Gene expression profiling reveals molecularly and clinically distinct subtypes of glioblastoma multiforme

Identification of diagnostic serum protein profiles of glioblastoma patients

Diagnosis of a glioblastoma (GBM) is triggered by the onset of symptoms and is based on cerebral imaging and histological examination. Serum-based biomarkers may support detection of GBM. Here, we explored serum protein concentrations of GBM patients and used data mining to explore profiles of biomarkers and determine whether these are associated with the clinical status of the patients. Gene and protein expression data for astrocytoma and GBM were used to identify secreted proteins differently expressed in tumors and in normal brain tissues. Tumor expression and serum concentrations of 14 candidate proteins were analyzed for 23 GBM patients and nine healthy subjects. Data-mining methods involving all 14 proteins were used as an initial evaluation step to find clinically informative profiles. Data mining identified a serum protein profile formed by BMP2, HSP70, and CXCL10 that enabled correct assignment to the GBM group with specificity and sensitivity of 89 and 96%, respectively (p < 0.0001, Fischer's exact test). Survival for more than 15 months after tumor resection was associated with a profile formed by TSP1, HSP70, and IGFBP3, enabling correct assignment in all cases (p < 0.0001, Fischer's exact test). No correlation was found with tumor size or age of the patient. This study shows that robust serum profiles for GBM may be identified by data mining on the basis of a relatively small study cohort. Profiles of more than one biomarker enable more specific assignment to the GBM and survival group than those based on single proteins, confirming earlier attempts to correlate single markers with cancer. These conceptual findings will be a basis for validation in a larger sample size.

Coexpression network analysis identifies transcriptional modules related to proastrocytic differentiation and sprouty signaling in glioma

Gliomas are primary brain tumors with high mortality and heterogeneous biology that is insufficiently understood. In this study, we performed a systematic analysis of the intrinsic organization of complex glioma transcriptome to gain deeper knowledge of the tumor biology. Gene coexpression relationships were explored in 790 glioma samples from 5 published patient cohorts treated at different institutions. We identified 20 coexpression modules that were common to all the data sets and associated with proliferation, angiogenesis, hypoxia, immune response, genomic alterations, cell differentiation phenotypes, and other features inherent to glial tumors. A collection of high-quality signatures for the respective processes was obtained using cross-data set summarization of the modules' gene composition. Individual modules were found to be organized into higher order coexpression groups, the two largest of them associated with glioblastoma and oligodendroglioma, respectively. We identified a novel prognostic gene expression signature (185 genes) linked to a proastrocytic pattern of tumor cell differentiation. This "proastrocytic" signature was associated with long survival and defined a subgroup of the previously established "proneural" class of gliomas. A strong negative correlation between proastrocytic and proneural markers across differentiated tumors underscored the distinction between these subtypes of glioma. Interestingly, one further novel signature in glioma was identified that was associated with EGFR (epidermal growth factor receptor) gene amplification and suggested that EGF signaling in glioma may be a subject to regulation by Sprouty family proteins. In summary, this integrated analysis of the glioma transcriptome provided several novel insights into molecular heterogeneity and pathogenesis of glial tumors.

A 4-gene signature associated with clinical outcome in high-grade gliomas

PURPOSE

Gene expression studies provide molecular insights improving the classification of patients with high-grade gliomas. We have developed a risk estimation strategy based on a combined analysis of gene expression data to search for robust biomarkers associated with outcome in these tumors.

EXPERIMENTAL DESIGN

We performed a meta-analysis using 3 publicly available malignant gliomas microarray data sets (267 patients) to define the genes related to both glioma malignancy and patient outcome. These biomarkers were used to construct a risk-score equation based on a Cox proportional hazards model on a subset of 144 patients. External validations were performed on microarray data (59 patients) and on RT-qPCR data (194 patients). The risk-score model performances (discrimination and calibration) were evaluated and compared with that of clinical risk factors, MGMT promoter methylation status, and IDH1 mutational status.

RESULTS

This interstudy cross-validation approach allowed the identification of a 4-gene signature highly correlated to survival (CHAF1B, PDLIM4, EDNRB, and HJURP), from which an optimal survival model was built (P < 0.001 in training and validation sets). Multivariate analysis showed that the 4-gene risk score was strongly and independently associated with survival (hazard ratio = 0.46; 95% CI, 0.26-0.81; P = 0.007). Performance estimations indicated that this score added beyond standard clinical parameters and beyond both the MGMT methylation status and the IDH1 mutational status in terms of discrimination (C statistics, 0.827 versus 0.835; P < 0.001).

CONCLUSION

The 4-gene signature provides an independent risk score strongly associated with outcome of patients with high-grade gliomas.

Detecting and quantifying multiple proteins in clinical samples in high-throughput using antibody microarrays

Many diagnostic and prognostic tests performed in the clinic today rely on the sensitive detection and quantification of a single protein, usually by means of an immunoassay. Even in the case of monogenic diseases, however, single markers are often insufficient to provide highly reliable predictions of disease onset, and the accuracy of these predictions only decreases for polygenic diseases and for very early detection or prediction. Recent studies have shown that predictive reliability increases dramatically when multiple markers are analyzed simultaneously. Antibody microarrays provide a powerful way to quantify the abundance of many different proteins simultaneously in a variety of sample types, including serum, urine, and tissue explants. Because the assay is highly miniaturized, very little sample is required and the assay can be performed in high-throughput. Using antibody microarrays, we have been able to identify prognostic markers of early mortality in patients with end-stage renal disease and have built multivariate models based on these markers. We anticipate that antibody microarrays will prove similarly useful in other discovery-based efforts and may ultimately enjoy routine use in clinical labs.

Gene pathways and subnetworks distinguish between major glioma subtypes and elucidate potential underlying biology

Molecular diagnostic tools are increasingly being used in an attempt to classify primary human brain tumors more accurately. While methods that are based on the analysis of individual gene expression prove to be useful for diagnostic purposes, they are devoid of biological significance since tumorgenesis is a concerted deregulation of multiple pathways rather than single genes. In a proof of concept, we utilize two large clinical data sets and show that the elucidation of enriched pathways and small differentially expressed sub-networks of protein interactions allow a reliable classification of glioblastomas and oligodendrogliomas. Applying a feature selection method, we observe that an optimized subset of pathways and subnetworks significantly improves the prediction accuracy. By determining the enrichment of altered genes in pathways and subnetworks we show that optimized subsets of genes rarely seem to be a target of genomic alteration. Our results suggest that groups of genes play a decisive role for the phenotype of the underlying tumor samples that can be utilized to reliably distinguish tumor types. In the absence of enrichment of genes that are genomically altered we assume that genetic changes largely exert an indirect rather than direct regulatory influence on a number of tumor-defining regulatory networks.

Gene expression profiling of gliomas: merging genomic and histopathological classification for personalised therapy

The development of DNA microarray technologies over the past decade has revolutionised translational cancer research. These technologies were originally hailed as more objective, comprehensive replacements for traditional histopathological cancer classification systems, based on microscopic morphology. Although DNA microarray-based gene expression profiling (GEP) remains unlikely in the near term to completely replace morphological classification of primary brain tumours, specifically the diffuse gliomas, GEP has confirmed that significant molecular heterogeneity exists within the various morphologically defined gliomas, particularly glioblastoma (GBM). Herein, we provide a 10-year progress report on human glioma GEP, with focus on development of clinical diagnostic tests to identify molecular subtypes, uniquely responsive to adjuvant therapies. Such progress may lead to a more precise classification system that accurately reflects the cellular, genetic, and molecular basis of gliomagenesis, a prerequisite for identifying subsets uniquely responsive to specific adjuvant therapies, and ultimately in achieving individualised clinical care of glioma patients.

