Search

SummaryAccelerating cures for children with cancer remains an immediate challenge due to extensive oncogenic heterogeneity between and within histologies, distinct molecular mechanisms evolving between diagnosis and relapsed disease, and limited therapeutic options. To systematically prioritize and rationally test novel agents in preclinical murine models, researchers within the Pediatric Preclinical Testing Consortium are continuously developing patient-derived xenografts (PDXs) from high-risk childhood cancers, many refractory to current standard-of-care treatments. Here, we genomically characterize 261 PDX models from 29 unique pediatric cancer malignancies and demonstrate faithful recapitulation of histologies, subtypes, and refine our understanding of relapsed disease. Expression and mutational signatures are used to classify tumors forTP53andNF1inactivation, as well as impaired DNA repair. We anticipate that these data will serve as a resource for pediatric oncology drug development and guide rational clinical trial design for children with cancer.HighlightsMultiplatform genomic analysis defines landscape of 261 pediatric cancer patient derived xenograft (PDX) modelsPediatric patient derived xenografts faithfully recapitulate relapsed diseaseInferredTP53pathway inactivation correlates with pediatric cancer copy number burdenSomatic mutational signatures predict impaired DNA repair across multiple histologies

BACKGROUND: Pediatric low-grade glioma (pLGG) is the most common central nervous system tumor of childhood. Although overall survival is very good, many children suffer from multiple progressions and functional morbidities. There is no one universally accepted therapy for children with recurrent disease, however, standard cytotoxic chemotherapies are often utilized by most practitioners. The Pediatric Brain Tumor Consortium conducted a multi-institutional phase II study evaluating selumetinib (AZD6244, ARRY-142886), a MAP/ERK Kinase I/II inhibitor, in patients with recurrent, refractory or progressive pLGG assigned to numerous strata. The aim of the study was to evaluate the efficacy of selumetinib in these patients. METHODS: Eligibility required age 3–21 y/o, a Lansky or Karnofsky performance score greater than 60 and the presence of recurrent, refractory or progressive pLGG after at least one standard therapy. Stratum 1 included children with World Health Organization (WHO) grade I pilocytic astrocytoma (PA) harboring either one of the two most common BRAF aberrations (KIAA1549-BRAF fusion or the BRAF(V600E) mutation). Stratum 3 included children with any neurofibromatosis type 1 (NF1)-associated pLGG (WHO grades I and II). Selumetinib was provided as capsules given orally at the recommended phase II dose of 25 mg/m(2) twice daily. The primary endpoint was stratum-specific objective response rate assessd by the local site and sustained for at least 8 weeks. All responses were reviewed centrally and statistical analyses were done as per protocol. Although the trial (NCT01089101) is still ongoing in other strata, enrollment and planned follow-up is compete on both strata 1 and 3. FINDINGS: Between July 25, 2013, and June 12, 2015, 25 eligible and evaluable children were accrued to stratum 1, and between August 28, 2013, and June 25, 2015, 25 eligible and evaluable children were accrued to stratum 3. On stratum 1, 9/25 (36%) patients achieved a partial response (PR). The median follow-up for the 11 patients who have not yet experienced an event is 36.4 months (4.4–50.5; IQR=23.9). On stratum 3, 10/25 (40%) patients achieved a PR with a median follow-up of 48.6 months (8.6–59.1; IQR=12.2) for the 17 subjects without progressions. All patients evaluable for visual acuity had improved or stable vision. The most common attributable toxicities on both strata were grade 1 and 2 CPK elevation, hypoalbuminemia, dyspnea, rash, duodenal ulcer, anemia, dry skin, fatigue and diarrhea. Rare grade 3 toxicities included elevated CPK (n=5), maculopapular rash (n=5), neutropenia (n=3), nausea (n=3), paronychia (n=3), acneiform rash (n=2), diarrhea (n=2), elevated ALT (n=1), decreased ejection fraction (n=1), gastric hemorrhage (n=1), headache (n=1), skin infection (n=1), tooth infection (n=1) and weight gain (n=1). There was only one grade 4 toxicity, lymphopenia. There were no treatment-realted deaths. Patient reported outcomes and quality of life assessments were not part of the current study. INTERPRETATION: Selumetinib is active against recurrent, refractory or progressive PA harboring common BRAF aberrations and NF1-associated pLGG. To our knowledge, this is one of the first prospectively tested and successful molecularly-targeted agents in pLGG. These data not only provide an alternative to standard chemotherapy for these subgroups of patients, but this success has led to an interest in exploring efficacy in patients as a first-line therapy. In fact, these data have directly led to the development of two Children’s Oncology Group phase III studies in newly diagnosed pLGG patients both with and without NF1 comparing standard chemotherapy to selumetinib. The current trial was funded by a National Cancer Institute (NCI) Cancer Therapy Evaluation Program (CTEP) PBTC U01 Grant: 2UM1CA081457 (UM1) and by the American Lebanese Syrian Associated Charities.

Many immunotherapies act by enhancing the ability of cytotoxic T cells to kill tumor cells. Killing depends on T cell recognition of antigens presented by class I major histocompatibility complex (MHC-I) proteins on tumor cells. In this study, we showed that medulloblastomas lacking the p53 tumor suppressor do not express surface MHC-I and are therefore resistant to immune rejection. Mechanistically, this is because p53 regulates expression of the peptide transporter Tap1 and the aminopeptidase Erap1, which are required for MHC-I trafficking to the cell surface. In vitro, tumor necrosis factor (TNF) or lymphotoxin-β receptor agonist can rescue expression of Erap1, Tap1 and MHC-I on p53-mutant tumor cells. In vivo, low doses of TNF prolong survival and synergize with immune checkpoint inhibitors to promote tumor rejection. These studies identified p53 as a key regulator of immune evasion and suggest that TNF could be used to enhance sensitivity of tumors to immunotherapy.

SUMMARY Accelerating cures for children with cancer remains an immediate challenge as a result of extensive oncogenic heterogeneity between and within histologies, distinct molecular mechanisms evolving between diagnosis and relapsed disease, and limited therapeutic options. To systematically prioritize and rationally test novel agents in preclinical murine models, researchers within the Pediatric Preclinical Testing Consortium are continuously developing patient-derived xenografts (PDXs)—many of which are refractory to current standard-of-care treatments—from high-risk childhood cancers. Here, we genomically characterize 261 PDX models from 37 unique pediatric cancers; demonstrate faithful recapitulation of histologies and subtypes; and refine our understanding of relapsed disease. In addition, we use expression signatures to classify tumors for TP53 and NF1 pathway inactivation. We anticipate that these data will serve as a resource for pediatric oncology drug development and will guide rational clinical trial design for children with cancer., Graphical Abstract, In Brief Rokita et. al provide an extensively annotated genomic dataset of somatic oncogenic regulation across 37 distinct pediatric malignancies. The 261 patient-derived xenograft models are available to the scientific community, and the genomic annotations will enable rational preclinical agent prioritization and acceleration of therapeutic targets for early-phase pediatric oncology clinical trials.

PBTC-042 was a phase I trial of palbociclib to determine the maximum tolerated dose (MTD) and describe toxicities in children. Palbociclib is an oral, selective cyclin dependent kinase 4/6 inhibitor. METHODS: A rolling-6 design was utilized. Eligible patients were children ≥4 and ≤21 years-old with a progressive/refractory CNS tumor with intact retinoblastoma protein, measurable disease, and ability to swallow capsules. Pharmacokinetic studies were performed during the first course. Here, we report on the heavily pretreated stratum, which included patients who received >4 prior treatment regimens (either chemotherapy or biologic agent), and/or craniospinal irradiation, and/or myeloablative chemotherapy plus stem cell rescue. Palbociclib was initiated at 50 mg/m2/day for 21 consecutive days of a 28-day course. This was one dosage level below the MTD for the less heavily pretreated stratum (75 mg/m2). RESULTS: Fourteen eligible patients were enrolled (median age 12.8 years; male 79%). Eleven patients (79%) had either ependymoma or medulloblastoma. Four eligible and evaluable patients were enrolled at 50 mg/m2 with no DLTs. This prompted a dosage increase to 75 mg/m2. Ten eligible subjects were enrolled and 7 were evaluable for DLT assessment. One of 7 evaluable patients experienced a DLT (grade 3 thrombocytopenia). This established 75 mg/m2 as the MTD for more heavily pretreated patients. Mean ± SD palbociclib apparent oral clearance was 34.6 ± 18.4 L/h/m2. CONCLUSION: The MTD for palbociclib on a 3 week on/1 week off schedule in children with brain tumors is 75 mg/m2 and does not appear to be influenced by the degree of prior therapy.

BACKGROUND Patients with chiasmatic-hypothalamic low grade glioma (CHLGG) have frequent MRIs with gadolinium based contrast agents (GBCA) for disease monitoring. Cumulative gadolinium deposition in children is a potential concern. The purpose of this research is to establish whether MRI with GBCA is necessary for determining tumor progression in children with CHLGG. METHODS Children with progressive CHLGG were identified from Texas Children’s Cancer Center between 2005–2019. Pre- and post-contrast MRI sequences were separately reviewed by one neuroradiologist who was blinded to the clinical course. Three dimensional measurements and tumor characteristics were collected. Radiographic progression was defined as a 25% increase in size (product of two largest dimensions) compared to baseline or best response after initiation of therapy. RESULTS A total of 28 patients with progressive CHLGG including 683 MRIs with GBCA (mean 24 MRIs/patient; range: 10–43 MRIs) were reviewed. No patients had a diagnosis of NF1. Progression was observed 92 times, 91 (98.9%) on noncontrast and 90 (97.8%) on contrast imaging. Sixty-seven radiographic and/or clinical progressions necessitating management changes were identified in all (100%) noncontrast sequences and 66 (98.5%) contrast sequences. Tumor growth >2 mm in any dimension was identified in 184/187(98.4%) on noncontrast and 181/187(96.8%) with contrast imaging. Non primary metastatic disease was seen in seven patients (25%), which were better visualized on contrast imaging in 4 (57%). CONCLUSION MRI without GBCA effectively identifies patients with progressive disease. One should consider eliminating contrast in imaging of children with CHLGG with GBCA reserved for monitoring those with metastatic disease.

