Identification of Let-7 miRNA Activity as a Prognostic Biomarker of SHH Medulloblastoma

Medulloblastoma (MB) is the most common pediatric embryonal brain tumor. The current consensus classifies MB into four molecular subgroups: sonic hedgehog-activated (SHH), wingless-activated (WNT), Group 3, and Group 4. MYCN and let-7 play a critical role in MB. Thus, we inferred the activity of miRNAs in MB by using the ActMiR procedure. SHH-MB has higher MYCN expression than the other subgroups. We showed that high MYCN expression with high let-7 activity is significantly associated with worse overall survival, and this association was validated in an independent MB dataset. Altogether, our results suggest that let-7 activity and MYCN can further categorize heterogeneous SHH tumors into more and less-favorable prognostic subtypes, which provide critical information for personalizing treatment options for SHH-MB. Comparing the expression differences between the two SHH-MB prognostic subtypes with compound perturbation profiles, we identified FGFR inhibitors as one potential treatment option for SHH-MB patients with the less-favorable prognostic subtype.

EXTH-69. FUNCTIONAL GENOMICS UNCOVER GENETIC DEPENDENCIES IN ATRTS

Brain tumors are the leading cause of cancer-related deaths in children. Embryonal brain tumors including medulloblastoma and atypical teratoid rhabdoid tumors (ATRTs) account for 15% of all primary brain and CNS tumors under the age of 14 years, with ATRTs being most prevalent in infants. Despite intensive research efforts, survival estimates for ATRT patients stay relatively low as compared to other tumor entities with a median survival of around 17 months. We here describe genome-wide CRISPR/Cas9 knockout screens in combination with small-molecule drug assays to identify targetable vulnerabilities in ATRTs. Based on functional genomic screening revealing ATRT context-specific genetic vulnerabilities (n = 671 genes), we successfully generated a small-molecule library that shows preferential activity in ATRT cells as compared to a broad selection of other human cancer cell lines. Of note, none of these drugs differentially affect ATRT cells from distinct molecular subgroups, suggesting that top candidate inhibitors might serve as pan-ATRT therapeutic avenues. CDK4/6 inhibitors, among the most potent drugs in our library, are capable of inhibiting tumor growth due to mutual exclusive dependency of ATRTs on either CDK4 or CDK6. Our approach might serve as a blueprint for fostering the identification of functionally-instructed therapeutic strategies in other incurable diseases beyond ATRT, whose genomic profiles also lack actionable alterations so far.

Maximizing the potential of aggressive mouse tumor models in preclinical drug testing

Atypical teratoid rhabdoid tumor (ATRT) is an aggressive embryonal brain tumor among infants and young children. Two challenges exist for preclinical testing in ATRT. First, genetically quiet, ATRT is a difficult tumor to target molecularly. Tumor cells need to divide to propagate tumor growth—intercepting the common crossroads in cell cycle progression is a feasible strategy. KIF11 is needed for bipolar spindle formation in metaphase. We identified KIF11 as a universal target of all ATRT-molecular-subtypes. Ispinesib, a KIF11-inhibitor, effectively inhibited tumor proliferation in all seven cell lines. A second challenge—a major challenge in preclinical drug testing in-vivo among aggressive tumor models, is the narrow therapeutic window to administer drugs within the limited murine lifespan. Our most aggressive ATRT tumor model was lethal in all mice within ~ 1 month of tumor implantation. Such short-surviving mouse models are difficult to employ for preclinical drug testing due to the narrow time window to administer drugs. To overcome this time restriction, we developed a clinical staging system which allowed physically-fit mice to continue treatment, in contrast to the conventional method of fixed drug-dose-duration regimen in preclinical testing which will not be feasible in such short-surviving mouse models. We validated this approach in a second embryonal brain tumor, medulloblastoma. This is a clinically relevant, cost-efficient approach in preclinical testing for cancer and non-cancer disease phenotypes. Widely used preclinical mouse models are not the most accurate and lack the aggressive tumor spectrum found within a single tumor type. Mice bearing the most aggressive tumor spectrum progress rapidly in the limited murine life-span, resulting in a narrow therapeutic window to administer drugs, and are thus difficult to employ in preclinical testing. Our approach overcomes this challenge. We discovered ispinesib is efficacious against two embryonal brain tumor types.

PATH-08. THE IMPORTANCE OF RE-DIAGNOSIS OF TUMORS PREVIOUSLY CLASSIFIED AS CENTRAL NERVOUS SYSTEM PRIMITIVE NEUROECTODERMAL TUMORS

BACKGROUND The recent molecular analyses have revealed that central nervous system primitive neuroectodermal tumors (CNS PNETs) those having clusters of small round tumor cells are genetically different tumors. However, the concepts of CNS PNET are complicated, and it is difficult to diagnose them appropriately in clinical field. To overcome this difficulty, we reviewed previous studies associated with CNS PNETs, and carried out several approaches, those are relatively easy access to use in clinics, for our 8 samples of embryonal brain tumors diagnosed CNS PNETs in our institution, initially. METHODS We used in combination with immunohistochemistry (IHC), Sanger sequence, Pyrosequence, polymerase chain reaction (PCR), real time PCR and copy number analysis referring recent reports. RESULTS In terms of the diagnosis three out of 8 cases were changed based on the results in this study from previous diagnoses. CONCLUSION In this review, it seemed that either the histopathological evaluation or molecular analyses would be not enough to make accurate diagnosis of CNS embryonal brain tumors, and it is essential to combine both of them including recent comprehensive analysis methods.