Genome-wide expression profiling identifies deregulated miRNAs in malignant astrocytoma

Malignant astrocytoma includes anaplastic astrocytoma (grade III) and glioblastoma (grade IV). Among them, glioblastoma is the most common primary brain tumor with dismal responses to all therapeutic modalities. We performed a large-scale, genome-wide microRNA (miRNA) (n=756) expression profiling of 26 glioblastoma, 13 anaplastic astrocytoma and 7 normal brain samples with an aim to find deregulated miRNA in malignant astrocytoma. We identified several differentially regulated miRNAs between these groups, which could differentiate glioma grades and normal brain as recognized by PCA. More importantly, we identified a most discriminatory 23-miRNA expression signature, by using PAM, which precisely distinguished glioblastoma from anaplastic astrocytoma with an accuracy of 95%. The differential expression pattern of nine miRNAs was further validated by real-time RT-PCR on an independent set of malignant astrocytomas (n=72) and normal samples (n=7). Inhibition of two glioblastoma-upregulated miRNAs (miR-21 and miR-23a) and exogenous overexpression of two glioblastoma-downregulated miRNAs (miR-218 and miR-219-5p) resulted in reduced soft agar colony formation but showed varying effects on cell proliferation and chemosensitivity. Thus we have identified the miRNA expression signature for malignant astrocytoma, in particular glioblastoma, and showed the miRNA involvement and their importance in astrocytoma development.

Brain fatty acid-binding protein and omega-3/omega-6 fatty acids: mechanistic insight into malignant glioma cell migration

Malignant gliomas (MG) are highly infiltrative tumors that consistently recur despite aggressive treatment. Brain fatty acid-binding protein (FABP7), which binds docosahexaenoic acid (DHA) and arachidonic acid (AA), localizes to sites of tumor infiltration and is associated with a poor prognosis in MG. Manipulation of FABP7 expression in MG cell lines affects cell migration, suggesting a role for FABP7 in tumor infiltration and recurrence. Here, we show that DHA inhibits and AA stimulates migration in an FABP7-dependent manner in U87 MG cells. We demonstrate that DHA binds to and sequesters FABP7 to the nucleus, resulting in decreased cell migration. This anti-migratory effect is partially dependent on peroxisome proliferator-activated receptor γ, a DHA-activated transcription factor. Conversely, AA-bound FABP7 stimulates cell migration by activating cyclooxygenase-2 and reducing peroxisome proliferator-activated receptor γ levels. Our data provide mechanistic insight as to why FABP7 is associated with a poor prognosis in MG and suggest that relative levels of DHA and AA in the tumor environment can make a profound impact on tumor growth properties. We propose that FABP7 and its fatty acid ligands may be key therapeutic targets for controlling the dissemination of MG cells within the brain.

An ANOCEF genomic and transcriptomic microarray study of the response to radiotherapy or to alkylating first-line chemotherapy in glioblastoma patients

BACKGROUND

The molecular characteristics associated with the response to treatment in glioblastomas (GBMs) remain largely unknown. We performed a retrospective study to assess the genomic characteristics associated with the response of GBMs to either first-line chemotherapy or radiation therapy. The gene expression (n = 56) and genomic profiles (n = 67) of responders and non-responders to first-line chemotherapy or radiation therapy alone were compared on Affymetrix Plus 2 gene expression arrays and BAC CGH arrays.

RESULTS

According to Verhaak et al.'s classification system, mesenchymal GBMs were more likely to respond to radiotherapy than to first-line chemotherapy, whereas classical GBMs were more likely to respond to first-line chemotherapy than to radiotherapy. In patients treated with radiation therapy alone, the response was associated with differential expression of microenvironment-associated genes; the expression of hypoxia-related genes was associated with short-term progression-free survival (< 5 months), whereas the expression of immune genes was associated with prolonged progression-free survival (> 10 months). Consistently, infiltration of the tumor by both CD3 and CD68 cells was significantly more frequent in responders to radiotherapy than in non-responders. In patients treated with first-line chemotherapy, the expression of stem-cell genes was associated with resistance to chemotherapy, and there was a significant association between response to treatment and p16 locus deletions. Consistently, in an independent data set of patients treated with either radiotherapy alone or with both radiotherapy and adjuvant chemotherapy, we found that patients with the p16 deletion benefited from adjuvant chemotherapy regardless of their MGMT promoter methylation status, whereas in patients without the p16 deletion, this benefit was only observed in patients with a methylated MGMT promoter.

CONCLUSION

Differential expression of microenvironment genes and p16 locus deletion are associated with responses to radiation therapy and to first-line chemotherapy, respectively, in GBM. Recently identified transcriptomic subgroups of GBMs seem to respond differently to radiotherapy and to first-line chemotherapy.

Maintenance of critical properties of brain tumor stem-like cells after cryopreservation

It would be very useful to be able to classify brain tumor stem cells (BTSCs) by certain criteria to afford the design of specific or individualized treatment. Here, we studied two BTSC lines with differing biological and molecular features and whose respective features were well preserved after cryopreservation as single cells in SFM or 90% serum with 10% DMSO, a method not previously reported. The resuscitated BTSCs shared properties indistinguishable from their respective parental cells, including tumor sphere forming potentials, growth and differentiation properties, and tumorigenesis in vivo. The two cell lines also had differing molecule profiles, which can be well preserved after cryopreservation, similar to that of their respective primary tumors. Therefore, BTSCs from different patients, that have their own properties, were well retained by the present cryopreservation method, which might be a useful and reliable method for preserving BTSCs for long-term studies, such as classification and specific therapy design.

Short-term temperature effect on the HRMAS spectra of human brain tumor biopsies and their pattern recognition analysis

OBJECT

To investigate the effect of temperature (0 versus 37 degrees C) in the high-resolution magic angle spinning spectroscopy (HRMAS) pattern of human brain tumor biopsies and its influence in recognition-based tumor type prediction. This proof-of-principle study addressed the bilateral discrimination between meningioma (MM) and glioblastoma multiforme (GBM) cases.

MATERIALS AND METHODS

Forty-three tumor biopsy samples were collected (20 MM and 23 GBM), kept frozen and later analyzed at 0 degrees C and 37 degrees C by HRMAS. Post-HRMAS histopathology was used to validate the tumor type. Time-course experiments (100 min) at both temperatures were carried out to monitor HRMAS pattern changes. Principal component analysis and linear discriminant analysis were used for classifier development with a training set of 20 biopsies.

RESULTS

Temperature-dependent, spectral pattern changes mostly affected mobile lipids and choline-containing compounds resonances and were essentially reversible. Incubation of 3 MM and 3 GBM at 37 degrees C during 100 minutes produced irreversible pattern changes below 13% in a few resonances. Classification performance of an independent test set of 7 biopsies was 100% for the pulse-and-acquire, CPMG at echo times (TE) of 30 ms and 144 ms and Hahn Echo at TE 30 ms at 0 degrees C and 37 degrees C. The performance for Hahn Echo spectra at 136 ms was 83.3% at 0 degrees C and 100% at 37 degrees C.

CONCLUSION

The spectral pattern of mobile lipids changes reversibly with temperature. HRMAS demonstrated potential for automated brain tumor biopsy classification. No advantage was obtained when acquiring spectra at 37 degrees C with respect to 0 degrees C in most of the conditions used for the discrimination addressed.

