Skip to main content
Free access
Neurooncology
June 20, 2008

Stem Cell–Related “Self-Renewal” Signature and High Epidermal Growth Factor Receptor Expression Associated With Resistance to Concomitant Chemoradiotherapy in Glioblastoma

Publication: Journal of Clinical Oncology

Abstract

Purpose

Glioblastomas are notorious for resistance to therapy, which has been attributed to DNA-repair proficiency, a multitude of deregulated molecular pathways, and, more recently, to the particular biologic behavior of tumor stem-like cells. Here, we aimed to identify molecular profiles specific for treatment resistance to the current standard of care of concomitant chemoradiotherapy with the alkylating agent temozolomide.

Patients and Methods

Gene expression profiles of 80 glioblastomas were interrogated for associations with resistance to therapy. Patients were treated within clinical trials testing the addition of concomitant and adjuvant temozolomide to radiotherapy.

Results

An expression signature dominated by HOX genes, which comprises Prominin-1 (CD133), emerged as a predictor for poor survival in patients treated with concomitant chemoradiotherapy (n = 42; hazard ratio = 2.69; 95% CI, 1.38 to 5.26; P = .004). This association could be validated in an independent data set. Provocatively, the HOX cluster was reminiscent of a “self-renewal” signature (P = .008; Gene Set Enrichment Analysis) recently characterized in a mouse leukemia model. The HOX signature and EGFR expression were independent prognostic factors in multivariate analysis, adjusted for the O-6-methylguanine-DNA methyltransferase (MGMT) methylation status, a known predictive factor for benefit from temozolomide, and age. Better outcome was associated with gene clusters characterizing features of tumor-host interaction including tumor vascularization and cell adhesion, and innate immune response.

Conclusion

This study provides first clinical evidence for the implication of a “glioma stem cell” or “self-renewal” phenotype in treatment resistance of glioblastoma. Biologic mechanisms identified here to be relevant for resistance will guide future targeted therapies and respective marker development for individualized treatment and patient selection.

Introduction

In glioblastoma, introduction of combined chemoradiotherapy of concomitant and adjuvant temozolomide (TMZ) and radiotherapy (TMZ/RT→TMZ) has allowed significant prolongation of survival,1,2 in particular in patients with an epigenetically silenced O-6-methylguanine-DNA methyltransferase (MGMT) DNA repair gene.3,4 However, outcome remains unsatisfactory, and ongoing clinical trials explore modulation of MGMT or the addition of targeted agents.1,5 Recognizing molecular tumor signatures of underlying biologic processes associated with resistance in patients treated with this new standard therapy will allow the identification of potential targets for improvement of therapy.
Recent concepts for cancer development suggest that a minority population of cancer stem-like cells may determine the biologic behavior of tumors, including response to therapy. Failure to cure cancer has been attributed to the fact that therapies are aimed at the tumor bulk without significantly harming tumor stem-like cells,6 supported by experimental evidence in a respective mouse model showing that this glioblastoma subpopulation of cells is more resistant to radiotherapy.7 Facilitated by markers differentiating stem-cells and progenitors of the different lineages, the origin of leukemic stem cells has been traced back to hematopoietic stem cells as well as progenitor populations that have acquired self-renewal properties.8,9 In contrast, the origin and concept of glioma stem-like cells remains to be fully elucidated. CD133 positivity has been postulated to be a glioma stem-cell marker, considering that this subpopulation of glioma-derived cells seems to have a higher potential to generate and maintain tumors in vivo.7,10
With the goal of identifying molecular mechanisms of treatment resistance, we report here on the analysis of glioblastoma-derived gene expression profiles from patients treated within two prospective clinical trials.1,2 We have identified several biologic processes associated with resistance or responsiveness to combined chemoradiotherapy that provide important information guiding novel treatment strategies and aiming at individualized therapy. Intriguingly, an expression signature associated with resistance shows high similarity with a stem cell–related self-renewal signature.9 Our data provide the first clinical evidence to our knowledge for the involvement of a tumor stem-cell phenotype in the escape of glioblastoma from chemoradiotherapy.

Patients and Methods

Tumor Samples and Patient Characteristics

Gene expression profiles were established from 80 frozen glioblastoma samples obtained from 76 patients, comprising 70 tumors from initial surgery and 10 samples resected at recurrence, and from four non-neoplastic brain tissue samples. All patients were treated within a phase II or a randomized phase III trial1,2 and provided written informed consent for molecular studies of their tumor. The protocol was approved by the ethics committee at each center. Sixty-eight patients with complete molecular and clinical information were included in survival analysis, with a median age of 51 years (range, 26 to 70 years) at enrollment. Thereof, 42 received TMZ/RT→TMZ, and 26 were randomly assigned to RT only. Second-line therapy frequently involved alkylating agents including TMZ. Patient characteristics are summarized in Table A1(online only). The validation set comprised 76 independent patients of the European Organisation for Research and Treatment of Cancer (EORTC)/National Cancer Institute of Canada (NCIC) study,1 39 randomly assigned to TMZ/RT→TMZ, and 37 to RT (median age, 54 years; range, 25-69 years), whose glioblastomas were available on a tissue microarray (TMA). There was no difference in survival compared with the general trial population, neither in the test population nor the validation set (P > .2).

Gene Expression Profiling

Total RNA was extracted from frozen tumor sections (10 to 15 mg; Qiagen RNeasy-Lipid Tissue Kit; Hilden, Germany). The first section served as reference for diagnosis and tumor content (> 70%). Probes were prepared with the Enzo BioArray-High Yield Kit (Enzo Life Sciences, Farmingdale, NY) for double amplification and were hybridized to Affymetrix HG-133Plus2.0 GeneChips (Affymetrix, Santa Clara, CA). The microarray data is deposited in the Gene Expression Omnibus (GEO) database at http://www.ncbi.nlm.nih.gov/geo/ (accession-number GSE7696). Quantitative reverse transcription polymerase-chain reaction (qRT-PCR) was performed on the ABI Prism7900 with SYBR Green (Applied Biosystems, Foster City, CA). Primers are listed in Table A2 (online only). Results were normalized to the expression of the EIF2C3, DNAJA4, and B2M genes that exhibit little variation in this data set.

Data Analysis and Statistical Methods

Analyses were carried out in R, a free software environment available at http://www.r-project.org/, SAS (V9.1.3; SAS Institute, Cary, NC), or Coupled Two Way Clustering (CTWC),11-13 available at http://ctwc.weizmann.ac.il. The expression intensities for all probe sets from Affymetrix CEL-files were estimated using robust multiarray average with probe-level quantile normalization followed by median polish summarization as implemented in the BioConductor software (http://www.bioconductor.org/).
The expression matrix of 84 samples and 3,860 most variable probe sets (standard deviation, > 0.75) was input into CTWC using default parameters and two levels of clustering. CTWC analysis can be viewed at: http://bcf.isb-sib.ch/projects/cancer/glio/. Probe sets comprising stable gene clusters emerging from CTWC served as input for supervised analyses.
Details on the Cox and partial least square models, and published data sets used9,14-16 are described in the Appendix and Table A3 (online only). The Benjamini-Hochberg procedure was applied for multiple testing correction (false-discovery rates).17

TMA and Immunohistochemistry

The TMA was constructed with glioblastoma from the EORTC/NCIC-trial.1 HOXA10 (Santa Cruz Biotechnology, Santa Cruz, CA; sc-17159; dilution, 1:200) expression was determined by immunohistochemistry (citrate buffer: pH 6.0, 15 minutes in pressure cooker) and scored without knowledge of clinical information on a scale of 0 to 6 (0, no expression; 1, weak nuclear expression in < 20% cells; 2, strong in < 20% cells; 3, weak in 20% to 50% cells; 4, strong in 20% to 50% cells; 5, weak in > 50% cells; 6, strong in > 50% cells). Dichotomization for survival analysis was no to low score (1, 2) v intermediate to high expression (scores 2-6). Frozen sections of neurospheres were fixed with acetone and stained for HOXA10 and CD133 (Santa Cruz; sc-17159, dilution 1:100; Miltenyi AC133-2, 293C3, dilution 1:50).

Glioblastoma-Derived Neurospheres

Fresh glioblastoma tissue was dissociated in presence of papain and DNase I basically as described previously.18 Cells were cultured under stem-cell conditions to form spheres using Dulbecco's modified Eagle medium/F12 medium containing B27 supplement and 20 ng/mL of both epidermal growth factor and fibroblast growth factor 2. Neurospheres from six glioblastomas propagated between 4 weeks and 14 months were used for immunostaining.

Results

Gene Expression Signatures Associated With Tumor Resistance

There was no obvious association of patient characteristics or survival with genetic subtypes evident from the sample dendrogram S1(G1), clustering all samples (S1) and all genes (G1) passing a variation filter (Fig 1). The methylation status of the MGMT gene promoter appears to be randomly distributed. All stable gene clusters emerging from this analysis are listed in Table 1, named by the predominant biologic function suggested by the genes they comprise, whereas their inter-relationship is visualized in Fig 2A. Similar gene clusters are obtained when using only the subset of glioblastoma clustered in S4 [G1(S4); Appendix Table A4, online only]. Cluster S4 comprises most glioblastoma (69 of 80), but not the nontumoral tissues that form their own stable cluster, S3 (Fig 1). Similar gene clusters are present in other glioblastoma data sets as visualized in Fig 2B for data sets published by the groups of Nelson and Aldape, respectively.14,15All 18 nonoverlapping, stable gene clusters [G1(S1)] were interrogated for association with survival using Cox proportional hazards, adjusted for age (> 50 years) and MGMT methylation status.3,4 Seven gene clusters were most influential for explaining survival in patients assigned to TMZ/RT→TMZ (Table 1; gene lists, Appendix Tables A5-A10, online only). A respective partial least square model yielded comparable results (Appendix Fig A1, online only). However, the MGMT methylation status was yet the most influential predictor of survival (Table 2; Appendix Fig A1, online only). For this study, we focused on the two most significant clusters characterized by HOX (homeobox) gene (G28/G98) and EGFR (epidermal growth factor receptor) gene expression (G25), respectively, and here we briefly comment on the biologic and clinical implications denoted by the other relevant clusters.

Self-Renewal Signature Associated With Resistance to Chemoradiotherapy in Glioblastoma

Increased expression of cluster G28, dominated by HOX genes and comprising the cell-cycle checkpoint gene GADD45G,19 was found to be associated with worse outcome (P = .004; hazard ratio [HR] = 2.69; 95% CI, 1.38 to 5.26; Table 1). HOX genes are essential in axis determination during embryonic development and are known to be involved in cancer including glioblastomas.20,21 The interaction term between this HOX cluster and chemoradiotherapy was significant (P = .001) in the Cox model when evaluating all 68 patients from the two treatment arms, implying that high expression of HOX genes may be predictive for resistance to TMZ/RT→TMZ therapy.
The HOX gene clusters G28 and G98 emerging from clustering all genes (G1) either with all samples (S1) or only with glioblastoma clustered in S4 are almost identical (Pearson correlation, 0.98; 19 of 21 probe-sets; Appendix Table A11, online only). Intriguingly, G98 in addition comprises Prominin 1, encoding the putative glioma stem–cell marker CD133 (Fig 3A), suggesting that in a subpopulation of glioblastoma, concerted upregulation of HOX genes might be associated with a tumor stem-like cell phenotype. In accordance, we find high HOXA10 protein expression in glioblastoma-derived neurospheres, cultured under stem-cell conditions, as displayed in Fig 3B together with CD133 expression. To include information on Prominin 1, we show results on G98 for all following analyses. Fig 3C visualizes the association of short survival with enhanced expression of G98.

Validation in Independent Data Sets

The association of the HOX signature with resistance to treatment was subsequently validated in a sample set of the trial not available for initial discovery, arrayed on TMA. HOXA10 was evaluated by immunohistochemistry (Fig 4) as a representative of the correlated set of HOX genes. Strong nuclear expression was often observed in patches of tumor cells situated in the vicinity of blood vessels. High HOXA10 expression was associated with worse outcome in patients randomly assigned to TMZ/RT→TMZ therapy (n = 39; HR = 2.57; 95% CI, 1.21 to 5.47; P = .014; Fig 5). The patients randomly assigned to RT only did not show such a relationship, suggesting that high HOXA10 expression may be predictive for resistance to a synergistic effect of concomitant chemoradiotherapy, in concordance with the significant interaction term between treatment and expression of G28 or G98 (Appendix Table A12, online only). Similar HOX clusters can be identified in the Nelson and Aldape glioblastoma data sets.14,15 Correlating G98 gene expression with outcome in these data sets totaling 102 glioblastoma revealed a trend for worse outcome (P = .09; HR = 1.29; 95% CI, 0.97 to 1.72; Appendix Fig A2, online only). Of note, in contrast to our data, these patients were treated before the TMZ/RT→TMZ regimen was established. In accordance with better survival, anaplastic glioma (WHO grade 3) profiled in these publications revealed significantly lower expression of G98 genes compared with glioblastoma (WHO grade 4, P < .001 for the Aldape data set14; P = .002 for the Nelson data set15; Wilcoxon rank sum test with continuity correction; Appendix Fig A3, online only). However, within grade 3 glioma of the two data sets increased expression of G98 genes was associated with worse outcome (n = 44; P = .007; HR = 3.35; 95% CI, 1.39 to 8.08; Appendix Fig A2).

