Preclinical Systemic Pharmacokinetics, Dose Proportionality, and Central Nervous System Distribution of the ATM Inhibitor WSD0628, a Novel Radiosensitizer for the Treatment of Brain Tumors

Abstract

Radiation therapy, a standard treatment option for many cancer patients, induces DNA double-strand breaks (DSBs), leading to cell death. Ataxia telangiectasia mutated (ATM) kinase is a key regulator of DSB repair, and ATM inhibitors are being explored as radiosensitizers for various tumors, including primary and metastatic brain tumors. Efficacy of radiosensitizers for brain tumors may be influenced by a lack of effective drug delivery across the blood-brain barrier. The objective of this study was to evaluate the systemic pharmacokinetics and mechanisms that influence the central nervous system (CNS) distribution of WSD0628, a novel and potent ATM inhibitor, in the mouse. Further, we have used these observations to form the basis of predicting effective exposures for clinical application. We observed a greater than dose proportional increase in exposure, likely due to saturation of clearance processes. Our results show that WSD0628 is orally bioavailable and CNS penetrant, with unbound partitioning in CNS (i.e., unbound tissue partition coefficient) between 0.15 and 0.3. CNS distribution is not limited by the efflux transporters P-glycoprotein and breast cancer resistant protein. WSD0628 is distributed uniformly among different brain regions. Thus, WSD0628 has favorable pharmacokinetic properties and potential for further exploration to determine the pharmacodynamics-pharmacokinetics efficacy relationship in CNS tumors. This approach will provide critical insights for the clinical translation of WSD0628 for the treatment of primary and secondary brain tumors.

SIGNIFICANCE STATEMENT

This study evaluates the preclinical systemic pharmacokinetics, dose proportionality, and mechanisms influencing CNS distribution of WSD0628, a novel ATM inhibitor for the treatment of brain tumors. Results indicate that WSD0628 is orally bioavailable and CNS penetrant without efflux transporter liability. We also observed a greater than dose proportional increase in exposure in both the plasma and brain. These favorable pharmacokinetic properties indicate WSD0628 has potential for further exploration for use as a radiosensitizer in the treatment of brain tumors.

Introduction

The majority of cancer patients receive ionizing radiation therapy as a standard treatment (Jaffray and Gospodarowicz, 2015; Gong et al., 2021). Ionizing radiation causes DNA damage by inducing DNA double-strand breaks (DSBs), leading to cell cycle arrest and DNA repair. A complex network of DNA damage response signaling pathways is activated to regulate and repair the induced damage to DNA. Ataxia telangiectasia mutated (ATM) kinase is a key orchestrator of DSB repair. Unrepaired DSBs ultimately trigger subsequent activation of various cell death mechanisms like apoptosis, mitotic catastrophe, necrosis, senescence, and autophagy (Barton et al., 2006, 2014; Tyldesley et al., 2011; Baskar et al., 2012). Thus, inhibiting ATM can prevent the repair of damaged DNA and increase the sensitivity of cells to radiation therapy (Fig. 1). Several ATM inhibitors are being developed as radiosensitizers to increase the efficacy of radiation therapy as well as other DSB-inducing agents, such as cytotoxic chemotherapy, in various tumors (Bindra et al., 2017; Waqar et al., 2022).

Fig. 1.

ATM-mediated DNA damage repair. DNA damaging radiation therapy and chemotherapy-induced DNA double-strand breaks are repaired by the ATM pathway. WSD0628 is a strong inhibitor of ATM and prevents the repair of damaged DNA, thereby increasing the sensitivity of cells to radiation and chemotherapy. Figure created using BioRender.com.

Standard treatment options for brain tumors include surgery, radiation therapy, and chemotherapy. However, brain tumors present additional challenges to efficacious therapy, such as constrained access to the tumor because of its anatomical location, limited drug penetration and drug efflux at the blood-brain barrier (BBB), heterogeneously disrupted brain-tumor barrier, and the infiltrative nature of some brain tumors like glioblastoma (GBM) (Van Tellingen et al., 2015; Phoenix et al., 2016; Aldape et al., 2019). GBM is the most common, aggressive type of primary malignant brain tumor with a median survival of about 14 months after diagnosis (Mohammed et al., 2022; Ostrom et al., 2022). Because of its infiltrative nature, GBM can be regarded as a disease of the whole brain with often invasive GBM cells residing behind an intact BBB (Agarwal et al., 2011b). Thus, a crucial step is to develop drugs and drug delivery systems with adequate BBB permeability for safe and efficacious activity. Additionally, brain tumors are the most common solid tumors in pediatrics and a major cause of death in this age group (Smith and Reaman, 2015; Ostrom et al., 2022). Thus, there is a critical unmet need to develop novel therapeutics and improve the efficacy of current brain tumor therapies.

ATM inhibitors such as KU-55933, KU-60019, KU-59403, CP-46672, AZ31, AZ32, AZD0156, AZD1390, and WSD0628 are being explored in combination with radiation therapy to develop efficacious treatments for primary brain tumors (Jin and Oh, 2019). KU-55933, KU-60019, KU-59403, and AZ31 are potent and selective ATM inhibitors but with poor BBB permeability (Hickson et al., 2004; Golding et al., 2009; Batey et al., 2013; Karlin et al., 2018). AZD0156 was discovered with significantly improved potency and oral bioavailability (F) and is currently being clinically explored for peripheral [non-central nervous system (CNS)] solid tumors (Pike et al., 2018). However, AZD0156 has limited CNS permeability, partly because it is a substrate for the efflux transporters P-glycoprotein (P-gp) and breast cancer resistant protein (Bcrp) (Durant et al., 2018). AZD1390 is a structural analog of AZD0156, with improved BBB penetration (Durant et al., 2018). AZD1390 is currently being explored as a radiosensitizer in various clinical trials for CNS and non-CNS applications (NCT05182905, NCT03215381, NCT 05116254, NCT03423628, NCT05678010) (ClinicalTrials.gov).

WSD0628 is a novel, highly potent (IC₅₀ against ATM < 1nM) and selective ATM inhibitor, with robust in vitro radiosensitization potential, low hERG liability, and lack of aldehyde oxidase metabolism liability, which can result in lower intersubject variability in drug clearance (Manevski et al., 2019; Zhong et al., 2021; Tuma et al., 2022). There are limited reports of preclinical systemic pharmacokinetics and CNS distribution of WSD0628. The objective of this study is to evaluate the preclinical systemic pharmacokinetics, dose proportionality, CNS distribution, and mechanisms influencing CNS distribution of WSD0628. We evaluated the role of major efflux transporters, P-gp and Bcrp, in limiting the CNS distribution and examined spatial distribution of WSD0628 in different regions of the brain. We also evaluated the potential for radiosensitization-associated toxicities to other tissues and the extent of binding to plasma and CNS tissues. Based on our results, we observed a greater than dose proportional increase in the plasma and CNS exposure of WSD0628. WSD0628 is CNS penetrant, and CNS distribution is not limited by P-gp and Bcrp. These results indicate that WSD0628 has potential as a radiosensitizer in patients with brain tumors. Modeling and simulation-based approaches were used to confirm saturable elimination of WSD0628 and predict dosing regimens that will most likely result in effective exposure in tumors in the CNS. Overall, this study provides critical insights into the pharmacokinetics and CNS penetration of WSD0628, which can be useful to make decisions about the clinical translation of this molecule for CNS tumors.