PATH-24. MOLECULAR CLASSIFICATION OF HIGH RISK INFANT EMBRYONAL BRAIN TUMORS ENROLLED IN THE ACNS0334 TRIAL: A REPORT FROM THE CHILDREN’S ONCOLOGY GROUP

Young children with embryonal brain tumors including medulloblastoma (MB), supratentorial primitive neuro-ectodermal tumor, or pineoblastoma have historically been considered high-risk patients with poor outcomes despite the use of intensive radiation-sparing treatment. In the ACNS0334 phase III trial, 91 consented children <36 months old with the above diagnoses were randomized to intensive induction chemotherapy with or without methotrexate followed by consolidation with stem cell rescue. Here we present the results of a centralized integrated molecular analysis including global methylation profiling (65/91), and whole exome sequencing of tumor (46/91) and germline (35/91) DNA. Unsupervised clustering analyses of methylation profiles using multiple orthogonal methods against a reference dataset of 1200 pediatric brain tumors, revealed known and new molecular entities. For tumors diagnosed as MB on central pathology review, 7.3% (3/41) had a non-MB molecular diagnosis (2 embryonal tumor with multiple rosettes/ETMR, 1 group MYC pineoblastoma), with the remainder as MB Group SHH (11/41), Group3 (25/41), and Group4 (2/41). Among histologic non-MBs, 3/24 (12.5%) were molecular entities not intended for trial inclusion (1 each for ATRT, pleomorphic xanthoastrocytoma, and high-grade glioma). ETMR, historically considered a rare entity, was molecularly identified in a significant proportion (14/65; 21.5%) of samples. Among MB-SHH, we detected deleterious PTCH1 mutations in 6/9 tumors but none among 5 germline samples tested; a germline SUFU frameshift mutation with tumor LOH was also observed in MB-SHH. Correlation of these and other molecular features to the parallel clinical analysis will yield important markers of risk stratification and predictors of treatment response.

MBRS-29. PROSPECTIVE MOLECULAR PROFILING IN PEDIATRIC MEDULLOBLASTOMA PATIENTS ENROLLED ON THE “HEAD START 4” PROTOCOL

Medulloblastoma is the most common malignant embryonal brain tumor in children with only modest improvements in outcomes achieved over the last 20 years. The implementation of irradiation-avoiding strategies, including trials by the “Head Start” consortium, have demonstrated improved cure rates along with enhanced quality of life. Simultaneously, the classification of medulloblastomas has undergone a dramatic shift as molecular testing has made it possible to divide these tumors into distinctive subtypes. Currently, the WHO recognizes four medulloblastoma molecular subgroups; however it remains unclear how patients within these subgroups respond to modern irradiation-avoiding therapies. This study aims to demonstrate the feasibility of prospective molecular profiling in medulloblastoma patients enrolled on the “Head Start 4” trial. Whole-exome sequencing (SureSelect Human All Exon V6+COSMIC) and DNA methylation (Illumina EPIC Array) profiling were performed on 10 paired tumor/blood samples and 4 tumor samples, respectively. High-quality mutational and copy number data were produced for each of the 10 subjects demonstrating well-described gene mutations (SUFU) and chromosomal losses (9q and 10q). Four subjects had methylation profiling which successfully separated them into the WHO subgroups (two SHH and two Group 3). These data showed the feasibility of prospective high-dimensional mutational and DNA methylation analysis using “Head Start 4” patients. Future work will focus on finalizing these profiling efforts, enabling the development of models that predict response to irradiation-avoiding treatment and, in general, a better understanding of the molecular mechanisms underlying treatment resistance and tumor progression, leading to more personalized approaches to treating children with medulloblastoma.

TBIO-02. IMMUNE PROFILING OF RARE EMBRYONAL BRAIN TUMORS REVEAL EVIDENCE OF DYSREGULATED INTERFERON SIGNALLING AS A POTENTIAL DETERMINANT OF IMMUNOLOGICAL HETEROGENEITY

Embryonal brain tumors (EBTs) remain the most common malignant pediatric brain tumors. Despite recent advances and improved understanding of the molecular biology of EBTs, clinical outcomes remain poor for rare EBTs. Previous large-scale genomic studies of rare EBTs have shed light on distinct genomic, transcriptomic and epigenomic profiles. Interestingly, these studies have revealed prominent tumor heterogeneity that provides opportunity to develop novel treatment strategies to improve patient outcomes. To examine the tumor microenvironment and identify tumor- specific biological dependencies, we performed deconvolution analysis of bulk gene expression (171 RNA-seq, 236 microarrays) and 586 methylation arrays, which revealed significant intra and inter-tumoral heterogeneity and implicated interferon (IFN)-mediated signalling as a determinant of a distinct immunological profile in rare EBTs. To further elucidate the importance of IFN signalling, we performed scRNA-seq on 20 primary samples, which provided evidence of a spectrum of IFN-immunological responses that vary from immunosuppressive to immunologically exhaustive that occur in a host dependent manner. To further validate our findings, we utilised a genetically engineered murine model of Atypical Teratoid Rhabdoid Tumor and primary xenografts in humanised mice to corroborate our in-silico profiles in vivo. Through amalgamation of our in-silico data with our in vivo data, we have identified evidence that dysregulated IFN responses represent a core element of the immunological heterogeneity present within subsets of rare EBTs. An improved understanding of the immune milieu in rare EBTs will provide avenues to develop specific onco-immune targets to address this clinical need.