HOX Signature Reminiscent of Self-Renewal

The core of the HOX cluster was found to be part of the top 20 genes of a self-renewal signature identified by Krivtsov et al9 in murine MLL-AF9-induced leukemic stem cells derived from committed progenitors. Our G98-derived signature was significantly enriched in genes discriminating self-renewal versus non–self-renewal in this expression data set according to Gene Set Enrichment Analysis (GSEA; P = .008; Appendix Fig A4 and Table A13, online only) and the Wilcoxon two-sample test (G98; P < .001).22 The relevance of our signature was further demonstrated in a human data set16 in which G98, similar to the original murine self-renewal signature,9 was able to significantly differentiate MLL-rearranged–acute myeloid leukemia from acute myeloid leukemia (GSEA, P < .001; Wilcoxon two-sample test P = .01; Appendix Table A14, online only).
Interestingly, a significant, although low, correlation between the mean DNA copy number of the two BAC (bacterial artificial chromosome) clones (GS1-213H12 and CTB-23D20) bordering the HOXA gene locus on chromosome 7 and the mean expression of the HOXA genes was observed in a set of 60 glioblastoma (Pearson correlation coefficient r = 0.27; P = .03; manuscript in preparation). These flanking BACs were more amplified than their neighbors (Appendix Fig A4, online only). Hence, the herein-proposed HOX-dominated self-renewal signature for glioblastoma may in part be acquired by increased gene dosage.

High EGFR Expression Is Associated With Tumor Resistance

Two gene clusters representing amplification-mediated overexpression of proto-oncogenes were associated with tumor resistance: G13 (P = .02; HR = 1.1; 95% CI, 1.0 to 1.2) characterized by coordinated upregulation of contiguous genes on chromosome 12q13-15, comprising the proto-oncogenes CDK4 and MDM2, and G25, dominated by EGFR probe sets (P = .002; HR = 2.8; 95% CI, 1.4 to 5.4; Table 1; Fig 6). The array-derived EGFR measurement (probe set 201983_s_at) was confirmed by qRT-PCR (Pearson correlation, 0.89). Expression of the constitutively activated EGFRvIII mutant (18 of 70; 26%), as measured by qRT-PCR, did not further influence outcome prediction (P = .94).
Cluster G18, associated with tumor resistance (P = .03; HR = 1.94; 95% CI, 1.07 to 3.51), displayed some correlation with EGFR expression (r = 0.57; Fig 2A) in particular with Aquaporin 4 (AQP4). AQP4 has been associated with brain tumor related edema.23 Aquaporins require activation of mitogen-activated protein kinase signaling that may be mediated by EGFR activation.24 Another family member, AQP1, has been linked with tumor angiogenesis and cell migration25 and was associated with worse outcome in this study (P = .003; HR = 2.44, 95% CI, 1.36 to 4.04) and in the two external data sets (combined P = .009; HR = 1.51; 95% CI, 1.11 to 2.06).

Blood Vessels Markers Associated With Better Outcome

G7 is characterized by genes associated with endothelial cells, basement membranes, signaling pathways of vascular development and angiogenesis, and tumor-derived endothelial markers26,27 (Table A5). Hence, G7 may differentiate tumors according to their angiogenic pattern that may be indicative for drug perfusion, and therefore show association with benefit from treatment (Table 1). G7 is in the center of a “super cluster” (Fig 2A) constituted of several stable gene clusters with biologically related features, such as hypoxia-regulated genes G9, and the myeloid progenitor/adhesion cluster G14, which is also correlated with better outcome (Table 1). Beside genes related to cell adhesion, mesenchymal stem cells (PRR16), and homeobox genes (MEOX1, MKX), G14 includes aldehyde dehydrogenase28 and bone morphogenetic protein 5 gene (BMP5)29 both associated with differentiation of stem cells. The cluster comprises positive (MEOX1) and negative regulators (SOSTDC1) of BMPs, which recently have been shown to inhibit tumorigenic potential of human brain tumor stem cells by promoting their differentiation.29 This cluster may reflect the perivascular microenvironment proposed recently to serve as niche for brain tumor stem cells.30

Innate Immune Response Associated With Better Survival

Alongside markers of innate immunity and macrophages (CD11b), G24 comprises numerous cell-surface receptors known as markers for M2-polarized macrophages,31 such as CD163 (Appendix Table A9, online only). M2 polarization mediates tolerance and downregulates inflammation, alleviating immune surveillance.31,32 A wide range of CD163-positive cells can be observed in glioblastoma (Appendix Fig A5, online only). The cluster also contains probes encoding major histocompatibility complex class II surface molecules, but lacks expression of costimulatory molecules critical for T-cell activation.33 This immune signature may be relevant for strategies of tumor vaccination.

HOX Signature and EGFR Expression Are Independent Prognostic Factors

Multivariate analysis suggests that the HOX signature and EGFR expression, respectively, are independent prognostic factors for poor outcome in TMZ/RT→TMZ–treated glioblastoma patients, explaining 67% of the survival outcome variations together with the MGMT status and age (Table 2).

Discussion

Gene expression signatures relevant for treatment resistance to TMZ/RT→TMZ have been identified in a prospectively treated population of glioblastoma patients. Although epigenetic inactivation of the MGMT gene promoter remained the most prominent predictive factor, expression signatures allowed identification of patient subgroups that may benefit from specific additional therapies targeting particular mechanisms of resistance.
As an independent predictive factor of resistance, we have identified a HOX-dominated gene cluster, evocative of a self-renewal signature recently described for leukemia.9 Strong HOXA10 expression of glioblastoma-derived neurospheres is in line with a role of HOX genes in the glioma stem-like cell compartment. These findings provide the first clinical evidence to our knowledge for the relevance of a stem-like cell phenotype in treatment resistance of glioblastoma.
In leukemia, expression of translocation-related fusion proteins led to MLL-mediated chromatin remodeling associated with re-expression of HOX genes.9,34,35 In glioblastoma, however, such fusion proteins have not been described, and no indications from the present data set link MLL expression with the self-renewal signature. We provide evidence that the HOX-dominated gene signature emerges with malignant progression to glioblastoma, and may be acquired in some glioblastomas by low-level amplification, the latter supporting the notion that gliomas may also arise from progenitors, in agreement with mouse models.36 The identification of GADD45G as part of the HOX signature may provide further evidence for an enhanced DNA repair potential that has recently been associated with radiation resistance of glioma stem cells.7
These molecular and clinical data underscore the importance of the self-renewal phenotype that could be explored as a potential treatment target.37 First efforts blunting glioma stem cell–related self-renewal properties of tumors suggest that strategies forcing differentiation (eg, mediated by cytokines such as BMP4) may be promising.29
Targeting resistance to TMZ/RT→TMZ associated with overexpression of the EGFR gene is of particular clinical interest because this alteration affects a large proportion of patients, and inhibitors of several key targets in downstream signaling pathways are already available or are in advanced clinical development.38 Similarly, inhibitors for CDK4 and MDM2 are under investigation, and both proto-oncogenes showed amplification-mediated overexpression, identified here as associated with worse outcome.
Surprisingly, good prognosis was associated with increased expression of a signature for tumor endothelium markers. This signature may predict improved cytotoxic activity by means of better perfusion of the tumor with the chemotherapy agent TMZ. This is in accordance with the concept suggesting that antiangiogenic agents may temporarily lead to “normalization” of aberrant tumor vasculature, resulting in more efficient delivery of drugs and oxygen to the tumor.39,40 In a recent trial, addition of the antiangiogenic integrin-inhibitor cilengitide appears to confer increased antitumor activity in conjunction with TMZ/RT→TMZ in patients with a methylated MGMT gene promoter.5
Another interesting insight of our study suggests infiltration of M2-polarized macrophages into the tumors. The altered capacity of these glioma-infiltrating macrophages to induce effective antitumor T-cell response may obstruct therapeutic strategies aimed at boosting adaptive immunity against the tumor. M2 polarization is driven by tumor- and T-cell–derived cytokines,31 consistent with the well-known expression of the immunosuppressive cytokines transforming growth factor-β and interleukin-10 in malignant glioma.41 Thus, for effective immunotherapy/vaccination, full resection of the tumor may be required to remove the microenvironment conferring immunosuppression and tolerance.
The molecular signatures identified in this study associated with outcome underline the need for development of multimodality treatments targeting not only the tumor cells, but including strategies aimed at the glioma stem-like cell compartment, and interfering with tumor-host interaction that provides the specialized microenvironment relevant for the maintenance of tumor stem-like cells (the stem-cell niche), angiogenesis, and immune response. The hypotheses generated in this study need to be tested in prospective trials. Our findings may guide a rational choice of agents, targets, trial design, and appropriate patient selection, incorporating biomarkers defining mechanisms of response and resistance.

Authors’ Disclosures of Potential Conflicts of Interest

Although all authors completed the disclosure declaration, the following author(s) indicated a financial or other interest that is relevant to the subject matter under consideration in this article. Certain relationships marked with a “Urdquo; are those for which no compensation was received; those relationships marked with a “C” were compensated. For a detailed description of the disclosure categories, or for more information about ASCO's conflict of interest policy, please refer to the Author Disclosure Declaration and the Disclosures of Potential Conflicts of Interest section in Information for Contributors.
Employment or Leadership Position: None Consultant or Advisory Role: Roger Stupp, Schering-Plough (C), Merck Serono (C); Monika E. Hegi, Oncomethylome Sciences (C) Stock Ownership: None Honoraria: Wolfgang Wick, Schering-Plough, Essex-Pharma; J. Gregory Cairncross, Schering-Plough; Roger Stupp, Schering-Plough, Merck Serono; Monika E. Hegi, Schering-Plough Research Funding: Monika E. Hegi, Schering-Plough, Oncomethylome Sciences Expert Testimony: None Other Remuneration: None

Author Contributions

Conception and design: Anastasia Murat, Eugenia Migliavacca, Monika E. Hegi
Financial support: Nicolas de Tribolet, Monika E. Hegi
Administrative support: Roger Stupp, Monika E. Hegi
Provision of study materials or patients: Nicolas de Tribolet, Luca Regli, Wolfgang Wick, Mathilde C.M. Kouwenhoven, Johannes A. Hainfellner, Frank L. Heppner, Pierre-Yves Dietrich, Yitzhak Zimmer, Gregory Cairncross, Roger Stupp
Collection and assembly of data: Anastasia Murat, Eugenia Migliavacca, Marie-France Hamou, Robert-Charles Janzer, Roger Stupp, Monika E. Hegi
Data analysis and interpretation: Anastasia Murat, Eugenia Migliavacca, Thierry Gorlia, Wanyu L. Lambiv, Tal Shay, Marie-France Hamou, Eytan Domany, Mauro Delorenzi, Monika E. Hegi
Manuscript writing: Anastasia Murat, Eugenia Migliavacca, Thierry Gorlia, Eytan Domany, Mauro Delorenzi, Roger Stupp, Monika E. Hegi
Final approval of manuscript: Anastasia Murat, Eugenia Migliavacca, Thierry Gorlia, Wanyu L. Lambiv, Tal Shay, Marie-France Hamou, Nicolas de Tribolet, Luca Regli, Wolfgang Wick, Mathilde C.M. Kouwenhoven, Johannes A. Hainfellner, Frank L. Heppner, Yitzhak Zimmer, Gregory Cairncross, Robert-Charles Janzer, Eytan Domany, Mauro Delorenzi, Roger Stupp, Monika E. Hegi

Appendix

Partial Least Square Regression

Partial least square (PLS) regression is a technique that combines features from principal component analysis and multiple linear regression (Wold H: Partial Least Squares. New York, NY Wiley, 1985). It is particularly useful to predict an outcome from a large set of highly correlated predictors. In this study, the PLS procedure in SAS was used to define combinations of the genes expressions (ie, factors) that attempt to explain the genes expression variability and the survival outcome at the same time. For each patient, the martingale residuals obtained from a Cox regression with a constant as the only predictor were used as outcome variable to be explained (Wold H: Partial Least Squares. New York, NY Wiley, 1985). This approach was derived from the works of Therneau et al (Biometrika 77:147-160, 1990) and Leblanc et al (Biometrics 48:411-425, 1992) with the classification and regression tree algorithm and applied to PLS regression. This allowed us to account for the effect of censoring on survival estimates.
The optimal number of PLS factors is generally obtained by cross validation, but in this small sample (n = 42 or 68), this method was not conclusive (ie, no optimal number of factors could be obtained). Therefore, we empirically chose to fit a PLS model with two factors to minimize overfitting and obtain interpretable results. The two factors explained 66% of the survival outcome variations. The contribution of each gene was assessed by their coefficients in the linear regression and their X weights plotted in a factorial plane. Arbitrarily, clusters were given X weights by averaging the X weights of their constituent genes.

Description of External Data Sets

Two external gene expression data sets (GeneChip U133A&B, Affymetrix) of malignant glioma were used for validation of survival factors, referred to as “Nelson” (Freije WA, Castro-Vargas FE, Fang Z, et al: Cancer Res 64:6503-6510, 2004) and “Aldape” data sets (Phillips HS, Kharbanda S, Chen R, et al: Cancer Cell 9:157-173, 2006), respectively. Only cases with diagnosis of glioblastoma, astrocytoma grade III, and mixed oligoastrocytoma grade III were included. The Nelson data set utilized here comprised 17 grade III and 54 grade IV gliomas. The median age was 43 years at diagnosis for all 71 patients, or 51 years for glioblastoma patients, respectively. The Aldape data set consisted of 75 patients, 27 grade III, 48 grade IV with a median age at diagnosis of 47 years for all cases, or 49 years for glioblastoma patients, respectively. This data set was enriched for long-term survivors according to the authors. Adjustments in the Cox models for these two external glioma validation sets comprised age as a dichotomous variable (> 50 years) and tumor grade.
In addition, we used a mouse expression data set published by Krivtsov et al (Krivtsov AV, Twomey D, Feng Z, et al: Nature 442:818-822, 2006) and a human leukemia data set by Ross et al (Ross ME, Mahfouz R, Onciu M, et al: Blood 104:3679-3687, 2004). For both human malignant glioma validation sets, only the chips that passed quality criteria based on the BioConductor library affyPLM (Gentleman R, Carey V, Huber W, et al: Bioinformatics and Computational Biology Solutions Using R and Bioconductor, Springer, 2005) were included. The external data sets were each separately normalized using robust multiarray average, when Affymetrix CEL files were available (Aldape, Nelson, and Ross). For the Krivtsov mouse expression data set (Krivtsov AV, Twomey D, Feng Z, et al: Nature 442:818-822, 2006), the Affymetrix Microarray Suite version 5.0 processed data, as submitted in the Gene Expression Omnibus database, were used. For the human-murine comparison, we mapped the human gene sets to homolog murine gene sets using the annotation provided by Affymetrix on http//www.affymetrix.com (accessed November 14, 2006; Table A13). For GSEA analysis, the data sets were filtered to retain only the most variant probe sets (standard deviation, > 0.5).