Genomic aberrations in the cell cycle and PI3K pathway are commonly observed in recurrent childhood brain tumors. Dual inhibition of CDK4/6 (ribociclib) and mTOR (everolimus) has strong biologic rationale, non-overlapping single-agent toxicities, and adult clinical experience. The maximum tolerated dosage (MTD) and/or recommended phase two dose (RP2D) of ribociclib and everolimus was determined in the Phase I study and ribociclib concentrations were characterized in plasma and tumor in children undergoing neurosurgical procedures. Following resection, eligible patients were enrolled in the Phase I study according to a rolling 6 design and received ribociclib and everolimus once daily for 21 days and 28 days, respectively. Patients undergoing surgery received ribociclib at the pediatric RP2D (350 mg/m2/day) for 7–10 days pre-operatively. Pharmacokinetic samples were collected on both cohorts and analyzed in nine patients on phase I study. Sixteen eligible patients enrolled on phase I study (median age 10.3 years; range: 3.9–20.4) and 5 patients were enrolled on the surgical cohort (median age 11.4 years; range: 7.2–17.1). Six patients enrolled at dose level 1 without dose limiting toxicities (DLT). Two of the three patients at dose level 2 experienced DLT (grade 3 hypertension and grade 4 ALT). The most common grade 3/4 toxicities were lymphopenia, neutropenia, and leucopenia. Everolimus concentrations following administration of everolimus alone were lower than those following drug combination, suggesting an impact of ribociclib on everolimus pharmacokinetics. The MTD/RP2D of ribociclib and everolimus in recurrent CNS tumors is 120 mg/m2 and 1.2 mg/ m2 daily for 21 days and 28 days, respectively.

BACKGROUND Pseudoprogression after proton therapy of CNS tumors is a challenging clinical situation. The rate of pseudoprogression after proton therapy of pediatric spinal tumors is unknown. METHODS Records of pediatric patients with spinal pilocytic astrocytoma (sPA; n = 9) or myxopapillary ependymoma (MPE; n = 6) with gross disease treated with proton therapy with at least 6 months of follow up from completion of proton therapy were retrospectively reviewed for demographics, treatment characteristics, and occurrence of pseudoprogression. Pseudoprogression was defined as a post-radiation increase in tumor size with subsequent decrease in size without additional tumor-directed therapy. RESULTS The median age at radiation for sPA patients was 10.1y (range, 7.0 – 16.2y) and 12.7y (range, 7.9 – 14.4y) for MPE patients. The median prescribed dose was 45 GyRBE (range, 39.6 – 50.4 GyRBE) for sPA patients and 50.4 GyRBE (range, 45 – 54 GyRBE) for MPE patients. One sPA patient received concurrent vincristine. Median follow up after proton therapy was 44 months (range, 9 – 99 months). Six of nine sPA patients (67%) had pseudoprogression occurring at a median of 81 days (range, 34 – 136 days) after proton therapy; no MPE patients developed pseudoprogression (0%; p < 0.03). Two sPA patients with pseudoprogression were symptomatic and improved with medical therapy. CONCLUSION Preliminary analysis suggests that pseudoprogression occurs frequently within 6 months after proton therapy for sPA and infrequently after proton therapy for MPE.

Genomic aberrations in the cell cycle and mTOR pathways have been reported in diffuse pontine gliomas (DIPG) and high-grade gliomas (HGG). Dual inhibition of CDK4/6 (ribociclib) and mTOR (everolimus) has strong biologic rationale, non-overlapping single-agent toxicities, and adult clinical experience. The maximum tolerated dose (MTD) and/or recommended phase two dose (RP2D) of ribociclib and everolimus administered during maintenance therapy following radiotherapy was determined in the phase I study as a rolling 6 design. Ribociclib and everolimus were administered once daily for 21 days and 28 days, respectively starting two-four weeks post completion of radiotherapy. All HGG patients and any DIPG patient who had undergone biopsy were screened for RB protein by immunohistochemistry. Eighteen eligible patients enrolled (median age 8 years; range: 2–18). Six patients enrolled at dose levels 1,2, and 3 without dose limiting toxicities (DLT). Currently, five patients are enrolled at dose level 3 expansion cohort. The median number of cycles are 4.5 (range: 1–20+). Among the expansion cohort, one dose limiting toxicity included a grade 3 infection and one patient required a dose reduction in course 3 due to grade 3 ALT and grade 4 hypokalemia. The most common grade 3/4 adverse events included neutropenia. Preliminary pharmacokinetic studies on 12 patients suggest an impact of ribociclib on everolimus pharmacokinetics. The MTD/RP2D of ribociclib and everolimus following radiotherapy in newly diagnosed DIPG and HGG is anticipated to be 170 mg/m2/day x 21 days and 1.5 mg/ m2/day every 28 days which is equivalent to the adult RP2D.

PURPOSE Sporadic optic pathway/hypothalamic gliomas (OP/HGs) represent a unique entity within pediatric low-grade glioma. Despite favorable survival, the location makes treatment difficult and local progression debilitating. We conducted longitudinal assessment of visual acuity (VA) among patients treated in the modern era with chemotherapy (CT) or early radiotherapy (RT). METHODS Clinical characteristics were abstracted for patients treated over a 15-year period (2000–2015) at a single institution. Comprehensive ophthalmologic data taken at three to six-month intervals was examined with age-appropriate VA metrics converted to LogMAR scale. Kaplan-Meir “blindness-free survival” (BFS) curves were calculated as time to bilateral functional blindness (i.e. LogMAR ≥ 0.8 in both eyes), stratified by treatment and compared using log-rank test. RESULTS Thirty-six patients with median follow-up of 7.6 years (range: 2–17) were identified. Median age at diagnosis was 2.5 years (IQR: CONCLUSIONS In a contemporary cohort, early RT, defined as initial or 1st line salvage therapy for OP/HGs manifested in superior VA. Children undergoing CT are at highest risk of functional blindness following five years of treatment.

Background. Meningiomas constitute one-third of all primary brain tumors. Although typically benign, about 20% of these tumors recur despite surgery and radiation, and may ultimately prove fatal. There are currently no effective chemotherapies for meningioma. We, therefore, set out to develop patient-derived orthotopic xenograft (PDOX) mouse models of human meningioma using tumor. Method. Of nine patients, four had World Health Organization (WHO) grade I tumors, five had WHO grade II tumors, and in this second group two patients also had recurrent (WHO grade III) meningioma. We also classified the tumors according to our recently developed molecular classification system (Types A, B, and C, with C being the most aggressive). We transplanted all 11 surgical samples into the skull base of immunodeficient (SCID) mice. Only the primary and recurrent tumor cells from one patient&mdash, both molecular Type C, despite being WHO grades II and III, respectively&mdash, led to the formation of meningioma in the resulting mouse models. We characterized the xenografts by histopathology and RNA-seq and compared them with the original tumors. We performed an in vitro drug screen using 60 anti-cancer drugs followed by in vivo validation. Results. The PDOX models established from the primary and recurrent tumors from patient K29 (K29P-PDOX and K29R-PDOX, respectively) replicated the histopathology and key gene expression profiles of the original samples. Although these xenografts could not be subtransplanted, the cryopreserved primary tumor cells were able to reliably generate PDOX tumors. Drug screening in K29P and K29R tumor cell lines revealed eight compounds that were active on both tumors, including three histone deacetylase (HDAC) inhibitors. We tested the HDAC inhibitor Panobinostat in K29R-PDOX mice, and it significantly prolonged mouse survival (p &lt, 0.05) by inducing histone H3 acetylation and apoptosis. Conclusion. Meningiomas are not very amenable to PDOX modeling, for reasons that remain unclear. Yet at least some of the most malignant tumors can be modeled, and cryopreserved primary tumor cells can create large panels of tumors that can be used for preclinical drug testing.