Materials and Methods

Chemicals and Reagents

WSD0628 (PCT number: No. PCT/US2020/051833) was obtained from WayShine Biopharma. The internal standard for the liquid chromatography-tandem mass spectrometry (LC-MS/MS) assay of WSD0628 was AZD1390. Regenerated cellulose semipermeable membrane rapid equilibrium dialysis (RED) inserts (8kDa molecular weight cutoff) and 96-well high-density polypropylene base plates were purchased from Thermo Fisher Scientific (Waltham, MA). Solvents, other chemicals, and reagents were high-performance liquid chromatography-grade or analytical grade and purchased from Thermo Fisher Scientific or Sigma-Aldrich (St. Louis, MO).

Animals

Pharmacokinetic investigations were conducted using an equivalent number of male and female (four mice at each time point) Friend Leukemia Virus strain B (FVB), age between 8 and 16 weeks, wild-type, and Mdr1a/b^−/−Bcr1p^−/− (P-gp and Bcrp triple knockout) mice. The breeding pairs were procured from Taconic Bioscience, Inc. (Germantown, NY). The mice were maintained and housed as per the established and approved breeding protocols in the Research Animal Resource facility located at the Academic Health Center, University of Minnesota. The mice were maintained on a 12 hour light/dark cycle with ad libitum access to food and water. Regular tail biopsies were conducted to determine and confirm their genotypes (TransnetYX, Cordova, TN). All the protocols for animal experiments were approved by the University of Minnesota Institutional Animal Care and Use Committee and performed in accordance with the principles outlined in the Guide for the Care and Use of Laboratory Animals established by the US National Institutes of Health (Bethesda, MD).

Drug distribution studies in tumors were performed using female athymic nude mice (Crl: NU–Foxn1^nu, Charles River, Wilmington, MA) between the age of 6 and 8 weeks with intracranially implanted eGFP/fLuc2 labeled GBM43, a TP53 mutant glioma patient-derived xenograft (PDX). Published reports indicate TP53 mutant gliomas are sensitive to the combination of ATM inhibitors with ionizing radiation (Biddlestone-Thorpe et al., 2013). PDX maintenance and intracranial injections were performed as previously described (Carlson et al., 2011). The mice were maintained on a 12 hour light/dark cycle with ad libitum access to food and water. Mice were randomized into treatment groups 3 weeks after inoculation with GBM43. This animal work was approved by the Institutional Animal Care and Use Committee, Mayo Clinic, Rochester.

In Vitro Assessment of Free (Unbound) Fractions

The unbound fraction (f_u) of WSD0628 was determined in mouse plasma, FVB brain homogenate, FVB spinal cord tissue homogenate, and human GBM tumor using the RED device as per the manufacturer’s protocol. Brain, spinal cord, and GBM tumor homogenates were prepared by adding three volumes of PBS (pH 7.4) and homogenizing using Omni THB-01 tissue homogenizer (Omni International, Kennesaw, GA) at medium speed for 30 seconds. Plasma, brain homogenate, spinal cord homogenate, and tumor homogenate were spiked with WSD0628 to attain a final concentration of 5 μM containing 0.3% v/v DMSO. An aliquot of 300 μL of spiked matrix was added to the donor chamber in the RED insert, and 500 μL of PBS with 0.3% v/v DMSO was added to the corresponding receiver chamber. The device was covered using an impermeable sealing tape and incubated at 37°C on a shaker at 600 rpm for 6 hours. Post incubation, samples were collected from the donor and receiver chamber and stored at –80°C until drug concentration analysis using LC-MS/MS. The f_u of WSD0628 in plasma was calculated as the ratio of drug concentration in the receiver chamber to the drug concentration in the donor chamber. Unbound fraction of WSD0628 in brain, spinal cord, and tumor was calculated using eq. 1 (Kalvass and Maurer, 2002), and f_u,diluted is the ratio of drug concentration in the receiver chamber to the donor chamber. The dilution factor, D, is 4 in this case.

Systemic and CNS Distribution and Dose Proportionality Following a Single Intravenous and Oral Dose

Wild-type mice (n = 4/time point) were dosed with 1, 5, or 10 mg/kg WSD0628 intravenously with a bolus injection in the tail vein or 0.1, 1, 5 or 10 mg/kg WSD0628 orally. The vehicle for the intravenous (IV) bolus dose was 30% v/v 1:1–ethanol: Cremophor EL and 70% v/v saline. The vehicle for the oral dose was 0.5% w/v hydroxypropyl methylcellulose, 0.1% w/v Tween 80 suspended in distilled water. To test if absorption was dissolution-rate limited, WSD0628 was also administered at a dose of 10 mg/kg as a solution orally, dissolved in 2% v/v DMSO, 50% v/v PEG400, 10% v/v absolute ethanol in water. Mice were euthanized using CO₂ inhalation at time 0.167, 0.5, 1, 2, 4, 8, 12, 16, 24, 32, and 48 hours post dose. The 32- and 48 hour time points were only studied for the 10 mg/kg dose. Blood was collected using heparinized syringes by cardiac puncture in heparinized tubes followed by centrifugation at 7500 rpm for 10 min at 4°C to separate plasma. The brain was removed from the skull and rinsed with saline, followed by blotting on Kimwipes to remove the superficial meninges. The spinal cord was collected by hydraulic extrusion from the distal end of the spinal column, washed with saline, and blotted on a Kimwipe (Overland et al., 2009; Richner et al., 2017). Plasma and tissue specimens were stored at –80°C until analysis by LC-MS/MS. Brain concentrations were corrected for residual blood by subtracting drug concentrations in the brain vascular space, estimated as 1.4% of the whole brain volume (Dai et al., 2003).

Spatial Distribution in Different Brain Regions

Spatial distribution of WSD0628 in different regions of the brain was studied in FVB wild-type mice after a 7.5 mg/kg oral dose. At 2 and 8 hours after dose, mice (n = 4/time point) were euthanized and blood, brain, and spinal cord were collected. Blood was collected by cardiac puncture using heparinized syringes and heparinized tubes followed by centrifugation at 7500 rpm for 10 min at 4°C to separate plasma. The brain was immediately dissected to separate cortex, thalamus and hypothalamus, midbrain, pons, cerebellum, and medulla. All tissue specimens were stored at –80°C until analysis by LC-MS/MS.

Role of Efflux Transporters at the BBB (P-gp and Bcrp) in CNS Distribution

Influence of efflux transporters at the BBB, particularly P-gp and Bcrp, in limiting CNS distribution of WSD0628 was evaluated using FVB Mdr1a/b^−/−Bcrp1^−/− (P-gp and Bcrp triple knockout) mice. Mdr1a/b^−/−Bcrp1^−/− mice were administered a 5 mg/kg IV bolus dose via a tail vein injection. Mice (n = 4/time point) were euthanized at 0.167, 0.5, 1, 2, 4, 8, 16, and 24 hours after dose and plasma, brain, and spinal cord were collected as described in the previous sections. Tissue specimens were stored at –80°C until analysis by LC-MS/MS.