Identification of a CpG island methylator phenotype that defines a distinct subgroup of glioma

We have profiled promoter DNA methylation alterations in 272 glioblastoma tumors in the context of The Cancer Genome Atlas (TCGA). We found that a distinct subset of samples displays concerted hypermethylation at a large number of loci, indicating the existence of a glioma-CpG island methylator phenotype (G-CIMP). We validated G-CIMP in a set of non-TCGA glioblastomas and low-grade gliomas. G-CIMP tumors belong to the proneural subgroup, are more prevalent among lower-grade gliomas, display distinct copy-number alterations, and are tightly associated with IDH1 somatic mutations. Patients with G-CIMP tumors are younger at the time of diagnosis and experience significantly improved outcome. These findings identify G-CIMP as a distinct subset of human gliomas on molecular and clinical grounds.

Integrated molecular genetic profiling of pediatric high-grade gliomas reveals key differences with the adult disease

PURPOSE

To define copy number alterations and gene expression signatures underlying pediatric high-grade glioma (HGG).

PATIENTS AND METHODS

We conducted a high-resolution analysis of genomic imbalances in 78 de novo pediatric HGGs, including seven diffuse intrinsic pontine gliomas, and 10 HGGs arising in children who received cranial irradiation for a previous cancer using single nucleotide polymorphism microarray analysis. Gene expression was analyzed with gene expression microarrays for 53 tumors. Results were compared with publicly available data from adult tumors.

RESULTS

Significant differences in copy number alterations distinguish childhood and adult glioblastoma. PDGFRA was the predominant target of focal amplification in childhood HGG, including diffuse intrinsic pontine gliomas, and gene expression analyses supported an important role for deregulated PDGFRalpha signaling in pediatric HGG. No IDH1 hotspot mutations were found in pediatric tumors, highlighting molecular differences with adult secondary glioblastoma. Pediatric and adult glioblastomas were clearly distinguished by frequent gain of chromosome 1q (30% v 9%, respectively) and lower frequency of chromosome 7 gain (13% v 74%, respectively) and 10q loss (35% v 80%, respectively). PDGFRA amplification and 1q gain occurred at significantly higher frequency in irradiation-induced tumors, suggesting that these are initiating events in childhood gliomagenesis. A subset of pediatric HGGs showed minimal copy number changes.

CONCLUSION

Integrated molecular profiling showed substantial differences in the molecular features underlying pediatric and adult HGG, indicating that findings in adult tumors cannot be simply extrapolated to younger patients. PDGFRalpha may be a useful target for pediatric HGG, including diffuse pontine gliomas.

Targeting brain cancer: advances in the molecular pathology of malignant glioma and medulloblastoma

Malignant brain tumours continue to be the cause of a disproportionate level of morbidity and mortality across a wide range of individuals. The most common variants in the adult and paediatric populations - malignant glioma and medulloblastoma, respectively - have been the subject of increasingly intensive research over the past two decades that has led to considerable advances in the understanding of their basic biology and pathogenesis. This Review summarizes these developments in the context of the evolving notion of molecular pathology and discusses the implications that this work has on the design of new treatment regimens.

Massively parallel signature sequencing and bioinformatics analysis identifies up-regulation of TGFBI and SOX4 in human glioblastoma

Background

A comprehensive network-based understanding of molecular pathways abnormally altered in glioblastoma multiforme (GBM) is essential for developing effective therapeutic approaches for this deadly disease.

Methodology/Principal Findings

Applying a next generation sequencing technology, massively parallel signature sequencing (MPSS), we identified a total of 4535 genes that are differentially expressed between normal brain and GBM tissue. The expression changes of three up-regulated genes, CHI3L1, CHI3L2, and FOXM1, and two down-regulated genes, neurogranin and L1CAM, were confirmed by quantitative PCR. Pathway analysis revealed that TGF- β pathway related genes were significantly up-regulated in GBM tumor samples. An integrative pathway analysis of the TGF β signaling network identified two alternative TGF−β signaling pathways mediated by SOX4 (sex determining region Y-box 4) and TGFBI (Transforming growth factor beta induced). Quantitative RT-PCR and immunohistochemistry staining demonstrated that SOX4 and TGFBI expression is elevated in GBM tissues compared with normal brain tissues at both the RNA and protein levels. In vitro functional studies confirmed that TGFBI and SOX4 expression is increased by TGF- β stimulation and decreased by a specific inhibitor of TGF- β receptor 1 kinase.

Conclusions/Significance

Our MPSS database for GBM and normal brain tissues provides a useful resource for the scientific community. The identification of non-SMAD mediated TGF−β signaling pathways acting through SOX4 and TGFBI (GENE ID:7045) in GBM indicates that these alternative pathways should be considered, in addition to the canonical SMAD mediated pathway, in the development of new therapeutic strategies targeting TGF−β signaling in GBM. Finally, the construction of an extended TGF- β signaling network with overlaid gene expression changes between GBM and normal brain extends our understanding of the biology of GBM.

Antiangiogenic therapy and mechanisms of tumor resistance in malignant glioma

Despite advances in surgery, radiation therapy, and chemotherapeutics, patients with malignant glioma have a dismal prognosis. The formations of aberrant tumour vasculature and glioma cell invasion are major obstacles for effective treatment. Angiogenesis is a key event in the progression of malignant gliomas, a process involving endothelial cell proliferation, migration, reorganization of extracellular matrix and tube formation. Such processes are regulated by the homeostatic balance between proangiogenic and antiangiogenic factors, most notably vascular endothelial growth factors (VEGFs) produced by glioma cells. Current strategies targeting VEGF-VEGF receptor signal transduction pathways, though effective in normalizing abnormal tumor vasculature, eventually result in tumor resistance whereby a highly infiltrative and invasive phenotype may be adopted. Here we review recent anti-angiogenic therapy for malignant glioma and highlight implantable devices and nano/microparticles as next-generation methods for chemotherapeutic delivery. Intrinsic and adaptive modes of glioma resistance to anti-angiogenic therapy will be discussed with particular focus on the glioma stem cell paradigm.

A role for fibrillar collagen deposition and the collagen internalization receptor endo180 in glioma invasion

BACKGROUND

Glioblastoma multiforme (GBM, WHO grade IV) is the most common and most malignant of astrocytic brain tumors, and is associated with rapid invasion into neighboring tissue. In other tumor types it is well established that such invasion involves a complex interaction between tumor cells and locally produced extracellular matrix. In GBMs, surprisingly little is known about the associated matrix components, in particular the fibrillar proteins such as collagens that are known to play a key role in the invasion of other tumor types.

METHODOLOGY/PRINCIPAL FINDINGS

In this study we have used both the Masson's trichrome staining and a high resolution multiple immunofluorescence labeling method to demonstrate that intratumoral fibrillar collagens are an integral part of the extracellular matrix in a subset of GBMs. Correlated with this collagen deposition we observed high level expression of the collagen-binding receptor Endo180 (CD280) in the tumor cells. Further, interrogation of multiple expression array datasets identified Endo180 as one of the most highly upregulated transcripts in grade IV GBMs compared to grade III gliomas. Using promoter analysis studies we show that this increased expression is, in part, mediated via TGF-beta signaling. Functionally, we demonstrate that Endo180 serves as the major collagen internalization receptor in GBM cell lines and provide the first evidence that this activity is critical for the invasion of GBM cells through fibrillar collagen matrices.

CONCLUSIONS/SIGNIFICANCE

This study demonstrates, for the first time, that fibrillar collagens are extensively deposited in GBMs and that the collagen internalization receptor Endo180 is both highly expressed in these tumors and that it serves to mediate the invasion of tumor cells through collagen-containing matrices. Together these data provide important insights into the mechanism of GBM invasion and identify Endo180 as a potential target to limit matrix turnover by glioma cells and thereby restrict tumor progression.

