Search

Epidermal growth factor receptor (EGFR) is frequently amplified, mutated and overexpressed in malignant gliomas. Our investigation of the proximal and distal responses to EGFR inhibition identified the molecular mechanisms for the therapeutic resistance of EGFR inhibition in gliomas. Oncogenic EGFRviii-dependent gliomas initially showed cytotoxic responses upon removal of oncogenic cue. However, during the initial regression, a subset of tumor cells with activated mesenchymal subtype gene expression program emerged by lineage reprogramming which promoted tumor relapse in the absence of EGFRviii signaling. This lineage switch is dependent on YAP1 activation as a response to therapy, and is a key to EGFRviii independent tumor growth. YAP signature stratifies overall survival of recurrent glioma patients. Inhibition of YAP1 activation suppressed mesenchymal gene expression and significantly delayed recurrence of gliomas. Our findings provide mechanisms underlying inefficacies of EGFR targeted therapy in glioblastoma and suggest a new combinatorial targeting of EGFR and YAP for deeper and more durable responses.

First-order radiomic features, such as metabolic tumor volume (MTV) and total lesion glycolysis (TLG), are associated with disease progression in early-stage classical Hodgkin lymphoma (HL). We hypothesized that a model incorporating first- and second-order radiomic features would more accurately predict outcome than MTV or TLG alone. We assessed whether radiomic features extracted from baseline PET scans predicted relapsed or refractory disease status in a cohort of 251 patients with stage I-II HL who were managed at a tertiary cancer center. Models were developed and tested using a machine-learning algorithm. Features extracted from mediastinal sites were highly predictive of primary refractory disease. A model incorporating 5 of the most predictive features had an area under the curve (AUC) of 95.2% and total error rate of 1.8%. By comparison, the AUC was 78% for both MTV and TLG and was 65% for maximum standardize uptake value (SUVmax). Furthermore, among the patients with refractory mediastinal disease, our model distinguished those who were successfully salvaged from those who ultimately died of HL. We conclude that our PET radiomic model may improve upfront stratification of early-stage HL patients with mediastinal disease and thus contribute to risk-adapted, individualized management.

We have previously reported that overexpression of Programmed Death -1 Homolog (PD-1H) in human monocytes leads to activation and spontaneous secretion of multiple pro inflammatory cytokines. Here we evaluate changes in monocytes gene expression after enforced PD-1H expression by gene array. The results show that there are significant alterations in 51 potential candidate genes that relate to immune response, cell adhesion and metabolism. Genes corresponding to pro-inflammatory cytokines showed the highest upregulation, 7, 3.2, 3.0, 5.8, 4.4 and 3.1 fold upregulation of TNF-α, IL-1 β, IFN-α, γ, λ and IL-27 relative to vector control. The data are in agreement with cytometric bead array analysis showing induction of proinflammatory cytokines, IL-6, IL-1β and TNF-α by PD-1H. Other genes related to inflammation, include transglutaminase 2 (TG2), NF-κB (p65 and p50) and toll like receptors (TLR) 3 and 4 were upregulated 5, 4.5 and 2.5 fold, respectively. Gene set enrichment analysis (GSEA) also revealed that signaling pathways related to inflammatory response, such as NFκB, AT1R, PYK2, MAPK, RELA, TNFR1, MTOR and proteasomal degradation, were significantly upregulated in response to PD-1H overexpression. We validated the results utilizing a standard inflammatory sepsis model in humanized BLT mice, finding that PD-1H expression was highly correlated with proinflammatory cytokine production. We therefore conclude that PD-1H functions to enhance monocyte activation and the induction of a pro-inflammatory gene expression profile.

Glioblastoma (GBM), the most malignant of primary brain tumors, is a devastating and deadly disease, with a median survival of 14 months from diagnosis, despite standard regimens of radical brain tumor surgery, maximal safe radiation, and concomitant chemotherapy. GBM tumors nearly always re-emerge after initial treatment and frequently display resistance to current treatments. One theory that may explain GBM re-emergence is the existence of glioma stem-like cells (GSCs). We sought to identify variant protein features expressed in low passage GSCs derived from patient tumors. To this end, we developed a proteomic database that reflected variant and non-variant sequences in the human proteome, and applied a novel retrograde proteomic workflow, to identify and validate the expression of 126 protein variants in 33 glioma stem cell strains. These newly identified proteins may harbor a subset of novel protein targets for future development of GBM therapy.

Although mesenchymal stem cells (MSCs) have been implicated as stromal components of several cancers, their ultimate contribution to tumorigenesis and their potential to drive cancer stem cells, particularly in the unique microenvironment of human brain tumors, remain largely undefined. Consequently, using established criteria, we isolated glioma-associated-human MSCs (GA-hMSCs) from fresh human glioma surgical specimens for the first time. We show that these GA-hMSCs are nontumorigenic stromal cells that are phenotypically similar to prototypical bone marrow-MSCs. Low-passage genomic sequencing analyses comparing GA-hMSCs with matched tumor-initiating glioma stem cells (GSCs) suggest that most GA-hMSCs (60%) are normal cells recruited to the tumor (group 1 GA-hMSCs), although, rarely (10%), GA-hMSCs may differentiate directly from GSCs (group 2 GA-hMSCs) or display genetic patterns intermediate between these groups (group 3 GA-hMSCs). Importantly, GA-hMSCs increase proliferation and self-renewal of GSCs in vitro and enhance GSC tumorigenicity and mesenchymal features in vivo, confirming their functional significance within the GSC niche. These effects are mediated by GA-hMSC-secreted interleukin-6, which activates STAT3 in GSCs. Our results establish GA-hMSCs as a potentially new stromal component of gliomas that drives the aggressiveness of GSCs, and point to GA-hMSCs as a novel therapeutic target within gliomas. Stem Cells 2015;33:2400–2415

ObjectThe authors' institution previously reported a 69% rate of crude local control for surgical management of lateral ventricle metastases at the University of Texas MD Anderson Cancer Center. For comparison, the authors here report their institutional experience with use of stereotactic radiosurgery (SRS) to treat intraventricular metastases.MethodsTo identify patients with intraventricular metastases for this retrospective review, the authors queried an institutional SRS database containing the medical records of 1962 patients with 5800 brain metastases who consecutively underwent SRS from June 2009 through October 2013. End points assessed were local control (crude and locoregional), distant failure–free survival, progression-free survival, and overall survival.ResultsOf the 1962 records examined, those for 25 (1.3%) patients with 30 (0.52%) intraventricular metastases were identified. Median patient age at SRS was 55.8 years. The most common primary malignancy was renal cell carcinoma (n = 13), followed by melanoma (n = 7) and breast adenocarcinoma (n = 5). Median tumor volume was 0.75 cm3 (range 0.01–5.6 cm3). Most lesions were located in the lateral ventricles (n = 25, 83.3%) and were treated to a median dose of 20 Gy (range 14–20 Gy). A total of 12 (48%) patients received whole-brain radiation therapy, most (n = 10) before SRS. With a median follow-up of 11.4 months (range 1.6–39.2 months), the rate of crude local control was 93.3%, and the rates of 6-month and 1-year actuarial locoregional control were 85.2% and 56.2%, respectively. The median overall survival time after SRS was 11.6 months (range 1.3–38.9 months), and the 6-month and 1-year actuarial rates were 87.1% and 46.7%, respectively. Disease dissemination developed in 7 (28%) patients as a second intraventricular metastatic lesion (n = 3, 12%), leptomeningeal disease (n = 3, 12%), or both (n = 1, 4%). Radiographic changes developed in 5 (20%) patients and included necrosis (n = 2, 8%) and hemorrhage (n = 3, 12%). A primary diagnosis of renal cell carcinoma was associated with an improved rate of distant failure–free survival (p = 0.05) and progression-free survival (p = 0.08).ConclusionsSRS provides excellent local control for intraventricular metastases, with acceptable treatment-related toxicity, thereby supporting nonsurgical treatment for these lesions. The propensity for intraventricular dissemination among intraventricular metastases seems to be histologically dependent.