Steady-State Tissue Distribution

An extensive tissue distribution study at steady-state was conducted by surgical implantation of Alzet osmotic pumps in the intraperitoneal cavity of FVB wild-type mice (n = 6), primed to release 7.5 μg/h of WSD0628. Twenty-five hours after implantation (>5 half-lives), mice were euthanized and the following tissues were collected: blood, brain, spinal cord, optic nerve, sciatic nerve, tongue, esophagus, small intestine, spleen, heart, bone marrow, lungs, kidneys, and liver. Blood was collected by cardiac puncture using heparinized syringes and heparinized tubes followed by centrifugation at 7500 rpm for 10 min at 4°C to separate plasma. The organs were rinsed with saline and blotted on a Kimwipe to remove superficial blood. All tissue specimens were stored at –80°C until analysis by LC-MS/MS. Brain concentrations were corrected for residual blood as described earlier. Concentrations in liver, heart, kidney, spleen, lungs, tongue (muscle), small intestine, and esophagus (gastrointestinal tract) were corrected using information about vascular volume of blood in these tissues obtained from the literature (Baxter et al., 1994; Pawaskar et al., 2013).

Evaluation of Drug Distribution in Orthotopically Implanted GBM

Drug distribution to tumor was evaluated in mice bearing intracranially implanted eGFP/fLuc2 labeled GBM43 PDX tumors. Mice (n = 5) were administered a dose of 7.5 mg/kg orally, and plasma and brain were harvested 2 hours postdose (approximately the time for peak plasma concentration) and brains were flash frozen until further processing. The frozen brain was placed in an adult mouse brain slicer matrix (Zivic Instruments, Pittsburgh, PA) and sliced into uniform slices of 1 mm thickness. A FHS/T01 miner’s lamp (BLS Biological Laboratory Equipment, Hungary) with an appropriate light source (FS/ULS-02B2) and emission filter (FS/TEF-3GY2) was used to dissect the fluorescent GFP-labeled tumor regions from the brain slices. The separated tumor and normal brain regions were stored at –80°C until analysis by LC-MS/MS.

LC-MS/MS Analysis

All tissue specimens were homogenized with three volumes of 5% w/v bovine serum albumin using an Omni THB-01 tissue homogenizer (Omni International, Kennesaw, GA) at medium speed for 30 seconds. Drug concentration in all samples was measured using a validated LC-MS/MS assay. One hundred microliter of analyte sample was spiked with 100 ng of the internal standard (AZD1390) and extracted by protein precipitation using 5X volume of ice-cold acetonitrile with 0.1% v/v formic acid. The sample mixture was vortexed for 5 min at room temperature and centrifuged at 13000 rpm for 10 min at 4°C. One hundred microliter of the supernatant was transferred into a glass vial insert and 2 μL was injected into a Synergi Polar-RP column (75 × 2 mm, 4 μm, 80 Å; Phenomenex, Torrance, CA) on a Thermo UltiMate 3000 high-performance liquid chromatography system coupled with a Thermo Scientific TSQ Vantage triple quadrupole mass spectrometer (Thermo Fisher Scientific, San Jose, CA). A gradient elution method with a mobile phase consisting of solvent A (0.1% v/v formic acid in water) and solvent B (0.1% v/v formic acid in acetonitrile) at a flow rate of 0.5 mL/min was used. Solvent B was held at 25% for 3.5 min, linearly ramped from 25% to 95% in 0.5 min, held at 95% for 1.5 min, and brought back down to 25% for 1.5 min. The mass spectrometer was operated with an electrospray ionization source, and analysis was performed in a single reaction monitoring mode detecting transitions of m/z 440.11 > 134.12 for WSD0628 and m/z 478.30 > 126.06 for AZD1390 in a positive electrospray ionization mode. The retention time was 1.3 min and 2.3 min for WSD0628 and AZD1390, respectively. The calibration curve was linear over the range of 2.5 to 1000 ng/mL (5.69 nM–2276 nM) (weighted 1/Y²) with a coefficient of variation of less than 20%, and the lower limit of quantification was 2.5 ng/mL. All measured specimens were diluted to fall in the range of the calibration curve for each sample analysis. The freeze-thaw, bench top, long-term, and autosampler stability of WSD0628 was evaluated in plasma, brain, and spinal cord, and WSD0628 was stable over the time course and conditions relevant to the study.

Pharmacokinetic Data Analysis

Noncompartmental Analysis

Noncompartmental analysis performed using Phoenix WinNonlin, version 8.3 (Certara USA, Inc., Princeton, NJ) to analyze and estimate pharmacokinetic parameters from concentration-time profiles after a single intravenous or oral dose of WSD028. The C_max and the time to reach C_max (T_max) after an oral dose was the highest observed concentration from the concentration-time profile. The terminal elimination rate constant (λ_z) was determined by log-linear regression of the terminal phase of the concentration-time profile. Drug exposure to the last measured time point was determined by calculating the area under the curve (AUC_0-last), determined by linear trapezoidal integration until the last measured time point. AUC_0-inf was extrapolated from the last measured time point to infinite time by dividing the last measured concentration (C_last) by the terminal elimination rate constant (λ_z). The systemic clearance (CL) and apparent oral clearance (CL/F) were estimated as the ratio of dose to AUC_0-inf. The steady-state volume of distribution after IV dosing (V_ss) was estimated as mean residence time MRT_inf x clearance and the apparent oral volume of distribution (V/F) was estimated as Dose/(λ_z x AUC_0-inf). MRT_inf is estimated as the AUMC_0-inf/AUC_0-inf. Half-life was determined as ln(2)/λ_z. Absolute oral bioavailability (F) was determined as the ratio of the dose-normalized AUC_0-inf after oral administration to the dose-normalized AUC_0-inf after intravenous administration of the same dose, except for 0.1 mg/kg oral dose where the bioavailability was determined relative to the exposure after a 1 mg/kg intravenous dose as shown in eq. 2.

Dose proportionality was assessed for AUC and C_max using the power model (Wixley, 1997; Lee et al., 2015). The proportionality between each parameter and dose was determined using eq. 3, where parameter is AUC or C_max, α is a coefficient and β is the proportionality constant.

Log transformation of this equation gives a linear relationship represented as eq. 4, where β represents the slope of the linear relationship between log(parameter) and log(Dose).

It was concluded that the parameters are dose proportional if the 90% confidence interval (CI) of β was contained in the range of (1+ log(0.8)/log(r), 1 + log(1.25)/log(r)), where r is the ratio of the highest dose to the lowest dose in the study.

To characterize the extent of drug distribution in the CNS, the total tissue partition coefficient (Kp) was calculated as the ratio of total exposure in the tissue to exposure in plasma.

The unbound tissue partition coefficient (Kp_uu,tissue) was calculated using the ratio of free fraction in the tissue to plasma multiplied with the tissue partition coefficient as shown in eq. 6 and was used to evaluate the extent of CNS permeability of the pharmacologically active (unbound) drug.