CD44 attenuates activation of the hippo signaling pathway and is a prime therapeutic target for glioblastoma

Glioblastoma multiforme (GBM) is the most aggressive brain tumor that, by virtue of its resistance to chemotherapy and radiotherapy, is currently incurable. Identification of molecules whose targeting may eliminate GBM cells and/or sensitize glioblastoma cells to cytotoxic drugs is therefore urgently needed. CD44 is a major cell surface hyaluronan receptor and cancer stem cell marker that has been implicated in the progression of a variety of cancer types. However, the major downstream signaling pathways that mediate its protumor effects and the role of CD44 in the progression and chemoresponse of GBM have not been established. Here we show that CD44 is upregulated in GBM and that its depletion blocks GBM growth and sensitizes GBM cells to cytotoxic drugs in vivo. Consistent with this observation, CD44 antagonists potently inhibit glioma growth in preclinical mouse models. We provide the first evidence that CD44 functions upstream of the mammalian Hippo signaling pathway and that CD44 promotes tumor cell resistance to reactive oxygen species-induced and cytotoxic agent-induced stress by attenuating activation of the Hippo signaling pathway. Together, our results identify CD44 as a prime therapeutic target for GBM, establish potent antiglioma efficacy of CD44 antagonists, uncover a novel CD44 signaling pathway, and provide a first mechanistic explanation as to how upregulation of CD44 may constitute a key event in leading to cancer cell resistance to stresses of different origins. Finally, our results provide a rational explanation for the observation that functional inhibition of CD44 augments the efficacy of chemotherapy and radiation therapy.

Functional genomics to explore cancer cell vulnerabilities

Our understanding of glioblastoma multiforme (GBM), the most common form of primary brain cancer, has been significantly advanced by recent efforts to characterize the cancer genome using unbiased high-throughput sequencing analyses. While these studies have documented hundreds of mutations, gene copy alterations, and chromosomal abnormalities, only a subset of these alterations are likely to impact tumor initiation or maintenance. Furthermore, genes that are not altered at the genomic level may play essential roles in tumor initiation and maintenance. Identification of these genes is critical for therapeutic development and investigative methodologies that afford insight into biological function. This requirement has largely been fulfilled with the emergence of RNA interference (RNAi) and high-throughput screening technology. In this article, the authors discuss the application of genome-wide, high-throughput RNAi-based genetic screening as a powerful tool for the rapid and cost-effective identification of genes essential for cancer proliferation and survival. They describe how these technologies have been used to identify genes that are themselves selectively lethal to cancer cells, or synthetically lethal with other oncogenic mutations. The article is intended to provide a platform for how RNAi libraries might contribute to uncovering glioma cell vulnerabilities and provide information that is highly complementary to the structural characterization of the glioblastoma genome. The authors emphasize that unbiased, systems-level structural and functional genetic approaches are complementary efforts that should facilitate the identification of genes involved in the pathogenesis of GBM and permit the identification of novel drug targets.

The potential of biologic network models in understanding the etiopathogenesis of ovarian cancer

Ovarian cancer is one of the most common gynecologic malignancies and is the 5th leading cause of cancer mortality in women in the United States. Understanding the biology and molecular pathogenesis of ovarian epithelial tumors is key to developing improved prognostic indicators and effective therapies. The selection of ovarian serous carcinomas as one of the three cancer types for extensive genomic and proteomic characterization of The Cancer Genome Atlas (TCGA) project offers an important opportunity to extend our knowledge of ovarian cancer. The data portal includes molecular characterization, high throughput sequencing, and clinical data. Models to determine which of these genes act as "key drivers" of ovarian carcinogenesis and which are innocent "passengers" are needed. Standard statistical approaches often fail to differentiate between these driver and passenger genes, given that the correlation between sets of genes or genes and endpoints alone does not establish causality. As contrasted to basic correlations analyses, biological network models offer the ability to resolve causality by elucidating the directional linkages between genetics, molecular characterizations of the system, and clinical measures. This article describes the use of a novel, supercomputer-driven approach named REFS to learn network models directly from the TGCA ovarian cancer data set and simulate these models to learn the "key drivers" of ovarian carcinogenesis. The model can be validated by out-of-sample testing, and may provide a powerful new tool for ovarian cancer research.

Automated brain tumor biopsy prediction using single-labeling cDNA microarrays-based gene expression profiling

AIMS

Gene signatures obtained from microarray experiments may be of use to improve the prediction of brain tumor diagnosis. Nevertheless, automated and objective prediction with accuracy comparable to or better than the gold standard should be convincingly demonstrated for possible clinician uptake of the new methodology. Herewith, we demonstrate that primary brain tumor types can be discriminated using microarray data in an automated and objective way.

METHODS

Postsurgical biopsies from 35 patients [17 glioblastoma multiforme (Gbm) and 18 meningothelial meningioma (Mm)] were stored in liquid nitrogen, total RNA was extracted, and cDNA was labeled with Cy3 fluorochrome and hybridized onto a cDNA-based microarray containing 11,500 cDNA clones representing 9300 loci. Scanned data were preprocessed, normalized, and used for predictor development. The predictive functions were fitted to a subset of samples and their performance evaluated with an independent subset. Expression results were validated by means of real time-polymerase chain reaction.

RESULTS

Some gene expression-based predictors achieved 100% accuracy both in training resampling validation and independent testing. One of them, composed of GFAP, PTPRZ1, GPM6B and PRELP, produced a 100% prediction accuracy for both training and independent test datasets. Furthermore, the gene signatures obtained, increased cell detoxification, motility and intracellular transport in Gbm, and increased cell adhesion and cytochrome-family genes in Mm, agree well with the expected biologic and pathologic characteristics of the studied tumors.

CONCLUSIONS

The ability of gene signatures to automate prediction of brain tumors through a fully objective approach has been demonstrated. A comparison of gene expression profiles between Gbm and Mm may provide additional clues about patterns associated with each tumor type.

Mutant EGFR is required for maintenance of glioma growth in vivo, and its ablation leads to escape from receptor dependence

Epidermal growth factor receptor (EGFR) gene amplification is the most common genetic alteration in high-grade glioma, and approximately 50% of EGFR-amplified tumors also harbor a constitutively active mutant form of the receptor, DeltaEGFR. Although DeltaEGFR greatly enhances tumor growth and is thus an attractive target for anti-glioma therapies, recent clinical experiences with EGFR kinase inhibitors have been disappointing, because resistance is common and tumors eventually recur. Interestingly, it has not been established whether DeltaEGFR is required for maintenance of glioma growth in vivo, and, by extension, if it truly represents a rational therapeutic target. Here, we demonstrate that in vivo silencing of regulatable DeltaEGFR with doxycycline attenuates glioma growth and, therefore, that it is crucial for maintenance of enhanced tumorigenicity. Similar to the clinical experience, tumors eventually regained aggressive growth after a period of stasis, but interestingly, without re-expression of DeltaEGFR. To determine how tumors acquired this ability, we found that a unique gene, KLHDC8, herein referred to as SDeltaE (Substitute for DeltaEGFR Expression)-1, is highly expressed in these tumors, which have escaped dependence on DeltaEGFR. SDeltaE-1 is also expressed in human gliomas and knockdown of its expression in DeltaEGFR-independent "escaper" tumors suppressed tumor growth. Taken together, we conclude that DeltaEGFR is required for both glioma establishment and maintenance, and that gliomas undergo selective pressure in vivo to employ alternative compensatory pathways to maintain aggressiveness in the event of EGFR silencing. Such alternative pathways function as substitutes for DeltaEGFR signaling and should therefore be considered as potential targets for additional therapy.