Supervised Analysis

For each covariate, the hazard ratio, its 95% CI, and the associated Wald P were examined. The Benjamini-Hochberg procedure (Benjamini Y, Hochberg Y: J R Stat Soc B 57:289-300, 1995) was applied as multiple testing correction method to convert P values into estimated false-discovery rates. The HRs for a gene (or the mean of a cluster) were standardized by referring them to a variation corresponding to an interquartile range, to obtain an interpretable value, for microarray as well as for real-time PCR data. Therefore, the interpretation of an estimated HR of 1.5 is that the hazard rate increases by 50% for an increase of 1 interquartile range of the log-scale gene expression, independently of the expression level at which it is calculated.
Fig A1. Partial least square (PLS) model for prediction of overall survival. Cluster X weights were obtained by averaging the X weights of their constituent genes. The clusters are numbered according to the legend and the description in Table 1. Age and MGMT methylation status were used as covariates. Clusters of genes with X weights that are nearest to the PLS factors, represented by the axes (PLS factor 1 and 2), and the farthest from the center of the plane contribute most to the PLS regression. The first two factors shown explain 66% of the survival outcome variations. Clusters of genes grouped in the upper and right side of the plane have a positive association with shorter survival (ie, higher hazards), while those in the lower and left side are positively associated with longer survival (ie, lower hazards).
Fig A2. Expression of G98 and association with survival in external data sets. High expression of the self-renewal cluster G98 is significantly associated with worse outcome in two independent data sets comprising a total of 146 malignant glioma (P = .007, hazard ratio = 1.46, 95% CI, 1.11 to 1.92; model adjusted for tumor grade WHO grade 3 and 4; age, and stratified for the data set, Nelson [Freije WA, Castro-Vargas FE, Fang Z, et al: Cancer Res 64:6503-6510, 2004]; Aldape [Phillips HS, Kharbanda S, Chen R, et al: Cancer Cell 9:157-173, 2006]). The forest plots visualize the relationship between expression of G98 and outcome in the two data sets separated for tumor grade. The hazard ratio for glioblastoma (WHO grade 4) is 1.29 (95% CI, 0.97 to 1.72; P = .09; left panel). A hazard ratio of 3.35 (95% CI, 1.39 to 8.08; P = .007) is observed for the subset of anaplastic glioma (WHO grade 3; right panel).
Fig A3. Box plot for expression of G98 in grade 3 and grade 4 glioma of external data sets. G98 expression significantly differentiates anaplastic glioma (WHO grade 3) from glioblastoma (WHO grade 4). Glioblastomas exhibit significantly higher expression of G98 in both external data sets (Wilcoxon rank sum test with continuity correction, Nelson [Freije WA, Castro-Vargas FE, Fang Z, et al: Cancer Res 64:6503-6510, 2004] P < .001; Aldape [Phillips HS, Kharbanda S, Chen R, et al: Cancer Cell 9:157-173, 2006] P = .002).
Fig A4. Array-based comparative genomic hybridization data for the chromosomal region around the HOXA gene cluster on chromosome 7. Every row is a marker, ordered by chromosomal location. Every column is a sample (n = 60), ordered by the copy number of BAC GS1-213H12 (bacterial artificial chromosome) using SPIN software (Tsafrir D, Tsafrir I, Ein-Dor L, et al: Bioinformatics 21:2301-2308, 2005). Color code represents the log2 (tumor/normal) ratio of the copy number, red corresponding to amplification. The HOXA gene cluster resides between markers GS1-213H12 and CTB-23D20 that appear to be amplified more frequently and more strongly than the neighboring markers.
Fig A5. CD163 expression in glioblastoma. Immunohistochemistry for CD163 (Novocastra, Newcastle upon Tyne, United Kingdom; NCL-163; dilution 1:800) on representative paraffin-embedded glioblastoma. Diameter of tissues in upper panel is 0.6 mm. CD163 is a marker for M2-polarized macrophages.
Table A1. Summary of Patient Information
Tissue*Patient IDRec IDSexMGMT StatusOverall Survival (months)Age (years)TreatmentCensoring
GBM12 MaleMethylated1738TMZ/RT→TMZUncensored
GBM15 FemaleUnmethylated645RTUncensored
GBM30 MaleMethylated761TMZ/RT→TMZUncensored
GBM55 MaleMethylated1050RTUncensored
GBM5757RMaleMethylated3653TMZ/RT→TMZCensored
GBM63 MaleMethylated1649RTUncensored
RecGBM 65RFemaleMethylated1346TMZ/RT→TMZUncensored
GBM70 MaleUnknown3453RTCensored
GBM80 FemaleMethylated3047TMZ/RT→TMZUncensored
GBM83 MaleUnmethylated1649RTUncensored
GBM84 MaleMethylated154RTUncensored
GBM90 FemaleMethylated3061TMZ/RT→TMZCensored
GBM93 MaleUnmethylated1070TMZ/RT→TMZUncensored
GBM94 MaleUnmethylated959RTUncensored
GBM97 FemaleUnmethylated462RTUncensored
RecGBM 100RMaleUnmethylated2464TMZ/RT→TMZUncensored
GBM104 MaleMethylated2056TMZ/RT→TMZUncensored
GBM106 MaleUnmethylated1027TMZ/RT→TMZUncensored
GBM129 MaleUnmethylated449TMZ/RT→TMZUncensored
GBM147 MaleMethylated1549RTUncensored
Normal brain148      NA
RecGBM 155RMaleMethylated2152TMZ/RT→TMZUncensored
GBM157 MaleUnmethylated849TMZ/RT→TMZUncensored
GBM159 FemaleUnmethylated1461RTUncensored
GBM161 MaleUnmethylated1253TMZ/RT→TMZUncensored
GBM190 MaleMethylated1761TMZ/RT→TMZUncensored
GBM211 MaleMethylated944RTUncensored
RecGBM 212RFemaleUnmethylated1659RTUncensored
GBM218 FemaleMethylated1648RTUncensored
GBM222 FemaleUnmethylated1262TMZ/RT→TMZUncensored
GBM236 MaleUnmethylated1157TMZ/RT→TMZUncensored
GBM242 MaleUnmethylated1454RTUncensored
GBM246 MaleMethylated2354RTUncensored
GBM254 MaleUnmethylated764RTUncensored
GBM269 MaleUnmethylated761TMZ/RT→TMZUncensored
GBM274 FemaleUnmethylated949TMZ/RT→TMZUncensored
GBM283 MaleMethylated757RTUncensored
GBM284 MaleMethylated2752RTUncensored
GBM288 MaleMethylated2745TMZ/RT→TMZCensored
GBM300 MaleUnknown947TMZ/RT→TMZUncensored
GBM323 MaleMethylated1653RTUncensored
GBM328 MaleUnmethylated1163RTUncensored
GBM344 MaleUnmethylated1657TMZ/RT→TMZUncensored
GBM345 MaleMethylated659RTUncensored
GBM349 FemaleMethylated1452RTUncensored
GBM356 MaleUnmethylated732TMZ/RT→TMZUncensored
GBM360 FemaleMethylated2356TMZ/RT→TMZCensored
GBM390 MaleUnmethylated1642RTUncensored
GBM396 MaleMethylated1650TMZ/RT→TMZCensored
GBM428 MaleMethylated957RTUncensored
GBM444 MaleMethylated663TMZ/RT→TMZUncensored
GBM458 FemaleUnmethylated1450TMZ/RT→TMZUncensored
GBM461 MaleUnmethylated1564TMZ/RT→TMZUncensored
GBM506 MaleUnmethylated657TMZ/RT→TMZUncensored
GBM527 FemaleMethylated2453RTCensored
GBM528 MaleUnmethylated2433TMZ/RT→TMZCensored
GBM534 FemaleMethylated2342RTCensored
GBM547 MaleUnmethylated1743TMZ/RT→TMZUncensored
GBM558 FemaleUnmethylated656TMZ/RT→TMZUncensored
GBM563 MaleUnmethylated953TMZ/RT→TMZUncensored
GBM566 MaleUnmethylated1644RTUncensored
GBM574 FemaleUnmethylated544RTUncensored
Normal brain1076      NA
GBM1284 FemaleMethylated4748TMZ/RT→TMZUncensored
GBM1297 MaleMethylated6036TMZ/RT→TMZUncensored
GBM1308 FemaleMethylated1745TMZ/RT→TMZUncensored
GBM1317 MaleUnmethylated1726TMZ/RT→TMZUncensored
GBM1357 MaleMethylated2836TMZ/RT→TMZUncensored
GBM1360 MaleMethylated965TMZ/RT→TMZUncensored
GBM1399 MaleMethylated7939TMZ/RT→TMZCensored
GBM14301497RMaleMethylated1537TMZ/RT→TMZUncensored
GBM1437 MaleUnmethylated1648TMZ/RT→TMZUncensored
GBM1453 FemaleUnmethylated953TMZ/RT→TMZUncensored
Normal brain1553      NA
GBM16211622RMaleMethylated2562TMZ/RT→TMZUncensored
Normal brain1780      NA
RecGBM1795R1869RRMaleMethylated7243TMZ/RT→TMZCensored
RecGBM 1854RMaleMethylated6352TMZ/RT→TMZCensored
GBMge197 MaleMethylated5346TMZ/RT→TMZCensored
GBMge205 FemaleMethylated2356TMZ/RT→TMZUncensored
Abbreviations: GBM, glioblastoma; Rec ID, identification of recurrent GBM; TMZ/RT→TMZ, concomitant and adjuvant temozolomide and radiotherapy; RecGBM, recurrent GBM; RT, radiotherapy; NA, not applicable.
*
Normal brain, non-neoplastic brain tissue.
Initial treatment.
Table A2. Primer Sequences Used in Quantitative Reverse Transcription Polymerase-Chain Reaction
TargetPrimer Sequences
EIF2C3 
    Forward primer5`-CTACATTGTAGTTCAGAAGAGACATCACA-3`
    Reverse primer5`-TGCCACTTCTTCCAACCCTTT-3`
DNAJA4 
    Forward primer5`-GGATGAGGGCGAGAAGTTTAAA-3`
    Reverse primer5`-GGTCATAAACATCCCTTTTCTTTGG-3`
B2M 
    Forward primer5-`TGCTCGCGCTACTCTCTCTTT-3`
    Reverse primer5`-TCTGCTGGATGACGTGAGTAAAC-3`
EGFRwt 
    Forward primer5`-GTGGTCCTTGGGAATTTGGA-3`
    Reverse primer5`-CACCTCCTGGATGGTCTTTAAGA-3`
EGFRvIII 
    Forward primer5`-CGCTGCTGGCTGCGCTCTG-3`
    Reverse primer5`-AGATGGAGGAAGACGGCGT-3`
Table A3. Cox Survival Models Used for Individual Data Sets
Data SetModel
Hegi42 (TMZ/RT)y∼ β1meanCluster + β2MGMT + β3age
Hegi42 (TMZ/RT)y∼ β1EGFRvIII + β2MGMT + β3age + β4EGFRwt
Hegi42 (TMZ/RT)y∼ β1meanG98 + β2EGFR + β3MGMT + β4age
Hegi68 (all, G98)y∼ β1meanG98 + β2MGMT + β3age + β4TMZ + β5meanG98:TMZ + β6MGMT:TMZ
Hegi68 (all, EGFR)y∼ β1EGFR + β2MGMT + β3age + β4TMZ + β5EGFR:TMZ + β6MGMT:TMZ
Nelsony∼ β1meanCluster + β2grade + β3age
Aldapey∼ β1meanCluster + β2grade + β3age
Combinedy∼ β1meanCluster + β2grade + β3age + strata(dataSet)
Abbreviations: MGMT, MGMT methylation status; TMZ, temozolomide; RT, radiotherapy; MGMT:TMZ, interaction factor with concomitant and adjuvant TMZ and RT.
Table A4. Cox Analysis Using G1(S4) Clusters in the Temozolomide- and Radiotherapy-Treated Patient Cohort (n = 42)
S1 v S4 Clusters*Stable Clusters in G1(S4)  No. of ProbesPFalse-Discovery RateHazard Ratio95% CI
 Main ClustersSubclustersDescription     
pG02G71 Astrocyte markers (?)17.030.110.580.36 to 0.96
pG02/pG07G72 Migration-related17.290.490.760.46 to 1.26
G23G74 Oligodendrocyte markers79.670.871.070.79 to 1.44
G24/pG07G80 Innate immune response152.030.110.620.41 to 0.94
G29G82> G77Proliferation98.470.721.260.67 to 2.36
G13G83 Chromosome 1237.020.111.081.01 to 1.15
G09G84 Hypoxia-induced genes54.510.720.830.47 to 1.45
 G88 ? TGF-β, MYC, SOD220.080.220.680.44 to 1.05
G07 pG14G90> G79 > G73;G85Tumor blood vessel markers/adhesion131.010.100.600.40 to 0.89
pG16/pG29G91> G86Membrane and cytoskeleton proteins89.910.940.960.52 to 1.79
pG16G92 Protein binding, glutamate metabolism, OLIG1 and OLIG2 (?)93.140.291.490.88 to 2.54
 G93 Regulation of transcription, EGR1-320.750.910.930.59 to 1.45
G21G94> G89 > G78Neural genes168.940.940.980.59 to 1.63
G20G95 ?? = G2019.250.481.390.79 to 2.44
G12G96> G87 > G81 > G75Interferon-induced genes75.820.931.040.73 to 1.48
G27G97> G76Spermatogenesis78.090.221.380.95 to 2.00
G28G98 HOX genes21.0040.072.561.34 to 4.88
Abbreviation: p-, p, partial G1(S1) cluster.
*
Corresponding clusters in G1(S1).
P value for the coefficient relative to the mean of the cluster in a multivariate Cox proportional hazards model adjusted for MGMT methylation status and age as additive independent risk factors (y∼ β1meanCluster + β2MGMT + β3 age)
Each continuous variable was scaled to have the interquartile range equal to 1 and median equal to 0.
Table A5. Cluster G07
Probe IDGene SymbolChromosomal LocationCommentsCitation
1556499_s_atCOL1A1chr17q21.33Tumor endothelium marker*
200700_s_atKDELR2chr7p22.1  
200771_atLAMC1chr1q31Basement membrane 
201438_atCOL6A3chr2q37Tumor endothelium marker*
201505_atLAMB1chr7q22Basement membrane 
201744_s_atLUMchr12q21.3-q22Angiogenesis
201852_x_atCOL3A1chr2q31Tumor endothelium marker*
202007_atNID1chr1q43Tumor endothelium marker*
202112_atVWFchr12p13.3Endothelial marker 
202202_s_atLAMA4chr6q21Basement membrane 
202291_s_atMGPchr12p13.1-p12.3Endothelial marker*
202310_s_atCOL1A1chr17q21.33Tumor endothelium marker*
202375_atSEC24Dchr4q26  
202403_s_atCOL1A2chr7q22.1Tumor endothelium marker*
202404_s_atCOL1A2chr7q22.1Tumor endothelium marker*
202465_atPCOLCEchr7q22  
202709_atFMODchr1q32  
202766_s_atFBN1chr15q21.1Endothelial marker 
202878_s_atCD93chr20p11.21  
202952_s_atADAM12chr10q26.3Cell adhesion 
202998_s_atLOXL2chr8p21.3-p21.2  
203325_s_atCOL5A1chr9q34.2-q34.3  
203851_atIGFBP6chr12q13Angiogenesis 
204017_atKDELR3chr22q13.1  
204083_s_atTPM2chr9p13.2-p13.1  
204114_atNID2chr14q21-q22Tumor endothelium marker*
204464_s_atEDNRAchr4q31.23Patterning of blood vesselsGO
204677_atCDH5chr16q22.1Endothelial marker 
204682_atLTBP2chr14q24  
204844_atENPEPchr4q25Angiogenesis
205499_atSRPX2chrXq21.33-q23  
205572_atANGPT2chr8p23.1Angiogenesis 
207714_s_atSERPINH1chr11q13.5  
208690_s_atPDLIM1chr10q22-q26.3  
209156_s_atCOL6A2chr21q22.3Endothelial marker 
209596_atMXRA5chrXp22.33  
210495_x_atFN1chr2q34Basement membrane 
211148_s_atANGPT2chr8p23.1Angiogenesis 
211161_s_atCOL3A1chr2q31Tumor endothelium marker*
211343_s_atCOL13A1chr10q22  
211651_s_atLAMB1chr7q22Basement membrane 
211719_x_atFN1chr2q34Basement membrane 
211964_atCOL4A2chr13q34Glioma endothelial marker§
211966_atCOL4A2chr13q34Glioma endothelial marker§
211980_atCOL4A1chr13q34Glioma endothelial marker§
211981_atCOL4A1chr13q34Glioma endothelial marker§
212298_atNRP1chr10p12Angiogenesis
212364_atMYO1Bchr2q12-q34  
212464_s_atFN1chr2q34Basement membrane 
212488_atCOL5A1chr9q34.2-q34.3  