The telomere reverse transcriptase promoter (TERTp) is frequently mutated in malignant gliomas, particularly glioblastomas (GBM, 85%), and is involved in maintaining telomere length. TERTp mutations are prognostic for patient survival. Given the evolving use of Infinium DNA methylation arrays in assaying biomarkers in glioma, we aimed to develop an accurate predictor of TERTp mutations in gliomas using this platform. We analyzed TCGA lower grade glioma (LGG) and GBM available with both TERTp sequencing and Infinium 450k data. Samples were split into training and testing set to determine predictor accuracy. Probes with potential quality issues were excluded and probe selection performed using an elastic net algorithm. Probes were selected across the genome, not limited to the TERTp locus. A logistic regression model with binomial distribution was built using the final optimized model, constructed based on 1000 CpG probes, all distinct from TERTp. Prediction accuracy was 100% (179/179, AUC=1) in the training set and 97.5% (116/119) in the test set. Comparison of the normalized frequency of selected probes by chromosome revealed enrichment for chromosomes 20 (0.4987%) and 18 (0.4859%). The majority of probes enriched in CpG islands (53.6%), followed by open sea (22.9%), shore (17.4%), and shelf (6.1%). Relative to known genes, probe enrichment occurred predominantly near transcriptional start sites (within 200–1500 bp, 31.0%), followed by gene bodies (26.3%) and intergenic regions (18.3%). Quite unexpectedly, probes within the predictor do not map near TERT or other telomere maintenance factors nor fell within known methylation signatures, such as G-CIMP, suggesting that TERTp mutation is associated with novel epigenetic changes. Moreover, this biomarker permits rapid and cost effective detection of TERTp mutation using the widely used Infinium platform.

Glioblastoma (GBM) is a primary brain cancer that contains populations of stem-like cancer cells (GSCs) that home to specialized perivascular niches comprised of vascular endothelial cells, pericytes and extracellular matrix (ECM)-rich basement membranes. Interactions with their niche influence GSC self-renewal, differentiation and drug resistance, although the pathways underlying these events remain largely unknown. Here, we report that the adhesion receptor alphavbeta8 integrin and its ECM-bound latent transforming growth factor beta1 (TGFbeta1) protein ligand play central roles in promoting niche co-option and GBM initiation. alphavbeta8 integrin is highly expressed in GSCs and is essential for self-renewal and lineage commitment in vitro. Inhibition of alphavbeta8 integrin adhesion and signaling using function-blocking antibodies inhibit sphere formation in vitro. Fractionation of beta8high cells from freshly resected human GBM samples also reveals a requirement for this integrin in sphere formation in vitro and tumor growth and invasion in vivo. In contrast, fractionated beta8low GBM cells fail to form spheroids in vitro and do not generate tumors after intracranial implantation. Whole transcriptome sequencing of primary GBM cells reveals that alphavbeta8 integrin promotes tumor initiation in the brain, in large part, by driving TGFbeta1-induced DNA replication and mitotic checkpoint progression. We show that beta8high cells activate latent TGFbeta1 in the microenvironment, resulting in enhanced TGFbeta receptor signaling in beta8low GBM cells and tumor endothelial cells via Smad transcription factors. Collectively, these data identify the alphavbeta8 integrin-TGFbeta1 signaling axis as crucial for exploitation of the perivascular niche and identify potential therapeutic targets for inhibiting tumor growth and progression in patients with GBM.

Glioblastoma expression subtypes have been previously been associated with genomic abnormalities, treatment response, and differences in tumor microenvironment. We leveraged IDH wild-type glioblastomas, derivative neurospheres, and single cell gene expression profiles to define three tumor-intrinsic transcriptional subtypes designated as proneural, mesenchymal, and classical, a revision of the previously reported TCGA subtypes. Transcriptomic subtype multiplicity correlated with increased intratumoral heterogeneity and the presence of tumor microenvironment. In silico cell sorting identified macrophages/microglia, CD4+ T lymphocytes, and neutrophils in the glioma microenvironment. NF1 deficiency resulted in increased tumor-associated macrophages/microglia infiltration. Comparison of matching primary and recurrent gliomas elucidated treatment-induced phenotypic tumor evolution, including expression subtype switching, in 45% of our cohort as well as associations between microenvironmental components and treatment response. Gene signature-based tumor microenvironment inference revealed a decrease in invading monocytes and a subtype-dependent increase in macrophages/microglia cells upon disease recurrence. Hypermutation at diagnosis or at recurrence was associated with CD8+ T cell enrichment. Frequency of M2 macrophage detection was associated with short-term relapse after radiation therapy. Our study provides a comprehensive transcriptional and cellular landscape of IDH wild-type glioblastoma during treatment modulated tumor evolution. Characterization of the evolving glioblastoma transcriptome and tumor microenvironment aids in designing more effective immunotherapy trials.

Patients with glioblastoma (GBM) have a universally poor prognosis and are in urgent need of effective treatment strategies. Recent advances in sequencing techniques unraveled the complete genomic landscape of GBMs and revealed profound heterogeneity of individual tumors even at the single cell level. Genomic profiling has detected epidermal growth factor receptor (EGFR) gene alterations in more than half of GBMs. Major genetic events include amplification and mutation of EGFR. Yet, treatment strategies targeting EGFR have thus far failed in clinical trials. In this review, we discuss the clonal and functional heterogeneity of EGFRs in GBM development and critically reassess the potential of EGFRs as therapeutic targets.

An integrated genomic and functional analysis to elucidate DNA damage signaling factors promoting self-renewal of glioma stem cells (GSCs) identified proliferating cell nuclear antigen (PCNA)-associated factor (PAF) up-regulation in glioblastoma. PAF is preferentially overexpressed in GSCs. Its depletion impairs maintenance of self-renewal without promoting differentiation and reduces tumor-initiating cell frequency. Combined transcriptomic and metabolomic analyses revealed that PAF supports GSC maintenance, in part, by influencing DNA replication and pyrimidine metabolism pathways. PAF interacts with PCNA and regulates PCNA-associated DNA translesion synthesis (TLS); consequently, PAF depletion in combination with radiation generated fewer tumorspheres compared with radiation alone. Correspondingly, pharmacological impairment of DNA replication and TLS phenocopied the effect of PAF depletion in compromising GSC self-renewal and radioresistance, providing preclinical proof of principle that combined TLS inhibition and radiation therapy may be a viable therapeutic option in the treatment of glioblastoma multiforme (GBM).

Glioblastoma (GBM) is a primary brain cancer that contains populations of stem-like cancer cells (GSCs) that home to specialized perivascular niches. GSC interactions with their niche influence self-renewal, differentiation and drug resistance, although the pathways underlying these events remain largely unknown. Here, we report that the integrin αvβ8 and its latent transforming growth factor β1 (TGFβ1) protein ligand have central roles in promoting niche co-option and GBM initiation. αvβ8 integrin is highly expressed in GSCs and is essential for self-renewal and lineage commitment in vitro. Fractionation of β8high cells from freshly resected human GBM samples also reveals a requirement for this integrin in tumorigenesis in vivo. Whole-transcriptome sequencing reveals that αvβ8 integrin regulates tumor development, in part, by driving TGFβ1-induced DNA replication and mitotic checkpoint progression. Collectively, these data identify the αvβ8 integrin-TGFβ1 signaling axis as crucial for exploitation of the perivascular niche and identify potential therapeutic targets for inhibiting tumor growth and progression in patients with GBM.

Fusion genes, particularly those involving kinases, have been demonstrated as drivers and are frequent therapeutic targets in cancer1. Here, we describe our results on detecting transcript fusions across 33 cancer types from The Cancer Genome Atlas (TCGA), totaling 9,966 cancer samples and 648 normal samples2. Preprocessing, including read alignment to both genome and transcriptome, and fusion detection were carried out using a uniform pipeline3. To validate the resultant fusions, we also called somatic structural variations for 561 cancers from whole genome sequencing data. A summary of the data used in this study is provided in Table S1. Our results can be accessed per our portal at http://www.tumorfusions.org.

We leveraged IDH wild-type glioblastomas, derivative neurospheres, and single-cell gene expression profiles to define three tumor-intrinsic transcriptional subtypes designated as proneural, mesenchymal, and classical. Transcriptomic subtype multiplicity correlated with increased intratumoral heterogeneity and presence of tumor microenvironment. In silico cell sorting identified macrophages/microglia, CD4+ T lymphocytes, and neutrophils in the glioma microenvironment. NF1 deficiency resulted in increased tumor-associated macrophages/microglia infiltration. Longitudinal transcriptome analysis showed that expression subtype is retained in 55% of cases. Gene signature-based tumor microenvironment inference revealed a decrease in invading monocytes and a subtype-dependent increase in macrophages/microglia cells upon disease recurrence. Hypermutation at diagnosis or at recurrence associated with CD8+ Tcell enrichment. Frequency of M2 macrophages detection associated with short-term relapse after radiation therapy.