Integrated genomic analysis identifies clinically relevant subtypes of glioblastoma characterized by abnormalities in PDGFRA, IDH1, EGFR, and NF1

The Cancer Genome Atlas Network recently cataloged recurrent genomic abnormalities in glioblastoma multiforme (GBM). We describe a robust gene expression-based molecular classification of GBM into Proneural, Neural, Classical, and Mesenchymal subtypes and integrate multidimensional genomic data to establish patterns of somatic mutations and DNA copy number. Aberrations and gene expression of EGFR, NF1, and PDGFRA/IDH1 each define the Classical, Mesenchymal, and Proneural subtypes, respectively. Gene signatures of normal brain cell types show a strong relationship between subtypes and different neural lineages. Additionally, response to aggressive therapy differs by subtype, with the greatest benefit in the Classical subtype and no benefit in the Proneural subtype. We provide a framework that unifies transcriptomic and genomic dimensions for GBM molecular stratification with important implications for future studies.

Reversing HOXA9 oncogene activation by PI3K inhibition: epigenetic mechanism and prognostic significance in human glioblastoma

HOXA genes encode critical transcriptional regulators of embryonic development that have been implicated in cancer. In this study, we documented functional relevance and mechanism of activation of HOXA9 in glioblastoma (GBM), the most common malignant brain tumor. Expression of HOXA genes was investigated using reverse transcription-PCR in primary gliomas and glioblastoma cell lines and was validated in two sets of expression array data. In a subset of GBM, HOXA genes are aberrently activated within confined chromosomal domains. Transcriptional activation of the HOXA cluster was reversible by a phosphoinostide 3-kinase (PI3K) inhibitor through an epigenetic mechanism involving histone H3K27 trimethylation. Functional studies of HOXA9 showed its capacity to decrease apoptosis and increase cellular proliferation along with tumor necrosis factor-related apoptosis-including ligand resistance. Notably, aberrant expression of HOXA9 was independently predictive of shorter overall and progression-free survival in two GBM patient sets and improved survival prediction by MGMT promoter methylation. Thus, HOXA9 activation is a novel, independent, and negative prognostic marker in GBM that is reversible through a PI3K-associated epigenetic mechanism. Our findings suggest a transcriptional pathway through which PI3K activates oncogenic HOXA expression with implications for mTOR or PI3K targeted therapies.

Aberrant fatty acid-binding protein-7 gene expression in cutaneous malignant melanoma

Fatty acid-binding protein-7 (FABP7) has been shown to be expressed in cutaneous melanoma; however, its role in tumor progression is unclear. Expression of FABP7 was assessed during melanoma progression through assessment of various clinicopathology stages of primary tumor progression and metastasis. FABP7 mRNA was highly expressed in 60 of 87 (69%) primary melanomas, compared with significant (P<0.0001) reduction in 13 of 68 (19%) metastatic melanomas. Analysis of 37 paired primary and metastatic melanomas by immunohistochemistry with anti-FABP7 Ab showed 73 and 27% positivity, respectively (P<0.001). FABP7 detection of metastatic tissues was inversely correlated with relapse-free (P<0.0001) and overall (P<0.0001) survival. To examine FABP7 expression loss in advanced melanomas, loss of heterozygosity (LOH) was assessed using microsatellite markers encompassing the FABP7 gene. LOH was identified in 10 of 20 (50%) metastatic melanomas at 6q22.31, compared with 0 of 14 primary melanomas (P=0.0017). FABP7 as a surrogate biomarker for circulating tumor cells (CTCs) in the blood was assessed by quantitative real-time (qRT)-PCR from melanoma patients' blood (n=134). Assessment of patients' blood showed that FABP7(+) CTC decreased with disease progression. FABP7 may function as a tumor progression gene and can be used as a potential diagnostic biomarker of early-stage melanoma systemic spreading in blood.

Intrinsic gene expression profiles of gliomas are a better predictor of survival than histology

Gliomas are the most common primary brain tumors with heterogeneous morphology and variable prognosis. Treatment decisions in patients rely mainly on histologic classification and clinical parameters. However, differences between histologic subclasses and grades are subtle, and classifying gliomas is subject to a large interobserver variability. To improve current classification standards, we have performed gene expression profiling on a large cohort of glioma samples of all histologic subtypes and grades. We identified seven distinct molecular subgroups that correlate with survival. These include two favorable prognostic subgroups (median survival, >4.7 years), two with intermediate prognosis (median survival, 1-4 years), two with poor prognosis (median survival, <1 year), and one control group. The intrinsic molecular subtypes of glioma are different from histologic subgroups and correlate better to patient survival. The prognostic value of molecular subgroups was validated on five independent sample cohorts (The Cancer Genome Atlas, Repository for Molecular Brain Neoplasia Data, GSE12907, GSE4271, and Li and colleagues). The power of intrinsic subtyping is shown by its ability to identify a subset of prognostically favorable tumors within an external data set that contains only histologically confirmed glioblastomas (GBM). Specific genetic changes (epidermal growth factor receptor amplification, IDH1 mutation, and 1p/19q loss of heterozygosity) segregate in distinct molecular subgroups. We identified a subgroup with molecular features associated with secondary GBM, suggesting that different genetic changes drive gene expression profiles. Finally, we assessed response to treatment in molecular subgroups. Our data provide compelling evidence that expression profiling is a more accurate and objective method to classify gliomas than histologic classification. Molecular classification therefore may aid diagnosis and can guide clinical decision making.

Glioblastoma subclasses can be defined by activity among signal transduction pathways and associated genomic alterations

BACKGROUND

Glioblastoma multiforme (GBM) is an umbrella designation that includes a heterogeneous group of primary brain tumors. Several classification strategies of GBM have been reported, some by clinical course and others by resemblance to cell types either in the adult or during development. From a practical and therapeutic standpoint, classifying GBMs by signal transduction pathway activation and by mutation in pathway member genes may be particularly valuable for the development of targeted therapies.

METHODOLOGY/PRINCIPAL FINDINGS

We performed targeted proteomic analysis of 27 surgical glioma samples to identify patterns of coordinate activation among glioma-relevant signal transduction pathways, then compared these results with integrated analysis of genomic and expression data of 243 GBM samples from The Cancer Genome Atlas (TCGA). In the pattern of signaling, three subclasses of GBM emerge which appear to be associated with predominance of EGFR activation, PDGFR activation, or loss of the RAS regulator NF1. The EGFR signaling class has prominent Notch pathway activation measured by elevated expression of Notch ligands, cleaved Notch receptor, and downstream target Hes1. The PDGF class showed high levels of PDGFB ligand and phosphorylation of PDGFRbeta and NFKB. NF1-loss was associated with lower overall MAPK and PI3K activation and relative overexpression of the mesenchymal marker YKL40. These three signaling classes appear to correspond with distinct transcriptomal subclasses of primary GBM samples from TCGA for which copy number aberration and mutation of EGFR, PDGFRA, and NF1 are signature events.

CONCLUSIONS/SIGNIFICANCE

Proteomic analysis of GBM samples revealed three patterns of expression and activation of proteins in glioma-relevant signaling pathways. These three classes are comprised of roughly equal numbers showing either EGFR activation associated with amplification and mutation of the receptor, PDGF-pathway activation that is primarily ligand-driven, or loss of NF1 expression. The associated signaling activities correlating with these sentinel alterations provide insight into glioma biology and therapeutic strategies.