212489_atCOL5A1chr9q34.2-q34.3  
213125_atOLFML2Bchr1q23.3  
213139_atSNAI2chr8q11  
213790_at  
214081_atPLXDC1chr17q21.1  
215076_s_atCOL3A1chr2q31Tumor endothelium marker*
216442_x_atFN1chr2q34Basement membrane 
218729_atLXNchr3q25.32  
219134_atELTD1chr1p33-p32  
219773_atNOX4chr11q14.2-q21Oxygen sensor activity
221729_atCOL5A2chr2q14-q32  
221730_atCOL5A2chr2q14-q32  
224833_atETS1chr11q23.3Angiogenesis#
225681_atCTHRC1chr8q22.3Vascular remodeling**
225799_atMGC4677///LOC541471chr2p11.2///chr2q13  
226311_at  
226731_atPELOchr5q11.2  
226777_at  
226804_atFAM20Achr17q24.2  
227628_atLOC493869chr5q11.2  
228776_atGJA7chr17q21.31  
229218_atCOL1A2chr7q22.1Tumor endothelium marker*
230061_atTM4SF18chr3q25.1  
232458_atCOL3A1chr2q31Tumor endothelium marker*
236034_at  
237261_at  
241981_atFAM20Achr17q24.2  
Abbreviation: GO, gene ontology.
*
St Croix B, Rago C, Velculescu V, et al: Science 289:1197-1202, 2000.
Hagedorn M, Javerzat S, Gilges D, et al: Proc Natl Acad Sci U S A 102:1643-1648, 2005.
Marchio S, Lahdenranta J, Schlingemann RO, et al: Cancer Cell 5:151-162, 2004.
§
Madden SL, Cook BP, Nacht M, et al: Am J Pathol 165:601-608, 2004.
Soker S, Takashima S, Miao HQ, et al: Cell 92:735-745, 1998.
Geiszt M, Kopp JB, Varnai P, et al: Proc Natl Acad Sci U S A 97:8010-8014, 2000.
#
Stamatovic SM, Keep RF, Mostarica-Stojkovic M, et al: J Immunol 177:2651-2661, 2006.
**
Pyagay P, Heroult M, Wang Q, et al: Circ Res 96:261-268, 2005.
Table A6. Cluster G13
Probe IDGene SymbolChromosomal Location
1555385_atB4GALNT1chr12q13.3
1568706_s_atAVILchr12q14.1
200714_x_atOS9chr12q13
201131_s_atCDH1chr16q22.1
202246_s_atCDK4chr12q14
203226_s_atTSPAN31chr12q13.3
203227_s_atTSPAN31chr12q13.3
204027_s_atMETTL1chr12q13
205386_s_atMDM2chr12q14.3-q15
205539_atAVILchr12q14.1
205676_atCYP27B1chr12q13.1-q13.3
206100_atCPMchr12q14.3
206435_atB4GALNT1chr12q13.3
211832_s_atMDM2chr12q14.3-q15
212656_atTSFMchr12q13-q14
213861_s_atFAM119Bchr12q14.1
214331_atTSFMchr12q13-q14
214332_s_atTSFMchr12q13-q14
214951_atSLC26A10chr12q13
215399_s_atOS9chr12q13
217373_x_atMDM2chr12q14.3-q15
217542_atCPMchr12q14.3
218768_atNUP107chr12q15
224489_atKIAA1267chr17q21.31
225160_x_atMGC5370chr12q14.3
226454_atMARCH9chr12q14.1
226546_at
227678_atXRCC6BP1chr12q14.1
229711_s_atMGC5370chr12q14.3
229917_at
230001_atMARCH9chr12q14.1
235019_atCPMchr12q14.3
235706_atCPMchr12q14.3
235721_atDTX3chr12q13.3
238733_atCPMchr12q14.3
238999_atAVILchr12q14.1
241765_atCPMchr12q14.3
244675_atRGS8chr1q25
Table A7. Cluster G14
Probe IDGene SymbolChromosomal LocationCommentsCitation
203180_atALDH1A3chr15q26.3Differentiation of stem cells*
203434_s_atMMEchr3q25.1-q25.2  
204776_atTHBS4chr5q13  
205430_atBMP5chr6p12.1Differentiation of stem cells
205431_s_atBMP5chr6p12.1Differentiation of stem cells
205619_s_atMEOX1chr17q21Regulator of BMPs
205713_s_atCOMPchr19p13.1  
205848_atGAS2chr11p14.3-p15.2Apoptosis 
212414_s_atSEPT6///N-PACchrXq24///chr16p13.3  
213456_atSOSTDC1chr7p21.1Regulator of BMPs§
217525_atOLFML1chr11p15.4  
218499_atRP6-213H19.1chrXq26.2  
219837_s_atCYTL1chr4p16-p15  
220014_atPRR16chr5q23.1Mesenchymal stem-cell marker
220065_atTNMDchrXq21.33-q23  
226834_at  
226930_atFNDC1chr6q25  
227070_atGLT8D2chr12q  
227850_x_atCDC42EP5chr19q13.42  
229839_atSCARA5chr8p21.1  
231766_s_atCOL12A1chr6q12-q13  
232090_atDNM3chr1q24.3  
235944_atHMCN1chr1q25.3-q31.1  
236035_at  
239468_atMKXchr10p12.1Homeobox gene
244885_at  
Abbreviation: BMP, bone morphogenetic protein.
*
Chute JP, Muramoto GG, Whitesides J, et al: Proc Natl Acad Sci U S A 103:11707-11712, 2006.
Piccirillo SG, Reynolds BA, Zanetti N, et al: Nature 444:761-765, 2006.
Gianakopoulos PJ, Skerjanc IS: J Biol Chem 280:21022-21028, 2005.
§
Laurikkala J, Kassai Y, Pakkasjarvi L, et al: Dev Biol 264:91-105, 2003.
Barberi T, Willis LM, Socci ND, et al: PLoS Med 2:e161, 2005.
Anderson DM, Arredondo J, Hahn K, et al: Dev Dyn 235:792-801, 2006.
Table A8. Cluster G18
Probe IDGene SymbolChromosomal Location
202800_atSLC1A3chr5p13
203723_atITPKBchr1q42.13
204041_atMAOBchrXp11.23
204363_atF3chr1p22-p21
205363_atBBOX1chr11p14.2
205741_s_atDTNAchr18q12
206022_atNDPchrXp11.4
207443_atNR2E1chr6q21
207455_atP2RY1chr3q25.2
209301_atCA2chr8q22
209921_atSLC7A11chr4q28-q32
210066_s_atAQP4chr18q11.2-q12.1
210067_atAQP4chr18q11.2-q12.1
210068_s_atAQP4chr18q11.2-q12.1
210906_x_atAQP4chr18q11.2-q12.1
213395_atMLC1chr22q13.33
213618_atCENTD1chr4p14
217678_atSLC7A11chr4q28-q32
223605_atSLC25A18chr22q11.2
226189_atITGB8chr7p15.3
226228_atAQP4chr18q11.2-q12.1
227084_atDTNAchr18q12
231925_atP2RY1chr3q25.2
235795_atPAX6chr11p13
238003_atFLJ25530chr11q24.2
Table A9. Cluster G24
Probe IDGene SymbolChromosomal LocationCommentsCitation
1552316_a_atGIMAP1chr7q36.1  
1552365_atSCINchr7p21.3  
1552367_a_atSCINchr7p21.3  
1552386_atC5orf29chr5q11.2  
1552807_a_atSIGLEC10chr19q13.3  
1554899_s_atFCER1Gchr1q23M2 marker*
1555349_a_atITGB2chr21q22.3Macrophage markers 
1555728_a_atMS4A4Achr11q12  
201137_s_atHLA-DPB1chr6p21.3  
201422_atIFI30chr19p13.1  
201487_atCTSCchr11q14.1-q14.3M2 marker
201631_s_atIER3chr6p21.3  
201721_s_atLAPTM5chr1p34  
201743_atCD14chr5q22-q32 5q31.1Innate immunity 
202546_atVAMP8chr2p12-p11.2  
202803_s_atITGB2chr21q22.3  
202833_s_atSERPINA1chr14q32.1  
202901_x_atCTSSchr1q21M2 marker
202902_s_atCTSSchr1q21M2 marker
202917_s_atS100A8chr1q21  
202953_atC1QBchr1p36.12M2 marker
202957_atHCLS1chr3q13  
203104_atCSF1Rchr5q33-q35  
203290_atHLA-DQA1chr6p21.3  
203305_atF13A1chr6p25.3-p24.3M2 marker
203416_atCD53chr1p13Macrophage markers 
203535_atS100A9chr1q21  
203561_atFCGR2Achr1q23  
203645_s_atCD163chr12p13.3M2 marker§
203665_atHMOX1chr22q12 22q13.1  
204006_s_atFCGR3A///FCGR3Bchr1q23  
204007_atFCGR3Bchr1q23  
204122_atTYROBPchr19q13.1Innate immunity 
204150_atSTAB1chr3p21.1M2 marker
204174_atALOX5APchr13q12M2 marker*
204232_atFCER1Gchr1q23M2 marker*
204416_x_atAPOC1chr19q13.2  
204430_s_atSLC2A5chr1p36.2  
204438_atMRC1///MRC1L1chr10p12.33M2 marker
204446_s_atALOX5chr10q11.2M2 marker*
204563_atSELLchr1q23-q25  
204670_x_atHLA-DRB1chr6p21.3  
204787_atVSIG4chrXq12-q13.3  
204834_atFGL2chr7q11.23  
204959_atMNDAchr1q22  
204971_atCSTAchr3q21  
205027_s_atMAP3K8chr10p11.23  
205119_s_atFPR1chr19q13.4  
205681_atBCL2A1chr15q24.3  
205786_s_atITGAMchr16p11.2  
205890_s_atUBDchr6p21.3  
205898_atCX3CR1chr3p21 3p21.3  
205997_atADAM28chr8p21.2  
206111_atRNASE2chr14q24-q31  
206420_atIGSF6chr16p12-p13  
206584_atLY96chr8q21.11  
207655_s_atBLNKchr10q23.2-q23.33  
208018_s_atHCKchr20q11-q12  
208146_s_atCPVLchr7p15-p14  
208306_x_atHLA-DRB1chr6p21.3  
208894_atHLA-DRAchr6p21.3  
208944_atTGFBR2chr3p22M2 marker#
209312_x_atHLA-DRB1chr6p21.3  
209619_atCD74chr5q32  
209823_x_atHLA-DQB1chr6p21.3  
210176_atTLR1chr4p14Innate immunity 
210314_x_atTNFSF13///TNFSF12-TNFSF13chr17p13.1  
210982_s_atHLA-DRAchr6p21.3  
211429_s_atSERPINA1chr14q32.1  
211654_x_atHLA-DQB1///LOC650557chr6p21.3  
211656_x_atHLA-DQB1chr6p21.3  
211742_s_atEVI2Bchr17q11.2  
211990_atHLA-DPA1chr6p21.3  
211991_s_atHLA-DPA1chr6p21.3  
212543_atAIM1chr6q21  
212588_atPTPRCchr1q31-q32Macrophage markers 
212671_s_atHLA-DQA1///HLA-DQA2chr6p21.3  
212998_x_atHLA-DQB1chr6p21.3  
213537_atHLA-DPA1chr6p21.3  
213566_atRNASE6chr14q11.2  
214467_atGPR65chr14q31-q32.1  
214511_x_atFCGR1A///chr1q21.2-q21.3///chr1p11.2  
214770_atMSR1chr8p22M2 marker*
215049_x_atCD163chr12p13.3M2 marker§
215193_x_atHLA-DRB1chr6p21.3  
216233_atCD163chr12p13.3M2 marker§
216950_s_atFCGR1Achr1q21.2-q21.3  
217388_s_atKYNUchr2q22.2  
217478_s_atHLA-DMAchr6p21.3  
217767_atC3///LOC653879chr19p13.3-p13.2Innate immunity 
217983_s_atRNASET2chr6q27  
218232_atC1QAchr1p36.12M2 marker
218854_atSART2chr6q22  
219386_s_atSLAMF8chr1q23.2  
219607_s_atMS4A4Achr11q12  
219666_atMS4A6Achr11q12.1  
219890_atCLEC5Achr7q33  
220005_atP2RY13chr3q24  
220146_atTLR7chrXp22.3Innate immunity 
220330_s_atSAMSN1chr21q11  
220491_atHAMPchr19q13.1Innate immunity 
220532_s_atLR8chr7q36.1  
221210_s_atNPLchr1q25  
221491_x_atHLA-DRB1///HLA-DRB3///HLA-DRB4///HLA-DRB5chr6p21.3  
221698_s_atCLEC7Achr12p13.2M2 marker**
223280_x_atMS4A6Achr11q12.1  
223343_atMS4A7chr11q12  
223620_atGPR34chrXp11.4-p11.3  
223660_atADORA3chr1p13.2  
224356_x_atMS4A6Achr11q12.1  
225353_s_atC1QCchr1p36.11M2 marker
225502_atDOCK8chr9p24.3  
225646_atCTSCchr11q14.1-q14.3M2 marker
225647_s_atCTSCchr11q14.1-q14.3M2 marker
226068_atSYKchr9q22  
227265_atFGL2chr7q11.23  
227266_s_atFYBchr5p13.1  
227346_atZNFN1A1chr7p13-p11.1  
227889_atAYTL1chr16q12.2  
228532_atC1orf162chr1p13.2  
229074_at  
229560_atTLR8chrXp22Innate immunity 
229723_atTAGAPchr6q25.3  
229937_x_atLILRB1chr19q13.4  
230252_atGPR92chr12p13.31  
230391_at  
230550_atMS4A6Achr11q12.1  
230925_atAPBB1IPchr10p12.1  
232617_atCTSSchr1q21M2 marker
232843_s_atDOCK8chr9p24.3  
234987_atC20orf118chr20q11.23  
236028_atIBSPchr4q21-q25  
244434_at  
38487_atSTAB1chr3p21.1M2 marker
*
Gordon S: Nat Rev Immunol 3:23-35, 2003.
Van Ginderachter JA, Meerschaut S, Liu Y, et al: Blood 108:525-535, 2006.
Biswas SK, Gangi L, Paul S, et al: Blood 107:2112-2122, 2006.
§
Kristiansen M, Graversen JH, Jacobsen C, et al: Nature 409:198-201, 2001.
Kzhyshkowska J, Gratchev A, Goerdt S: J Cell Mol Med 10:635-649, 2006.
Stein M, Keshav S, Harris N, et al: J Exp Med 176:287-292, 1992.
#
Bogdan C, Nathan C: Ann N Y Acad Sci 685:713-739, 1993.
**
Willment JA, Lin HH, Reid DM, et al: J Immunol 171:4569-4573, 2003.
Table A10. Cluster G25
Probe IDGene SymbolChromosomal Location
201983_s_atEGFRchr7p12
201984_s_atEGFRchr7p12
203373_atSOCS2chr12q
203484_atSEC61Gchr7p11.2
209692_atEYA2chr20q13.1
210135_s_atSHOX2chr3q25-q26.1
224999_atEGFRchr7p12
228307_atEMILIN3chr20q11.2-q12
232120_atEGFRchr7p12
232224_atMASP1chr3q27-q28
232539_at
232541_atEGFRchr7p12
232882_atFOXO1Achr13q14.1
232925_atEGFRchr7p12
232935_atLHFPchr13q12
233025_atPDZD2chr5p13.3
233044_atEGFRchr7p12
243327_atEGFRchr7p12
Table A11. G28 Versus G98 Probe Sets
G1(S1)G1(S4)Probe SetAnnotation*
G28204121_atGADD45G Chr:9q22.1-q22.2
G28G981557051_s_atHOXA2 Chr:7p15-p14
G28G98204362_atSCAP2 Chr:7p21-p15
G28G98205522_atHOXD4 Chr:2q31.1
G28G98206858_s_atHOXC6 Chr:12q13.3
G28G98209905_atHOXA9 Chr:7p15-p14
G28G98213150_atHOXA10 Chr:7p15-p14
G28G98213844_atHOXA5 Chr:7p15-p14
G28G98214651_s_atHOXA9 Chr:7p15-p14
G28G98225639_atSCAP2 Chr:7p21-p15
G28G98226582_atLOC400043 Chr:12q13.13
G28G98228564_atLOC375295 Chr:2q31.2
G28G98228642_atHOXA2 Chr:7p15-p14
G28G98229400_atHOXD10 Chr:2q31.1
G28G98231906_atHOXD8 Chr:2q31.1
G28G98235521_atHOXA3 Chr:7p15-p14
G28G98235753_atHOXA7 Chr:7p15-p14
G28G98238847_atHOXD10 Chr:2q31.1
G28G98239153_atFLJ41747 Chr:12q13.13
G28G98244521_atTSHZ2 Chr:20q13.2
G98204304_s_atPROM1 Chr:4p15.33
G98226863_atFAM110C Chr:2p25.3
*
Updated Affymetrix (Santa Clara, CA) annotation, November 2007.
Table A12. Estimated Hazard Ratios for Treatment Effect in MGMT-Methylated Patients at Different Expression Levels of G98: Multivariate Cox Proportional Hazards Model (n = 68)
Expression LevelHazard Ratio95% CITreatment Interaction Term P
G98  .002
First quartile0.130.05 to 0.37 
Median0.250.11 to 0.60 
Third quartile0.470.20 to 1.13 
NOTE. The log-hazard function was defined as:
g(t,X,TMZ,MGMT,age,β) = ln[h0(t)] + β1X + β2TMZ + β3MGMT + β4age + β5(X * TMZ) + β6(TMZ * MGMT), where X represents the expression level of G98 (summarized as the mean of the cluster), TMZ is the dichotomous variable representing the regimen chemoradiotherapy versus radiotherapy alone, MGMT is a dichotomous variable representing MGMT promoter methylation status and age is a dichotomous variable (> 50 years at the time of randomization).
Table A13. Mouse Probe Sets Used for GSEA in the Krivtsov Data Set: G98 Mouse Homologues
Probe Sets*Gene TitleGene Symbol
1431475_a_atHomeobox A10HoxA10
1427433_s_atHomeobox A3HoxA3
1448926_atHomeobox A5HoxA5
1452421_atHomeobox A3HoxA3
1455626_atHomeobox A9HoxA9
1421579_atHomeobox A9HoxA9
1427454_atHomeobox C6HoxC6
1418879_atRIKEN cDNA 9030611O19 gene9030611O19Rik
1418895_atsrc family–associated phosphoprotein 2Scap2
1419700_a_atProminin 1Prom1
1449252_atRIKEN cDNA 9030611O19 gene9030611O19Rik
1450209_atHomeobox D4HoxD4
NOTE. Only most variant probe sets (standard deviation, > 0.5) were retained for GSEA. Krivtsov AV, Twomey D, Feng Z, et al: Nature 442:818-822, 2006.
Abbreviation: GSEA, Gene Set Enrichment Analysis.
*
Affymetrix (Santa Clara, CA) annotation, November 2006.
Table A14. Human Probe Sets Used for GSEA in the Ross Data Set: G98 U133A Corresponding Probe Sets to U133 Plus 2.0
Probe Sets*Gene TitleGene Symbol
214651_s_atHomeobox A9HOXA9
213150_atHomeobox A10HOXA10
209905_atHomeobox A9HOXA9
213844_atHomeobox A5HOXA5
204362_atsrc family–associated phosphoprotein 2SCAP2
204304_s_atProminin 1PROM1
NOTE. Only most variant probe sets (standard deviation, > 0.5) were retained for GSEA. Ross ME, Mahfouz R, Onciu M, et al: Blood 104:3679-3687, 2004.
Abbreviation: GSEA, Gene Set Enrichment Analysis.
*
Affymetrix (Santa Clara, CA) annotation, November 2006.