Brain tumor is leading cause of cancer-related death in children. While significant advances have been made in molecularly subgrouping tumors of same pathologic diagnosis, little is known about the biologic differences among racial/ethnic populations, and there is a lack of animal models that represent different racial/ethnic patients. Here, we report our analysis of tumorigenicity of a total of 215 pediatric brain tumors in SCID mice. All surgical tumor tissues were obtained from cryo lab and directly implanted into the anatomically matched locations in mouse brains, i.e., cerebral tumors (such as GBM) into mouse right cerebra, and cerebellar tumors (such as medulloblastoma) into mouse cerebella. The animals were closely monitored following institutional-approved animal protocols. Tumor formation was validated either through the harvesting of visible tumors or via histopathologic examination of paraffin-embedded whole mouse brains. From the 215 tumors, racial/ethnic information was validated in 180 tumors. Overall tumor formation was 41.2% (52/126) in white, 26.9% (7/26) in black, 50% (8/16) in more than one race, and 20% (2/5) in Asian patients. When different tumor types were compared, children with medulloblastoma exhibited similar tumor take rate, ranging from 50% (3/6) in black, to 54% (15/28) in whites and 67% (2/3) of American Indian or Alaska Native, whereas in GBM, it was 79% (11/14) in white, and 1/1 in other racial groups, and in ependymoma, it ranged from 14% (1/7) in American Indian or Alaska to 18% (4/22) in white patients. The tumor take of atypical teratoid/rhabdoid tumor (ATRT) was 63% (5/8) in white but 0% (0/3) in black patient. Low-grade gliomas (total 26) did not form xenografts, and sample size was small ( Note: This abstract was not presented at the conference. Citation Format: Lin Qi, Mari Kogiso, Yuchen Du, Yulun Huang, Huiyuan Zhang, Frank Braun, Holly Lindsay, Sibo Zhao, Sarah Injac, Lazlo Perlaky, Patricia Baxter, Wan-Yee Teo, Zhigang Liu, Xiumei Zhao, Yujing Zhang, Jack M.F. Su, Xiao-Nan Li. Racial/ethnic differences of pediatric brain tumors in the development of orthotopic PDX models [abstract]. In: Proceedings of the Eleventh AACR Conference on the Science of Cancer Health Disparities in Racial/Ethnic Minorities and the Medically Underserved; 2018 Nov 2-5; New Orleans, LA. Philadelphia (PA): AACR; Cancer Epidemiol Biomarkers Prev 2020;29(6 Suppl):Abstract nr C121.

Glioblastoma multiforme (GBM) is the most lethal brain tumor. Diffuse invasion is one of the important biologic features that make GBM particularly difficult to treat. While existing studies on GBM invasion are primarily conducted in tumor core tissues from surgical resections, it is unclear whether infiltrative GBM cells would be more informative for studying their invasive nature, and little is known if and which cancer stem cell populations are driving GBM invasion. To address these issues, we utilized 6 patient tumor-derived orthotopic xenograft mouse models to isolate invasive GBM cells (GBM(INV), infiltrating normal mouse brain parenchyma) and tumor core GBM cells (GBM(TC)) and compared their biological features. Our result showed that the GBM(INV) cells have stronger neurosphere forming efficiency in vitro, more tumorigenic capacity in vivo, even some invasion-related genes also be detected to be changed in GBM(INV) cells. In the further study, gene profiling showed that CD57 is higher expressed in mostly GBM models and in patient tissues; CD57+ cells have stronger neurosphere formation ability with compared to CD57- cells; The survival time of mice injected with CD57+ cells decreases compared with mice with CD57- cells. Furthermore, CD57+ cells in GBM(INV) is higher expressed in PDOX invasive tumor cells than the expression of GBM(TC) cells; CD57+ cells in GBM(INV) cells are enriched (>2 folds) and have stronger neurosphere formation ability and more invasive with compared to GBM(TC) cells in mostly GBM models; The survival time of mice injected with GBM(INV) cells decreases compared with mice with GBM(TC) cells. We found that CD57+ cells expressed high levels of self-renewal genes (BMI1 and Nanog). In conclusion, we showed that invasive GBM cells were not biologically identical to the matched tumor core cells and identified CD57 as a novel stem cell marker that was associated with GBM infiltration.

Medulloblastoma (MB) is the most common malignant brain tumor of childhood. Although outcomes have improved in recent decades, new treatments are still needed to improve survival and reduce treatment-related complications. The MB subtypes groups 3 and 4 represent a particular challenge due to their intragroup heterogeneity, which limits the options for "rational" targeted therapies. Here, we report a systems biology approach to drug repositioning that integrates a nonparametric, bootstrapping-based simulated annealing algorithm and a 3D drug functional network to characterize dysregulated driver signaling networks, thereby identifying potential drug candidates. From more than 1300 drug candidates studied, we identified five members of the cardiac glycoside family as potentially inhibiting the growth of groups 3 and 4 MB and subsequently confirmed this in vitro. Systemic in vivo treatment of orthotopic patient-derived xenograft (PDX) models of groups 3 and 4 MB with digoxin, a member of the cardiac glycoside family approved for the treatment of heart failure, prolonged animal survival at plasma concentrations known to be tolerated in humans. These results demonstrate the power of a systematic drug repositioning method in identifying a potential treatment for MB. Our strategy could potentially be used to accelerate the repositioning of treatments for other human cancers that lack clearly defined rational targets.

The PBTC is conducting a phase II study (NCT01089101) evaluating selumetinib (AZD6244, ARRY-142886), a MEK I/II inhibitor, in children with recurrent/refractory LGG assigned to 6 strata. We present the updated data on Stratum 2 and 5. Also, data on subsequent progression after treatment completion in patients enrolled on Stratum 1 and 3 will be discussed. Finally, we present details on the re-treatment study (PBTC-029C). Both stratum 2 (pilocytic astrocytoma [PA] without common BRAF aberrations) and Stratum 5 (non-pilocytic LGG with BRAF aberrations) met response criteria for expansion (> 2 objective responses in 16 patients), and accrual to a total of 25 patients on each stratum is ongoing. Among 50 patients treated on Stratum 1 (PA with BRAF aberrations) or Stratum 3 (NF-associated LGG), 21 have progressed. Thirteen of 21 have progressed after stopping therapy. The median time to progression for these 13 patients is 119 days (10–928). The re-treatment study has enrolled 25 patients who received a median of 12 re-treatment courses (2–36). The most common attributable toxicities after re-treatment were grade 1 CPK elevation (44%), diarrhea (44%), hypoalbuminemia (40%), elevated AST (36%), rash (36%) and fatigue (32%). The most common grade 3/4 attributable toxicities were grade 3 paronychia (8%), CPK elevation (4%), AST elevation (4%), decreased ejection fraction (4%), neutropenia (4%), elevated triglycerides (4%), peripheral neuropathy (4%) and grade 4 CPK elevation (4%). There is not a significant difference between the toxicities observed during original therapy versus re-treatment. The most current response and patient demographic data will be presented.

Diffuse invasion is one of the key features that make GBM particularly difficult to treat. We hypothesize that direct comparison of matched invasive (GBMINV) and tumor core GBM cells (GBMTC) would facilitate the discovery of drivers of pediatric GBM (pGBM) invasion. However, GBMINV cells are extremely difficult to obtain from normal brain tissues because aggressive surgical resection of normal tissue carries the risk of serious neurological deficits. Most past and current studies on GBM invasion were and are forced to utilize the resected primary tumor masses. To overcome this barrier, we utilized a panel of 6 pediatric patient tumor-derived orthotopic xenograft (PDOX) mouse models to isolate matching pairs of GBMTC cells and GBMINV cells and confirmed a significantly elevated invasive capacity in GBMINV cells both in vitro and in vivo. Global profiling of 768 human microRNA using a real-time PCR-based Taqman system identified 23 microRNAs were upregulated in the GBMINV cells in at least 4 of the 6 pGBM models as compared with the matching GBMTC cells. We subsequently showed that silencing the top three miRNAINV, miR-126, miR-369-5p, and miR-487b, suppressed tumor cell migration in vitro (both as neurospheres and monolayer cultures) without affecting cell proliferation, and blocked pGBM invasion in mouse brains. Integrated analysis of the mRNA profiling of the same set of GBMTC and GBMINV cells revealed the affected signaling pathways and identified KCNA1 as the sole common computational target gene of the three miRNAINV. Treatment of three pairs of GBMTC and GBMINV cells with two KCNA1 inhibitors, ADWX1 and Agitoxin 2, caused significant suppression of pGBM cell migration in vitro. In conclusion, this study revealed an intrinsically elevated invasive phenotype in GBMINV cells, identified miR-126, -369-5p, and -487b as novel drivers of pGBM invasion, and characterized KCNA1 as a potential therapeutic target for arresting pGBM invasion.

Background We conducted a phase II study of oral capecitabine rapidly disintegrating tablets given concurrently with radiation therapy (RT) to assess progression-free survival (PFS) in children with newly diagnosed diffuse intrinsic pontine gliomas (DIPG). Patients and methods Children 3-17 years with newly diagnosed DIPG were eligible. Capecitabine, 650 mg/m2 /dose BID (maximum tolerated dose [MTD] in children with concurrent radiation), was administered for 9 weeks starting the first day of RT. Following a 2-week break, three courses of capecitabine, 1,250 mg/m2 /dose BID for 14 days followed by a 7-day rest, were administered. As prospectively designed, 10 evaluable patients treated at the MTD on the phase I trial were included in the phase II analyses. The design was based on comparison of the PFS distribution to a contemporary historical control (n = 140) with 90% power to detect a 15% absolute improvement in the 1-year PFS with a type-1 error rate, α = 0.10. Results Forty-four patients were evaluable for the phase II objectives. Capecitabine and RT was well tolerated with low-grade palmar plantar erythrodyesthesia, increased alanine aminotransferase, cytopenias, and vomiting the most commonly reported toxicities. Findings were significant for earlier progression with 1-year PFS of 7.21% (SE = 3.47%) in the capecitabine-treated cohort versus 15.59% (SE = 3.05%) in the historical control (P = 0.007), but there was no difference for overall survival (OS) distributions (P = 0.30). Tumor enhancement at diagnosis was associated with shorter PFS and OS. Capecitabine was rapidly absorbed and converted to its metabolites. Conclusion Capecitabine did not improve the outcome for children with newly diagnosed DIPG.