Drug exposure between wild-type and Mdr1a/b^−/−Bcrp1^−/− (P-gp and Bcrp triple knockout) mice was compared using a distribution advantage (DA) given as the ratio of tissue partition coefficient between the knockout to wild-type.

Compartmental Modeling and Simulation

An open two-compartment model with first-order absorption and saturable Michaelis–Menten elimination from the central compartment (Fig. 8A) was simultaneously fit to the observed plasma concentration-time profiles following a single oral dose of 0.1 mg/kg, 1 mg/kg, 5 mg/kg, and 10 mg/kg using SAAM II (version 2.3.3; Nanomath LLC, Spokane, WA). Parameters estimated with the model include the first-order rate constant of absorption (k_abs), parameters from the Michaelis–Menten equation, V_max, Km, intercompartmental rate constants (k₁₂ and k₂₁), and volume of distribution of the central compartment (V_c). Unbound concentrations in plasma were determined by multiplying the model-predicted total concentrations with the unbound fraction in plasma determined using RED. The concentration versus time profiles in brain and tumor were predicted by accounting for the observed unbound drug partitioning to brain (Kp_uu,brain) and tumor (Kp_uu,tumor) and assuming the ratio of concentrations in tissue to plasma stays constant over time (Fig. 8B). The model was used to predict unbound plasma, brain, and CNS tumor concentration-time profiles following multiple dosing regimens, and to gain insights into potentially effective dosing regimens for subsequent testing in in vivo efficacy studies. The equations for the rate of change of amount (x_i) of drug in i^th compartment in the model (eqs. 8–10) and subsequent prediction of unbound plasma, brain, and tumor concentrations (eqs. 11–13) are summarized next.

Statistical Analysis

All experimental data are presented as mean ± S.D. except for AUCs, which are presented as mean ± S.E.M., where variances for AUC_0-last were calculated using the Bailer method (Bailer, 1988) and variances for AUC_0-inf were calculated using the Yuan method (Yuan, 1993). Dose proportionality was evaluated using power regression model as described earlier. One-way ANOVA followed by Tukey’s multiple comparisons test with significance level set as 0.05 was used to test statistical differences in unbound fractions in different matrices and plasma and brain data collected for brain regions spatial drug distribution. All statistical comparisons were performed using GraphPad Prism 9.1.1 (GraphPad Software, La Jolla, CA).

Results

In Vitro Assessment of Free (Unbound) Fractions of WSD0628

The free (unbound) fractions of WSD0628 in mouse plasma, brain homogenate, spinal cord homogenate, and human GBM tumor homogenate determined using RED following a 6-hour incubation are summarized in Table 1. WSD0628 is highly bound to plasma, brain, and spinal cord. We observe similar unbound fractions of WSD0628 in plasma (0.020 ± 0.002), brain homogenate (0.019 ± 0.002), and spinal cord homogenate (0.018 ± 0.001) (p-value > 0.05). A higher unbound fraction was observed in the human GBM tumor tissue (0.054 ± 0.002, p-value < 0.0001). WSD0628 was stable in plasma, brain homogenate, spinal cord homogenate, and tumor homogenate for 6 hours at 37°C (data not shown), and recovery from all matrices was between 89% and 110%. Assessment of free fractions facilitates measurement of the extent of brain penetration of the pharmacologically active drug, that is, Kp_,uu,brain.

TABLE 1.

Unbound fraction of WSD0628 in plasma, brain homogenate, spinal cord homogenate, and GBM tumor homogenate measured using rapid equilibrium dialysis. Data represent mean ± S.D. (n = 5).

Matrix	Unbound fraction (fu)	% Recovery
Plasma	0.020 ± 0.002	91.75 ± 0.06
Brain homogenate	0.019 ± 0.002	109.28 ± 7.72
Spinal cord homogenate	0.018 ± 0.001	104.41 ± 3.24
GBM tumor	0.054 ± 0.002	89.97 ± 3.96

Systemic Pharmacokinetics and Dose Proportionality

The systemic pharmacokinetics and dose proportionality of WSD0628 were evaluated after single intravenous or oral doses ranging from 1 mg/kg to 10 mg/kg and 0.1 mg/kg to 10 mg/kg, respectively. Examining dose proportionality is necessary to predict the effect of dose adjustments. The concentration-time profiles after a single intravenous dose or oral dose are shown in Fig. 2, A and B, and the corresponding pharmacokinetic parameters determined from these concentration-time profiles using noncompartmental analysis are presented in Table 2. After a single intravenous dose, we observe a rapid distribution phase followed by rapid decline in the concentrations at the lower doses (1 mg/kg and 5 mg/kg) as compared with a less rapid decline after the highest administered dose (10 mg/kg). This increase in the terminal phase half-life is characteristic of nonlinear, saturable elimination. When administered orally, we observed that maximal plasma concentrations were attained at 4 hours postdose after a 0.1 mg/kg and 1 mg/kg dose and there was more rapid absorption at higher doses of 5 mg/kg and 10 mg/kg (T_max = 1–2 hours). Dose normalized plasma concentration-time profiles after an intravenous or oral dose are shown in Fig. 2, C and D. The dose-normalized plasma concentration-time profiles were non-superimposable, indicating that WSD0628 exhibits nonlinear pharmacokinetics, implying that the pharmacokinetic parameters are concentration dependent. Clearance decreases with an increase in dose (0.27 L/h/kg after 1 mg/kg IV to 0.03 L/h/kg after 10 mg/kg IV), resulting in a greater than proportional increase in dose normalized AUC. Decrease in the observed clearance also results in an increase in the observed half-life of WSD0628, and the half-life increases from 3.42 h to 18.4 h within the dose range studied. Similar observations were observed after oral administration of WSD0628. Absorption appeared to plateau at doses higher than 5 mg/kg, since the maximum plasma concentration observed after a 5 mg/kg, 7.5 mg/kg (from spatial distribution in different brain regions study) and 10 mg/kg oral dose of the suspension were similar, and we can speculate that this nonlinearity could be attributed to dissolution rate limited absorption (Supplemental Fig. 1). A 10 mg/kg dose was administered orally as a solution to test this hypothesis, and we observed similar concentration-time profiles regardless of the dosing vehicle. Hence, absorption may not be limited by the dissolution rate but could be solubility limited as highlighted by the Developability Classification System (Kataoka et al., 2013; Takano et al., 2008; Butler and Dressman, 2010). Oral bioavailability was determined to be 74% at 0.1 mg/kg (relative to 1 mg/kg IV dose), 100% at 1 mg/kg, 64% at 5 mg/kg, and approximately 30% at 10 mg/kg. The decrease in bioavailability with an increase in dose may be because of limited absorption of the drug at the higher doses. No trends were observed in the concentrations attained between both the sexes.

TABLE 2.