Repeated hepatocyte growth factor neutralizing antibody treatment leads to HGF/SF unresponsiveness in human glioblastoma multiforme cells

The purpose of this work is to seek putative markers for multi-targeted therapeutic treatment of human glioblastoma. We previously developed an anti-HGF neutralizing antibody cocktail Amix that inhibits human glioblastoma growth in mouse xenograft models. When these treated tumors were re-injected into nude mice and treatment with the neutralizing antibody cocktail plus heparin was repeated, the growth of the twice-treated tumors became HGF-independent, suggesting a possible switch in dominant signaling pathways. Microarray of the tumor cells revealed a number of genes elevated in the twice-treated tumor cells relative to untreated control tumors, including BAI1, CASP8, IL8, IGF1, TGFB1 and TNF. Our analyses provide a series of putative markers for additional evaluation in treating glioblastoma. Multi-targeted therapeutic approach might be a better solution for treating this disease.

Molecular predictors of progression-free and overall survival in patients with newly diagnosed glioblastoma: a prospective translational study of the German Glioma Network

PURPOSE

The prognostic value of genetic alterations characteristic of glioblastoma in patients treated according to present standards of care is unclear.

PATIENTS AND METHODS

Three hundred one patients with glioblastoma were prospectively recruited between October 2004 and December 2006 at the clinical centers of the German Glioma Network. Two hundred fifty-eight patients had radiotherapy, 199 patients had temozolomide, 189 had both, and seven had another chemotherapy as the initial treatment. The tumors were investigated for TP53 mutation, p53 immunoreactivity, epidermal growth factor receptor, cyclin-dependent kinase CDK 4 or murine double minute 2 amplification, CDKN2A homozygous deletion, allelic losses on chromosome arms 1p, 9p, 10q, and 19q, O(6)-methylguanine methyltransferase (MGMT) promoter methylation, and isocitrate dehydrogenase 1 (IDH1) mutations.

RESULTS

Median progression-free (PFS) and overall survival (OS) were 6.8 and 12.5 months. Multivariate analysis revealed younger age, higher performance score, MGMT promoter methylation, and temozolomide radiochemotherapy as independent factors associated with longer OS. MGMT promoter methylation was associated with longer PFS (relative risk [RR], 0.5; 95% CI, 0.38 to 0.68; P < .001) and OS (RR, 0.39; 95% CI, 0.28 to 0.54; P < .001) in patients receiving temozolomide. IDH1 mutations were associated with prolonged PFS (RR, 0.42; 95% CI, 0.19 to 0.91; P = .028) and a trend for prolonged OS (RR, 0.43; 95% CI, 0.15 to 1.19; P = .10). No other molecular factor was associated with outcome.

CONCLUSION

Molecular changes associated with gliomagenesis do not predict response to therapy in glioblastoma patients managed according to current standards of care. MGMT promoter methylation and IDH1 mutational status allow for stratification into prognostically distinct subgroups.

Recent insights into PDGF-induced gliomagenesis

Gliomas are aggressive and almost incurable glial brain tumors which frequently display abnormal platelet-derived growth factor (PDGF) signaling. Evidence gained from studies on several in vivo animal models has firmly established a causal connection between aberrant PDGF signaling and the formation of some gliomas. However, only recently has significant knowledge been gained regarding crucial issues such as the glioma cell of origin and the relationship between the transforming stimulus and the cellular characteristics of the resulting tumor. Based on recent evidence, we propose that PDGF can bias cell-fate decisions, driving the acquisition of cell type-specific features by the progeny of multipotent neural progenitors, thus determining the shape and direction of the transformation path. Furthermore, recent data about the cellular mechanisms of PDGF-driven glioma progression and maintenance indicate that PDGF may be required, unexpectedly, to override cell contact inhibition and promote glioma cell infiltration rather than to stimulate cell proliferation.

Gene expression analysis of an EGFR indirectly related pathway identified PTEN and MMP9 as reliable diagnostic markers for human glial tumor specimens

In this study the mRNA levels of five EGFR indirectly related genes, EGFR, HB-EGF, ADAM17, PTEN, and MMP9, have been assessed by Real-time PCR in a panel of 37 glioblastoma multiforme specimens and in 5 normal brain samples; as a result, in glioblastoma, ADAM17 and PTEN expression was significantly lower than in normal brain samples, and, in particular, a statistically significant inverse correlation was found between PTEN and MMP9 mRNA levels. To verify if this correlation was conserved in gliomas, PTEN and MMP9 expression was further investigated in an additional panel of 16 anaplastic astrocytoma specimens and, in parallel, in different human normal and astrocytic tumor cell lines. In anaplastic astrocytomas PTEN expression was significantly higher than in glioblastoma multiforme, but no significant correlation was found between PTEN and MMP9 expression. PTEN and MMP9 mRNA levels were also employed to identify subgroups of specimens within the different glioma malignancy grades and to define a gene expression-based diagnostic classification scheme. In conclusion, this gene expression survey highlighted that the combined measurement of PTEN and MMP9 transcripts might represent a novel reliable tool for the differential diagnosis of high-grade gliomas, and it also suggested a functional link involving these genes in glial tumors.

Beyond grade: molecular pathology of malignant gliomas

High-grade gliomas (HGGs) represent a heterogenous group of tumors and account for most primary brain tumors. Despite aggressive therapies, they are invariably associated with poor patient outcome. These tumors include the anaplastic (World Health Organization [WHO] grade III) histologies of astrocytomas, oligodendrogliomas, and ependymomas and the WHO grade IV glioblastoma multiforme (GBM). The recent elucidation of the fundamental molecular alterations associated with these tumors has begun to unravel the critical events in their tumorigenesis but for the most part has done little to alter patient survival. Prognostication for patients with these tumors has relied principally on tumor grade and clinical factors (age, performance status, and so on) and has been inexact at best in identifying those with long-term survival potential. An even greater challenge has been to identify predictive biomarkers of therapy in the hope of tailoring a patient's therapy based on their tumor's molecular characteristics. This review discusses the molecular pathology of high-grade gliomas, with particular emphasis on anaplastic astrocytomas and GBMs because these represent the most common forms of malignant gliomas. It also focuses on the molecular signatures defined by large-scale gene expression profiling experiments because these studies are at the forefront in developing new biomarkers and identifying new therapeutic targets.

A systems biology-based gene expression classifier of glioblastoma predicts survival with solid tumors

Accurate prediction of survival of cancer patients is still a key open problem in clinical research. Recently, many large-scale gene expression clusterings have identified sets of genes reportedly predictive of prognosis; however, those gene sets shared few genes in common and were poorly validated using independent data. We have developed a systems biology-based approach by using either combined gene sets and the protein interaction network (Method A) or the protein network alone (Method B) to identify common prognostic genes based on microarray gene expression data of glioblastoma multiforme and compared with differential gene expression clustering (Method C). Validations of prediction performance show that the 23-prognostic gene classifier identified by Method A outperforms other gene classifiers identified by Methods B and C or previously reported for gliomas on 17 of 20 independent sample cohorts across five tumor types. We also find that among the 23 genes are 21 related to cellular proliferation and two related to response to stress/immune response. We further find that the increased expression of the 21 genes and the decreased expression of the other two genes are associated with poorer survival, which is supportive with the notion that cellular proliferation and immune response contribute to a significant portion of predictive power of prognostic classifiers. Our results demonstrate that the systems biology-based approach enables to identify common survival-associated genes.