// Keeratikarn Boonyawan 1,3 , Kenneth R. Hess 2 , Jie Yang 3 , Lihong Long 3 , Qianghu Wang 3 , Ravesanker Ezhilarasan 3 , Alessandra Auia 4 , Kristin D. Alfaro-Munoz 5 , John F. de Groot 5 , Krishna P. Bhat 4 and Erik P. Sulman 3 1 Department of Radiation Oncology, Ramathibodi Hospital, Mahidol University, Bangkok, Thailand 2 Department of Biostatistics, The University of Texas MD Anderson Cancer Center, Houston, Texas, USA 3 Department of Radiation Oncology, The University of Texas MD Anderson Cancer Center, Houston, Texas, USA 4 Department of Translational Molecular Pathology and Neurosurgery, The University of Texas MD Anderson Cancer Center, Houston, Texas, USA 5 Department of Neuro-Oncology, The University of Texas MD Anderson Cancer Center, Houston, Texas, USA Correspondence to: Erik P. Sulman, email: // Keywords : glioblastoma, platelets, thrombocytosis, prognosis, survival Received : June 05, 2017 Accepted : August 07, 2017 Published : October 12, 2017 Abstract Background: Thrombocytosis is triggered by and promotes tumor growth. The relationship between the change in circulating platelets after chemoradiation therapy (CRT) or adjuvant temozolomide (TMZ) and survival in glioblastoma remains unclear. We hypothesized that an increase in platelets after these treatments would be predictive of a shorter survival. Methods: We retrospectively reviewed data on 122 patients with newly diagnosed, pathologically proven glioblastoma who had been treated with surgery, followed by CRT and adjuvant TMZ, from 2007 to 2016. The association between the changes in blood count levels and survival was analyzed by the log-rank test. To adjust for confounding, we performed a multivariate analysis using known prognostic co-variates. Results: Patients were dichotomized on the basis of the relative change in platelets after CRT from the baseline: ≤30% increase, low ( n = 101) vs >30% increase, high ( n = 12). The median survival for high vs . low platelets were 11 vs 28 months ( p = 0.0062). No significant survival differences were observed on the basis of platelet changes during adjuvant TMZ. Similarly, changes in lymphocyte counts were not significantly prognostic. On multivariate analysis, MGMT, performance status, and an increase in platelets after CRT were significantly associated with survival (HR for platelets, 4.5; 95% confidence interval, 1.6-12.6). Conclusions: Increased platelet counts after CRT are predictive of poor survival in glioblastoma. The effect is platelet specific and does not reflect bone marrow changes, as lymphocyte changes were not significantly prognostic. These results suggest an interaction between platelets and tumor aggressiveness. Thus, platelets serve as a novel, minimally invasive liquid biopsy for predicting outcome.

Cancer cells survive cellular crisis through telomere maintenance mechanisms. We report telomere lengths in 18,430 samples, including tumors and non-neoplastic samples, across 31 cancer types. Telomeres were shorter in tumors than in normal tissues and longer in sarcomas and gliomas than in other cancers. Among 6,835 cancers, 73% expressed telomerase reverse transcriptase (TERT), which was associated with TERT point mutations, rearrangements, DNA amplifications and transcript fusions and predictive of telomerase activity. TERT promoter methylation provided an additional deregulatory TERT expression mechanism. Five percent of cases, characterized by undetectable TERT expression and alterations in ATRX or DAXX, demonstrated elongated telomeres and increased telomeric repeat-containing RNA (TERRA). The remaining 22% of tumors neither expressed TERT nor harbored alterations in ATRX or DAXX. In this group, telomere length positively correlated with TP53 and RB1 mutations. Our analysis integrates TERT abnormalities, telomerase activity and genomic alterations with telomere length in cancer.

Novel proteoforms with single amino acid variations represent proteins that often have altered biological functions but are less explored in the human proteome. We have developed an approach, searching high quality shotgun proteomic data against an extended protein database, to identify expressed mutant proteoforms in glioma stem cell (GSC) lines. The systematic search of MS/MS spectra using PEAKS 7.0 as the search engine has recognized 17 chromosome 19 proteins in GSCs with altered amino acid sequences. The results were further verified by manual spectral examination, validating 19 proteoforms. One of the novel findings, a mutant form of branched-chain aminotransferase 2 (p.Thr186Arg), was verified at the transcript level and by targeted proteomics in several glioma stem cell lines. The structure of this proteoform was examined by molecular modeling in order to estimate conformational changes due to mutation that might lead to functional modifications potentially linked to glioma. Based on our initial findings, we believe that our approach presented could contribute to construct a more complete map of the human functional proteome.

An increasing number of experiments have been designed to detect intracellular and intercellular molecular interactions. Based on these molecular interactions (especially protein interactions), molecular networks have been built for using in several typical applications, such as the discovery of new disease genes and the identification of drug targets and molecular complexes. Because the data are incomplete and a considerable number of false-positive interactions exist, protein interactions from different sources are commonly integrated in network analyses to build a stable molecular network. Although various types of integration strategies are being applied in current studies, the topological properties of the networks from these different integration strategies, especially typical applications based on these network integration strategies, have not been rigorously evaluated. In this paper, systematic analyses were performed to evaluate 11 frequently used methods using two types of integration strategies: empirical and machine learning methods. The topological properties of the networks of these different integration strategies were found to significantly differ. Moreover, these networks were found to dramatically affect the outcomes of typical applications, such as disease gene predictions, drug target detections, and molecular complex identifications. The analysis presented in this paper could provide an important basis for future network-based biological researches.

Glioblastoma multiforme (GBM), the most common and deadliest brain tumor with a median survival of 12–15 months, has been characterized by robust angiogenesis, high invasiveness, and universal recurrence. Genomics and transcriptomics studies defined three GBM subtypes (proneuronal, classical, mesenchymal), which are associated with genomic abnormalities, treatment response, and diversity in tumor microenvironment. However, molecular mechanisms of subtype-specific genotype in contribution to tumor phenotype are less understood. We have previously generated a de novo human GBM model by inactivation of p53 and activation of AKT signaling pathways in human neural stem/progenitor cells (hNSCs). Further characterization of this model indicates that the tumors recapitulate GBM’s classical histopathogical features and exhibit a molecular profile resembling the mesenchymal subtype. Comparative transcriptomic analysis revealed that YKL-40/CHI3L1, a secreted glycoprotein, is significantly upregulated during AKT activation induced malignant transformation of hNSCs in vitro and in vivo. Pharmacological inhibition of PI3K/AKT/mTOR signaling pathway results in decreasing YKL-40 mRNA expression and protein secretion level. Mechanistically, motif enrichment analysis and functional validation revealed that YKL-40 is bidirectionally regulated by a gene regulatory network with two transcription factors, BACH2 and CEBPB. Furthermore, YKL-40 significantly activates AKT-S6 and MAPK-ERK kinase signaling pathways, resulting in enhancing tumor cells proliferation, neurosphere and soft-agar colony formation, and tumor progression. In vivo silencing YKL-40 attenuates tumor angiogenesis and prolongs animal survival in the orthotopic xenograft mouse models. Interestingly, knocking down YKL-40 expression decreases microglia/macrophages infiltration in tumor mass. Both in vitro and in vivo validation indicate that PI3K/AKT/YKL-40 mediates tumor-associated microglia/macrophages recruitment and activation during tumor progression. Taken together, these results suggest that auto- and paracrine signaling of PI3K/AKT/YKL-40 affect tumor cells and tumor microenvironment contributing to glioblastoma progression, indicating YKL-40 as a predictive biomarker for evaluation of anti- PI3K/AKT/mTOR therapy, and a potential drug target for the treatment of this devastating disease.

Although significant progresses in molecular oncology are being made using conventional cell lines, most therapies still fail in phase III clinical trials. Patient-derived models are being used more frequently as they are more faithfully representing the genomic features of primary tumors. However, one by one test of each model from large biobanks is extremely economy and time consuming. In this study, each of a panel of patient-derived glioblastoma stem cell (GSC) models was uniquely tagged by a lentiviral Cas9D10A and paired-gRNA targetable unique reporter (CAPTURE) barcoding system. Barcoded GSCs were then pooled evenly and following by radiation treatment (RT) in vitro. Amplicon sequencing was employed to count the barcodes distribution, which represent the relative cell number. The results showed that this approach faithfully identified the RT resistant GSCs from a mixing pool when comparing to the results from canonical clonogenic assay. In addition, a fluorescence marker will be switched by delivery of corresponding barcodes targeting CRISPR so that we can re-isolate interested cell models from the treated pool for investigating the treatment sensitivity and resistance mechanism. This study will provide a robust approach for therapeutic discovery take advantage of patient-derived models from large biobanks. Citation Format: Ze-yan Zhang, Yingwen Ding, Ravesanker Ezhilarasan, Jie Yang, Lihong Long, Lawrence Bronk, Qianghu Wang, Erik P. Sulman. High-throughput evaluation of treatment response in patient-derived glioma stem cell models [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2019; 2019 Mar 29-Apr 3; Atlanta, GA. Philadelphia (PA): AACR; Cancer Res 2019;79(13 Suppl):Abstract nr 1917.