Systemic pharmacokinetic parameters from NCA after a single IV or PO dose

Route	Units	IV	IV	IV	PO	PO	PO	PO	PO
Dose	mg/kg	1	5	10	0.1	1	5	10 (suspension)	10 (solution)
Half-life (t1/2)	h	3.42	4.60	18.4	2.55	3.90	3.90	13.55	16.82
CL or CL/F	L/h/kg	0.27	0.12	0.03	0.36	0.27	0.18	0.12	0.09
Vss or V/F	L/kg	1.21	0.72	0.86	1.34	1.54	1.04	2.44	2.26
Tmax	h	—	—	—	4	4	2	1	2
Cmaxa	μM	—	—	—	0.087 (0.01)	0.73 (0.09)	9.06 (1.59)	8.65 (0.75)	10.72 (1.46)
Cmax/Dose	μM*kg/mg	—	—	—	0.87	0.73	1.81	0.87	1.07
AUC(0-∞)b	μM*h	8.38 (0.82)	96.68 (14.68)	684.87 (207.68)	0.62 (0.18)	8.44 (1.71)	62.23 (7.67)	180.99 (26.08)	248.24 (59.93)
AUC(0-∞)/Dose	μM*h*kg/mg	8.38	19.34	68.49	6.2	8.44	12.45	18.10	24.82
F (AUCp.o./AUCi.v.)		1	1	1	0.74 (relative to 1 mg/kg IV)	1.01	0.64	0.26	0.36
% extrapolatedc		0.70	2.70	16.00	2.11	2.50	1.33	20.89	24.92

CL, clearance; CL/F, apparent oral clearance; IV, intravenous; NCA, noncompartmental analysis; PO, oral; T_max, time to reach C_max; V/F, apparent oral volume of distribution; Vss, volume of distribution at steady state.

^aC_max represented as mean (S.D.), n = 4.

^bAUC represented as mean (S.E.M.).

^c% extrapolated = (AUC_(0-∞) – AUC_0-t,last)/AUC_(0-∞)*100.

Fig. 2.

Systemic pharmacokinetics and dose proportionality of WSD0628 in FVB wild-type mice. Plasma concentration versus time profile of WSD0628 after (A) an intravenous bolus injection of 1 mg/kg, 5 mg/kg, and 10 mg/kg; (B) oral administration of 0.1 mg/kg, 1 mg/kg, 5 mg/kg, 10 mg/kg dose. Oral administration formulation was a suspension unless otherwise mentioned. Dose-normalized plasma concentration versus time profiles after (C) an intravenous bolus injection of 1 mg/kg, 5 mg/kg, and 10 mg/kg; (D) oral administration of 0.1 mg/kg, 1 mg/kg, 5 mg/kg, 10 mg/kg dose. Data represent mean ± S.D., n = 4.

Dose proportionality was also evaluated using the power model regression analysis (FDA Guidance for Industry, 2006). Based on the results from the linear regression of the log-transformed data, the 90% CI of the slope (β) for AUC_0-inf was outside the 80% to 125% acceptable linear range [0.903, 1.097] for intravenous and [0.952, 1.048] for oral dosing (Supplemental Fig. 1). The 90% CI for the slope of linear regression of the log-transformed AUC_0-inf on log (Dose) was (–0.058, 3.173) in the case of intravenous dosing and (1.098, 1.345) for oral dosing, and the 90% CI of the slope for the log-transformed C_max was (0.799, 1.298) for oral dosing. Thus, we concluded that there was a lack of dose proportionality over the dose range tested.

Distribution of WSD0628 to CNS Tissues

Distribution of WSD0628 to CNS tissues, namely, brain (Fig. 3) and spinal cord (Fig. 4), were studied after a single intravenous or oral dose ranging from 1 mg/kg to 10 mg/kg. A complete concentration-time profile in the CNS after a 0.1 mg/kg dose could not be established because concentrations at time points before 1 h and after 8 h of dose administration were below the lower limit of quantification (LLOQ). We observed similar half-lives in the CNS as in the plasma and a greater than dose proportional increase in CNS exposure, corresponding to the nonlinear increase in plasma exposure (Supplemental Fig. 2). Hence, this confirms that plasma concentration is the driving force concentration for concentrations in brain (Fig. 3, A and B) and in spinal cord (Fig. 4, A and B). The pharmacokinetic parameters in the CNS are summarized in Tables 3 and 4. WSD0628 attains rapid equilibrium in the CNS as depicted from the brain to plasma and spinal cord to plasma concentration ratios over time (Fig. 3, C and D and Fig. 4, C and D, respectively). The total partition coefficient (Kp) in the CNS was determined as the ratio of exposure (AUC) in the tissue to exposure (AUC) in plasma and was determined to be between 0.16 and 0.31 in the CNS tissues after intravenous or oral administration of the various doses. The total partition coefficient was corrected for binding to plasma proteins and CNS tissues using free fractions from RED to determine the unbound partition coefficient (Kp_uu). This was similar to the total partition coefficient due to similar extent of binding to plasma and CNS tissues. The total and unbound partition coefficients in the spinal cord trended slightly higher than the brain total and unbound partition coefficients across all doses.

TABLE 3.

Pharmacokinetic parameters in brain after a single IV or PO dose

Route	Units	IV	IV	IV	PO	PO	PO	PO
Dose	mg/kg	1	5	10	1	5	10	10 (solution)
Half-life (t1/2)	h	2.62	5.00	22.62	3.80	3.81	13.49	17.74
AUC(0-∞)a	μM*h	1.34 (0.23)	21.33 (3.71)	179.60 (68.87)	2.60 (0.66)	17.55 (2.60)	35.07 (4.39)	52.37 (14.27)
Kp$\left(\frac{\mathit{\text{AU}}{\mathit{\text{C}}}_{\mathit{\text{tissue}}}}{\mathit{\text{AU}}{\mathit{\text{C}}}_{\mathit{\text{plasma}}}}\right)$		0.16	0.22	0.26	0.31	0.28	0.19	0.21
Kpuu $\left(\frac{\mathit{f}{\mathit{u}}_{\mathit{\text{tissue}}}{\mathit{\text{*AUC}}}_{\mathit{\text{tissue}}}}{\mathit{f}{\mathit{u}}_{\mathit{\text{plasma}}}\mathit{\text{*AU}}{\mathit{\text{C}}}_{\mathit{\text{plasma}}}}\right)$		0.15	0.20	0.24	0.27	0.26	0.17	0.19
% extrapolatedb		12.09	3.13	20.87	2.22	0.97	19.19	30.45

IV, intravenous; PO, oral.

^aAUC represented as mean (S.E.M.).

^b% extrapolated = (AUC_(0-∞) – AUC_0-t,last)/AUC_(0-∞)*100.

TABLE 4.