OMICS and brain tumour biomarkers

Currently, brain tumours are diagnosed by surgical biopsy and light microscopic examination of tissue, with immunohistochemistry in difficult cases. We review research in the field of brain tumour diagnosis and discuss several new approaches. In future, tumour type, optimal treatment, and prognosis could be obtained by studying the gene (genomics), protein (proteomics) or metabolite (metabolomics) content of tumour cells. These techniques generate complex data, analysed using techniques such as pattern recognition software to identify biomarker signatures of different tumours. Compared with individual biomarkers, biomarker signatures appear to increase diagnostic accuracy and may produce an improved brain tumour classification system.

Nuclear factor I regulates brain fatty acid-binding protein and glial fibrillary acidic protein gene expression in malignant glioma cell lines

Glial fibrillary acidic protein (GFAP), an intermediate filament protein normally found in astrocytes, and the radial glial marker brain fatty acid-binding protein (B-FABP; also known as FABP7) are co-expressed in malignant glioma cell lines and tumors. Nuclear factor I (NFI) recognition sites have been identified in the B-FABP and GFAP promoters, and transcription of both genes is believed to be regulated by NFI. Here, we study the role of the different members of the NFI family in regulating endogenous and ectopic B-FABP and GFAP gene transcription in human malignant glioma cells. We show by gel shifts that all four members of the NFI family (NFIA, NFIB, NFIC, and NFIX) bind to B-FABP and GFAP NFI consensus sites. Over-expression of NFIs, in conjunction with mutation analysis of NFI consensus sites using a reporter gene assay, supports a role for all four NFIs in the regulation of the GFAP and B-FABP genes. Knock-down of single or combined NFIs reveals promoter-dependent and promoter-context-dependent interaction patterns and suggests cross talk between the different members of the NFI family. Our data indicate that the NFI family of transcription factors plays a key role in the regulation of both the B-FABP and GFAP genes in malignant glioma cells.

ROCKs are expressed in brain tumors and are required for glioma-cell migration on myelinated axons

The interactions between migrating glioma cells and myelinated fiber tracts are poorly understood. We identified that C6 glioma cells can migrate along myelinated chicken retinal axons in a novel coculture, thereby expressing small GTPases of the Rho family and serine/threonine Rho-associated kinases (ROCKs). We found that the ROCK1 isoform is also highly expressed in native human high-grade gliomas. Glioma cells migrated faster in vitro along myelinated axons than on laminin-1, with the former but not the latter being specifically and reversibly blocked by the ROCK inhibitor Y27632. These data suggest that the mechanisms underlying the migration of glioma cells on myelinated axons differ from those underlying the migration on extracellular matrix molecules such as laminin-1.

Radiogenomics: creating a link between molecular diagnostics and diagnostic imaging

Studies employing high-throughput biological techniques have recently contributed to an improved characterization of human cancers, allowing for novel sub-classification, better diagnostic accuracy, and more precise prognostication. However, requirement of surgical procurement of tissue among other things limits the clinical application of such methods in everyday patient care. Radiographic imaging is routine in clinical practice but is currently histopathology based. The use of routine radiographic imaging provides a potential platform for linking specific imaging traits with specific gene expression patterns that inform the underlying cellular pathophysiology; imaging features could then serve as molecular surrogates that contribute to the diagnosis, prognosis, and likely gene-expression-associated treatment response of various forms of human cancer. This review focuses on high-throughput methods such as microarray analysis of gene expression, their role in cancer research, and in particular, on novel methods of associating gene expression patterns with radiographic imaging phenotypes, known as "radiogenomics." These findings underline a potential future role of both diagnostic and interventional radiologists in genetic assessment of cancer patients with radiographic imaging studies.

APRIL is overexpressed in cancer: link with tumor progression

Background

BAFF and APRIL share two receptors – TACI and BCMA – and BAFF binds to a third receptor, BAFF-R. Increased expression of BAFF and APRIL is noted in hematological malignancies. BAFF and APRIL are essential for the survival of normal and malignant B lymphocytes, and altered expression of BAFF or APRIL or of their receptors (BCMA, TACI, or BAFF-R) have been reported in various B-cell malignancies including B-cell non-Hodgkin's lymphoma, chronic lymphocytic leukemia, Hodgkin's lymphoma, multiple myeloma, and Waldenstrom's macroglobulinemia.

Methods

We compared the expression of BAFF, APRIL, TACI and BAFF-R gene expression in 40 human tumor types – brain, epithelial, lymphoid, germ cells – to that of their normal tissue counterparts using publicly available gene expression data, including the Oncomine Cancer Microarray database.

Results

We found significant overexpression of TACI in multiple myeloma and thyroid carcinoma and an association between TACI expression and prognosis in lymphoma. Furthermore, BAFF and APRIL are overexpressed in many cancers and we show that APRIL expression is associated with tumor progression. We also found overexpression of at least one proteoglycan with heparan sulfate chains (HS), which are coreceptors for APRIL and TACI, in tumors where APRIL is either overexpressed or is a prognostic factor. APRIL could induce survival or proliferation directly through HS proteoglycans.

Conclusion

Taken together, these data suggest that APRIL is a potential prognostic factor for a large array of malignancies.

Unsupervised analysis of transcriptomic profiles reveals six glioma subtypes

Gliomas are the most common type of primary brain tumors in adults and a significant cause of cancer-related mortality. Defining glioma subtypes based on objective genetic and molecular signatures may allow for a more rational, patient-specific approach to therapy in the future. Classifications based on gene expression data have been attempted in the past with varying success and with only some concordance between studies, possibly due to inherent bias that can be introduced through the use of analytic methodologies that make a priori selection of genes before classification. To overcome this potential source of bias, we have applied two unsupervised machine learning methods to genome-wide gene expression profiles of 159 gliomas, thereby establishing a robust glioma classification model relying only on the molecular data. The model predicts for two major groups of gliomas (oligodendroglioma-rich and glioblastoma-rich groups) separable into six hierarchically nested subtypes. We then identified six sets of classifiers that can be used to assign any given glioma to the corresponding subtype and validated these classifiers using both internal (189 additional independent samples) and two external data sets (341 patients). Application of the classification system to the external glioma data sets allowed us to identify previously unrecognized prognostic groups within previously published data and within The Cancer Genome Atlas glioblastoma samples and the different biological pathways associated with the different glioma subtypes offering a potential clue to the pathogenesis and possibly therapeutic targets for tumors within each subtype.

Glioblastoma inhibition by cell surface immunoglobulin protein EWI-2, in vitro and in vivo

EWI-2, a cell surface IgSF protein, is highly expressed in normal human brain but is considerably diminished in glioblastoma tumors and cell lines. Moreover, loss of EWI-2 expression correlated with a shorter survival time in human glioma patients, suggesting that EWI-2 might be a natural inhibitor of glioblastoma. In support of this idea, EWI-2 expression significantly impaired both ectopic and orthotopic tumor growth in nude mice in vivo. In vitro assays provided clues regarding EWI-2 functions. Expression of EWI-2 in T98G and/or U87-MG malignant glioblastoma cell lines failed to alter two-dimensional cell proliferation but inhibited glioblastoma colony formation in soft agar and caused diminished cell motility and invasion. At the biochemical level, EWI-2 markedly affects the organization of four molecules (tetraspanin proteins CD9 and CD81 and matrix metalloproteinases MMP-2 and MT1-MMP), which play key roles in the biology of astrocytes and gliomas. EWI-2 causes CD9 and CD81 to become more associated with each other, whereas CD81 and other tetraspanins become less associated with MMP-2 and MT1-MMP. We propose that EWI-2 inhibition of glioblastoma growth in vivo is at least partly explained by the capability of EWI-2 to inhibit growth and/or invasion in vitro. Underlying these functional effects, EWI-2 causes a substantial molecular reorganization of multiple molecules (CD81, CD9, MMP-2, and MT1-MMP) known to affect proliferation and/or invasion of astrocytes and/or glioblastomas.