Cancers are composed of heterogeneous cell populations and the clonal evolution of these cells is one of the key reasons for treatment resistance and tumor recurrence. A fundamental challenge in studying clonal evolution in these tumors is the difficulty in capturing the phenotype-associated (e.g. treatment resistant) sub-populations from the heterogeneous population. We developed an approach to individually barcode and isolate specific cell subpopulations by constructing a Cas9D10A and paired-gRNA targetable unique reporter (CAPTURE) barcoding library with up to 36 million unique barcodes. This approach enabled us to uniquely barcode > 1 million cells, track barcode distribution following treatment and then isolate the resistant subpopulation using the subpopulation-specific barcode. Proof of principle studies showed that specific barcoding cells, even at as low as 0.1%, could be efficiently isolated using barcode targeting and cell sorting. Applied our barcoding approach, we found that A375 cell acquired BRAF inhibitor resistance both from pre-exiting and late-emerging mechanisms. Colony formation experiments showed that the isolated subpopulations were indeed resistant clones. Whole exome, transcriptome and methylome analysis were applied to study the captured subpopulations. Our CAPTURE barcoding approach will enable the identification of the both pre-exiting and late-emerging genetic or epigenetic changes driving treatment resistance. Citation Format: Ze-yan Zhang, Ravesanker Ezhilarasan, Yingwen Ding, Qianghu Wang, Jie Yang, Lihong Long, Roel G. Verhaak, Erik P. Sulman. Development of a CRISPR-Cas9D10A targetable, high-complexity, single-cell barcoding approach for capture of treatment resistant subclones from heterogeneous cancers [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2019; 2019 Mar 29-Apr 3; Atlanta, GA. Philadelphia (PA): AACR; Cancer Res 2019;79(13 Suppl):Abstract nr 4698.

PURPOSE: Genomic alterations classify cancers into subtypes with distinct clinical management and prognoses. Molecular classification of gliomas, the most common and lethal primary brain tumor, define tumors into distinct biologic and clinical entities. Mutation of isocitrate dehydrogenese (IDH) is associated with hypermethylation of CpG sites in gene promotors. Other alterations, including telomerase (TERTp) mutation, ATRX mutation, chromosome 1p/19q co-deletion (1p19q codel), and gene expression subtype (Classical/CL, Mesenchymal/MES, Proneural/PN), have yet to be associated with an epigenetic signature. We hypothesized that DNA methylation signatures would classify gliomas based on genetic alterations and give insight into the development of each subtype. The resulting platform functions as a unified diagnostic (UniD) with high processivity applicable to clinical diagnosis and generalizable across molecular oncology. METHODS: Machine learning algorithms were applied to whole methylome data to build classifiers for IDH, TERTp, and ATRX mutations; 1p19q codel and gene expression subtype. Models were validated with data from a phase III trial of anaplastic gliomas. RESULTS: Individual models were generated and prediction accuracies for IDH, TERTp, and ATRX mutations, and 1p19q codel were 100%, 98.3%, 90.48%, and 99.21% in test set and 89.9%, 82.8%, 92.47%, and 89.99% in the validation clinical trial data. The prediction model for gene expression subtype, a previously reported classifier enriched for characteristic somatic alterations, achieved 72% accuracy. Analysis of the misclassified cases revealed that the characteristic alterations associated with the expression subtypes were more correctly classified by methylation than by gene expression. Methylation-determined CL subtype showed high EGFR (p-value = 0.04) and amplification (p-value = 0.00019) compared to transcriptional MES samples, and low expression (p-value = 0.08) and amplification (p-value = 0.056) in methylation-determined MES but transcriptional CL samples. CONCLUSION: Distinct DNA methylation signatures were associated with key somatic genomic alterations in gliomas. It improved the existing classifiers based on gene expression and provided a unique clinical diagnostic platform for rapid determination of glioma subtype at the time of patient diagnosis. The extensive and significant relationship between cancer epigenetic signatures indicates that this approach would have broad applicability to other tumor types and lead to similar unified diagnostic platforms. A R package (UniD) is provided for implantation of this diagnostic platform. Citation Format: Jie Yang, Qianghu Wang, Ravesanker Ezhilarasan, Lihong Long, Benedikt Wiestler, Wolfgang Wick, Yinsen Miao, Erik Sulman. DNA-methylation based epigenetic signatures predict and deconvolute somatic genomic alterations in gliomas [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2019; 2019 Mar 29-Apr 3; Atlanta, GA. Philadelphia (PA): AACR; Cancer Res 2019;79(13 Suppl):Abstract nr 4238.

Glioblastoma stem cells (GSCs) are implicated in tumor neovascularization, invasiveness, and therapeutic resistance. To illuminate mechanisms governing these hallmark features, we developed a de novo glioblastoma multiforme (GBM) model derived from immortalized human neural stem/progenitor cells (hNSCs) to enable precise system-level comparisons of pre-malignant and oncogene-induced malignant states of NSCs. Integrated transcriptomic and epigenomic analyses uncovered a PAX6/DLX5 transcriptional program driving WNT5A-mediated GSC differentiation into endothelial-like cells (GdECs). GdECs recruit existing endothelial cells to promote peritumoral satellite lesions, which serve as a niche supporting the growth of invasive glioma cells away from the primary tumor. Clinical data reveal higher WNT5A and GdECs expression in peritumoral and recurrent GBMs relative to matched intratumoral and primary GBMs, respectively, supporting WNT5A-mediated GSC differentiation and invasive growth in disease recurrence. Thus, the PAX6/DLX5-WNT5A axis governs the diffuse spread of glioma cells throughout the brain parenchyma, contributing to the lethality ofGBM.

Astrocytomas are the most common primary human brain tumors. Receptor tyrosine kinases (RTKs), including tyrosine receptor kinase B (TrkB, also known as tropomyosin-related kinase B; encoded by neurotrophic tyrosine kinase receptor type 2 [NTRK2]), are frequently mutated by rearrangement/fusion in high-grade and low-grade astrocytomas. We found that activated TrkB can contribute to the development of astrocytoma and might serve as a therapeutic target in this tumor type.To identify RTKs capable of inducing astrocytoma formation, a library of human tyrosine kinases was screened for the ability to transform murine Ink4aActivated TrkB cooperated with Ink4a/Arf loss to induce the formation of astrocytomas through a mechanism mediated by activation of signal transducer and activator of transcription 3 (STAT3). TrkB activation positively correlated with Ccl2 expression. TrkB-induced astrocytomas remained dependent on TrkB signaling for survival, highlighting a role of NTRK2 as an addictive oncogene. Furthermore, the QKI-NTRK2 fusion associated with human astrocytoma transformed Ink4aOur findings provide evidence that constitutively activated NTRK2 alleles, notably the human tumor-associated QKI-NTRK2 fusion, can cooperate with Ink4a/Arf loss to drive astrocytoma formation. Therefore, we propose NTRK2 as a potential therapeutic target in the subset of astrocytoma patients defined by QKI-NTRK2 fusion.

SummaryWe leveraged IDH wild type glioblastomas and derivative neurospheres to define tumor-intrinsic transcription phenotypes. Transcriptomic multiplicity correlated with increased intratumoral heterogeneity and tumor microenvironment presence. In silico cell sorting demonstrated that M2 macrophages/microglia are the most frequent type of immune cells in the glioma microenvironment, followed by CD4 T lymphocytes and neutrophils. Hypermutation associated with CD8+ T cell enrichment. Longitudinal transcriptome analysis of 124 pairs of primary and recurrent gliomas showed expression subtype is retained in 53% of cases with no proneural to mesenchymal transition being apparent. Inference of the tumor microenvironment through gene signatures revealed a decrease in invading monocytes but a subtype dependent increase in M2 macrophages/microglia cells after disease recurrence. All expression datasets are accessible through http://recur.bioinfo.cnio.es/.SignificanceIDH wild type glioblastoma expression phenotypes have been related to tumor characteristics including genomic abnormalities and treatment response. We explored the intratumoral transcriptomic landscape, including a definition of tumor-intrinsic gene expression subtypes and how they relate to the different cellular components of the tumor immune environment. Comparison of matching primary and recurrent gliomas provided insights into the treatment-induced phenotypic tumor evolution. Proneural to mesenchymal transitions have long been suspected but were not apparent, while intratumoral heterogeneity was a predictor of subtype transition upon recurrence. Characterizing the evolving glioblastoma transcriptome en tumor microenvironment aids in designing more effective immunotherapy trials. Our study provides a comprehensive transcriptional and cellular landscape of IDH wild type GBM during treatment modulated tumor evolution.HighlightsNext generation GBM-intrinsic transcriptional subtypes: proneural, classical, mesenchymalM2 macrophages, CD4+ T-lymphocytes and neutrophils dominate glioblastoma microenvironmentSensitivity to radiotherapy may associate with M2 macrophage presenceCD8+ T cells are enriched in hypermutated GBMs at diagnosis and recurrence