Pharmacokinetic parameters in spinal cord after a single IV or PO dose

Route	Units	IV	IV	IV	PO	PO	PO	PO
Dose	mg/kg	1	5	10	1	5	10	10 (solution)
Half-life (t1/2)	h	1.69	5.07	18.39	3.78	3.80	13.84	15.89
AUC(0-∞)a	μM*h	1.39 (0.11)	29.66 (7.31)	232.52 (51.14)	2.60 (0.82)	20.87 (8.07)	40.98 (5.99)	75.90 (9.97)
Kp$\left(\frac{\mathit{\text{AU}}{\mathit{C}}_{\mathit{\text{tissue}}}}{\mathit{\text{AU}}{\mathit{\text{C}}}_{\mathit{\text{plasma}}}}\right)$		0.17	0.31	0.34	0.31	0.34	0.26	0.31
Kpuu $\left(\frac{\mathit{f}{\mathit{u}}_{\mathit{\text{tissue}}}{\mathit{\text{*AUC}}}_{\mathit{\text{tissue}}}}{\mathit{f}{\mathit{u}}_{\mathit{\text{plasma}}}\mathit{\text{*AU}}{\mathit{\text{C}}}_{\mathit{\text{plasma}}}}\right)$		0.15	0.27	0.30	0.27	0.29	0.23	0.27
% extrapolatedb		3.13	3.39	14.56	2.00	0.99	19.13	26.25

IV, intravenous; PO, oral.

^aAUC represented as mean (S.E.M.).

^b% extrapolated = (AUC_(0-∞) – AUC_0-t,last)/AUC_(0-∞)*100.

Fig. 3.

Brain distribution of WSD0628 in FVB wild-type mice. Brain concentration versus time profile of WSD0628 after (A) an intravenous bolus injection of 1 mg/kg, 5 mg/kg, and 10 mg/kg; (B) oral administration of 1 mg/kg, 5 mg/kg, 10 mg/kg dose. Only concentrations measured above lower limit of quantification (LLOQ) shown after oral administration of 0.1 mg/kg. Brain to plasma concentration ratios versus time after (C) an intravenous bolus injection of 1 mg/kg, 5 mg/kg, and 10 mg/kg; (D) oral administration of 1 mg/kg, 5 mg/kg, 10 mg/kg dose. Data represent mean ± S.D., n = 4.

Fig. 4.

Spinal cord distribution of WSD0628 in FVB wild-type mice. Spinal cord concentration versus time profile of WSD0628 after (A) an intravenous bolus injection of 1 mg/kg, 5 mg/kg, and 10 mg/kg; (B) oral administration of 1 mg/kg, 5 mg/kg, 10 mg/kg dose. Only concentrations measured above lower limit of quantification (LLOQ) shown after oral administration of 0.1 mg/kg. Spinal cord to plasma concentration ratios versus time after (C) an intravenous bolus injection of 1 mg/kg, 5 mg/kg, and 10 mg/kg; (D) oral administration of 1 mg/kg, 5 mg/kg, 10 mg/kg dose. Data represent mean ± S.D., n = 4.

Spatial Distribution in Different Brain Regions

Differences in drug distribution across the BBB in various regions of the brain could potentially impact the therapeutic effect of the drug, given that the location of the tumor within the brain can vary widely. Spatial drug distribution in different regions of the brain, namely the cortex, thalamus and hypothalamus, midbrain, pons, cerebellum, and medulla, was studied at 2 h and 8 h after a single 7.5 mg/kg oral dose, and the observed concentrations and the resultant tissue to plasma partition coefficients are shown in Fig. 5, A and B, respectively. Medulla samples were pooled due to the limited sample weight from each mouse. WSD0628 was homogenously distributed across different brain regions (p-value > 0.05) at both time points in this study. The mean of the tissue to plasma partition coefficients in the different brain regions was 0.31 ± 0.03 at both 2- and 8-h postdose. Similar to the trend from the CNS distribution in the dose-escalation study, we observed a slightly higher partition coefficient in the spinal cord (0.40 ± 0.05) as compared with different brain regions (Supplemental Table 1). Thus, our results indicate that there was no difference in the distribution of WSD0628 among different anatomical regions of the brain.

Fig. 5.

Spatial distribution of WSD0628 in different anatomical regions of the CNS. (A) Concentrations of WSD0628 in plasma and different regions of the CNS; (B) tissue to plasma partition coefficient in different regions of the CNS at 2 h and 8 h after oral administration of 7.5 mg/kg. Data represent mean ± S.D., n = 4 (p-value > 0.05).

Role of Efflux Transporters at the BBB (P-gp and Bcrp) in CNS Distribution

The combined impact of P-gp and Bcrp at the BBB in limiting the brain distribution of WSD0628 was studied in a transgenic Mdr1a/b^−/−Bcrp1^−/− (P-gp and Bcrp knockout) mouse model. The observed concentration-time profiles in plasma, brain, and spinal cord after a 5 mg/kg intravenous dose as shown in Fig. 6A. We observed a rapid distribution followed by first-order elimination. The exposure (AUC) and total partition coefficient (Kp) in the brain in the knockout mice (0.33) was similar to that in the wild-type mice (0.22) (unpaired t test, p-value > 0.2) (Table 5). Similar to wild-type mice, WSD0628 attains rapid equilibrium in the CNS of the knockout mice as observed from the brain to plasma and spinal cord to plasma concentration ratios over time in wild-type mice and transporter knockout mice (Fig. 6B–F). The distribution advantage (DA) in transporter knockout mice was determined to be 1.50 in brain and 1.26 in spinal cord, which suggests that P-gp and Bcrp play a limited role in influencing the CNS distribution of WSD0628.

TABLE 5.

Influence of P-gp and Bcrp in CNS distribution of WSD0628

Parameter (unit)	Wild-type	Mdr1a/b−/−Bcrp1−/−
AUC0-∞, plasma (μM*h)a	96.68 (14.68)	121.01 (14.16)
AUC0-∞, brain (μM*h)a	21.33 (3.71)	39.91 (6.53)
AUC0-∞, spinal cord (μM*h)a	29.66 (7.31)	47.56 (7.94)
Kp, brain	0.22	0.33
Kp, spinal cord	0.31	0.39
DAbrain	1	1.50
DAspinal cord	1	1.26

^aAUC represented as mean (S.E.M.).

Fig. 6.

Influence of efflux transporters P-gp and Bcrp in CNS distribution of WSD0628. (A) Concentration versus time profile in plasma, brain, and spinal cord after intravenous administration of a single 5 mg/kg in Mdr1a/b^−/−Bcrp1^−/− mice. (B) Brain to plasma and (C) spinal cord to plasma concentration ratios versus time after intravenous administration of a single 5 mg/kg dose in wild-type and Mdr1a/b^−/−Bcrp1^−/− mice. Concentration versus time profile in (D) plasma, (E) brain, (F) spinal cord in wild-type and Mdr1a/b^−/−Bcrp1^−/− mice. Data represent mean ± S.D., n = 4.

Steady-State Tissue Distribution

Tissue partition coefficients after a constant infusion of WSD0628 to steady state were determined in brain, spinal cord, optic nerve, sciatic nerve, tongue, esophagus, small intestine, spleen, heart, bone marrow, lungs, kidneys, and liver. Optic nerve, sciatic nerve, and esophagus specimens were pooled (n = 6) due to the limited sample weight obtained from each mouse. The observed concentrations in the tissue and the resulting tissue to plasma partition coefficients are shown in Fig. 7, A and B, respectively. Tissue concentrations were corrected for concentration of the drug in the vascular space for tissues where information about vascular volume in the specific tissue was available. Tissue drug partition coefficients are listed in Supplemental Table 2.