Heterogeneous activation of the TGFbeta pathway in glioblastomas identified by gene expression-based classification using TGFbeta-responsive genes

BACKGROUND

TGFbeta has emerged as an attractive target for the therapeutic intervention of glioblastomas. Aberrant TGFbeta overproduction in glioblastoma and other high-grade gliomas has been reported, however, to date, none of these reports has systematically examined the components of TGFbeta signaling to gain a comprehensive view of TGFbeta activation in large cohorts of human glioma patients.

METHODS

TGFbeta activation in mammalian cells leads to a transcriptional program that typically affects 5-10% of the genes in the genome. To systematically examine the status of TGFbeta activation in high-grade glial tumors, we compiled a gene set of transcriptional response to TGFbeta stimulation from tissue culture and in vivo animal studies. These genes were used to examine the status of TGFbeta activation in high-grade gliomas including a large cohort of glioblastomas. Unsupervised and supervised classification analysis was performed in two independent, publicly available glioma microarray datasets.

RESULTS

Unsupervised and supervised classification using the TGFbeta-responsive gene list in two independent glial tumor gene expression data sets revealed various levels of TGFbeta activation in these tumors. Among glioblastomas, one of the most devastating human cancers, two subgroups were identified that showed distinct TGFbeta activation patterns as measured from transcriptional responses. Approximately 62% of glioblastoma samples analyzed showed strong TGFbeta activation, while the rest showed a weak TGFbeta transcriptional response.

CONCLUSION

Our findings suggest heterogeneous TGFbeta activation in glioblastomas, which may cause potential differences in responses to anti-TGFbeta therapies in these two distinct subgroups of glioblastomas patients.

The role of DNA synthesis imaging in cancer in the era of targeted therapeutics

Non-specific targets such as DNA and microtubules have been the mainstay of cancer therapeutics and the most effective clinical agents until a decade ago. Advances in genetics, molecular and cellular biology over the past decade led to the development of a new generation of agents that are far more specific and effective. In contrast to progress seen with therapeutic agents, general monitoring targets such as proliferation imaging are just gaining momentum and targeted imaging is still in its infancy. In these paradoxical times, this review assesses the role of proliferation imaging in monitoring the efficacy of targeted therapeutics.

Clustering cancer gene expression data: a comparative study

Background

The use of clustering methods for the discovery of cancer subtypes has drawn a great deal of attention in the scientific community. While bioinformaticians have proposed new clustering methods that take advantage of characteristics of the gene expression data, the medical community has a preference for using "classic" clustering methods. There have been no studies thus far performing a large-scale evaluation of different clustering methods in this context.

Results/Conclusion

We present the first large-scale analysis of seven different clustering methods and four proximity measures for the analysis of 35 cancer gene expression data sets. Our results reveal that the finite mixture of Gaussians, followed closely by k-means, exhibited the best performance in terms of recovering the true structure of the data sets. These methods also exhibited, on average, the smallest difference between the actual number of classes in the data sets and the best number of clusters as indicated by our validation criteria. Furthermore, hierarchical methods, which have been widely used by the medical community, exhibited a poorer recovery performance than that of the other methods evaluated. Moreover, as a stable basis for the assessment and comparison of different clustering methods for cancer gene expression data, this study provides a common group of data sets (benchmark data sets) to be shared among researchers and used for comparisons with new methods. The data sets analyzed in this study are available at .

The influence of feeding linoleic, gamma-linolenic and docosahexaenoic acid rich oils on rat brain tumor fatty acids composition and fatty acid binding protein 7 mRNA expression

Background

Experimental studies indicate that gamma linolenic acid (GLA) and docosahexaenoic acid (DHA) may inhibit glioma cells growth but effects of oral consumption of these fatty acids on brain tumor fatty acid composition have not been determined in vivo.

Methods

GLA oil (GLAO; 72% GLA), DHA oil (DHAO; 73% DHA) were fed to adult wistar rats (1 mL/rat/day) starting one week prior to C6 glioma cells implantation and continued for two weeks after implantation. Control group were fed same amount of high linoleic acid safflower oil (74–77% linoleic acid). Fatty acid composition of tumor samples was determined in a set of 8–12 animals in each group and serum fatty acid in 6 animals per each group. Gene expression of tumor fatty acid binding protein 7 (FABP7), epidermal growth factor receptor (EGFR), peroxisome proliferator activated receptor γ (PPAR-γ) and retinoid × receptor-α (RXR-α) were determined in a set of 18 animals per group.

Results

DHAO feeding increased EPA of brain tumors and decreased ratio of n-6/n-3 fatty acids. Serum levels of EPA were also increased in DHAO group. A similar trend in serum and tumor levels of DHA were observed in DHAO group but it did not achieve statistical significance. GLAO increased serum concentration of GLA but had no significant effect on tumor GLA or dihomo-gamma linolenic acid (DGLA) concentrations. Gene expression of FABP7 was up-regulated in tumors of DHAO group but no other significant effects were observed on EGFR, PPAR-γ or RXR-α expression, and expression of these genes in tumors of GLAO were not different from SFO group.

Conclusion

Dietary supplementation of DHA containing oil could be an effective way to increase levels of long chain n-3 fatty acids in brain tumors and this increase may be mediated partly by up-regulation of FABP7 expression.

Development of a novel ozone- and photo-stable HyPer5 red fluorescent dye for array CGH and microarray gene expression analysis with consistent performance irrespective of environmental conditions

Background

Array-based comparative genomic hybridization (CGH) and gene expression profiling have become vital techniques for identifying molecular defects underlying genetic diseases. Regardless of the microarray platform, cyanine dyes (Cy3 and Cy5) are one of the most widely used fluorescent dye pairs for microarray analysis owing to their brightness and ease of incorporation, enabling high level of assay sensitivity. However, combining both dyes on arrays can become problematic during summer months when ozone levels rise to near 25 parts per billion (ppb). Under such conditions, Cy5 is known to rapidly degrade leading to loss of signal from either "homebrew" or commercial arrays. Cy5 can also suffer disproportionately from dye photobleaching resulting in distortion of (Cy5/Cy3) ratios used in copy number analysis. Our laboratory has been active in fluorescent dye research to find a suitable alternative to Cy5 that is stable to ozone and resistant to photo-bleaching. Here, we report on the development of such a dye, called HyPer5, and describe its' exceptional ozone and photostable properties on microarrays.

Results

Our results show HyPer5 signal to be stable to high ozone levels. Repeated exposure of mouse arrays hybridized with HyPer5-labeled cDNA to 300 ppb ozone at 5, 10 and 15 minute intervals resulted in no signal loss from the dye. In comparison, Cy5 arrays showed a dramatic 80% decrease in total signal during the same interval. Photobleaching experiments show HyPer5 to be resistant to light induced damage with 3- fold improvement in dye stability over Cy5. In high resolution array CGH experiments, HyPer5 is demonstrated to detect chromosomal aberrations at loci 2p21-16.3 and 15q26.3-26.2 from three patient sample using bacterial artificial chromosome (BAC) arrays. The photostability of HyPer5 is further documented by repeat array scanning without loss of detection. Additionally, HyPer5 arrays are shown to preserve sensitivity and data quality from gene expression experiments.

Conclusion

HyPer5 is a red fluorescent dye that behaves functionally similar to Cy5 except in stability to ozone and light. HyPer5 is demonstrated to be resistant to ozone at up to 300 ppb, levels significantly higher than commonly observed during summer months. Consequently, HyPer5 dye can be used in parallel with Cy3 under any environmental conditions in array experiments.