Glossary Terms

HOX (homeobox) genes:
The proteins encoded by these genes, the homeodomain proteins, play important roles in the developmental processes.
MGMT:
The DNA repair enzyme, O6-methylguanine DNA methyltransferase, which confers resistance to alkylating agents. Thus, cells are protected from the toxicity of alkylating agents, which frequently target the O6 position of guanine in DNA.
Clustering:
Organization of data consisting of many variables (multivariate data) into classes with similar patterns. Hierarchical clustering creates a dendrogram based on pairwise similarities in gene expression within a set of samples. Samples within a cluster are more similar to one another than to samples outside the cluster. The vertical length of branches in the tree represents the extent of similarity between the samples. Thus, shorter the branch length, the fewer the differences between the samples.
Tissue microarray:
Used to analyze the expression of genes of interest simultaneously in multiple tissue samples, tissue microarrays consist of hundreds of individual tissue samples placed on slides ranging from 2 to 3 mm in diameter. Using conventional histochemical and molecular detection techniques, tissue microarrays are powerful tools to evaluate the expression of genes of interest in tissue samples. In cancer research, tissue microarrays are used to analyze the frequency of a molecular alteration in different tumor type, to evaluate prognostic markers, and to test potential diagnostic markers.
EGFR (epidermal growth factor receptor):
Also known as HER-1, EGFR belongs to a family of receptors (HER-2, HER-3, HER-4 are other members of the family) and binds to the EGF, TGF-α, and other related proteins, leading to the generation of proliferative and survival signals within the cell. It also belongs to the larger family of tyrosine kinase receptors and is generally overexpressed in several solid tumors of epithelial origin.
CD133:
CD133, encoded by the Prominin-1 gene, is used as a marker for stem cells of normal tissues such as neural and hematopoietic stem cells, but also as a marker for tumor stem-like cells of distinct origin such as brain tumors and breast and colon cancer.
Cancer stem-like cells:
A cancer cell that has the potential to transfer disease or to form tumors after transplantation. Cancer stem-like cells have the potential to self-renew (see self-renewal), forming additional tumorigenic cancer cells of similar phenotype, and to give rise to phenotypically diverse cancer cells with more limited potential.
Self-renewal:
Process by which a progenitor gives rise to daughter progenitors of equivalent developmental potential (ie, multipotent stem cells self-renew by dividing to generate one or two multipotent daughter cells).
Tumor-host interaction:
The evolving cross talk between different cell types within the tumor and its host environment. Cancer cells can subvert the normal tissue architecture and reprogram the activity of stromal cells to their advantage, such as promoting the formation of new blood vessels or the recruitment of inflammatory cells, which are key events promoting cancer cell survival, invasion, and metastasis.
Fig 1. Sample dendrogram S1(G1) by Coupled Two-Way Clustering (CTWC). Sample dendrogram S1(G1) emerging from clustering all 84 samples S1 with all genes G1. Stable sample clusters S2 to S6 emerge. Clinical information: age at diagnosis (green, < 50 years; brown, > 50 years), overall survival in months (green, < 9; red, 9-18; pink > 18); MGMT methylation status (gray, methylated; black, unmethylated; white, unknown); sex (red, female); and sample Id# (sample identification number, nontumoral brain tissue, yellow; recurrent glioblastoma, black; color code tumor pairs from same patient, black only recurrent glioblastoma).
Fig 2. (A): The distance matrix of all filtered genes G1 (Euclidian distance) heatmap; blue indicates short distance (more similar), red indicates large distance. The distance ranges from 0 (corresponding to Pearson correlation = 1; ie, full similarity of the two expression profiles) to 2 (corresponding to anticorrelated profiles, Pearson = −1). Zero correlation indicated by orange. Annotation is based on gene dendrogram G1(S1), not shown. The distance matrix visualizes relationships between clusters as indicated by lines and arrows (eg, G7 is in the center of a supercluster, reflecting close relationship with G9, G12, G14 and G2). In addition G7 is related to G29, but not G23 or G21. (B) Respective distance matrices for two published data sets with 48 and 54 glioblastomas14,15 comprising the common probe sets (3,649) and ordered according to the matrix in Figure 2A. All three data sets appear similar, particularly visible for the largest clusters annotated in Figure 2A.
Fig 3. Relation of glioblastoma derived HOX gene signature with survival. (A) Stable cluster G98 was found by clustering all genes G1 with samples clustered in S4 (69 glioblastomas). Red indicates relative overexpression and blue relative underexpression. (B) Glioblastoma-derived neurospheres (40× magnification) cultured under stem-cell conditions exhibit strong nuclear staining for HOXA10 and express CD133 as determined by immunohistochemistry. Representative neurospheres from two glioblastoma are shown. (C) Kaplan-Meier survival estimates of the 42 patients treated with concomitant and adjuvant temozolomide (TMZ) and radiotherapy (TMZ/RT→TMZ) separated into high (High HOX) versus low (Low HOX) expressors of G98 genes (dichotomized according to Coupled Two-Way Clustering sample dendrogram). The P value of the log-rank test is shown for the two groups, stratified by the MGMT methylation status. M-MGMT, methylated MGMT; U-MGMT, unmethylated MGMT.
Fig 4. HOXA10 expression determined by immunohistochemistry on tumor microarray. Diameter of tissues in upper panel is 0.6 mm. (A) High nuclear expression; (B) focal high nuclear expression; and (C) no expression.
Fig 5. HOXA10 expression and overall survival in independent patient cohort. Kaplan-Meier survival estimates of (A) 39 patients randomly assigned to concomitant and adjuvant temozolomide (TMZ) and radiotherapy (TMZ/RT→TMZ) and (B) 37 patients randomly assigned to RT, separated into high versus low expressors of HOXA10 as determined by immunohistochemistry on tumor microarray.
Fig 6. Association of EGFR expression with survival. Kaplan-Meier curves of the 42 patients in the concomitant and adjuvant temozolomide (TMZ) and radiotherapy (TMZ/RT→TMZ) cohort divided between low and high expression of EGFR (probe set 201983_s_at, dichotomized according to a Gaussian mixture model). The P value from log-rank test is shown for two groups defined by EGFR expression and stratified according to the MGMT methylation status. M-MGMT, methylated MGMT; U-MGMT, unmethylated MGMT.
Table 1. Cox Analysis Using Stable Gene Clusters Identified in G1(S1) in the Cohort of Treated Patients (n = 42)
From G1(S1) DescriptionNo. of Probes in ClusterP*False-Discovery RateHazard Ratio95% CI
Main ClustersSubclusters      
G2 Migration-related19.210.320.730.45 to 1.19
G7 Tumor blood vessel markers77.040.110.730.54 to 0.99
G9 Hypoxia-induced32.700.780.890.49 to 1.61
G11 SOX genes15.250.341.460.77 to 2.77
G12 Interferon-induced genes80.780.831.060.72 to 1.55
G13 Chromosome 1239.020.091.131.02 to 1.24
G14 Myeloid lineage/adhesion26.030.090.610.39 to 0.95
G16 Glial lineage49.070.161.640.96 to 2.81
G18 Brain physiology25.030.091.941.07 to 3.51
G20 ??24.200.321.360.85 to 2.19
G21> G17 > G10 > G3Neural genes213.990.991.000.69 to 1.46
G22 Imprinted genes34.200.320.650.34 to 1.25
G23> G19 > G4Oligodendrocyte markers110.660.781.070.78 to 1.47
G24> G15 > G8Innate immune response134.030.090.650.44 to 0.96
G25 EGFR18.0020.032.781.44 to 5.36
G27> G5Spermatogenesis79.100.201.380.94 to 2.03
G28 HOX genes20.0040.032.691.38 to 5.26
G29> G26 > G6Proliferation101.430.551.270.70 to 2.32
NOTE. Bold text indicates statistical significance, P < .05.
*
P value for the coefficient relative to the mean of the cluster in a multivariate Cox proportional hazards model adjusted for MGMT methylation status and age as additive independent risk factors (y∼ β1meanCluster + β2MGMT + β3age).
False discovery rate to correct for multiple testing (Benjamini-Hochberg procedure).17
Each continuous variable was scaled to have the interquartile range equal to 1 and median equal to 0.
Table 2. Overall Cox Proportional Hazards Models for the Cohort of 42 Patients Treated With Concomitant and Adjuvant Temozolomide and Radiotherapy
VariableUnivariate Model   Multivariate Model   
 Hazard Ratio95% CIPR2Hazard Ratio95% CIPR2
HOX cluster (G98), mean expression1.891.03 to 3.46.040.103.321.61 to 6.82.0010.67*
EGFR expression1.871.09 to 3.22.020.123.131.62 to 6.04< .001 
MGMT methylation0.150.06 to 0.37< .0010.380.060.0 to 0.20< .001 
Age > 50 years1.890.93 to 3.83.080.072.611.22 to 5.57.01 
*
R2 for the overall multivariate Cox proportional hazards model.