Glioblastomas are highly infiltrated by diverse immune cells, including microglia, macrophages, and myeloid-derived suppressor cells (MDSCs). Understanding the mechanisms by which glioblastoma-associated myeloid cells (GAMs) undergo metamorphosis into tumor-supportive cells, characterizing the heterogeneity of immune cell phenotypes within glioblastoma subtypes, and discovering new targets can help the design of new efficient immunotherapies. In this study, we performed a comprehensive battery of immune phenotyping, whole-genome microarray analysis, and microRNA expression profiling of GAMs with matched blood monocytes, healthy donor monocytes, normal brain microglia, nonpolarized M0 macrophages, and polarized M1, M2a, M2c macrophages. Glioblastoma patients had an elevated number of monocytes relative to healthy donors. Among CD11b+ cells, microglia and MDSCs constituted a higher percentage of GAMs than did macrophages. GAM profiling using flow cytometry studies revealed a continuum between the M1- and M2-like phenotype. Contrary to current dogma, GAMs exhibited distinct immunological functions, with the former aligned close to nonpolarized M0 macrophages.

Cancer-testis (CT) genes represent the similarity between the processes of spermatogenesis and tumorigenesis. It is possible that their selective expression pattern can help identify driver genes in cancer. In this study, we integrate transcriptomics data from multiple databases and systematically identify 876 new CT genes in 19 cancer types. We explore their relationship with testis-specific regulatory elements. We propose that extremely highly expressed CT genes (EECTGs) are potential drivers activated through epigenetic mechanisms. We find mutually exclusive associations between EECTGs and somatic mutations in mutated genes, such as PIK3CA in breast cancer. We also provide evidence that promoter demethylation and close non-coding RNAs (namely, CT-ncRNAs) may be two mechanisms to reactivate EECTG gene expression. We show that the meiosis-related EECTG (MEIOB) and its nearby CT-ncRNA have a role in tumorigenesis in lung adenocarcinoma. Our findings provide methods for identifying epigenetic-driver genes of cancer, which could serve as targets of future cancer therapies., Genes normally expressed in the testis but aberrantly expressed in cancer are termed cancer testis antigens. In this study, the authors catalogue the expression of these genes in 19 different cancer types and correlate expression with some somatically mutated oncogenes.

Analysis of the biological pathways involved in complex human diseases is an important step in elucidating the pathogenesis and mechanism of diseases. Most pathway analysis approaches identify disease-related biological pathways using overlapping genes between pathways and diseases. However, these approaches ignore the functional biological association between pathways and diseases. In this paper, we designed a novel computational framework for prioritising disease-risk pathways based on functional profiling. The disease gene set and biological pathways were translated into functional profiles in the context of GO annotations. We then implemented a semantic similarity measurement for calculating the concordance score between a functional profile of disease genes and a functional profile of pathways (FPP); the concordance score was then used to prioritise and infer disease-risk pathways. A freely accessible web toolkit, 'Functional Profiling-based Pathway Prioritisation' (FPPP), was developed (http://bioinfo.hrbmu.edu.cn/FPPP). During validation, our method successfully identified known disease-pathway pairs with area under the ROC curve (AUC) values of 96.73 and 95.02% in tests using both pathway randomisation and disease randomisation. A robustness analysis showed that FPPP is reliable even when using data containing noise. A case study based on a dilated cardiomyopathy data set indicated that the high-ranking pathways from FPPP are well known to be linked with this disease. Furthermore, we predicted the risk pathways of 413 diseases by using FPPP to build a disease similarity landscape that systematically reveals the global modular organisation of disease associations.

Previous studies have shown that insertion/deletion polymorphisms in the angiotensin-converting enzyme (ACE) gene and the endothelial nitric oxide synthase (eNOS) gene are associated with Henoch-Schönlein purpura nephritis (HSPN). However, further studies are required to prove the relationship between HSPN and ACE and eNOS single nucleotide polymorphisms (SNPs). We studied six ACE SNPs and two eNOS SNPs by genotyping HSPN patients. Statistical analyses indicate that four ACE SNPs and two eNOS SNPs are associated with HSPN susceptibility. A cumulative effect analysis suggested that an increased number of unfavourable genotypes may lead to an increased risk of HSPN. By comparing alleles, genotypes and haplotypes that are associated with lupus nephritis (LN) and HSPN, we found genetic heterogeneity between HSPN and LN.

One important challenge in the post-genomic era is uncovering the relationships among distinct pathophenotypes by using molecular signatures. Given the complex functional interdependencies between cellular components, a disease is seldom the consequence of a defect in a single gene product, instead reflecting the perturbations of a group of closely related gene products that carry out specific functions together. Therefore, it is meaningful to explore how the community of protein complexes impacts disease associations. Here, by integrating a large amount of information from protein complexes and the cellular basis of diseases, we built a human disease network in which two diseases are linked if they share common disease-related protein complex. A systemic analysis revealed that linked disease pairs exhibit higher comorbidity than those that have no links, and that the stronger association two diseases have based on protein complexes, the higher comorbidity they are prone to display. Moreover, more connected diseases tend to be malignant, which have high prevalence. We provide novel disease associations that cannot be identified through previous analysis. These findings will potentially provide biologists and clinicians new insights into the etiology, classification and treatment of diseases.

The World Health Organization (WHO) classification and grading system attempts to predict the clinical course of meningiomas based on morphological parameters. However, because of high interobserver variation of some criteria, more reliable prognostic markers are required. Here, we assessed the TERT promoter for mutations in the hotspot regions C228T and C250T in meningioma samples from 252 patients. Mutations were detected in 16 samples (6.4% across the cohort, 1.7%, 5.7%, and 20.0% of WHO grade I, II, and III cases, respectively). Data were analyzed by t test, Fisher's exact test, log-rank test, and Cox proportional hazard model. All statistical tests were two-sided. Within a mean follow-up time in surviving patients of 68.1 months, TERT promoter mutations were statistically significantly associated with shorter time to progression (P < .001). Median time to progression among mutant cases was 10.1 months compared with 179.0 months among wild-type cases. Our results indicate that the inclusion of molecular data (ie, analysis of TERT promoter status) into a histologically and genetically integrated classification and grading system for meningiomas increases prognostic power. Consequently, we propose to incorporate the assessment of TERT promoter status in upcoming grading schemes for meningioma.

With the advances in high-throughput genome/transcriptome sequencing technologies and mass spectrometry (MS)-based proteomics, thousands of gene-protein pairs can be matched and merged in a single experiment. It is of interest to perform a correlative analysis of gene and protein expression data and investigate the nature of their similarity/dissimilarity as it could harbour potential biomarkers or drug targets. Manual determination of data points of interest quickly becomes a very complex and laborious process. Thus, there is a high demand for automated ‘omics’ data integration tools that can not only routinely match and combine gene and protein expression values but also provide a measure to highlight meaningful biological insights. In this work, we applied a fast and easy approach to integrate large proteomic and transcriptomic data derived from the deep analysis of glioma cancer stem cells (GSCs). The proposed algorithm provides a mathematical distance between two data sets and asignes a direction of their interrelation based on the abundancies. We distinguished three types of the data correlation: concordant, anticoncordant where protein abundance was higher than that of the corresponding RNA and anticoncordant where protein abundance was lower. We investigated the nature of the observed discordances and were able to separate different, phenotypically divergent, classes of GSC lines.

