Patient-derived models recapitulate heterogeneity of molecular signatures and drug response in pediatric high-grade glioma

Pediatric high-grade glioma (pHGG) is a major contributor to cancer-related death in children. In vitro and in vivo disease models reflecting the intimate connection between developmental context and pathogenesis of pHGG are essential to advance understanding and identify therapeutic vulnerabilities. Here we report establishment of 21 patient-derived pHGG orthotopic xenograft (PDOX) models and eight matched cell lines from diverse groups of pHGG. These models recapitulate histopathology, DNA methylation signatures, mutations and gene expression patterns of the patient tumors from which they were derived, and include rare subgroups not well-represented by existing models. We deploy 16 new and existing cell lines for high-throughput screening (HTS). In vitro HTS results predict variable in vivo response to PI3K/mTOR and MEK pathway inhibitors. These unique new models and an online interactive data portal for exploration of associated detailed molecular characterization and HTS chemical sensitivity data provide a rich resource for pediatric brain tumor research.

Oncoprint for pediatric high-grade glioma signature mutations in these samples. Sample Name: SJ-HGGX42

Copy Number Variation
DNA copy number variation analysis was performed from methylation array data using Conumee [2]. The Y axis shows the log2 copy number ratio of the tumor sample compared to a panel of normal reference brain tissues. Copy number ratios are plotted across chromosomes with the dotted vertical lines representing centromeres. Chromosomal gains or losses are detected as significant positive or negative deviations from genomic baseline. Brain tumor relevant gene regions are highlighted for easier assessment of chromosomal or focal alterations.

Supplementary Fig 10: Pharmacokinetic (PK) and pharmacodynamic analyses of Paxalisib and Mirdametinib
show lack of appreciable plasma or brain PK drug interaction, high brain exposure, and effective pathway inhibition a. Mirdametinib population mean and 90% prediction interval plasma concentration-time profiles alone (mir) and in combination with Paxalisib (mir+pax). A minor increase in Mirdametinib AUC (1.30-fold) was observed in combination. b. Paxalisib population mean and 90% prediction interval plasma concentration-time profiles alone (pax) and in combination with Mirdametinib (pax+mir). A minor increase in Paxalisib AUC (1.63-fold) was observed in combination.    Chemietek (Lot#03), CAS# 391210-10-9 Supplementary Note 1

Detailed information for drug-drug interaction pharmacokinetic study of mirdametinib and paxalisib in non-tumor bearing female CD-1 nude mice In Vivo Pharmacokinetic (PK) Study Design
The plasma pharmacokinetics (PK) of the MEK inhibitor mirdametinib (PD-0325901) and PI3K inhibitor paxalisib (GDC-0084) were studied to determine whether a PK drug-drug interaction (DDI) exists between these agents when co-administered orally in mice. A moderate interaction, defined as a ≥2-fold difference in plasma area under the concentration-time curve (AUC) or apparent oral clearance (CL/F), was considered as impactful and practically significant.
In the main study (Study 1), 3 groups of 9 mice each were studied with a mixed, staggered

Bioanalysis
Plasma and brain homogenate (Dilution Factor = 6, with ultrapure water) were subjected to deproteinization and analyzed for mirdametinib and paxalisib concentrations using a

25
The method qualification and bioanalytical runs all passed acceptance criteria for non-GLP assay performance. A linear model (1/X 2 weighting) fit the calibrators across the 1 to 100 ng/mL range, with a correlation coefficient (R) of ≥0.9959. Sample dilution integrity was confirmed. The lower limit of quantitation (LLOQ), defined as a peak area signal-to-noise ratio of 5 or greater verses a matrix blank with IS, was 1 ng/mL for plasma and 6 ng/mL for brain homogenate secondary to the dilution. The intra-run precision and accuracy was ≤ 7.36% CV and 88.2% to 113%, respectively.

Pharmacokinetic (PK) Analysis
Summary statistics for mirdametinib and paxalisib concentration-time (Ct) data in plasma and brain were generated by study, occasion ( were derived from the model parameter estimates using standard formulae for the relevant compartmental model 2 .