Acknowledgments

We thank A.-C. Diserens, P. Descombes, and D. Chollet for excellent technical support; D.R. Macdonald, A. Guha, and A. Merlo for providing frozen tumor tissues; and the clinical colleagues and nurses involved in the studies at the participating hospitals.
Supported by the Swiss National Science Foundation (Grant No. 3100AO-108266/1; M.E.H.); Jacqueline Seroussi Memorial Foundation for Cancer Research (M.E.H.), Translational Research Fund of the EORTC (M.E.H.); the Nélia and Amadeo Barletta Foundation (M.E.H., R.S.); National Center of Competence in Research (NCCR) Molecular Oncology at the Swiss Institute for Experimental Cancer Research (ISREC) (M.E.H., M.D.); Ridgefield Foundation (E.D.); EC FP6 (E.D.); and the VITAL-IT project of the Swiss Institute of Bioinformatics (E.M., M.D.).
A.M. and E.M. contributed equally to this article.
Terms in blue are defined in the glossary, found at the end of this article and online at www.jco.org.
Authors’ disclosures of potential conflicts of interest and author contributions are found at the end of this article.

References

1.
Stupp R, Mason WP, van den Bent MJ, et al: Radiotherapy plus concomitant and adjuvant temozolomide for glioblastoma. N Engl J Med 352::987,2005-996,
2.
Stupp R, Dietrich P, Ostermann Kraljevic S, et al: Promising survival for patients with newly diagnosed glioblastoma multiforme treated with concomitant radiation plus temozolomide followed by adjuvant temozolomide. J Clin Oncol 20::1375,2002-1382,
3.
Hegi ME, Diserens AC, Godard S, et al: Clinical trial substantiates the predictive value of O-6-methylguanine-DNA methyltransferase promoter methylation in glioblastoma patients treated with temozolomide. Clin Cancer Res 10::1871,2004-1874,
4.
Hegi ME, Diserens AC, Gorlia T, et al: MGMT gene silencing and benefit from temozolomide in glioblastoma. N Engl J Med 352::997,2005-1003,
5.
Stupp R, Hegi ME, Gilbert MR, et al: Chemoradiotherapy in malignant glioma: Standard of care and future directions. J Clin Oncol 25::4127,2007-4136,
6.
Reya T, Morrison SJ, Clarke MF, et al: Stem cells, cancer, and cancer stem cells. Nature 414::105,2001-111,
7.
Bao S, Wu Q, McLendon RE, et al: Glioma stem cells promote radioresistance by preferential activation of the DNA damage response. Nature 444::756,2006-760,
8.
Bonnet D, Dick JE: Human acute myeloid leukemia is organized as a hierarchy that originates from a primitive hematopoietic cell. Nat Med 3::730,1997-737,
9.
Krivtsov AV, Twomey D, Feng Z, et al: Transformation from committed progenitor to leukaemia stem cell initiated by MLL-AF9. Nature 442::818,2006-822,
10.
Singh SK, Hawkins C, Clarke ID, et al: Identification of human brain tumour initiating cells. Nature 432::396,2004-401,
11.
Getz G, Domany E: Coupled two-way clustering server. Bioinformatics 19::1153,2003-1154,
12.
Getz G, Levine E, Domany E: Coupled two-way clustering analysis of gene microarray data. Proc Natl Acad Sci U S A 97::12079,2000-12084,
13.
Blatt M, Wiseman S, Domany E: Superparamagnetic clustering of data. Phys Rev Lett 76::3251,1996-3254,
14.
Freije WA, Castro-Vargas FE, Fang Z, et al: Gene expression profiling of gliomas strongly predicts survival. Cancer Res 64::6503,2004-6510,
15.
Phillips HS, Kharbanda S, Chen R, et al: Molecular subclasses of high-grade glioma predict prognosis, delineate a pattern of disease progression, and resemble stages in neurogenesis. Cancer Cell 9::157,2006-173,
16.
Ross ME, Mahfouz R, Onciu M, et al: Gene expression profiling of pediatric acute myelogenous leukemia. Blood 104::3679,2004-3687,
17.
Benjamini Y, Hochberg Y: Controlling the false discovery rate: A practical and powerful approach to multiple testing. J R Stat Soc B 57::289,1995-300,
18.
Clement V, Sanchez P, de Tribolet N, et al: HEDGEHOG-GLI1 signaling regulates human glioma growth, cancer stem cell self-renewal, and tumorigenicity. Curr Biol 17::165,2007-172,
19.
Vairapandi M, Balliet AG, Hoffman B, et al: GADD45b and GADD45g are cdc2/cyclinB1 kinase inhibitors with a role in S and G2/M cell cycle checkpoints induced by genotoxic stress. J Cell Physiol 192::327,2002-338,
20.
Abate-Shen C: Deregulated homeobox gene expression in cancer: Cause or consequence? Nat Rev Cancer 2::777,2002-785,
21.
Abdel-Fattah R, Xiao A, Bomgardner D, et al: Differential expression of HOX genes in neoplastic and non-neoplastic human astrocytes. J Pathol 209::15,2006-24,
22.
Subramanian A, Tamayo P, Mootha VK, et al: Gene set enrichment analysis: A knowledge-based approach for interpreting genome-wide expression profiles. Proc Natl Acad Sci U S A 102::15545,2005-15550,
23.
Manley GT, Fujimura M, Ma T, et al: Aquaporin-4 deletion in mice reduces brain edema after acute water intoxication and ischemic stroke. Nat Med 6::159,2000-163,
24.
Herrlich A, Leitch V, King LS: Role of proneuregulin 1 cleavage and human epidermal growth factor receptor activation in hypertonic aquaporin induction. Proc Natl Acad Sci U S A 101::15799,2004-15804,
25.
Saadoun S, Papadopoulos MC, Hara-Chikuma M, et al: Impairment of angiogenesis and cell migration by targeted aquaporin-1 gene disruption. Nature 434::786,2005-792,
26.
St Croix B, Rago C, Velculescu V, et al: Genes expressed in human tumor endothelium. Science 289::1197,2000-1202,
27.
Madden SL, Cook BP, Nacht M, et al: Vascular gene expression in nonneoplastic and malignant brain. Am J Pathol 165::601,2004-608,
28.
Chute JP, Muramoto GG, Whitesides J, et al: Inhibition of aldehyde dehydrogenase and retinoid signaling induces the expansion of human hematopoietic stem cells. Proc Natl Acad Sci U S A 103::11707,2006-11712,
29.
Piccirillo SG, Reynolds BA, Zanetti N, et al: Bone morphogenetic proteins inhibit the tumorigenic potential of human brain tumour-initiating cells. Nature 444::761,2006-765,
30.
Calabrese C, Poppleton H, Kocak M, et al: A perivascular niche for brain tumor stem cells. Cancer Cell 11::69,2007-82,
31.
Mantovani A, Sozzani S, Locati M, et al: Macrophage polarization: Tumor-associated macrophages as a paradigm for polarized M2 mononuclear phagocytes. Trends Immunol 23::549,2002-555,
32.
Van Ginderachter JA, Movahedi K, Hassanzadeh Ghassabeh G, et al: Classical and alternative activation of mononuclear phagocytes: Picking the best of both worlds for tumor promotion. Immunobiology 211::487,2006-501,
33.
Hussain SF, Yang D, Suki D, et al: The role of human glioma-infiltrating microglia/macrophages in mediating antitumor immune responses. Neuro-oncol 8::261,2006-279,
34.
Dorrance AM, Liu S, Yuan W, et al: Mll partial tandem duplication induces aberrant Hox expression in vivo via specific epigenetic alterations. J Clin Invest 116::2707,2006-2716,
35.
Krivtsov AV, Armstrong SA: MLL translocations, histone modifications and leukaemia stem-cell development. Nat Rev Cancer 7::823,2007-833,
36.
Bachoo RM, Maher EA, Ligon KL, et al: Epidermal growth factor receptor and Ink4a/Arf: Convergent mechanisms governing terminal differentiation and transformation along the neural stem cell to astrocyte axis. Cancer Cell 1::269,2002-277,
37.
Stupp R, Hegi ME: Targeting brain-tumor stem cells. Nat Biotechnol 25::193,2007-194,
38.
Sathornsumetee S, Rich JN: New treatment strategies for malignant gliomas. Expert Rev Anticancer Ther 6::1087,2006-1104,
39.
Jain RK: Normalization of tumor vasculature: An emerging concept in antiangiogenic therapy. Science 307::58,2005-62,
40.
Batchelor TT, Sorensen AG, di Tomaso E, et al: AZD2171, a pan-VEGF receptor tyrosine kinase inhibitor, normalizes tumor vasculature and alleviates edema in glioblastoma patients. Cancer Cell 11::83,2007-95,
41.
Kjellman C, Olofsson SP, Hansson O, et al: Expression of TGF-beta isoforms, TGF-beta receptors, and SMAD molecules at different stages of human glioma. Int J Cancer 89::251,2000-258,

Information & Authors

Information

Published In

Journal of Clinical Oncology
Pages: 3015 - 3024
PubMed: 18565887

History

Published in print: June 20, 2008
Published online: September 21, 2016

Permissions

Request permissions for this article.