Motivation: In the functional genomic era, a large number of gene sets have been identified via high-throughput genomic and proteomic technologies. These gene sets of interest are often related to the same or similar disorders or phenotypes, and are commonly presented as differentially expressed gene lists, co-expressed gene modules, protein complexes or signaling pathways. However, biologists are still faced by the challenge of comparing gene sets and interpreting the functional relationships between gene sets into an understanding of the underlying biological mechanisms. Results: We introduce a novel network-based method, designated corrected cumulative rank score (CCRS), which analyzes the functional communication and physical interaction between genes, and presents an easy-to-use web-based toolkit called GsNetCom to quantify the functional relationship between two gene sets. To evaluate the performance of our method in assessing the functional similarity between two gene sets, we analyzed the functional coherence of complexes in functional catalog and identified protein complexes in the same functional catalog. The results suggested that CCRS can offer a significant advance in addressing the functional relationship between different gene sets compared with several other available tools or algorithms with similar functionality. We also conducted the case study based on our method, and succeeded in prioritizing candidate leukemia-associated protein complexes and expanding the prioritization and analysis of cancer-related complexes to other cancer types. In addition, GsNetCom provides a new insight into the communication between gene modules, such as exploring gene sets from the perspective of well-annotated protein complexes. Availability and Implementation: GsNetCom is a freely available web accessible toolkit at http://bioinfo.hrbmu.edu.cn/GsNetCom. Contact: lixia@hrbmu.edu.cn Supplementary information: Supplementary data are available at Bioinformatics online.

MicroRNomics is a novel genomics that studies the identification, targets, biological functions, etc., of microRNAs (miRNAs) on a genomic scale. Computational target prediction algorithms are important applications in microRNomics. However, the overlaps between target sets predicted by different algorithms for one miRNA are often small. Our work is initiated to find the reasons causing "heterogeneity" and investigate whether the heterogeneous targets are homogeneous on functional levels by integrating similarity metrics. The results suggest that most human miRNAs own heterogeneous targets. The dissimilarity of thermodynamic characteristics and the different treatment of 3'-compensatory sites adopted by algorithms are the main reasons for target "heterogeneity." Meanwhile, we find most miRNA heterogeneous targets are functional homogeneity because of the common principles such as sites conservation and G:U wobble pairs in different algorithms. Our findings reveal the "functional homogeneity in miRNA target heterogeneity." The conclusions provide a perspective of microRNomics on functional levels, which introduce a new sight into human miRNA targets.

MicroRNAs (miRNAs) play important roles in post-transcriptional gene expression control. To gain new insight into human miRNAs, we performed comprehensive sequence-based homology search for known human miRNAs to study the evolutionary distribution of human miRNAs. Furthermore, we carried out a series of studies to compare various features for different lineage-specific human miRNAs. Our results showed that major expansions of human miRNA genes coincide with the advent of vertebrates, mammals and primates. Further system-level analysis revealed significant differences in human miRNAs that arose from different evolutionary time points for a number of characteristics, implicating genetic and functional diversification for different human miRNAs during evolution. Our finds provide more useful knowledge for further exploring origins and evolution of human miRNA genes.

Adult diffuse gliomas are a diverse class of brain tumors that ultimately result in fatal malignant progression. The highly variable disease progression intervals and poor treatment response observed in glioma have been explained, in part, by the extent of heterogeneity within and across gliomas. Somatic structural variants such as deletions, duplications, insertions, inversions, translocations, and extrachromosomal DNA contribute to glioma heterogeneity by generating genomic instability, a hallmark of cancer genomes. While comprehensive approaches such as whole genome sequencing have begun to catalog the glioma structural variant landscape, the molecular mechanisms by which they contribute to tumor progression remains poorly understood. Here, we identified genomic rearrangements in the whole genomes of 147 gliomas and glioma cell lines. Among the whole-genome sequencing samples were 62 glioblastomas (TCGA), 52 low-grade gliomas (TCGA), and our own set of 33 patient-derived glioma sphere-forming cells. On average, glioblastomas demonstrated a greater frequency of high-confidence structural variants (162 per tumor) than low-grade gliomas (130 per tumor, P=0.02) with two glioblastomas displaying evidence of chromothripsis. Across the World Heath Organization's molecular classifications, the IDH-mutant 1p/19q co-deleted tumors demonstrated the lowest structural variant burden when compared with both IDH-wildtype and IDH-mutant non-co-deleted tumors. Structural variant frequencies were found to be similar between patient tumors and an independent set of patient-derived cell lines. Across all samples, a majority of the structural variant breakpoints were found in intronic (42%) and intergenic regions (55%) suggesting that rearrangements predominantly impact distal regulatory regions. To investigate the functional consequences of structural variants in these regions we integrated these samples with available RNA-sequencing and epigenetic data to characterize deregulated transcriptional circuits. Analyses examining the association between genomic rearrangements and chromatin states to determine potential mechanisms of oncogene activation in select tumor and cell line samples are now underway. Together, our study demonstrates the importance of integrating structural variant calls from whole genome sequencing with expression and epigenetic data to define functional genomic rearrangements. Citation Format: Kevin C. Johnson, Floris P. Barthel, Ming Tang, Samirkumar Amin, Qianghu Wang, Erik P. Sulman, Kunal Rai, Roel G.W Verhaak. Identification and functional characterization of glioma structural variants [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 4355.

Glioma sphere-forming cells (GSCs) are important models for understanding glioma biology and development of novel therapies. We performed clonogenic assays to explore the relationship between methylation status and radiation response in twelve IDH wild type GSCs irradiated with 2-, 4-, and 6-Gy ionizing γ-radiation. The survival fraction at 4Gy and 6Gy (SF4 and SF6, respectively) was used to dichotomize GSCs as either radiation-sensitive or resistant. Seven (out of 12, 58%) GSCs were sensitive to radiation and five (out of 12, 42%) were resistant with SF4 > 30% and SF6 > 10%. DNA CpG methylomes of the GSCs were profiled using Illumina Infinium 450K methylation bead arrays. We observed that 304,458 out of 465,844 methylation probes (65.4%) showed increased methylation in radiation-resistant GSCs relative to radiation-sensitive GSCs (Fisher Test p-value < 1e-15). This translated to significant increases in DNA methylation of the promoter regions of 17,735 out of 21,214 (83.6%) genes in radiation-resistant compared to radiation-sensitive GSCs. RNA sequencing of the GSCs showed that 57% of genes were expressed at lower levels in radiation-resistant relative to radiation-sensitive GSCs. Through GSEA, we observed that fifteen of sixteen oxidative stress genes were methylated in radiation-resistant GSCs (p-value = 0.019), suggesting an association between radiation-resistance and reactive oxygen species metabolism (ROS). To validate our finding, we derived a methylation signature from the two GSC radiation response groups and used this to classify TCGA cases that received radiotherapy into a responders and non-responders group. We found that survival was significantly different between the two groups (median survival 84 vs. 61 weeks; HR 1.64 adjusting for patient age, p-value < 0.008), suggesting that a methylation signature could be used to predict clinical response to radiation treatment. Our study shows that GSCs can be used to determine clinical and functional properties of GBM.

Glioblastoma (GBM) is the most frequently occurring malignant primary tumor of the central nervous system, and despite aggressive clinical intervention, GBM is uniformly fatal. Data from multiple expression profiling experiments suggest that the type-I integral membrane glycoprotein podoplanin (PDPN) is an important prognostic marker in astrocytic gliomas independent of a number of relevant clinical factors. We recently discovered that radiotherapy (RT) resistance of GBM is associated with global DNA hypermethylation and hypothesized that PDPN may be involved in this phenomenon. Indeed, we found that PDPN knockdown sensitized glioma sphere-forming cells (GSCs) to RT in vitro. Furthermore, upon profiling PDPN-sorted GSC subpopulations, enrichment analysis revealed that PDPN upregulates polycomb repressive complex (PRC) activity to promote EZH2-driven global DNA hypermethylation. EZH2 is known to directly control DNA methylation, and we have previously shown that EZH2 protects GSCs from radiation-induced cell death. This is the first report implicating PDPN in EZH2-mediated RT resistance, and we believe the membrane protein may be a meaningful therapeutic target in RT resistant GBM. In summary, PDPN is a novel prognostic marker for individuals with astrocytic gliomas and promotes EZH2-mediated RT resistance in GSCs.