Fig. 7.

Steady-state tissue distribution of WSD0628. (A) Concentration of WSD0628 at steady-state in plasma, CNS, and other peripheral tissues after constant infusion of 7.5 μg/h for 25 h. (B) Tissue to plasma partition coefficient of WSD0628 at steady-state. Data represent mean ± S.D., n = 6 (except pooled samples from optic nerve, sciatic nerve, and esophagus, n = 6).

Evaluation of Drug Distribution in Orthotopically Implanted GBM Tumors

Drug concentrations in intracranially implanted GBM tumors and surrounding regions of normal brain were evaluated 2 h after a single oral dose of 7.5 mg/kg. The tissue distribution coefficient in normal brain and tumor were determined and are summarized in Supplemental Table 3. Drug concentrations in the tumor were higher than the concentrations in the regions of normal brain (p-value = 0.017). We observed similar concentrations in plasma and “normal brain” between the FVB mice used in the CNS drug distribution studies and the athymic nude mice in this study and do not anticipate any strain differences in the pharmacokinetics of WSD0628 in mice (Barr et al., 2020).

Compartmental Modeling and Simulation

An open two-compartment model was simultaneously fit to all observed plasma concentration-time profiles from the dose proportionality study. The model accounted for the observed decrease in clearance at higher doses using the Michaelis–Menten equation to describe elimination from the central compartment. The model describing the total plasma concentration-time profiles was then used to predict unbound drug concentrations in plasma, brain, and brain tumor. Weighted residuals between the observed and predicted data were randomly distributed between –2 to 2 and are depicted in Supplemental Fig. 3. The predicted unbound plasma concentration-time profiles predicted by the model, using eq. 11, fit the observed unbound plasma concentration data (Fig. 8C). The model estimated parameters are reported in Supplemental Table 4. This model was used to predict concentration-time profiles and exposures in plasma, brain, and brain tumor from oral doses between 0.1 mg/kg to 10 mg/kg (Fig. 9). Model predictions indicate that effective exposures in the CNS tumors, that is an unbound concentration over the set threshold of minimum effective target concentration of 20 nM for 12 to 16 h (Tuma et al., 2022), are attained at doses over 3.5 mg/kg administered once daily and are thus predicted to be effective in vivo when combined with radiation therapy. Simulations also indicate that drug accumulation is likely at doses higher than 5 mg/kg, administered at a frequency of once daily or higher. These simulations will aid in designing effective dosage regimens for future preclinical in vivo efficacy studies.

Fig. 8.

Compartmental pharmacokinetic modeling. (A) Open two-compartment model with first-order absorption and saturable Michaelis–Menten elimination from the central compartment was used to describe the systemic pharmacokinetics of WSD0628, which was subsequently extended to include a brain and tumor compartment; (B) schematic representation of extent of unbound drug (pharmacologically active drug) transport across membranes of tissues of interest, that is, brain and brain tumor; (C) model fits of observed plasma concentrations versus time after an oral dose of 0.1 mg.kg, 1 mg/kg, 5 mg/kg, and 10 mg/kg, external verification through concentrations at 2 and 8 h after an oral dose of 7.5 mg/kg.

Fig. 9.

Model predictions of concentrations in tissues of interest after multiple dosing. Predicted unbound concentrations versus time in (A) plasma, (B) brain, and (C) tumor after multiple daily doses. The dotted line represents the set unbound minimum effective target concentration, 20 nM, for efficacious radiosensitization.

Discussion

ATM is an attractive target for combination therapy with radiation, and WSD0628, a novel and potent ATM inhibitor, is being evaluated for use as a radiosensitizer in the treatment of brain tumors. A critical aspect of efficacious therapy is the ability to attain adequate exposure at the target site. This is particularly challenging in the case of tumors in the CNS due to limited passive delivery across the BBB and the expression of efflux transporters like P-gp and Bcrp at the BBB (Sarkaria et al., 2018; Arvanitis et al., 2019; Rathi et al., 2022). Additionally, it is important to characterize the extent of CNS distribution of the pharmacologically active (unbound) drug concentration by determining the extent of binding to the tissue and plasma. While maintaining adequate exposure in the CNS, limiting exposure to other organs may reduce the risk of toxicities. In the present study, we evaluate the systemic pharmacokinetics, dose proportionality, CNS distribution, drug binding, and steady-state tissue distribution of WSD0628 in mice and draw insights to help eventually predict human disposition and resultant efficacy.

The systemic pharmacokinetics and dose proportionality of WSD0628 was evaluated over a 100-fold range of 0.1 mg/kg to 10 mg/kg administered orally and a 10-fold range of 1 mg/kg to 10 mg/kg administered intravenously. We observed a decrease in clearance at higher doses (10-fold decrease in clearance) and a gradual, prolonged decrease in plasma concentrations resulting in an increased half-life of ∼18 h in mice after an intravenous dose of 10 mg/kg. Liver blood flow is reported to be 5.4 L/h/kg in mice (Davies and Morris, 1993, Mehvar, 2016), and the observed clearance in blood (using blood to plasma ratio, Supplemental Table 5) for WSD0628 is significantly lower than the reported liver blood flow. Assuming that the total clearance of WSD0628 is primarily driven by hepatic clearance, the estimated hepatic extraction ratio (E_H) for the studied doses ranges was between 0.007 and 0.06; thus, WSD0628 is likely to have a low extraction ratio in the liver. A low extraction ratio indicates that the hepatic clearance of the drug is independent of liver blood flow and is capacity limited by the intrinsic activity of the metabolizing enzymes in the liver. The observed greater-than-dose-proportional increase in plasma exposure is likely due to saturation of the metabolic clearance processes. This is a commonly observed phenomenon, and many clinically approved drugs exhibit this behavior (Levy, 1965; Ludden, 1991; Fraschini et al., 1993; Sonnichsen and Relling, 1994; Rost and Roots, 1996; Takeuchi et al., 2001; Theuretzbacher et al., 2006). Another potential cause of decreased clearance could be due to autoinhibition of the enzyme responsible for metabolizing WSD0628. Autoinhibition of metabolizing enzymes has been reported for many drugs (Filppula et al., 2013; Concheiro et al., 2014; Luo et al., 2019; Kaartinen et al., 2020; Wu et al., 2020).

We also observed a plateau in the C_max attained after a single oral dose at the higher doses and a decrease in oral bioavailability (F = Fa x Fg x Fh) with an increase in dose. WSD0628 was administered as an oral suspension, and we hypothesized that absorption was dissolution-rate limited and administered the same dose as an oral solution to test the hypothesis. However, we did not see a significant increase in systemic exposure or bioavailability when WSD0628 was administered as a solution. Only the drug in solution in the gastrointestinal tract will be absorbed across the membrane into the systemic circulation. Thus, the reduced bioavailability could be attributed to the low solubility of the drug in the gastrointestinal tract fluids (data on file, IB), leading to a decrease in the fraction of dose absorbed (Fa) with an increase in dose (Sjövall et al., 1985; York et al., 2000; Takano et al., 2008). The observed nonlinear absorption and clearance of WSD0628 will have implications for the determination of efficacious dosing regimens and dose adjustments and for establishing a therapeutic window, and it remains to be seen if similar nonlinearity is also observed in humans. This is particularly crucial for drugs with a narrow therapeutic window because an increase in dose can result in a supratherapeutic exposure leading to adverse events.

