentration; T½, el, elimination half-life, calculated from 1 to 2 hr; Cl, plasma clearance. F, oral bioavailability, calculated as the dose-corrected AUCp.o. divided by AUCi.v. of corresponding drug formulations.
AUC (0–24 hr)
5800 ± 3660
23880 ± 3020
22224 ± 4422.09
14250 ± 3660
1100 ± 240
4820 ± 1330
2710 ± 810
2540 ± 670
2.87 ± 1.15
3.12 ± 0.45
3.51 ± 0.32
2.76 ± 0.88
9.67 ± 2.9
0.84 ± 0.19
2.70 ± 0.54
1.49 ± 0.45
4.9 ± 1.0
54.0 ± 6.8
17.8 ± 3.7
32.2 ± 8.3
Palomid 529 tumor distribution and in vivo efficacy in intracranial GBM model
To investigate the roles of Abcb1 and Abcg2 in the tumor distribution of Palomid 529, we performed a pharmacokinetic analysis using the FVB nude mice bearing an intracranial U87 GBM tumor. In line with our previous experiments using nontumor bearing FVB mice, the Palomid 529 concentrations in plasma and left (tumor free) brain hemisphere of Abcb1a/1b−/−;Abcg2−/−versus WT mice were similar at 1 and 4 hr after administration of micronized formulation. Similarly, Abcb1 and Abcg2 also do not affect the Palomid 529 tumor distribution since the tumor concentrations in WT and Abcb1a/1b−/−;Abcg2−/− mice were also not significantly different (Figs. 5a–5c). Interestingly, the Palomid 529 concentrations in plasma, brain and tumor were very similar, suggesting this agent has excellent brain permeability. Overall, we noted that the Palomid 529 concentrations in plasma and brain were about twofold lower than in the previous studies (Figs. 3 and 5). This difference in clearance may be strain (FVB nude mice vs. FVB) and/or tumor related.
Finally, we evaluated the efficacy of Palomid 529 in the same GBM model. Treatment composed of daily i.p. administration of micronized Palomid 529 at a dose of 54 mg/kg was well tolerated and significantly inhibited orthotopic U87 tumor growth relative to untreated mice (Fig. 5d).
In this study, we show that Palomid 529 is not or just a very weak substrate of Abcb1 and/or Abcg2 and accordingly its systemic availability and brain and brain tumor delivery is hardly affected by these two transporters. This is an interesting finding given the broad substrate ranges of these two ABC transporters that alone or together severely limit the brain entry of many substances. This pharmacokinetic feature of Palomid 529 provides a unique advantage for potential application of this drug against intracranial tumors as also demonstrated in a U87 orthotopic GBM model. Oral application of this drug will be possible, as we found that an acceptable oral availability (>50%) could be achieved by optimization of the formulation.
ABCB1 and ABCG2 are two of the most important ABC drug transporters frequently being associated with drug resistance. This includes expression at the cell membrane of tumor cells as well as expression at barrier sites to build pharmacological sanctuary sites where tumor cells can flourish. ABCB1 and ABCG2 have well-established roles in cooperatively limiting the brain penetration of many cytotoxic anticancer drugs such as topotecan, but also for novel targeted agents. In recent years, a series of small molecular inhibitors including erlotinib[23-25], gefitinib, sunitinib, sorafenib, dasatinib,[28, 29] imatinib, tamoxifen, lapatinib, GDC-0941, vemurafenib have been identified as substrates of Abcb1 and/or Abcg2 and the brain penetrations of these compounds are markedly restricted by these two transporters.
Despite that Abcb1 transports a wide variety of substrates, Palomid 529 was not translocated by murine and human Abcb1, even when using a concentration equilibrium setup, which was claimed as an appropriate and more sensitive assay for highly cell membrane permeable drugs. Similarly, we found that Palomid 529 was not transported by human ABCG2, and that murine Abcg2 might transport Palomid 529 only at a marginal level. Because of the poor aqueous solubility of Palomid 529, 10% fetal calf serum was added to the medium of the apical and basolateral compartments in both models to avoid loss of drug by precipitation and/or adsorption. With about 25% translocation of Palomid 529 in the classical Transwell assay, it is shown here that Palomid 529 can still freely cross the cell-monolayer in the presence of serum proteins. Consequently, the absence of directional transport of Palomid 529 in both ABC-transporter transduced LLC and MDCK cell lines can only be due the fact that the affinity of Palomid 529 for both ABC-transporters is weak. We, next, confirmed these findings by using Abcb1a/b−/−;Abcg2−/− mice.