Authors

Affiliations

Anastasia Murat
From the Laboratory of Tumor Biology and Genetics, Centre Universitaire Romand de Neurochirurgie; Division of Neuropathology; Multidisciplinary Oncology Center, Centre Hospitalier Universitaire Vaudois and University of Lausanne, Lausanne; the National Center of Competence in Research Molecular Oncology at the Swiss Institute of Experimental Cancer Research; the Swiss Institute of Bioinformatics, Epalinges; the Centre Universitaire Romand de Neurochirurgie, the Service of Oncology, University Hospital Geneva, Geneva; the Department of Clinical Research and Radiation Oncology, Inselspital and University of Berne, Berne; the Institute of Neuropathology, University Hospital Zurich, Zurich, Switzerland; Data Center, European Organisation for Research and Treatment of Cancer (EORTC), Brussels, Belgium; Department of Physics of Complex Systems, Weizmann Institute of Science, Rehovot, Israel; the Department of General Neurology and Hertie Institute for Clinical Brain Research, University of Tübingen, Tübingen; Department of Neurooncology, University Clinic Heidelberg, Heidelberg; Institute of Neuropathology, Charité –Universitätsmedizin Berlin, Berlin, Germany; Department of Neurology, Erasmus, Rotterdam, the Netherlands; Institute of Neurology, Medical University of Vienna, Vienna, Austria; and the Department of Clinical Neurosciences, University of Calgary, Calgary, Canada
Eugenia Migliavacca
From the Laboratory of Tumor Biology and Genetics, Centre Universitaire Romand de Neurochirurgie; Division of Neuropathology; Multidisciplinary Oncology Center, Centre Hospitalier Universitaire Vaudois and University of Lausanne, Lausanne; the National Center of Competence in Research Molecular Oncology at the Swiss Institute of Experimental Cancer Research; the Swiss Institute of Bioinformatics, Epalinges; the Centre Universitaire Romand de Neurochirurgie, the Service of Oncology, University Hospital Geneva, Geneva; the Department of Clinical Research and Radiation Oncology, Inselspital and University of Berne, Berne; the Institute of Neuropathology, University Hospital Zurich, Zurich, Switzerland; Data Center, European Organisation for Research and Treatment of Cancer (EORTC), Brussels, Belgium; Department of Physics of Complex Systems, Weizmann Institute of Science, Rehovot, Israel; the Department of General Neurology and Hertie Institute for Clinical Brain Research, University of Tübingen, Tübingen; Department of Neurooncology, University Clinic Heidelberg, Heidelberg; Institute of Neuropathology, Charité –Universitätsmedizin Berlin, Berlin, Germany; Department of Neurology, Erasmus, Rotterdam, the Netherlands; Institute of Neurology, Medical University of Vienna, Vienna, Austria; and the Department of Clinical Neurosciences, University of Calgary, Calgary, Canada
Thierry Gorlia
From the Laboratory of Tumor Biology and Genetics, Centre Universitaire Romand de Neurochirurgie; Division of Neuropathology; Multidisciplinary Oncology Center, Centre Hospitalier Universitaire Vaudois and University of Lausanne, Lausanne; the National Center of Competence in Research Molecular Oncology at the Swiss Institute of Experimental Cancer Research; the Swiss Institute of Bioinformatics, Epalinges; the Centre Universitaire Romand de Neurochirurgie, the Service of Oncology, University Hospital Geneva, Geneva; the Department of Clinical Research and Radiation Oncology, Inselspital and University of Berne, Berne; the Institute of Neuropathology, University Hospital Zurich, Zurich, Switzerland; Data Center, European Organisation for Research and Treatment of Cancer (EORTC), Brussels, Belgium; Department of Physics of Complex Systems, Weizmann Institute of Science, Rehovot, Israel; the Department of General Neurology and Hertie Institute for Clinical Brain Research, University of Tübingen, Tübingen; Department of Neurooncology, University Clinic Heidelberg, Heidelberg; Institute of Neuropathology, Charité –Universitätsmedizin Berlin, Berlin, Germany; Department of Neurology, Erasmus, Rotterdam, the Netherlands; Institute of Neurology, Medical University of Vienna, Vienna, Austria; and the Department of Clinical Neurosciences, University of Calgary, Calgary, Canada
Wanyu L. Lambiv
From the Laboratory of Tumor Biology and Genetics, Centre Universitaire Romand de Neurochirurgie; Division of Neuropathology; Multidisciplinary Oncology Center, Centre Hospitalier Universitaire Vaudois and University of Lausanne, Lausanne; the National Center of Competence in Research Molecular Oncology at the Swiss Institute of Experimental Cancer Research; the Swiss Institute of Bioinformatics, Epalinges; the Centre Universitaire Romand de Neurochirurgie, the Service of Oncology, University Hospital Geneva, Geneva; the Department of Clinical Research and Radiation Oncology, Inselspital and University of Berne, Berne; the Institute of Neuropathology, University Hospital Zurich, Zurich, Switzerland; Data Center, European Organisation for Research and Treatment of Cancer (EORTC), Brussels, Belgium; Department of Physics of Complex Systems, Weizmann Institute of Science, Rehovot, Israel; the Department of General Neurology and Hertie Institute for Clinical Brain Research, University of Tübingen, Tübingen; Department of Neurooncology, University Clinic Heidelberg, Heidelberg; Institute of Neuropathology, Charité –Universitätsmedizin Berlin, Berlin, Germany; Department of Neurology, Erasmus, Rotterdam, the Netherlands; Institute of Neurology, Medical University of Vienna, Vienna, Austria; and the Department of Clinical Neurosciences, University of Calgary, Calgary, Canada
Tal Shay
From the Laboratory of Tumor Biology and Genetics, Centre Universitaire Romand de Neurochirurgie; Division of Neuropathology; Multidisciplinary Oncology Center, Centre Hospitalier Universitaire Vaudois and University of Lausanne, Lausanne; the National Center of Competence in Research Molecular Oncology at the Swiss Institute of Experimental Cancer Research; the Swiss Institute of Bioinformatics, Epalinges; the Centre Universitaire Romand de Neurochirurgie, the Service of Oncology, University Hospital Geneva, Geneva; the Department of Clinical Research and Radiation Oncology, Inselspital and University of Berne, Berne; the Institute of Neuropathology, University Hospital Zurich, Zurich, Switzerland; Data Center, European Organisation for Research and Treatment of Cancer (EORTC), Brussels, Belgium; Department of Physics of Complex Systems, Weizmann Institute of Science, Rehovot, Israel; the Department of General Neurology and Hertie Institute for Clinical Brain Research, University of Tübingen, Tübingen; Department of Neurooncology, University Clinic Heidelberg, Heidelberg; Institute of Neuropathology, Charité –Universitätsmedizin Berlin, Berlin, Germany; Department of Neurology, Erasmus, Rotterdam, the Netherlands; Institute of Neurology, Medical University of Vienna, Vienna, Austria; and the Department of Clinical Neurosciences, University of Calgary, Calgary, Canada
Marie-France Hamou
From the Laboratory of Tumor Biology and Genetics, Centre Universitaire Romand de Neurochirurgie; Division of Neuropathology; Multidisciplinary Oncology Center, Centre Hospitalier Universitaire Vaudois and University of Lausanne, Lausanne; the National Center of Competence in Research Molecular Oncology at the Swiss Institute of Experimental Cancer Research; the Swiss Institute of Bioinformatics, Epalinges; the Centre Universitaire Romand de Neurochirurgie, the Service of Oncology, University Hospital Geneva, Geneva; the Department of Clinical Research and Radiation Oncology, Inselspital and University of Berne, Berne; the Institute of Neuropathology, University Hospital Zurich, Zurich, Switzerland; Data Center, European Organisation for Research and Treatment of Cancer (EORTC), Brussels, Belgium; Department of Physics of Complex Systems, Weizmann Institute of Science, Rehovot, Israel; the Department of General Neurology and Hertie Institute for Clinical Brain Research, University of Tübingen, Tübingen; Department of Neurooncology, University Clinic Heidelberg, Heidelberg; Institute of Neuropathology, Charité –Universitätsmedizin Berlin, Berlin, Germany; Department of Neurology, Erasmus, Rotterdam, the Netherlands; Institute of Neurology, Medical University of Vienna, Vienna, Austria; and the Department of Clinical Neurosciences, University of Calgary, Calgary, Canada
Nicolas de Tribolet
From the Laboratory of Tumor Biology and Genetics, Centre Universitaire Romand de Neurochirurgie; Division of Neuropathology; Multidisciplinary Oncology Center, Centre Hospitalier Universitaire Vaudois and University of Lausanne, Lausanne; the National Center of Competence in Research Molecular Oncology at the Swiss Institute of Experimental Cancer Research; the Swiss Institute of Bioinformatics, Epalinges; the Centre Universitaire Romand de Neurochirurgie, the Service of Oncology, University Hospital Geneva, Geneva; the Department of Clinical Research and Radiation Oncology, Inselspital and University of Berne, Berne; the Institute of Neuropathology, University Hospital Zurich, Zurich, Switzerland; Data Center, European Organisation for Research and Treatment of Cancer (EORTC), Brussels, Belgium; Department of Physics of Complex Systems, Weizmann Institute of Science, Rehovot, Israel; the Department of General Neurology and Hertie Institute for Clinical Brain Research, University of Tübingen, Tübingen; Department of Neurooncology, University Clinic Heidelberg, Heidelberg; Institute of Neuropathology, Charité –Universitätsmedizin Berlin, Berlin, Germany; Department of Neurology, Erasmus, Rotterdam, the Netherlands; Institute of Neurology, Medical University of Vienna, Vienna, Austria; and the Department of Clinical Neurosciences, University of Calgary, Calgary, Canada
Luca Regli
From the Laboratory of Tumor Biology and Genetics, Centre Universitaire Romand de Neurochirurgie; Division of Neuropathology; Multidisciplinary Oncology Center, Centre Hospitalier Universitaire Vaudois and University of Lausanne, Lausanne; the National Center of Competence in Research Molecular Oncology at the Swiss Institute of Experimental Cancer Research; the Swiss Institute of Bioinformatics, Epalinges; the Centre Universitaire Romand de Neurochirurgie, the Service of Oncology, University Hospital Geneva, Geneva; the Department of Clinical Research and Radiation Oncology, Inselspital and University of Berne, Berne; the Institute of Neuropathology, University Hospital Zurich, Zurich, Switzerland; Data Center, European Organisation for Research and Treatment of Cancer (EORTC), Brussels, Belgium; Department of Physics of Complex Systems, Weizmann Institute of Science, Rehovot, Israel; the Department of General Neurology and Hertie Institute for Clinical Brain Research, University of Tübingen, Tübingen; Department of Neurooncology, University Clinic Heidelberg, Heidelberg; Institute of Neuropathology, Charité –Universitätsmedizin Berlin, Berlin, Germany; Department of Neurology, Erasmus, Rotterdam, the Netherlands; Institute of Neurology, Medical University of Vienna, Vienna, Austria; and the Department of Clinical Neurosciences, University of Calgary, Calgary, Canada
Wolfgang Wick
From the Laboratory of Tumor Biology and Genetics, Centre Universitaire Romand de Neurochirurgie; Division of Neuropathology; Multidisciplinary Oncology Center, Centre Hospitalier Universitaire Vaudois and University of Lausanne, Lausanne; the National Center of Competence in Research Molecular Oncology at the Swiss Institute of Experimental Cancer Research; the Swiss Institute of Bioinformatics, Epalinges; the Centre Universitaire Romand de Neurochirurgie, the Service of Oncology, University Hospital Geneva, Geneva; the Department of Clinical Research and Radiation Oncology, Inselspital and University of Berne, Berne; the Institute of Neuropathology, University Hospital Zurich, Zurich, Switzerland; Data Center, European Organisation for Research and Treatment of Cancer (EORTC), Brussels, Belgium; Department of Physics of Complex Systems, Weizmann Institute of Science, Rehovot, Israel; the Department of General Neurology and Hertie Institute for Clinical Brain Research, University of Tübingen, Tübingen; Department of Neurooncology, University Clinic Heidelberg, Heidelberg; Institute of Neuropathology, Charité –Universitätsmedizin Berlin, Berlin, Germany; Department of Neurology, Erasmus, Rotterdam, the Netherlands; Institute of Neurology, Medical University of Vienna, Vienna, Austria; and the Department of Clinical Neurosciences, University of Calgary, Calgary, Canada
Mathilde C.M. Kouwenhoven
From the Laboratory of Tumor Biology and Genetics, Centre Universitaire Romand de Neurochirurgie; Division of Neuropathology; Multidisciplinary Oncology Center, Centre Hospitalier Universitaire Vaudois and University of Lausanne, Lausanne; the National Center of Competence in Research Molecular Oncology at the Swiss Institute of Experimental Cancer Research; the Swiss Institute of Bioinformatics, Epalinges; the Centre Universitaire Romand de Neurochirurgie, the Service of Oncology, University Hospital Geneva, Geneva; the Department of Clinical Research and Radiation Oncology, Inselspital and University of Berne, Berne; the Institute of Neuropathology, University Hospital Zurich, Zurich, Switzerland; Data Center, European Organisation for Research and Treatment of Cancer (EORTC), Brussels, Belgium; Department of Physics of Complex Systems, Weizmann Institute of Science, Rehovot, Israel; the Department of General Neurology and Hertie Institute for Clinical Brain Research, University of Tübingen, Tübingen; Department of Neurooncology, University Clinic Heidelberg, Heidelberg; Institute of Neuropathology, Charité –Universitätsmedizin Berlin, Berlin, Germany; Department of Neurology, Erasmus, Rotterdam, the Netherlands; Institute of Neurology, Medical University of Vienna, Vienna, Austria; and the Department of Clinical Neurosciences, University of Calgary, Calgary, Canada
Johannes A. Hainfellner
From the Laboratory of Tumor Biology and Genetics, Centre Universitaire Romand de Neurochirurgie; Division of Neuropathology; Multidisciplinary Oncology Center, Centre Hospitalier Universitaire Vaudois and University of Lausanne, Lausanne; the National Center of Competence in Research Molecular Oncology at the Swiss Institute of Experimental Cancer Research; the Swiss Institute of Bioinformatics, Epalinges; the Centre Universitaire Romand de Neurochirurgie, the Service of Oncology, University Hospital Geneva, Geneva; the Department of Clinical Research and Radiation Oncology, Inselspital and University of Berne, Berne; the Institute of Neuropathology, University Hospital Zurich, Zurich, Switzerland; Data Center, European Organisation for Research and Treatment of Cancer (EORTC), Brussels, Belgium; Department of Physics of Complex Systems, Weizmann Institute of Science, Rehovot, Israel; the Department of General Neurology and Hertie Institute for Clinical Brain Research, University of Tübingen, Tübingen; Department of Neurooncology, University Clinic Heidelberg, Heidelberg; Institute of Neuropathology, Charité –Universitätsmedizin Berlin, Berlin, Germany; Department of Neurology, Erasmus, Rotterdam, the Netherlands; Institute of Neurology, Medical University of Vienna, Vienna, Austria; and the Department of Clinical Neurosciences, University of Calgary, Calgary, Canada
Frank L. Heppner
From the Laboratory of Tumor Biology and Genetics, Centre Universitaire Romand de Neurochirurgie; Division of Neuropathology; Multidisciplinary Oncology Center, Centre Hospitalier Universitaire Vaudois and University of Lausanne, Lausanne; the National Center of Competence in Research Molecular Oncology at the Swiss Institute of Experimental Cancer Research; the Swiss Institute of Bioinformatics, Epalinges; the Centre Universitaire Romand de Neurochirurgie, the Service of Oncology, University Hospital Geneva, Geneva; the Department of Clinical Research and Radiation Oncology, Inselspital and University of Berne, Berne; the Institute of Neuropathology, University Hospital Zurich, Zurich, Switzerland; Data Center, European Organisation for Research and Treatment of Cancer (EORTC), Brussels, Belgium; Department of Physics of Complex Systems, Weizmann Institute of Science, Rehovot, Israel; the Department of General Neurology and Hertie Institute for Clinical Brain Research, University of Tübingen, Tübingen; Department of Neurooncology, University Clinic Heidelberg, Heidelberg; Institute of Neuropathology, Charité –Universitätsmedizin Berlin, Berlin, Germany; Department of Neurology, Erasmus, Rotterdam, the Netherlands; Institute of Neurology, Medical University of Vienna, Vienna, Austria; and the Department of Clinical Neurosciences, University of Calgary, Calgary, Canada
Pierre-Yves Dietrich
From the Laboratory of Tumor Biology and Genetics, Centre Universitaire Romand de Neurochirurgie; Division of Neuropathology; Multidisciplinary Oncology Center, Centre Hospitalier Universitaire Vaudois and University of Lausanne, Lausanne; the National Center of Competence in Research Molecular Oncology at the Swiss Institute of Experimental Cancer Research; the Swiss Institute of Bioinformatics, Epalinges; the Centre Universitaire Romand de Neurochirurgie, the Service of Oncology, University Hospital Geneva, Geneva; the Department of Clinical Research and Radiation Oncology, Inselspital and University of Berne, Berne; the Institute of Neuropathology, University Hospital Zurich, Zurich, Switzerland; Data Center, European Organisation for Research and Treatment of Cancer (EORTC), Brussels, Belgium; Department of Physics of Complex Systems, Weizmann Institute of Science, Rehovot, Israel; the Department of General Neurology and Hertie Institute for Clinical Brain Research, University of Tübingen, Tübingen; Department of Neurooncology, University Clinic Heidelberg, Heidelberg; Institute of Neuropathology, Charité –Universitätsmedizin Berlin, Berlin, Germany; Department of Neurology, Erasmus, Rotterdam, the Netherlands; Institute of Neurology, Medical University of Vienna, Vienna, Austria; and the Department of Clinical Neurosciences, University of Calgary, Calgary, Canada
Yitzhak Zimmer
From the Laboratory of Tumor Biology and Genetics, Centre Universitaire Romand de Neurochirurgie; Division of Neuropathology; Multidisciplinary Oncology Center, Centre Hospitalier Universitaire Vaudois and University of Lausanne, Lausanne; the National Center of Competence in Research Molecular Oncology at the Swiss Institute of Experimental Cancer Research; the Swiss Institute of Bioinformatics, Epalinges; the Centre Universitaire Romand de Neurochirurgie, the Service of Oncology, University Hospital Geneva, Geneva; the Department of Clinical Research and Radiation Oncology, Inselspital and University of Berne, Berne; the Institute of Neuropathology, University Hospital Zurich, Zurich, Switzerland; Data Center, European Organisation for Research and Treatment of Cancer (EORTC), Brussels, Belgium; Department of Physics of Complex Systems, Weizmann Institute of Science, Rehovot, Israel; the Department of General Neurology and Hertie Institute for Clinical Brain Research, University of Tübingen, Tübingen; Department of Neurooncology, University Clinic Heidelberg, Heidelberg; Institute of Neuropathology, Charité –Universitätsmedizin Berlin, Berlin, Germany; Department of Neurology, Erasmus, Rotterdam, the Netherlands; Institute of Neurology, Medical University of Vienna, Vienna, Austria; and the Department of Clinical Neurosciences, University of Calgary, Calgary, Canada
J. Gregory Cairncross
From the Laboratory of Tumor Biology and Genetics, Centre Universitaire Romand de Neurochirurgie; Division of Neuropathology; Multidisciplinary Oncology Center, Centre Hospitalier Universitaire Vaudois and University of Lausanne, Lausanne; the National Center of Competence in Research Molecular Oncology at the Swiss Institute of Experimental Cancer Research; the Swiss Institute of Bioinformatics, Epalinges; the Centre Universitaire Romand de Neurochirurgie, the Service of Oncology, University Hospital Geneva, Geneva; the Department of Clinical Research and Radiation Oncology, Inselspital and University of Berne, Berne; the Institute of Neuropathology, University Hospital Zurich, Zurich, Switzerland; Data Center, European Organisation for Research and Treatment of Cancer (EORTC), Brussels, Belgium; Department of Physics of Complex Systems, Weizmann Institute of Science, Rehovot, Israel; the Department of General Neurology and Hertie Institute for Clinical Brain Research, University of Tübingen, Tübingen; Department of Neurooncology, University Clinic Heidelberg, Heidelberg; Institute of Neuropathology, Charité –Universitätsmedizin Berlin, Berlin, Germany; Department of Neurology, Erasmus, Rotterdam, the Netherlands; Institute of Neurology, Medical University of Vienna, Vienna, Austria; and the Department of Clinical Neurosciences, University of Calgary, Calgary, Canada
Robert-Charles Janzer
From the Laboratory of Tumor Biology and Genetics, Centre Universitaire Romand de Neurochirurgie; Division of Neuropathology; Multidisciplinary Oncology Center, Centre Hospitalier Universitaire Vaudois and University of Lausanne, Lausanne; the National Center of Competence in Research Molecular Oncology at the Swiss Institute of Experimental Cancer Research; the Swiss Institute of Bioinformatics, Epalinges; the Centre Universitaire Romand de Neurochirurgie, the Service of Oncology, University Hospital Geneva, Geneva; the Department of Clinical Research and Radiation Oncology, Inselspital and University of Berne, Berne; the Institute of Neuropathology, University Hospital Zurich, Zurich, Switzerland; Data Center, European Organisation for Research and Treatment of Cancer (EORTC), Brussels, Belgium; Department of Physics of Complex Systems, Weizmann Institute of Science, Rehovot, Israel; the Department of General Neurology and Hertie Institute for Clinical Brain Research, University of Tübingen, Tübingen; Department of Neurooncology, University Clinic Heidelberg, Heidelberg; Institute of Neuropathology, Charité –Universitätsmedizin Berlin, Berlin, Germany; Department of Neurology, Erasmus, Rotterdam, the Netherlands; Institute of Neurology, Medical University of Vienna, Vienna, Austria; and the Department of Clinical Neurosciences, University of Calgary, Calgary, Canada
Eytan Domany
From the Laboratory of Tumor Biology and Genetics, Centre Universitaire Romand de Neurochirurgie; Division of Neuropathology; Multidisciplinary Oncology Center, Centre Hospitalier Universitaire Vaudois and University of Lausanne, Lausanne; the National Center of Competence in Research Molecular Oncology at the Swiss Institute of Experimental Cancer Research; the Swiss Institute of Bioinformatics, Epalinges; the Centre Universitaire Romand de Neurochirurgie, the Service of Oncology, University Hospital Geneva, Geneva; the Department of Clinical Research and Radiation Oncology, Inselspital and University of Berne, Berne; the Institute of Neuropathology, University Hospital Zurich, Zurich, Switzerland; Data Center, European Organisation for Research and Treatment of Cancer (EORTC), Brussels, Belgium; Department of Physics of Complex Systems, Weizmann Institute of Science, Rehovot, Israel; the Department of General Neurology and Hertie Institute for Clinical Brain Research, University of Tübingen, Tübingen; Department of Neurooncology, University Clinic Heidelberg, Heidelberg; Institute of Neuropathology, Charité –Universitätsmedizin Berlin, Berlin, Germany; Department of Neurology, Erasmus, Rotterdam, the Netherlands; Institute of Neurology, Medical University of Vienna, Vienna, Austria; and the Department of Clinical Neurosciences, University of Calgary, Calgary, Canada
Mauro Delorenzi
From the Laboratory of Tumor Biology and Genetics, Centre Universitaire Romand de Neurochirurgie; Division of Neuropathology; Multidisciplinary Oncology Center, Centre Hospitalier Universitaire Vaudois and University of Lausanne, Lausanne; the National Center of Competence in Research Molecular Oncology at the Swiss Institute of Experimental Cancer Research; the Swiss Institute of Bioinformatics, Epalinges; the Centre Universitaire Romand de Neurochirurgie, the Service of Oncology, University Hospital Geneva, Geneva; the Department of Clinical Research and Radiation Oncology, Inselspital and University of Berne, Berne; the Institute of Neuropathology, University Hospital Zurich, Zurich, Switzerland; Data Center, European Organisation for Research and Treatment of Cancer (EORTC), Brussels, Belgium; Department of Physics of Complex Systems, Weizmann Institute of Science, Rehovot, Israel; the Department of General Neurology and Hertie Institute for Clinical Brain Research, University of Tübingen, Tübingen; Department of Neurooncology, University Clinic Heidelberg, Heidelberg; Institute of Neuropathology, Charité –Universitätsmedizin Berlin, Berlin, Germany; Department of Neurology, Erasmus, Rotterdam, the Netherlands; Institute of Neurology, Medical University of Vienna, Vienna, Austria; and the Department of Clinical Neurosciences, University of Calgary, Calgary, Canada
Roger Stupp
From the Laboratory of Tumor Biology and Genetics, Centre Universitaire Romand de Neurochirurgie; Division of Neuropathology; Multidisciplinary Oncology Center, Centre Hospitalier Universitaire Vaudois and University of Lausanne, Lausanne; the National Center of Competence in Research Molecular Oncology at the Swiss Institute of Experimental Cancer Research; the Swiss Institute of Bioinformatics, Epalinges; the Centre Universitaire Romand de Neurochirurgie, the Service of Oncology, University Hospital Geneva, Geneva; the Department of Clinical Research and Radiation Oncology, Inselspital and University of Berne, Berne; the Institute of Neuropathology, University Hospital Zurich, Zurich, Switzerland; Data Center, European Organisation for Research and Treatment of Cancer (EORTC), Brussels, Belgium; Department of Physics of Complex Systems, Weizmann Institute of Science, Rehovot, Israel; the Department of General Neurology and Hertie Institute for Clinical Brain Research, University of Tübingen, Tübingen; Department of Neurooncology, University Clinic Heidelberg, Heidelberg; Institute of Neuropathology, Charité –Universitätsmedizin Berlin, Berlin, Germany; Department of Neurology, Erasmus, Rotterdam, the Netherlands; Institute of Neurology, Medical University of Vienna, Vienna, Austria; and the Department of Clinical Neurosciences, University of Calgary, Calgary, Canada
Monika E. Hegi
From the Laboratory of Tumor Biology and Genetics, Centre Universitaire Romand de Neurochirurgie; Division of Neuropathology; Multidisciplinary Oncology Center, Centre Hospitalier Universitaire Vaudois and University of Lausanne, Lausanne; the National Center of Competence in Research Molecular Oncology at the Swiss Institute of Experimental Cancer Research; the Swiss Institute of Bioinformatics, Epalinges; the Centre Universitaire Romand de Neurochirurgie, the Service of Oncology, University Hospital Geneva, Geneva; the Department of Clinical Research and Radiation Oncology, Inselspital and University of Berne, Berne; the Institute of Neuropathology, University Hospital Zurich, Zurich, Switzerland; Data Center, European Organisation for Research and Treatment of Cancer (EORTC), Brussels, Belgium; Department of Physics of Complex Systems, Weizmann Institute of Science, Rehovot, Israel; the Department of General Neurology and Hertie Institute for Clinical Brain Research, University of Tübingen, Tübingen; Department of Neurooncology, University Clinic Heidelberg, Heidelberg; Institute of Neuropathology, Charité –Universitätsmedizin Berlin, Berlin, Germany; Department of Neurology, Erasmus, Rotterdam, the Netherlands; Institute of Neurology, Medical University of Vienna, Vienna, Austria; and the Department of Clinical Neurosciences, University of Calgary, Calgary, Canada