Mirdametinib Pharmacokinetics in Mice
The plasma PK of mirdametinib was well-described using a linear, two-compartment model with zero-order absorption. Absorption was rapid, with the Tmax generally occurring at the 0.25 hr time point. As there was no observable data in the absorption phase, the zeroorder absorption rate (Tk0) was fixed to 0.125 hr. The plasma Ct profile showed a distribution phase lasting approximately 1 hr, followed by an apparent terminal phase with a half-life of ~ 4 hrs. No accumulation occurred with daily dosing for 5 days in mice.
Apparent oral clearance (CL/F) was low at 12.9 mL/min/kg or ~14% of hepatic blood flow.
The apparent volume of distribution was large and greater than total body water. The bioavailability of mirdametinib was not evaluated in this study, but has previously been reported to be ~30% in rats (data not shown).
A linear two-compartment model with inter-individual and inter-occasion variability on both apparent oral clearance (CL/F) and apparent oral volume of distribution (Vc/F), and proportional residual error best described the overall plasma Ct data. The precision of the  Supplementary Fig. 10a.

Paxalisib Pharmacokinetics in Mice
The plasma PK of paxalisib was well-described using a linear, one-compartment model with Apparent oral clearance (CL/F) was low-to-moderate at 19.3 mL/min/kg or ~21.5% of hepatic blood flow. The apparent volume of distribution was large and greater than total body water. The bioavailability of paxalisib was not evaluated in this study, but has been reported to be high in various preclinical species.
A linear one-compartment model with inter-individual variability on first-order absorption rate (ka), apparent oral clearance (CL/F), apparent oral volume of distribution (Vc/F), and inter-occasion variability on apparent oral clearance (CL/F), and proportional residual error best described the overall plasma Ct data. The precision of some variability estimates was poor likely due to model overparameterization; however, the goodness of fit plots and post hoc individual visual predictive checks indicated adequate performance.
Combination status was tested as a covariate on CL/F inter-occasion variability, and its addition significantly improved the model fit (-2LL P=0.0005042182), suggesting a statistically significant effect of either mirdametinib and/or DrugX co-administration or study day on paxalisib plasma PK.
There was a 38.6% reduction in paxalisib CL/F in combination with mirdametinib on Day 5 (P=0.000218137), resulting in a statistically significant 1.63-fold increase in paxalisib AUC.
This failed to meet the ≥2-fold difference of AUC or CL/F criteria, and therefore we conclude that there is no practical effect of mirdametinib upon paxalisib PK. This is also supported visually by comparison of Ct plots. The parameter estimates from the full paxalisib model are presented in Supplementary Data 5c with the model predicted median, 90% prediction interval, and observed plasma paxalisib concentrations presented in Supplementary Fig. 10b.

Pharmacokinetic summary
The plasma PK of mirdametinib 14 mg/kg PO in mice is similar to that previously reported with respect to the Ct curve shape. However, the AUC increased less than proportionally compared with clinically relevant doses of 0.5 and 1.5 mg/kg PO 3 , suggesting saturable absorption, lower bioavailability, or higher clearance in our mice at the 14 mg/kg dose.
Mirdametinib plasma PK after multiple doses appeared similar (Day 1 vs Day 5), suggesting it has time-invariant PK in mice. While a higher mirdametinib plasma Cmax was observed 29 in combination, the AUCs were not practically different (1.30-fold) -as this failed to meet the ≥2-fold criteria, paxalisib has no practical effect upon mirdametinib PK in mice. The brain penetration of mirdametinib appeared similar alone and in combination on Day 5 (Kp,last 0.502 and 0.524, respectively), and similar to published data in mice 4 . The plasma PK of paxalisib 8 and 10 mg/kg PO in mice, after a single dose on Day 1, is similar to that reported at 25 mg/kg, assuming dose proportional PK 5 .
The brain penetration of paxalisib on Day 5 in combination (Kp,last = 1.42) was similar to that reported previously (single dose Kp,6hr = 1.39) 5 . There was no difference in paxalisib brain concentration either alone or in combination with mirdametinib or DrugX. Paxalisib's plasma AUC was 1.63-fold higher than expected in combination with mirdametinib. While this difference was statistically significant, indicating a possible weak drug interaction, it failed to meet the ≥2-fold critera for a practical interaction. Therefore mirdametinib had no practical effect on the PK properties of paxalisib.
Mirdametinib and paxalisib total plasma AUCs at these dose levels in mice exceed those observed clinically in humans. A PK-guided clinically relevant dose for mirdametinib, equivalent to 4 mg PO BID in humans 6 , would be 0.5 mg/kg PO BID in mice. Likewise for paxalisib, a PK-guided dose similar to 45 mg PO QD in humans 7 would be approximately 2 mg/kg to 4.5 mg/kg PO QD in mice.