Background The majority of glioblastomas have aberrant receptor tyrosine kinase (RTK)/RAS/phosphoinositide 3 kinase (PI3K) signaling pathways and malignant glioma cells are thought to be addicted to these signaling pathways for their survival and proliferation. However, recent studies suggest that monotherapies or inappropriate combination therapies using the molecular targeted drugs have limited efficacy possibly because of tumor heterogeneities, signaling redundancy and crosstalk in intracellular signaling network, indicating necessity of rationale and methods for efficient personalized combination treatments. Here, we evaluated the growth of colonies obtained from glioma tumor-initiating cells (GICs) derived from glioma sphere culture (GSC) in agarose and examined the effects of combination treatments on GICs using targeted drugs that affect the signaling pathways to which most glioma cells are addicted. Methods Human GICs were cultured in agarose and treated with inhibitors of RTKs, non-receptor kinases or transcription factors. The colony number and volume were analyzed using a colony counter, and Chou-Talalay combination indices were evaluated. Autophagy and apoptosis were also analyzed. Phosphorylation of proteins was evaluated by reverse phase protein array and immunoblotting. Results Increases of colony number and volume in agarose correlated with the Gompertz function. GICs showed diverse drug sensitivity, but inhibitions of RTK and RAF/MEK or PI3K by combinations such as EGFR inhibitor and MEK inhibitor, sorafenib and U0126, erlotinib and BKM120, and EGFR inhibitor and sorafenib showed synergy in different subtypes of GICs. Combination of erlotinib and sorafenib, synergistic in GSC11, induced apoptosis and autophagic cell death associated with suppressed Akt and ERK signaling pathways and decreased nuclear PKM2 and β-catenin in vitro, and tended to improve survival of nude mice bearing GSC11 brain tumor. Reverse phase protein array analysis of the synergistic treatment indicated involvement of not only MEK and PI3K signaling pathways but also others associated with glucose metabolism, fatty acid metabolism, gene transcription, histone methylation, iron transport, stress response, cell cycle, and apoptosis. Conclusion Inhibiting RTK and RAF/MEK or PI3K could induce synergistic cytotoxicity but personalization is necessary. Examining colonies in agarose initiated by GICs from each patient may be useful for drug sensitivity testing in personalized cancer therapy. Electronic supplementary material The online version of this article (doi:10.1186/s12967-016-0803-2) contains supplementary material, which is available to authorized users.

Background Brainstem metastases (BSMs) represent a significant treatment challenge. Stereotactic radiosurgery (SRS) is often used to treat BSM. We report our experience in the treatment of BSM with Gamma Knife SRS (GK_SRS). Methods The records of 1962 patients with brain metastases treated with GK_SRS between 2009 and 2013 were retrospectively reviewed. Seventy-four patients with 77 BSMs and follow-up brain imaging were identified. Local control (LC), overall survival (OS), progression-free survival (PFS), and toxicity were assessed. Results Median follow-up was 5.5 months (range, 0.2–48.5 months). Median tumor volume was 0.13 cm3 (range, 0.003–5.58 cm3). Median treatment dose was 16 Gy (range, 10–20 Gy) prescribed to 50% isodose line (range, 40%–86%). Crude LC was 94% (72/77). Kaplan-Meier estimate of median OS was 8.5 months (95% CI, 5.6–9.4 months). Symptomatic lesions and larger lesions, especially size ≥2 cm3, were associated with worse LC (HR = 8.70, P = .05; HR = 14.55, P = .02; HR = 62.81, P < .001) and worse OS (HR = 2.00, P = .02; HR = 2.14, P = .03; HR = 2.81, P = .008). Thirty-six percent of BSMs were symptomatic, of which 36% (10/28) resolved after SRS and 50% (14/28) had stable or improved symptoms. Actuarial median PFS was 3.9 months (95% CI, 2.7–4.9 months). Midbrain location was significant for worse PFS (HR = 2.29, P = .03). Toxicity was low (8%, 6/74), with size and midbrain location associated with increased toxicity (HR 1.57, P = .05; HR = 5.25, P = .045). Conclusions GK_SRS is associated with high LC (94%) and low toxicity (8%) for BSMs. Presence of symptoms or lesion size ≥ 2 cm3 was predictive of worse LC and OS.

The authors' institution previously reported a 69% rate of crude local control for surgical management of lateral ventricle metastases at the University of Texas MD Anderson Cancer Center. For comparison, the authors here report their institutional experience with use of stereotactic radiosurgery (SRS) to treat intraventricular metastases.To identify patients with intraventricular metastases for this retrospective review, the authors queried an institutional SRS database containing the medical records of 1962 patients with 5800 brain metastases who consecutively underwent SRS from June 2009 through October 2013. End points assessed were local control (crude and locoregional), distant failure-free survival, progression-free survival, and overall survival.Of the 1962 records examined, those for 25 (1.3%) patients with 30 (0.52%) intraventricular metastases were identified. Median patient age at SRS was 55.8 years. The most common primary malignancy was renal cell carcinoma (n = 13), followed by melanoma (n = 7) and breast adenocarcinoma (n = 5). Median tumor volume was 0.75 cm(3) (range 0.01-5.6 cm(3)). Most lesions were located in the lateral ventricles (n = 25, 83.3%) and were treated to a median dose of 20 Gy (range 14-20 Gy). A total of 12 (48%) patients received whole-brain radiation therapy, most (n = 10) before SRS. With a median follow-up of 11.4 months (range 1.6-39.2 months), the rate of crude local control was 93.3%, and the rates of 6-month and 1-year actuarial locoregional control were 85.2% and 56.2%, respectively. The median overall survival time after SRS was 11.6 months (range 1.3-38.9 months), and the 6-month and 1-year actuarial rates were 87.1% and 46.7%, respectively. Disease dissemination developed in 7 (28%) patients as a second intraventricular metastatic lesion (n = 3, 12%), leptomeningeal disease (n = 3, 12%), or both (n = 1, 4%). Radiographic changes developed in 5 (20%) patients and included necrosis (n = 2, 8%) and hemorrhage (n = 3, 12%). A primary diagnosis of renal cell carcinoma was associated with an improved rate of distant failure-free survival (p = 0.05) and progression-free survival (p = 0.08).SRS provides excellent local control for intraventricular metastases, with acceptable treatment-related toxicity, thereby supporting nonsurgical treatment for these lesions. The propensity for intraventricular dissemination among intraventricular metastases seems to be histologically dependent.

Glioma sphere-forming cells (GSCs) are important in glioblastoma (GBM) initiation, maintenance, and treatment resistance. We performed whole exome and transcriptome sequencing, DNA methylation profiling, DNA copy number determination, and functional characterization of 43 GSCs and matching tumors. Comparative analyses revealed that GSCs recapitulate the molecular landscape of GBM and provide a unique means for discovering the inter- and intra-tumor heterogeneity of GBM. We performed clonogenic assays to explore the relationship between methylation status and radiation response in twelve IDH wild type GSCs irradiated with 2-, 4-, and 6-Gy ionizing radiation. The survival fraction at 4Gy and 6Gy (SF4 and SF6, respectively) were used to dichotomize GSCs as either radiation-sensitive or resistant. DNA CpG methylomes of the GSCs were profiled using Infinium 450K methylation beadchip arrays. We observed that 304,458 out of 465,844 methylation probes (65.4%) showed increased methylation in radiation-resistant relative to radiation-sensitive GSCs (Fisher’s Exact Test, p < 1e-15). Using GSEA, we observed that fifteen of sixteen oxidative stress genes were methylated in radiation-resistant GSCs (p-value=0.019), suggesting an association between radiation-resistance and reactive oxygen species metabolism (ROS). To validate our finding, we derived a methylation signature differentiating the two GSC radiation response groups and used this to classify TCGA cases that received radiotherapy into a responder and non-responder group. We found that survival was significantly different between the two groups (median survival 84 vs. 61 weeks; HR 1.64 adjusting for patient age, p-value

Malignant gliomas are composed of heterogeneous cell populations and the clonal evolution of these cells is one of the key reasons for treatment resistance and tumor recurrence. A fundamental challenge in studying clonal evolution in these tumors is the difficulty in isolating the phenotype-associated (e.g. treatment resistant) sub-populations from the heterogeneous population. We developed an approach to individually barcode and isolate specific cell subpopulations using a CRISPR (clustered, regularly interspaced, short palindromic repeat)-Cas9D10A targetable, single-cell barcoding library with a complexity of 36 million unique barcodes. This approach enabled us to uniquely barcode > 1 million cells, identify enriching barcoded sub-populations following treatment and then isolate the resistant subpopulation using the subpopulation-specific barcode. Each barcode sequence carrying two guide RNAs so that CRISPR-Cas9D10A targeting can lead to “switching” (i.e. frameshift) of an EGFP (Enhanced Green Fluorescent Protein) reporter. Thus, targeted cells lose their EGFP signal and can be specifically isolated by cell sorting. Proof of principle studies showed that specific barcoding cells could be efficiently targeted to turn off EGFP and subpopulations isolated. Subsequent large-scale implementation of this approach has been performed to isolate resistant subpopulations following radiation and temozolomide. Our approach will lead to identification of the both pre-existing and acquired specific somatic genetic or epigenetic changes driving treatment resistance in patients with malignant gliomas.