Notes

Corresponding author: Monika E. Hegi, PhD, Laboratory of Tumor Biology and Genetics, Centre Universitaire Romand de Neurochirurgie, Centre Hospitalier Universitaire Vaudois and University of Lausanne, 46, rue du Bugnon, Lausanne 1011/Switzerland; e-mail: [email protected]

Metrics & Citations

Metrics

Altmetric

Citations

Article Citation

Download Citation

If you have the appropriate software installed, you can download article citation data to the citation manager of your choice. Simply select your manager software from the list below and click Download.

For more information or tips please see 'Downloading to a citation manager' in the Help menu.

Format





Download article citation data for:
Anastasia Murat, Eugenia Migliavacca, Thierry Gorlia, Wanyu L. Lambiv, Tal Shay, Marie-France Hamou, Nicolas de Tribolet, Luca Regli, Wolfgang Wick, Mathilde C.M. Kouwenhoven, Johannes A. Hainfellner, Frank L. Heppner, Pierre-Yves Dietrich, Yitzhak Zimmer, J. Gregory Cairncross, Robert-Charles Janzer, Eytan Domany, Mauro Delorenzi, Roger Stupp, Monika E. Hegi
Journal of Clinical Oncology 2008 26:18, 3015-3024

View Options

View options

PDF

View PDF

Get Access

Login options

Check if you have access through your login credentials or your institution to get full access on this article.

Personal login Institutional Login

Purchase Options

Purchase this article to get full access to it.

Purchase this Article

Subscribe

Subscribe to this Journal
Renew Your Subscription
Become a Member

Media

Figures

Other

Tables

Share

Share

Share article link

Share