The CNS distribution of WSD0628 was studied in mice after a single intravenous or oral dose. We observed similar greater than dose proportional increase in CNS exposure as observed in plasma, the driving force for CNS distribution. WSD0628 is CNS penetrant based on the observed total (bound + unbound) CNS partitioning coefficient. According to the free drug hypothesis, only the unbound drug can distribute across cell membranes into tissues and only the unbound drug can interact with the target (Recant and Riggs, 1952; Hammarlund-Udenaes et al., 2008; Heffron, 2018). Thus, it is necessary to correct for binding to plasma and CNS tissues to determine the extent of pharmacologically active drug available in the CNS, described by Kp_uu. The Kp_uu for WSD0628 in CNS tissues was determined to be between 0.15 to 0.30, indicating that WSD0628 is CNS penetrant. The Kp_uu value must be considered along with the desired target concentration at the site of action to evaluate if adequate exposure has been attained for a robust biological response. Efflux transporters like P-gp and Bcrp are expressed at the BBB and limit the entry of most drugs into the brain (Löscher and Potschka, 2005). Thus, molecules that are avid substrates of the efflux transporters may not be able to pass the BBB to attain effective drug levels at targets in the brain and their pharmacological effect may be limited to the peripheral organs (Agarwal et al., 2011a). P-gp and Bcrp substrate status of WSD0628 was evaluated using P-gp and Bcrp transporter knockout mice. We observed a less than 1.5-fold improvement (DA) in CNS penetration of WSD0628 in the absence of P-gp and Bcrp, which indicates that WSD0628 CNS distribution is not influenced by P-gp and Bcrp. Drugs that are avid substrates of P-gp, and Bcrp can exhibit DAs greater than 15 and as high as 100 (Parrish et al., 2015; Laramy et al., 2018). Some CNS tumors show preferred incidence in select brain regions (Larjavaara et al., 2007; Wu et al., 2012; Hill et al., 2015), and some tumors infiltrate throughout the brain. In either case, information about the spatial distribution of the drug in the brain provides critical insights for developing targeted CNS drug delivery systems. The data indicate homogenous distribution of WSD0628 throughout all the studied anatomical brain regions. We observed slightly higher drug distribution to intracranially implanted tumors as compared with the region of healthy brain tissue surrounding the tumor. This can potentially be attributed to a heterogeneously disrupted BBB, resulting in higher drug accumulation in regions of the tumor (Sarkaria et al., 2018).

An important aspect for efficacious drug therapy is maintaining adequate exposure of pharmacologically active drug concentrations for a time frame required for efficient biological activity, in this case, inhibition of DSB repair. Based on clonogenic survival studies, a minimum effective unbound target concentration of 20 nM is required to be maintained for 12 to 16 h following irradiation to provide robust radiosensitization (Penninckx et al., 2021). We used modeling and simulation approaches to determine the relevant parameters to characterize the saturable elimination of WSD0628 and predict exposures at various dosing regimens. Model predictions indicate that adequate exposure of WSD0628 is maintained in the CNS tumor compartment at a dose of 3.5 mg/kg or higher; hence we anticipate that WSD0628 will be effective in radiosensitizing tumors in the CNS. However, we also predicted drug accumulation at higher doses due to the decrease in clearance and increase in half-life of WSD0628 at higher doses. Thus, increasing the dose or dosing frequency to increase or maintain the exposure in the tumor or CNS tissues needs to be balanced with exposure in plasma and other organs with the potential for off-target toxicity to maintain drug levels within the therapeutic window. This model informs the design of efficacious dosing regimens that can be subsequently tested in preclinical in vivo efficacy studies.

Previously published reports about the efficacy of ATM inhibitors like AZ32, AZD0156, and AZD1390 have demonstrated the potential of the use of ATM inhibitors as radiosensitizers (Ronco et al., 2017; Durant et al., 2018; Pike et al., 2018). However, understanding differential distribution within organs at risk for enhanced radiation toxicity will help guide dosing and radiation planning techniques. To this end, we evaluated tissue distribution to various organs including optic nerve, sciatic nerve, bone marrow, small intestine, and esophagus (Olcina and Giaccia, 2016; Menolfi and Zha, 2020). Radiation dose and drug exposure within the optic nerve is specifically relevant for brain tumor planning, and exposure within the other organs, coupled with studies evaluating potentiation of radiation injury in animal models, could provide critical insights into the use of WSD0628 as a radiosensitizer for other peripheral tumors.

In conclusion, our results indicate that WSD0628 has favorable pharmacokinetic properties for the treatment of brain tumors, in conjunction with radiation or cytotoxic chemotherapy. WSD0628 is orally bioavailable, and we observed a greater than dose proportional increase in systemic and CNS exposure at doses higher than 5 mg/kg in mice. The CNS distribution of WSD0628 is not limited by P-gp and Bcrp, distributes homogenously throughout the different anatomical regions in the brain, and maintains effective exposures for sustained inhibition of ATM required for DSB repair. Collectively, these results will enable the development of a pharmacokinetic-pharmacodynamic-efficacy relationship of WSD0628 in CNS tumors (Fig. 10). Our preclinical investigations have been instrumental in identifying potential challenges in the distribution and clearance of WSD0628 and pivotal in optimizing dosing regimens and anticipating potential challenges in the distribution and clearance of WSD0628 in humans, thereby aiding decisions for clinical trials, such as the trial recently opened at the Mayo Clinic (NCT05917145).

Fig. 10.

Strategic approach to drug development of therapies targeted to the CNS. A systematic approach to understanding systemic and CNS distributional pharmacokinetics-pharmacodynamics-efficacy relationship to guide rational decision-making during drug development. Figure created using BioRender.com

Abbreviations

ATM

ataxia telangiectasia mutated

AUC

area under the curve

BBB

blood-brain barrier

Bcrp

breast cancer resistant protein

CI

confidence interval

CNS

central nervous system

DA

distribution advantage

DSB

double-strand break

F

oral bioavailability

f_u

unbound fraction

FVB

Friend Leukemia Virus strain B

GBM

glioblastoma

IV

intravenous

LC-MS/MS

liquid chromatography-tandem mass spetrometry

Kp

tissue partition coefficient

Kp_uu

unbound tissue partition coefficient

MRT

mean residence time

PDX

patient-derived xenograft

P-gp

P-glycoprotein

RED

rapid equilibrium dialysis

Authorship Contributions

Participated in research design: Rathi, Oh, Zhong, Sarkaria, Elmquist.

Conducted experiments: Rathi, Oh, W.J. Zhang, Mladek, Burgenske, Xue, W.Q. Zhang, Le.

Performed data analysis: Rathi, Oh, Mladek, Garcia, Elmquist.

Wrote or contributed to the writing of the manuscript: Rathi, Zhong, Sarkaria, Elmquist.