Glioblastoma (GBM) is one of the deadliest brain tumors both in adults and children. The efficacy of targeting mutated genes/pathways has not been systematically analyzed. Matching pairs of neurosphere and monolayer cell cultures from 4 molecularly characterized PDOX mouse models established from distinct clinical stages of pediatric glioma were treated with 7,913 drugs (mostly in clinical use or trials) at 4 doses (0-10 uM) for 3 and 7 days. Relatively long-term treatment was achieved for drug screening in most of the 4,629 compounds that were active in at least one culture at one time point. When gene mutations were druggable, not all inhibitors of the same family were active. In IC-4687GBM (a treatment-naïve GBM), high NF1 mutation (>76-99% allele) frequency was found in patient tumor, xenografts, and cultured cells, but only 3/17 MEK inhibitors were active in neurospheres and 2/17 in monolayer cells on day 7. In IC-R0315GBM (from an autopsied terminal GBM) that carried PI3KCA mutation (allele frequency 22-27%), 5/33 PI3K inhibitors were active in neurospheres and 8/33 in monolayer cells after 7 days. In IC-3752GBM (recurrent GBM) and ICb-1227AA (radiation-induced anaplastic astrocytoma), no druggable mutations were detected. The number of active drugs on day 7 was 366 in IC-4768GBM, 406 in IC-3752GBM, 284 in IC-R0315GBM, and 305 in ICb-1227AA. When the 4 matching pairs of neurospheres and monolayers were compared, the agents active in both cultures ranged from 36% to 60%, active only in neurosphere from 10.3% to 25%, and active only in monolayer cells from 14.8% to 53%. We showed that long-term treatment is feasible for high-throughput drug screening. Targeting druggable mutations can be achieved. Neurospheres and monolayer cells do not always respond equally toward the same drugs, and effective targeting of both subpopulations is needed to generate prolonged animal survival times.

BACKGROUNDMeningioma is the most common brain tumors in adults. Cranial-base meningioma are difficult to achieve complete resection while others exhibit progression and aggressive profiles characterized by high recurrence rates and resistance to standard treatment. The lack of clinically relevant animal models is blocking the development of novel therapies. Here, we report our establishment of orthotopic xenograft mouse models and in vitro culture systems from surgical specimens of primary and recurrent meningiomas.METHODS6 primary surgical samples and 4 recurrent samples were obtained from meningioma patients. Tumor tissues were dissociated into single cells and directly implanted into right cranial base of NOD/SCID mice (1x105 cells/mouse). Primary cultures were initiated in serum-free and traditional FBS-based media. Tumor growth was monitored by small animal MRI. Pathologic features of the PDOX models and the matched patient tumors were compared with standard H&E and immunohistochemical staining.RESULTSThree of the 10 tumors were not tumorigenic, and xenograft tumor formation from 5 additional samples is pending. Growth of intracranial (cranial base) xenograft was confirmed in two samples derived from the same patient diagnosed as atypical meningioma (K029MEN) and progressed as anaplastic meningioma at recurrence (K037MEN). These PDOX have since been serially subtransplanted in mouse brains for generation 2. The xenograft tumors replicated histopathological features (invasion, high proliferation and increased microvessel density) of their parental tumors. In vitro growth of K029MEN and K037MEN as neurospheres and monolayer were maintained for passaging for 10 times. Additionally, cells from K030MEN (WHO grade I meningioma) has been passaged as monolayer for more than 30 times.CONCLUSIONA novel set of meningioma PDOX models derived from matching primary and recurrent tumor was established. The xenograft tumors replicated the histopathological of the original patient tumors, providing the opportunities to understand the biology of meningiomas and to conduct preclinical drug testing.

Glioblastoma (GBM) is one of the deadliest brain tumors both in adults and children. The efficacy of targeting mutated genes/pathways has not been systematically analyzed. Additionally, it remains unknown if neurospheres and non-stem monolayer tumor cells will respond equally to the same target therapies. Relatively long-term treatment was achieved for high-throughput drug screening and revealed time- and dose-dependent effects in most of the 4,629 compounds that were active in at least one culture at one time point. When gene mutations were druggable, not all inhibitors of the same family were active. In IC-4687GBM, high NF1 mutation (>76–99% allele) frequency was found in patient tumor, xenografts, and cultured cells, but only 3/17 MEK inhibitors were active in neurospheres and 2/17 in monolayer cells on day 7. In IC-R0315GBM that carried PI3KCA mutation (allele frequency 22–27%), 5/33 PI3K inhibitors were active in neurospheres and 8/33 in monolayer cells after 7 days. In IC-3752GBM (recurrent GBM) and ICb-1127AA, no druggable mutations were detected. The number of active drugs on day 7 was 366 in IC-4768GBM, 406 in IC-3752GBM, 284 in IC-R0315GBM, and 305 in ICb-1277AA. When the 4 matching pairs of neurospheres and monolayers were compared, the agents active in both cultures ranged from 36% to 60%, active only in neurosphere from 10.3% to 25%, and active only in monolayer cells from 14.8% to 53%. Subsequent in vivo validation using MLN8327 in IC-4687GBM and IC-R0315GBM showed that effective targeting of both neurosphere and monolayer was required for significantly-improved animal survival times. We showed that long-term treatment is feasible for high-throughput drug screening. Targeting druggable mutations can be achieved but only by a fraction of specific agents. Neurospheres and monolayer cells do not always respond equally toward the same drugs, and effective targeting of both subpopulations is needed to generate prolonged animal survival times.

BACKGROUND Meningioma is the most common brain tumor in adults. Despite the overall benign nature of meningioma, skull base tumors can be difficult to completely resect while others exhibit progression and aggressive profiles. The lack of clinically relevant animal models is blocking the development of novel therapies. MATERIAL AND METHODS Twelve surgical specimens (1 × 105) from 11 adult meningioma patients were implanted into the frontal cranial-base of the brain of SCID mice. Mice were then followed and assessed for tumor formation. Tumor growth was confirmed by small animal MRI. Pathologic features of the PDOX models and the matched patient tumors were compared with standard H&E and immunohistochemical staining. RNAseq was performed to examine the molecular fidelity of PDOX tumors and to identify new therapeutic targets. A panel of 60 clinically-relevant drugs was developed for screening drug sensitivity. In vivo examination of therapeutic efficacy of Panobinostat was performed in two models by treating preformed PDOX tumors with i.p. injection (10 mg/kg), 5 days on, 5 days off for 2 cycles. RESULTS Intracranial xenograft formation was confirmed in two samples derived from the same patient, the first an atypical meningioma (K029MEN-P) and the second, which progressed to anaplastic meningioma at recurrence (K029MEN-R). MRI scanning revealed that the PDOX tumors grew from the skull base. These patient tumor cells can be cryopreserved for long-term maintenance of tumorigenicity. The xenograft tumors replicated histopathological features of parental tumors. Overall gene expression profiles of PDOX were similar to the original patient tumors. Using MEN primary culture cells, we screened 60 drugs and identified 12 (20%) active compounds. Panobinostat also significantly prolonged survival of mice bearing orthotopic meningiomas. CONCLUSION A set of meningioma PDOX models derived from primary and recurrent tumor was established. Our data further demonstrate that panobinostat exerts potent antitumor activity against high-grade meningioma.

Metastatic intracranial germinoma is difficult to treat. Although the proto-oncogene KIT is recognized as one of the most frequent genetic abnormalities in CNS germinoma, the development of new target therapeutic agents for CNS germinoma is hampered by the lack of clinically-relevant animal models that replicate the mutated or over-expressed KIT. CNS germinoma tumor cells from five pediatric patients were directly implanted into the brains of Rag2/severe combined immune deficiency mice. Once established, the xenograft tumors were sub-transplanted in vivo in mouse brains. Characterization of xenograft tumors were performed through histologic and immunohistochemical staining, and KIT mutation analysed with quantitative pyro-sequencing. Expression of putative cancer stem cell markers (CD133, CD15, CD24, CD44, CD49f) was analyzed through flow cytometry. Two patient-derived orthotopic xenograft (PDOX) models (IC-6999GCT and IC-9302GCT) were established from metastatic germinoma and serially sub-transplanted five times in mouse brains. Similar to the original patient tumors, they both exhibited faint expression (+) of PLAP, no expression (-) of β-HCG and strong (+++) expression of KIT. KIT mutation (D816H), however, was only found in IC-9320GCT. This mutation was maintained during the five in vivo tumor passages with an increased mutant allele frequency compared to the patient tumor. Expression of putative cancer stem cell markers CD49f and CD15 was also detected in a small population of tumor cells in both models. This new pair of PDOX models replicated the key biological features of pediatric intracranial germinoma and should facilitate the biological and pre-clinical studies for metastatic intracranial germinomas.

We previously showed that primary tumor-based orthotopic xenograft mouse models of medulloblastoma replicated the histopathological phenotypes of patients' original tumors. Here, we performed global gene expression profiling of 11 patient-specific xenograft models to further determine whether the xenograft tumors were molecularly accurate during serial subtransplantations in mouse brains and whether they represented all the molecular subtypes of medulloblastoma that were recently described. Analysis of the transcriptomes of 9 pairs of matched passage I xenografts and patients' tumors revealed high correlation coefficients (r(2) > 0.95 in 5 models, > 0.9 in 3 models, and > 0.85 in 1 model) and only identified 69 genes in which expressions were altered (FDR = 0.0023). Subsequent pair-wise comparisons between passage I, III, and V xenografts from the 11 models further showed that no dramatic alterations were introduced (r(2) > 0.9 in 8 models and > 0.8 in 3 models). The genetic abnormalities of each model were then identified through comparison with control RNAs from 5 normal cerebella and 2 fetal brains. Hierarchical clustering using 3 previously published molecular signatures showed that our models span the whole spectrum of molecular subtypes, including SHH (n = 2), WNT (n = 2), and the most recently identified group C (n = 4) and group D (n = 3). In conclusion, we demonstrated that the 11 orthotopic medulloblastoma xenograft models were molecularly faithful to the primary tumors, and our comprehensive collection of molecularly distinct animal models should serve as a valuable resource for the development of new targeted therapies for medulloblastoma.