Abcb1 was the first ABC transporter being identified as an important factor limiting the brain penetration of substances and usually it is the most dominant efflux transporter restricting the brain delivery of many chemotherapeutics. Only for some drugs, for example, sorafenib, its brain accumulation is predominantly limited by Abcg2 instead of Abcb1. In many cases, however, the absence of both transporters results in a more than proportional increase in brain uptake compared to the absence of each transporters alone.[13-15, 29-31, 33] The marked potency of Abcb1 to limit the brain entry of substances is also observed to some extent with Palomid 529. Although the in vitro transwell experiments for Palomid 529 was not capable to show transport by Abcb1, the brain concentration was still significantly, albeit only very moderately, increased in Abcb1 deficient mice. In contrast, the absence of Abcg2 alone did not statistically change the brain concentration and brain–plasma of Palomid 529 ratio in comparison with WT mice. Besides the differences in substrate affinities, the expression of Abcg2 in mouse BBB is also lower than that of Abcb1 as demonstrated by a proteomic based quantitative analysis of protein expression of transporters expressed in human and murine BBB, where Abcg2 expression was only 31.3% of Abcb1 expression in mouse brain microvessels. Expression of ABCG2 in human BBB is 1.85-fold higher than in mouse BBB, but due to the fact that Palomid 529 is not or just a very weak substrate of human ABCG2, it is probably not a limiting factor in the brain distribution in humans as well. Moreover, the effect of ABCB1 in human brain might also be smaller than in mouse brain, because ABCB1 expression in human brain microvessels was only 43.0% of its mouse counterparts. Consequently, ABCB1 and ABCG2 are not expected to exert a meaningful impact on the brain penetration of Palomid 529 in humans, making this compound potentially useful for treatment of intracranial tumors. Notably, at a therapeutically useful dose of 50 mg/kg7, the transport capacity of Abcb1 and Abcg2 might be saturated. Therefore we included a lower dose of 5.4 mg/kg to further determine the role of transporters. Indeed, the difference in the brain concentration of Palomid 529 between WT and Abcb1a/b−/−;Abcg2−/− mice became more evident at the lower dose. Nevertheless, the brain-to-plasma ratio was about twofold higher at the 54 mg/kg dose compared to 5.4 mg/kg in both WT and Abcb1a/b−/−;Abcg2−/− strains, which may implicate saturation of another, unknown efflux system, independent of Abcb1 and Abcg2.
Besides the BBB, ABCB1 and ABCG2 are also expressed on the blood tumor barrier. In fact, overexpression of ABCB1 and ABCG2 on tumor cells often confers multidrug resistance (MDR) and are linked clinically to poor prognosis.[35, 36] This is presumably due to a reduction of cellular drug accumulation caused by ABC-transporter mediated drug efflux. For drugs that are excellent substrates of Abcb1 such as doxorubicin, even a moderate increase of expression of Abcb1 in tumors is sufficient to confer resistance. In addition, ABC-transporter mediated acquired drug resistance is less likely when non/poor-substrate drugs are applied. Consequently, it may be preferred to use active substances that are not or only very weak substrates of ABCB1 and ABCG2. This may be even more valid for targeted agents that need to be administered chronically, as prolonged exposure increases the chance to cause an upregulation of expression of ABC-transporters. Using mice bearing orthotopic U87 tumor, we were able to show that both Abcb1 and Abcg2 did not impair the Palomid 529 distribution in tumor.
A major obstacle of the current standard of care of patients with malignant gliomas is that after surgical resection of the bulk tumor mass, there are still many tumor cells remaining that have invaded into the normal surrounding brain. Because of the protection by an intact BBB, these tumor cells are poorly accessible by most commonly used chemotherapeutic agents and can escape from therapy. Interestingly, other drugs, which poorly penetrate the BBB demonstrate a much more profound accumulation into brain tumor tissue relative to normal brain because of the leakiness of the vessels in those tumors. Consequently, the antitumor efficacy of these agents results from the fact that these experimental tumors are exposed to a very high drug concentration. For example, the brain tumor accumulation of vincristine was tenfold higher compared to normal brain as shown by Wang et al. In contrast, the concentration of Palomid 529 in brain and brain tumor was similar, despite the fact that U87 tumors also possess a leaky vasculature. These results strongly suggest that this agent will also be delivered at pharmacologically active levels to the invading GBM cells residing in normal brain with functional BBB.
High-grade glioma is a disease with a very dismal prognosis. Although the PI3K-mTOR signaling pathway is very frequently activated in this disease, it is not very likely that a single agent directed against this pathway by itself can exert a potent anti-proliferative effect. In this respect, however, Palomid 529 has been shown to synergize with radiation in experimental models including orthotopic gliomas.[8, 9] Importantly, in this study, we demonstrate that pharmacologically active levels of Palomid 529 can be delivered throughout the brain, which is important as this drug is being considered for clinical evaluation in high-grade glioma.
In short, Palomid 529 has an advantageous pharmacokinetic feature as being not or just a very weak substrate of Abcb1/ABCB1 and/or Abcg2/ABCG2. This feature enables Palomid 529 to enter the brain more easily than most other anti-cancer drugs whose brain penetrations are severely limited by Abcb1 and Abcg2. Consequently, pharmacologically active levels are reached throughout the brain. The systemic exposure of Palomid 529 is not affected by ABC-transporters and the availability of an orally applicable spray-dried dosing formulation allows convenient employment of this drug. Together, these preclinical results warrant further clinical testing of this drug.
The authors gratefully acknowledge the statistical advice of Dr. Marta Lopez Yurda and Dr. Catharina M. Korse.