Gliomas are the most common type of brain tumor in both children and adults. Several low-grade gliomas (LGG) have the ability to progress into more aggressive tumors -high-grade gliomas (HGG) including glioblastoma (GB). Although patients harboring a LGG may survive for years, after the tumor transforms to HGG, life expectancy rapidly declines to 12 to 15 months in adults and 40 months in children. Thus, inhibiting this process of malignant transformation (MT) is an attractive therapeutic strategy because of the more indolent course associated with LGGs. Immune response plays a critical role in surveillance against malignant transformation. Our previous study shows that fibrinogen-like protein 2 (FGL2) is a key hub of tumor-mediated immune suppression. Hence, we investigated the role of FGL2 in promoting tumor progression from LGG to HGG in glioma. Analysis of TCGA expression data showed that increased FGL2 expression is associated with poorer survival in LGG and GB patients. And there is a positive correlation of expression level between FGL2 and mesenchymal glioma marker CD44, and a negative correlation between FGL2 and proneural glioma marker OLIG2. Engineered expression of FGL2 in a PDGFB-dependent mouse model of oligodendroglioma, a common glioma subtype, yielded a significantly higher rate of HGGs (72% vs 29%, p=0.034) and poorer-symptom free survival (63 vs 90 days, p=0.003) than PDGFB expression alone. And HGGs from FGL2 + PDGFB expressing mice exhibited a distinct mesenchymal phenotype validating TCGA data. Further, FGL2 induced high numbers of CD4+FoxP3+ cells from an early time point of tumor formation underscoring its role in tumor progression. And FGL2 overexpression educated M2 skew in the tumors characterized by high expression of Iba1 and Arginase1 in macrophages. Finally, treatment with anti-FGL2 antibody significantly improves survival in mice, shifts the phenotype from mesenchymal HGG to proneural LGG, and rescues M2 macrophage skewing. Our results show that FGL2 is critical for malignant progression of glioma by inducing immunosuppression in tumor microenvironment, and raise the potential of FGL2 to be a promising target to suppress/reverse glioma progression and provide survival benefit in clinical. Citation Format: Khatri Latha, Jun Yan, Yuhui Yang, Loyola V. Gressot, Lingyuan Kong, Ganiraju Manyam, Ravesanker Ezhilarasan, Qianghu Wang, Erik P. Sulman, Jingda Xu, Richard E. Davis, Suyun Huang, Gregory N. Fuller, Arvind Rao, Amy B. Heimberger, Shulin Li, Ganesh Rao. Fibrinogen-like protein 2 drives malignant tumor progression in glioma [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 5606. doi:10.1158/1538-7445.AM2017-5606

Prognostic and predictive molecular diagnostics for patients with gliomas typically rely on multiple assays, requiring large amounts of tissue and high cost. DNA methylation has been utilized to identify prognostic subsets and high-throughput platforms exist that are suitable for archival tissue. Therefore, we developed a unified and integrated diagnostic pipeline that can assess multiple prognostic and predictive biomarkers using only the Illumina Infinium Methylation array. This pipeline includes two parts: data processing and diagnostic biomarkers. Data processing starts from the raw data and followed with quantitative sample, probe, and batch quality control. The diagnostic biomarkers include a glioma methylation assay that predicts radiation response (GaMA); tumor classification enriched for TCGA expression subclasses; copy number alterations including phosphatase and tensin homolog (PTEN) loss, epidermal growth factor receptor (EGFR) amplification, and chromosomes 1p/19q co-deletion; CpG island methylation phenotype (G-CIMP); isocitrate dehydrogenase (IDH) mutation, and O6-methylguanine-DNA methyltransferase (MGMT) promoter hypermethylation. WHO grade II-IV gliomas were analyzed in both publically available and institutional datasets. A signature was identified to effectively distinguished radiation resistant from radiation sensitive glioma stem-like cells (GSCs). Signature has been applied to 272 TCGA GBM samples from patients who received standard radiotherapy (RT). The survival analysis showed that the subgroup with RT-sensitive and RT-resistant have significant difference in survival time (log-rank test p-value = 0.0016). Gene expression subclasses prediction biomarker was build by using the revised TCGA gene expression subclasses as gold standard. The prediction accuracy in test data set was 83.5% in the homogeneous subgroup, 71.0% in the semi-heterogeneous subgroup, and 62.1% in the heterogeneous subgroup. The prediction accuracy decreased as tumor heterogeneity increased. Certain copy number alteration events were predicted by developing specific signatures. Revised methylation signatures were developed for IDH mutation and G-CIMP status respectively, which can identified 99% of those samples. 230 GBM samples with 450k data available were tested with MGMT methylation-specific real-time PCR for MGMT methylation status. The methylation-based MGMT prediction accuracy reached about 90%. In summary, we have developed a single, FFPE-based pipeline for unified and integrated determination of multiple biomarkers of malignant glioma. Citation Format: Jie Yang, Qianghu Wang, Lihong Long, Ravesanker Ezhilarasan, Erik Sulman. UniD: unified and integrated diagnostic pipeline for malignant gliomas based on DNA methylation data [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 3348. doi:10.1158/1538-7445.AM2017-3348

e13527 Background: Thrombocytosis is associated with poor survival in several malignancies and is both triggered by and promoting of tumor growth. We hypothesized that an increase in circulating platelets following standard temozolamide (TMZ) CRT and/or adjuvant TMZ would predict for decreased survival in patients with glioblastoma (GBM). Methods: We reviewed data of 122 patients with newly diagnosed, pathological-proven GBM from 2007 to 2016 with documented complete blood counts before the start of CRT, after CRT, before adjuvant TMZ, and after a minimum 2 cycles of adjuvant TMZ. The association between changes in blood count levels and patient survival was analyzed by log-rank test. Multivariate analysis was performed using known prognostic co-variates, including MGMT methylation, patient age, and performance status (PS). Results: The mean change in platelets before and after CRT was -46 k/µl (range -387 to 179 k/µl) and before and after adjuvant TMZ was -25 k/µl (range -408 to 143 k/µl). As a reference, the mean change in lymphocytes was -5.7 k/µl (range -0.7 to 1.3 k/µl) and -0.0 k/µl (range -1.6 to 0.9 k/µl) for CRT and adjuvant TMZ, respectively. Patients were dichotomized based on the relative change in platelets and lymphocytes from baseline (≤30% increase, “low”, n = 101 vs > 30% increase, “high”, n = 12). The median survival of low patients vs. high patients was 30 vs 13 months (p = 0.006). Potential confounders, such as infection or thrombosis, were not associated with the high group. However, no significant survival difference was observed between groups based on platelet changes during adjuvant TMZ. Similarly, changes in lymphocyte counts were not significantly prognostic. In multivariate analysis, extent of resection, MGMT status, PS and the increase in platelets after CRT were significantly associated with survival (HR for platelets 4.0, 95% CI 1.6-10.1). Conclusions: Increased platelet counts after CRT predict for poor survival in patients with GBM. The effect is platelet specific and does not reflect general bone marrow changes, as lymphocyte changes were not significantly prognostic. These results suggest a potential interaction between platelets and tumor aggressiveness.

Glioma sphere-forming cells (GSCs) play important roles in tumor maintenance and chemotherapy and radiation resistance, and thus serve as important models for the development of novel therapies and for understanding tumor biology. We hypothesized that genomic analyses of GSCs in the absence of normal stroma would identify novel somatic alterations not previously reported from whole tumor specimen analyses. We analyzed 42 GSCs, derived at the time of surgical resection, and their matched tumors of origin. Unsupervised consensus clustering by non-negative matrix factorization (CNMF) using whole transcriptome sequencing data identified 5 transcriptional classes. Comparison to published TCGA GBM subtypes using single-sample gene set enrichment analysis (ssGSEA) showed that 9 GSCs perfectly matched with TCGA mesenchymal subtype. The other 4 subgroups of GSCs mapped to proneural and/or classical subtypes, while none of the GSC classes matched with the neural TCGA subtype. Overall, 73% of GSCs matched the TCGA subtype of paried tumor, suggesting transcriptional consistency between GSCs and their parental tumor. A total of 88 candidate gene fusions were also identified in the GSCs, including the previously reported FGFR3-TACC3 and EGFR-SEP14 fusions. DNA copy number analysis indicated recurrent regions of frequent loss or gain, such as EGFR, MYC, and MYCN amplification, and deletion of PTEN and CDKN2A. Whole exome sequencing based analysis identified 3,918 somatic mutations in 22 GSCs. PTEN, TP53, EGFR and 17 additional genes were identified as significantly altered. Our results indicate genomic consistency between GSCs and their tumors of origin, suggesting the suitability of GSCs as a model for preclinical testing. Additional gene fusions and somatic mutations were detected in GSCs suggesting the important role of these cell models in identifying infrequent events only detectable in pure tumor cell populations.