A greater understanding of the Ras-MAP kinase-signaling pathway in pediatric low-grade glioma (LGG) has led to the ability to target this pathway with therapeutic intent. The PBTC conducted a multi-institutional phase II study (NCT01089101) evaluating selumetinib (AZD6244, ARRY-142886), a MEK I/II inhibitor, in children with recurrent/refractory LGG assigned to 6 strata and treated at 25 mg/m2/dose PO BID for up to two years. Here we present the data from three of these strata. Stratum I included children with non-NF-1 and non-optic pathway recurrent/refractory pilocytic astrocytoma (PA) harboring BRAF aberrations (BRAF V600e mutation or the BRAF-KIAA 1549 fusion). Eight of 25 (32%) patients achieved a partial response (PR) with 2-year PFS of 66+11%. Two of 7 (29%) patient tumors with a BRAF V600e mutation and 6/18 (33%) with a BRAF KIAA-1549 fusion had a PR. Stratum 3 enrolled NF-1-associated LGG. Tissue for tumor BRAF evaluation was not required for eligibility. Ten of 25 (40%) achieved PR with a 2-year PFS of 96+4%. Only one patient progressed while on treatment. Stratum 4 included children with non-NF-1 optic pathway/hypothalamic LGG. Tissue for tumor BRAF evaluation was not required for eligibility. Two of 16 (12.5%) had a PR with a 2-year PFS of 65+13%. The BRAF aberration status of the responders in strata 3 and 4 is mostly unknown. All responses were confirmed centrally and seven patients remain on treatment. The most common toxicities were grade 1/2 CPK elevation, diarrhea, hypoalbuminemia, elevated AST and rash. Rare grade 3/4 toxicities included elevated CPK, rash, neutropenia, emesis and paronychia. Selumetinib was effective in treating children with recurrent/refractory LGG, including those with NF-1 associated LGG and PA harboring BRAF V600e mutation or BRAF-KIAA 1549 fusion. Larger prospective studies are necessary to determine the future, specific role of this agent in treating children with various LGG molecular subtypes.

Objective: Glioblastoma and Medulloblastoma are deadliest brain tumors both in adults and children. While next-generation sequencing rapidly identifies whole exome/genome gene mutations, the efficacy of targeting mutated genes/pathways has not been systematically analyzed. Additionally, it remains unknown if neurospheres (enriched with cancer stem cells) and non-stem monolayer tumor cells will respond equally to the same target therapies. Methods: Matching pairs of neurosphere and monolayer cell cultures from 4 molecularly characterized PDOX mouse models established from distinct clinical stages of pediatric GBM and MB were treated with 7,913 drugs (mostly in clinical use or trials) at 4 doses (0-10 uM) for 3 and 7 days. Dose-response curves and active area under the curve were constructed to identify active drugs and correlate with model-specific gene mutations. Results: Relatively long-term treatment (7 day screening) was achieved for high-throughput drug screening and revealed time- and dose-dependent effects in most of the 4,629 compounds that were active in at least one culture at one time point. When gene mutations were druggable, not all inhibitors of the same family were active. In IC-4687GBM (a treatment-naïve GBM), high NF1 mutation (>76-99% allele) frequency was found in patient tumor, xenografts, and cultured cells, but only 3/17 MEK inhibitors were active in neurospheres and 2/17 in monolayer cells (GDC0980 and PI-103 were active in both) on day 7. In IC-R0315GBM (from an autopsied terminal GBM) that carried PI3KCA mutation (allele frequency 22-27%), 5/33 PI3K inhibitors were active in neurospheres and 8/33 in monolayer cells (only Tremetinib was active in both) after 7 days. In IC-3752GBM (recurrent GBM) and ICb-1127AA (radiation-induced anaplastic astrocytoma), no druggable mutations were detected. The number of active drugs on day 7 was 366 in IC-4768GBM, 406 in IC-3752GBM, 284 in IC-R0315GBM, and 305 in ICb-1277AA. When the 4 matching pairs of neurospheres and monolayers were compared, the agents active in both cultures ranged from 36% to 60%, active only in neurosphere from 10.3% to 25%, and active only in monolayer cells from 14.8% to 53%. The response of MB to the drugs is on the way for analysis. Conclusion: We showed that long-term treatment is feasible for high-throughput drug screening. Targeting druggable mutations can be achieved but only by a fraction of specific agents. Neurospheres and monolayer cells do not always respond equally toward the same drugs, and effective targeting of both subpopulations is needed to generate prolonged animal survival times. Citation Format: Lin Qi, Yuchen Du, Goeun Bae, Mari Mari, Frank Braun, Holly Lindsay, Huiyuan Zhang, Sibo Zhao, Sarah Injac, Patricia Baxter, Jack Su, Michael Mancini, Oliver Hampton, William Parsons, Mural Chintagumpala, Clifford Stephan, Peter Davies, Xiaonan Li. Identification of personalized active agents in pediatric medulloblastoma through high-throughput drug screening in matching pairs of patient derived orthotopic xenograft neurosphere and monolayer cells [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2018; 2018 Apr 14-18; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2018;78(13 Suppl):Abstract nr LB-069.

Background: Ependymoma (EPN) is one of the most malignant tumors and surgical resection is the mainstay of treatment followed by re-irradiation and chemotherapy. However the benefit of this modality have not been clearly demonstrated. Posterior fossa A (PFA) EPN have the worst prognosis and driven by epigenetic changes, suggesting targeting epigenetic changes in PFA EPN can potentially be effective. The limited availability of a reliable translational platform has hampered efforts to understand tumor biology and the preclinical and clinical trials for EPN. In this study, we established a series of orthotopic EPN xenograft mouse models from fresh surgical specimens. We also examined the therapeutic efficacy of SYC-836, a novel LSD-1 inhibitor developed at Baylor College of Medicine, both in vitro and in vivo in these PDOX models. Material and Methods: Clinically and pathologically available PFA EPN samples were obtained from patients who had treatment in Texas Children's Hospital. Parts of the surgical samples were dissociated into single cells and implanted into the cerebellum of SCID mice (1x105 cells /2 uL FBS). Once tumorigenicity confirmed, xenograft tumor would be serially subtransplanted in mouse brains and be cultured in vitro to develop the permanent cell line. To test if xenograft tumors precisely reflect morphologic and pathologic characteristics of their parental tumors, we compared histology, and several key parameters in invasiveness, growth rate and microvessel density by IHC. To examine in vitro anti-tumor activities, primary cultured cells from an established PDOX model of PFA EPN (4423EPN) were subjected to SYC-836 at various concentrations (0-25uM). Cell viability was measured at 5 different time points over 14 days. To validate the drug's in vivo efficacy, two established PFA EPN PDOX models, 4423EPN and 2002EPN, were utilized. 40 mice per model were implanted and divided into 4 treatment groups: 1) control , 2) radiation/standard therapy (2 Gy XRT x 5 days), 3) SYC-836 only (15mg/kg IP x 28 days), and 4) combination (radiation + SYC-836 per regimen above). Animal survival times were analyzed using log rank analysis. Results: By now we have successfully developed 6 intracerebellum xenografts. It has since been serially subtransplanted in mouse for more than 3 generations and can be cryopreserved for long-term maintenance of tumorigenicity. We showed xenograft tumors precisely replicated morphologic and pathologic characteristics of their parental tumors. SYC-836 demonstrated effective cell killing in vitro in both time- and dose-dependent manner. IC50 was ~7.5uM. In vivo experiment was completed in 2 PFA EPN PDOX models. There were no survival benefit with either XRT only or SYC-836 only compared to control group; however, the combination treatment lead to significant improvement in animal survival. SYC-836 was well tolerated in mice. Conclusion: We have successfully established a series of clinically orthotopic EPN xenograft models and they well-matched with those of parental EPNs both preclinically and clinically. Our data showed that combining SYC-836 with radiation synergistically prolongs animal survival significantly, suggesting that SYC-836 may have a role in the clinical setting by either reducing radiation dosages, or be a potential adjuvant agent to other chemotherapy drugs in treatment for EPN. Citation Format: Huiyuan Zhang, Sibo Zhao, Yuchen Du, Lin Qi, Frank K. Braun, Mari Kogiso, Holly B. Lindsay, Sarah G. Injac, Laszlo Perlaky, Patricia Baxter, Jack M. Su, Xiao-Nan Li. Novel LSD-1 inhibitor validation in new established clinically PFA patient-derived orthotopic xenograft (PDOX) models of ependymoma [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2018; 2018 Apr 14-18; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2018;78(13 Suppl):Abstract nr LB-167.

BACKGROUND: Diffuse intrinsic pontine glioma (DIPG) is a lethal childhood cancer with a median overall survival (OS) of 10 months and a post-progression survival (PPS) of 2–3 months. Panobinostat (LBH589) is a multi-histone deacetylase inhibitor with activity against preclinical models of DIPG. The aim of this study is to determine the safety, tolerability, maximum tolerated dose (MTD), toxicity profile and pharmacokinetics of panobinostat in children with DIPG. METHODS: In the initial cohort, panobinostat was administered orally three days per week for three weeks on, one week off to children with progressive DIPG. The starting dose was 10mg/m(2)/dose with dose escalation following a two-stage continual reassessment method. RESULTS: Nineteen eligible subjects were enrolled, of whom 17 were evaluable for toxicity. Dose-limiting toxicities were observed at dose level 2 (16 mg/m2/dose) and included thrombocytopenia (Grade 3, n=1; Grade 4, n=1), and neutropenia (Grade 4, n=1). Grade 3 posterior reversible encephalopathy syndrome was reported in one patient. Subjects received a median of 2 courses (range, 1–5). Several subjects remain alive at the time of abstract submission; the median survival from initiation of treatment at the time of progression (PPS) is 5.1 months (range 0.7 – 11.4 months). CONCLUSIONS: The MTD of panobinostat administered to children with progressive DIPG on a schedule of 3 weeks on, 1 week off, was 10mg/m(2)/dose. Dose-limiting toxicities (DLTs) are hematological, with thrombocytopenia the most common DLT. This trial was amended to evaluate the MTD of panobinostat in children with non-progressive DIPG, treated on alternate weeks, and is on-going.

MP470 is a multi-targeted tyrosine kinase inhibitor with potent activity against mutant c-Kit, PDGFRα, Flt3, c-Met and c-Ret that is being evaluated as an anticancer agent. The plasma and cerebrospinal fluid (CSF) pharmacokinetics of MP470 were studied in a non-human primate model that is highly predictive of CSF penetration in humans. Oral MP470, 300 mg, was administered to four non-human primates. Serial samples of blood were collected from four animals and CSF samples from three animals for pharmacokinetic studies. Plasma and CSF concentrations were measured using an LC–MS/MS assay. Both model-independent and model-dependent methods were used to analyze the pharmacokinetic data. Following a one-time oral dose of 300 mg, the MP470 plasma area under the curve (AUC) was 1,690 ± 821 nM h (mean ± SD). The half-life of MP470 in the plasma was 11.0 ± 3.4 h. There was no measurable MP470 in the CSF. Although CSF penetration is minimal, MP470 has demonstrated potent activity against cancer cell lines in vitro and in vivo, and further clinical investigation is warranted.

e14051Background: Diffuse intrinsic pontine glioma (DIPG) is a universally fatal brainstem tumor with survival < 1 year despite multimodal therapy. Similarly, other childhood high-grade gliomas (HG...

Recent identification of cancer stem cells in medulloblastoma (MB) and high-grade glioma has stimulated an urgent need for animal models that will not only replicate the biology of these tumors, but also preserve their cancer stem cell pool. We hypothesize that direct injection of fresh surgical specimen of MB and high-grade glioma tissues into anatomically equivalent locations in immune-deficient mouse brains will facilitate the formation of clinically accurate xenograft tumors by allowing brain tumor stem cells, together with their non-stem tumor and stromal cells, to grow in a microenvironment that is the closest to human brains. Eight of the 14 MBs (57.1%) and two of the three high-grade gliomas (66.7%) in this study developed transplantable (up to 12 passages) xenografts in mouse cerebellum and cerebrum, respectively. These xenografts are patient specific, replicating the histopathologic, immunophenotypic, invasive/metastatic, and major genetic (analyzed with 10K single nucleotide polymorphism array) abnormalities of the original tumors. The xenograft tumor cells have also been successfully cryopreserved for long-term preservation of tumorigenicity, ensuring a sustained supply of the animal models. More importantly, the CD133+ tumor cells, ranging from 0.2%–10.4%, were preserved in all the xenograft models following repeated orthotopic subtransplantations in vivo. The isolated CD133+ tumor cells formed neurospheres and displayed multi-lineage differentiation capabilities in vitro. In summary, our study demonstrates that direct orthotopic transplantation of fresh primary tumor cells is a powerful approach in developing novel clinical relevant animal models that can reliably preserve CD133+ tumor cell pools even during serial in vivo subtransplantations.Disclosure of potential conflicts of interest is found at the end of this article.

We hypothesize that direct comparison of the matched invasive and tumor core GBM cells would facilitate the discovery of drivers of GBM invasion. However, GBM cells that infiltrate deep into normal brain tissues (GBMinv) are extremely difficult to obtain from patients. To overcome this barrier, we utilized a panel of 7 (1 adult and 6 pediatric) patient-tumor derived orthotopic xenograft (PDOX) mouse models to isolate matched GBMinv and GBMTC cells by cutting whole mouse brains into 10 slices (1 mm thick), followed by microscopic dissection of tumor mass (to harvest GBMTC cells) and the "normal" mouse brains (to harvest GBMinv cells), respectively. After the mouse cells were decontaminated through FACS, matched GBMinv and GBMTC cells were subjected to microRNA profiling using Taqman MicroRNA array (768 microRNAs). Compared with the GBMTC cells, 23 microRNAs were upregulated in the GBMinv cells in at least 5 of the 7 GBM models. Functional validation showed that silencing miR-126 and mir-369-5p, the top two overexpressed microRNA in GBMinv, did not significantly alter cell proliferation but suppressed tumor cell migration (both as neurospheres and as monolayer cultures) in vitro. To confirm their role as drivers of GBM invasion in vivo, purified GBMinv cells were transduced with lentiviruses (silencing miR-126 and mir-369-5p, respectively) and implanted into the brains of NOD/SCID mice. Compared with the untreated parental cells and tumor cells transduced with non-target control lentiviruses, silencing miR-126 or miR-369-5p significantly suppressed GBM invasion. Not only the areas but also the depth of the invasive front were remarkably decreased, exhibiting sharp or well-demarcated margins between tumor and normal mouse brains. In conclusion, we identified a set of microRNAs that are selectively over-expressed in the invasive front of GBM cells and functionally confirmed that miR-126 and miR-369-5p as drivers of GBM invasion.

INTRODUCTION: Despite comprising up to 10% of pediatric CNS tumors, an animal model of intracranial germinoma has not been previously established. Standard of care for germinoma is radiation therapy, with recent evidence that neo-adjuvant chemotherapy may allow reduction of radiation dose, but concern for neuro-cognitive toxicity remains despite the reduction of radiation. A clinically relevant animal model of germinoma would facilitate investigation and discovery of novel agents which may obviate the need for radiation in the future. METHODS: To develop xenograft mouse models, we implanted germinoma tumor cells from two children with metastatic disease, one from a surgical specimen of a cervical metastatic tumor and the second from disseminated tumor cells in the CSF, directly into the right cerebral hemispheres of NOD-SCID mice. RESULTS: The germinoma tumors have been serially passaged in xenografts 5 times, with each passaged tumor maintaining the ability to generate tumors, suggestive of the persistence of viable tumorigenic/tumor stem cells. Both patient tumor samples have whole exome sequencing; one of the two models has a germinoma-associated mutation in kit. This mutation was confirmed by pyrosequencing and was maintained in all 5 xenograft tumor passages. In the kit wildtype model, kit was over-expressed as documented by IHC staining. Histopathologic staining demonstrated that both models maintain similarity with the original patient tumors. Cultured cells were grown as a monolayer from one model, and a cell line was established. CONCLUSIONS: Our orthotopic xenograft murine models of two metastatic pediatric germinomas are the first to be established for this tumor type. These models will allow much needed pre-clinical drug testing of chemotherapeutics and combination therapies as well as further biologic characterization of pediatric germinomas.

INTRODUCTION: Atypical teratoid rhabdoid tumors (ATRTs) are molecularly-defined by deletions or mutations in tumor suppressor INI1. INI1 deficiency causes dysfunction of chromatin-remodeling complexes which normally oppose EZH2, a epigenetic gene silencing protein. In the setting of INI1 deficiency, EZH2 is over-expressed, leading to ATRT oncogenesis. Using novel patient-derived orthotopic xenograft (PDOX) models, we investigated the pre-clinical utility of therapeutic EZH2 inhibition in ATRT using the drug DZNep. METHODS: Two PDOX models of ATRT were established by directly implanting tumor cells into mouse brains in the matched location with patient tumor. For in vitro testing, ATRT cells were exposed to DZNep at doses from 78 to 5000 nM for 2 weeks; cell viability was measured every 1- 3 days. For in vivo testing, 40 mice were divided into 4 treatment groups: control, DZNep (5 mg/kg i.p. daily x 14 days), standard of care (cranial radiation 2 Gy daily x 5 days, cisplatin 5 mg/kg i.p. on days 8 and 11), and combination therapy. Survival times were compared using log-rank analysis. RESULTS: DZNep showed dose- and time-dependent in vitro inhibition of tumor cell viability in both ATRT models tested. One of these models (IC-L1115ATRT) was then subjected to in vivo studies. ATRT tumors were present at the time of xenograft death in all 4 treatment groups including DZNep- and combination therapy-treated animals. Survival time did not differ between the control- and DZNep-treated groups or the standard of care- and combination therapy-treated groups (P > 0.05). CONCLUSIONS: While the EZH2 inhibitor DZNep is effective in vitro in killing ATRT tumor cells, it does not appear to have effects in vivo when tested in PDOX models of pediatric ATRT acting alone or in combination with combined chemo- and radio-therapies. Extensive further pre-clinical evidence should be obtained before an EZH2 inhibitor is brought into pediatric clinical trials.

Glioblastoma (GBM) is the most malignant brain tumor that occurs in both adults and children. Diffuse invasion into normal brain tissue is one of the major features that makes GBM particularly difficult to treat. While direct comparison of matched invasive (GBMINV) and tumor core (GBMTC) cells would facilitate the discovery of molecular drivers of GBM invasion, GBMINV cells are extremely difficult to obtain from patients. To overcome this barrier, 6 patient derived orthotopic xenograft (PDOX) mouse models of pediatric GBM were utilized to isolate GBMTC and GBMINV cells. Global profiling of 768 human microRNA using a real-time PCR-based Taqman system identified 23 microRNAs were upregulated in the GBMINV cells as compared with the matching GBMTC cells. Functional validation was performed by lentivirus-mediated silencing of miR-126, miR-487b and miR-369-5p, the top three overexpressed microRNA in the GBMINV cells. Compared with the control cells, silencing of miR-126, miR-487b or miR-369-5p in GBMINV cells resulted in significant inhibition of cell migration in vitro (without affecting tumor proliferation) (P

Background: Mutation in isocitrate dehydrogenase 1 (IDH1) occurs in > 70% of WHO grades II and III astrocytomas and oligodendrogliomas and secondary glioblastoma (GBM). The mutant enzyme catalyze the reduction of α-ketoglutaric acid to D-2-hydroxyglutaric acid, leading to cancer initiation. In this study, we examined the therapeutic efficacy of SYC-435 (1-hydroxypyridin-2-one), a newly developed mutant IDH1inhibitor, both in vitro and in vivo in IDH1 mutant gliomas as compared with IDH1 wild type GBMs. Methods: An established neurosphere line (BT142) of anaplastic oligoastrocytoma (AOA), and a patient-derived orthotopic xenograft (PDOX) model of recurrent AOA (IC-V0914AOA), and 2 PDOX models of GBM (IC-4687GBM and IC-3752GBM) were included. IDH1 mutations (R132H and R132C) were analyzed by pyrosequencing. To determine the in vitro antitumor activities, tumor cells were exposed to SYC-435 (0.02 to 20 µM) and examining for changes of cell proliferation every 3-4 days till day 13 by Cell Counting Kit-8 assay. For in vivo effects, orthotopic xenograft mouse models of IC-BT142AOA and IC-V0914AOA were treated with vehicle (as control), SYC-435 (i.p., 15 mg/kg/day x 28 days), temozolomide (TMZ, oral, 50 mg/kg/day x 5 days) + fractionated radiation (XRT, 2 Gy/day x 5 days) (as standard therapy), and combination of SYC-435 with standard therapy starting 2 weeks after intracranial tumor implantation. Animal survival times were analyzed by log rank analysis. Results: IDH1 R132H mutation (homozygous) was detected in BT142AOA neurosphere line and R132C mutation (mutant allele frequency 39-42%) in IC-V0914AOA xenograft cells, while the two GBM models (IC-4687GBM and IC-3752GBM) carried wild-type IDH1. Suppression of cell growth was observed in time- and dose-dependent manner by SYC-435, particularly at the IDH1 mutant models. At 0.5 µM, SYC-435 inhibited cell growth by 90% in BT142 and 60% in IC-V0914AOA cells, whereas in IDH wild-type GBMs only by 17% in IC-4687GBM and 19% in IC-3752GBM cells at day 13, indicating the high selectivity of SYC-435 of mutant over wild type IDH1. Systematic in vivo treatment with SYC-435 alone did not alter survival times in neither IC-BT142AOA nor IC-V0914AOA models when compared with the control group. Although standard therapy significantly prolonged animal survival times in both models (P Conclusion: SYC-435 possesses antitumor effects that are highly selective in IDH1 mutant gliomas, and generated strong synergistic activities with standard therapies in vivo. Our data support the clinical testing of SYC-435 in patients with IDH1 mutant glioma. Citation Format: Mari Kogiso, Lin Qi, Huiyuan Zhang, Frank K. Braun, Yuchen Du, Yulun Huang, Holly Lindsay, Sibo Zhao, Sarah G. G. Injac, Zhen Liu, Patricia A. Baxter, Jack M. Su, Laszlo Perlaky, D. Williams Parsons, Murali Chintagumpala, Adekunle Adesina, Jialiang Wang, Yongcheng Song, Xiao-Nan Li. Mutant isocitrate dehydrogenase 1 (IDH1) inhibitor synergistically prolongs animal survival with standard therapies in patient-derived IDH1 mutant glioma xenograft mouse models [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2017; 2017 Apr 1-5; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2017;77(13 Suppl):Abstract nr 4031. doi:10.1158/1538-7445.AM2017-4031

Objective: Glioblastoma (GBM) is one of the deadliest brain tumors both in adults and children. While next-generation sequencing rapidly identifies whole exome/genome gene mutations, the efficacy of targeting mutated genes/pathways has not been systematically analyzed. Additionally, it remains unknown if neurospheres (enriched with cancer stem cells) and non-stem monolayer tumor cells will respond equally to the same target therapies. Methods: Matching pairs of neurosphere and monolayer cell cultures from 4 molecularly characterized PDOX mouse models established from distinct clinical stages of pediatric glioma (treatment-naïve, early recurrent, terminal, and radiation-induced) were treated with 7,913 drugs (mostly in clinical use or trials) at 4 doses (0-10 uM) for 3 and 7 days. Dose-response curves and active area under the curve (AAUC) were constructed to identify active drugs and correlate with model-specific gene mutations. Results: Relatively long-term treatment (7 day screening) was achieved for high-throughput drug screening and revealed time- and dose-dependent effects in most of the 4,629 compounds that were active in at least one culture at one time point. When gene mutations were druggable, not all inhibitors of the same family were active. In IC-4687GBM (a treatment-naïve GBM), high NF1 mutation (>76-99% allele) frequency was found in patient tumor, xenografts, and cultured cells, but only 3/17 MEK inhibitors were active in neurospheres and 2/17 in monolayer cells (GDC0980 and PI-103 were active in both) on day 7. In IC-R0315GBM (from an autopsied terminal GBM) that carried PI3KCA mutation (allele frequency 22-27%), 5/33 PI3K inhibitors were active in neurospheres and 8/33 in monolayer cells (only Tremetinib was active in both) after 7 days. In IC-3752GBM (recurrent GBM) and ICb-1127AA (radiation-induced anaplastic astrocytoma), no druggable mutations were detected. The number of active drugs on day 7 was 366 in IC-4768GBM, 406 in IC-3752GBM, 284 in IC-R0315GBM, and 305 in ICb-1277AA. When the 4 matching pairs of neurospheres and monolayers were compared, the agents active in both cultures ranged from 36% to 60%, active only in neurosphere from 10.3% to 25%, and active only in monolayer cells from 14.8% to 53%. Subsequent in vivo validation using MLN8327 in IC-4687GBM and IC-R0315GBM showed that effective targeting of both neurosphere and monolayer was required for significantly-improved animal survival times. Conclusion: We showed that long-term treatment is feasible for high-throughput drug screening. Targeting druggable mutations can be achieved but only by a fraction of specific agents. Neurospheres and monolayer cells do not always respond equally toward the same drugs, and effective targeting of both subpopulations is needed to generate prolonged animal survival times. Citation Format: Lin Qi, Yuchen Du, Mari Kogiso, Goeun Bae, Frank Braun, Holly Lindsay, Huiyuan Zhang, Sibo Zhao, Sarah Injac, Patricia Baxter, Jack Su, Michael Mancini, Oliver Hampton, William Parsons, Murali Chintagumpala, Clifford Stephan, Peter Davies, Xiao-Nan Li, Xiao-Nan Li. Identification of personalized active agents in pediatric gliomas through high-throughput drug screening in matching pairs of patient derived orthotopic xenograft (PDOX) neurosphere and monolayer cells [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2017; 2017 Apr 1-5; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2017;77(13 Suppl):Abstract nr 4214. doi:10.1158/1538-7445.AM2017-4214

Background: Meningioma is the most common brain tumors in adults. Despite the overall benign nature of meningioma, cranial-base tumors are difficult to achieve complete resection while others exhibit progression and aggressive profiles characterized by high recurrence rates, pleomorphic histology, and resistance to standard treatment. The lack of clinically relevant animal models is blocking the development of novel therapies. Here, we report our establishment of orthotopic xenograft mouse models and in vitro culture systems from surgical specimens of primary and recurrent meningiomas. Material and Methods: 6 primary surgical samples (3 WHO grade I and 3 atypical), and 4 recurrent samples (1 WHO grade I, 2 atypical and 1 anaplastic meningiomas) were obtained from meningioma patients. Tumor tissues were dissociated into single cells and directly implanted into right cranial base of NOD/SCID mice (1x105 cells/mouse). Primary cultures were initiated in serum-free and traditional FBS-based media. Tumor growth was monitored by small animal MRI. Pathologic features of the PDOX models and the matched patient tumors were compared with standard H&E and immunohistochemical staining. Molecular fidelity of PDOX tumor was examined through genomic analysis and comparison with the parental tumors. Results: Three of the 10 tumors were not tumorigenic, and xenograft tumor formation from 5 additional samples is pending. Growth of intracranial (cranial base) xenograft was confirmed in two samples derived from the same patient diagnosed as atypical meningioma (K029MEN) and progressed as anaplastic meningioma at recurrence (K037MEN). These patient derived orthotopic xenografts (PDOX) have since been serially subtransplanted in mouse brains for generation 2 and can be cryopreserved for long-term maintenance of tumorigenicity. The xenograft tumors replicated histopathological features (invasion, high proliferation and increased microvessel density) of their parental tumors. Genomic analysis is being performed to examine the similarities between parental tumors and the corresponding orthotopic xenograft tumors and the discrepancies between the primary and the recurrent tumor-derived models. In vitro growth of K029MEN and K037MEN as neurospheres and monolayer were maintained for 2 months and passage for 10 times. Additionally, cells from K030MEN, a WHO grade I meningioma, has been passaged as monolayer for more than 30 times. Conclusion: A novel set of meningioma PDOX models derived from matching primary and recurrent tumor was established. The xenograft tumors replicated the histopathological and key molecular features of the original patient tumors, providing a unique opportunity to understand the biology of malignant meningiomas and to conduct preclinical drug testing. Note: This abstract was not presented at the meeting. Citation Format: Huiyuan Zhang, Lin Qi, Yuchen Du, Frank K. Braun, Mari Kogiso, Sibo Zhao, Holly B. Lindsay, Sarah G. Injac, Patricia A. Baxter, Jack M. Su, Akash J. Patel, Xiao-nan Li, Baylor College of Medicine. A novel set of patient-derived orthotopic xenograft (PDOX) models of primary and recurrent intracranial meningioma [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2017; 2017 Apr 1-5; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2017;77(13 Suppl):Abstract nr 4812. doi:10.1158/1538-7445.AM2017-4812

Brain tumor is the leading cause of cancer related death in children. To develop new therapies and to understand tumor biology, many efforts are made to develop patient derived orthotopic xenograft mouse models (PDOX). There are, however, some concerns about the potential impact of animal genders on tumor take rate, xenograft growth and in vivo drug responses. To systematically analyze such impact, we retrospectively examined >1,000 mice of 59 PDOX models, including glioblastoma (GBM), medulloblastoma (MB), primitive neuroectodermal tumor (PENT) and ependymoma (EPN) and diffuse intrinsic pontine glioma (DIPG), in which there were age-matched (within 2 wks differences) with relatively even numbers of male and female animals. All the animals were injected with identical number of tumor cells (1x10^5) in the mouse brains matching that in human patients. Animals were monitored daily and euthanized when they were moribund. Differences between male and female mice were analyzed by t-test. In 25 GBM models (15 adult and 10 pediatric) involving 419 mice, there was an average of number of mice was 17.8±9.5 mice per model (male: female = 9.88±5.8:7.93±3.6) (P>0.05). In 26 MB models with total 488 mice, the average animal number was 17.24±8.8 mice per model (male:female= 9.54±5.34 :7.71±3.45) (P>0.05). The tumor take rate was near >95% in both male and female mice (P>0.05). In GBM models, the survival times was 94.1±21.7 days in male mice and 92.3±20.8 days in female (P>0.05); wherease in MB models, they were 134.95±39.86 day in male and 121.08±31.68 days in female mice (P>0.05). The impact of animal gender on drug responses in vivo is essential to the preclinical drug testing. A total of 468 mice from 11 GBM, 5 MB, 3 EPN, 3 DIPG and 2 PNET were treated a series of standard and investigational drugs/compounds, including radiation, oncolytic virus (SVV-001), ABT888, BMI1-inhibitor, MLN8237, TMZ, Flavopiridol, SYC-435, SYC-719, SYC-836, Olig 2 inhibitor, VPA, SAHA, Echinomycin, MCB613, and PARPi. The overall survival times were 104.55±23.63 day in male mice, and 98.64±25.66 in female mice (P>0.05). The only model that exhibited the increased responses in male mice was found in IC-2305GBM treated with TMZ and PTC-596 (P Citation Format: Lin Qi, Mari Kogiso, Yuchen Du, Huiyuan Zhang, Frank Braun, Holly Lindsay, Sibo Zhao, Sarah Injac, Patricia Baxter, Zhigang Liu, Yujing Zhang, Jack Su, Adekunle Adesina, Andrew Walter, Jeffery Murray, Javad Nazarian, Will Parsons, Murali Chintagumpala, Xiao-nan Li. Impact of SCID mouse gender on tumorigenicity, xenograft growth and drug-responses in patient derived orthotopic xenograft models (PDOX) of malignant brain tumors [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2017; 2017 Apr 1-5; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2017;77(13 Suppl):Abstract nr 2821. doi:10.1158/1538-7445.AM2017-2821

10504 Background: A greater understanding of the Ras-MAP kinase-signaling pathway in pediatric low-grade glioma (LGG) paired with the availability of potent selective inhibitors has enhanced the ability to target this pathway with therapeutic intent. Methods: The PBTC conducted a multi-institutional phase II study (NCT01089101) evaluating selumetinib (AZD6244, ARRY-142886), a MEK I/II inhibitor, in children with recurrent/refractory LGG assigned to 6 strata and treated at 25 mg/m2/dose PO BID for up to two years. Here we present the data from three of these strata. The remaining strata are still accruing patients. Results: Stratum I included children with non-NF-1 and non-optic pathway recurrent/refractory pilocytic astrocytoma (PA) harboring BRAF aberrations (BRAF V600e mutation or the BRAF-KIAA 1549 fusion). Eight of 25 (32%) patients achieved a partial response (PR) with 2-year PFS of 66+/-11%. Two of 7 (29%) patient tumors with a BRAF V600e mutation and 6/18 (33%) with a BRAF KIAA-1549 fusion had a PR. Stratum 3 enrolled NF-1-associated LGG. Tissue for tumor BRAF evaluation was not required for eligibility. Ten of 25 (40%) achieved PR with a 2-year PFS of 96+/-4%. Only one patient progressed while on treatment. Stratum 4 included children with non-NF-1 optic pathway/hypothalamic LGG. Tissue for tumor BRAF evaluation was not required for eligibility. Two of 16 (12.5%) had a PR with a 2-year PFS of 65+/-13%. The BRAF aberration status of the responders in strata 3 and 4 is mostly unknown. All responses were confirmed centrally and seven patients remain on treatment. The most common toxicities were grade 1/2 CPK elevation, diarrhea, hypoalbuminemia, elevated AST and rash. Rare grade 3/4 toxicities included elevated CPK, rash, neutropenia, emesis and paronychia. Conclusions: Selumetinib was effective in treating children with recurrent/refractory LGG, including those with NF-1 associated LGG and PA harboring BRAF V600e mutation or BRAF-KIAA 1549 fusion. Larger prospective studies are necessary to determine the future, specific role of this agent in treating children with LGG harboring specific molecular aberrations. Clinical trial information: NCT01089101.

10543 Background: CNS malignancies are the most common solid tumors among children. Novel therapies are needed to help improve the survival outcomes in children with recurrent disease. Pomalidomide is an immunomodulatory agent thought to also function through a combination of anti-angiogenic, anti-inflammatory and cytotoixic activity making it a good candidate to explore in pediatric CNS tumors. Methods: A Phase I trial of pomalidomide was conducted among children ≥ 3 to < 21 years old with recurrent, progressive/refractory CNS tumors using the “rolling 6” dose escalation design. The primary objective was to determine the maximum tolerated dose (MTD) and/or recommended Phase II dose (RP2D) when given orally once daily for 21 consecutive days of a 28-day course. Once the MTD was established, 12 additional patients were enrolled on expansion cohorts based on age and steroid use. Results: 29 children were enrolled and 25 were evaluable for dose limiting toxicity (DLT) evaluation. The MTD was 2.6 mg/m2 (Dose level 2). DLTs observed at dose level 3 (3.4 mg/m2) included diarrhea (n = 1), thrombocytopenia (n = 1) and lung infection (n = 1), all grade 3. The most common toxicities were grade 1 lymphopenia (55%), leukopenia (62%) and thrombocytopenia (38%). There were no obvious differences in tolerability based on age or steroid use. Pharmacokinetics were similar to those observed in adults and increased in a dose-dependent manner. At the RP2D of 2.6 mg/m2, the Cmax was 97.4 ng/mL and the t½ was 4.1 hours. The median number of treatment cycles was 1.6 (0.2-12.3). Two patients, one with an oligodendroglioma and one with anaplastic pleomorphic xanthoastrocytoma, had long term stable disease for 9 and 18+ cycles. No objective responses were observed. Twelve month progression-free and overall-survivals were 5.2+/-3.6% and 12.8+/-8.5%, respectively. Immunologic correlate analyses are ongoing. Conclusions: The RP2D of pomalidomide is 2.6 mg/m2 in children with recurrent brain tumors. Further prospective evaluation of this agent alone or in combination will be necessary to better understand its efficacy in specific pediatric CNS tumor populations. Clinical trial information: NCT02415153.

Diffuse invasion into normal brain is one of the hallmark features that make GBM difficult to treat. Due to the lack of biologically accurate invasive GBM cells from patients, most of the existing studies on GBM invasion were conducted in surgical samples that were primarily tumor core tissues. Although cancer stem cells are critical in tumor initiation and therapy-resistance, their role in GBM invasion has not been well understood. To identify the cancer stem cell subpopulations that drive GBM invasion, To solve we utilized a panel of 7 (6 pediatric and 1 adult) patient tumor-derived orthotopic xenograft (PDOX) mouse models to isolate matched pairs of invasive GBM cells (from the "normal" mouse brains) and tumor core GBM cells and directly compared their stem cell features. The invasive cells exhibited stronger neurosphere forming efficiency in vitro and displayed higher tumorigenic capacity in vivo (particularly at 100 cells/mouse). A systemic screening of putative cancer stem cell markers (CD133, CD15, CD24/CD44, CD57 and CD117) showed that GBM cells in the invasive front were enriched with CD57+ cells (>2 folds than the tumor core), and most of them were CD57+ CD133-. Direct implantation of FACS purified CD57 + /CD133- cells (100 - 10,000 per mouse) into the brains of NOD/SCID mice confirmed their tumor forming capacity in vivo. We also showed that CD57+ cells expressed high levels of self-renewal genes and tend to stay in G0/G1 phases. Additionally, we found that CD133+ cells were frequently dual-positive with CD57 (CD33 + CD57+) both in the xenograft tumors and in patient GBMs. In conclusion, we identified CD57 as a novel GBM stem cell marker and showed that CD57+ GBM cells are particular enriched in the invasive front of GBMs. Our findings suggest that new anti-invasion therapies should be developed to target CD57+ cells in addition to CD133+ cells.