| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Molecular Biology, Pathobiology, and Genetics |
Departments of 1 Medicine, 2 Neurobiology, 3 Pathology, and 4 Molecular Genetics and Microbiology, 5 W.M. Keck Center for Neuro-Oncogenomics, Institutes of 6 Statistics and Decision Sciences, and 7 Genome Sciences and Policy, Duke University Medical Center, Durham, North Carolina; and 8 Institute of Information and Mathematical Sciences, Massey University, New Zealand
Requests for reprints: Jeremy N. Rich, Duke University Medical Center, Box 2900, Durham, NC 27710. Phone: 919-681-1693; Fax: 919-684-6514; E-mail: rich0001{at}mc.duke.edu.
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
At least two genetic pathways have been delineated in glioblastoma development: de novo and secondary glioblastomas (9). De novo glioblastomas represent the most frequent presentation with an initial diagnosis of glioblastoma without evidence of preexistent lower grade tumor. These patients are commonly of older age and have a high rate of epidermal growth factor receptor (EGFR) amplification, p16INK4A deletion, and phosphatase and tensin homologue deleted on chromosome 10 (PTEN; mutated in multiple advanced cancers 1) mutations. In contrast, secondary glioblastomas arise after a preceding diagnosis of lower grade tumors. TP53 and RB mutations are thought to be more common in the development of secondary glioblastomas (9). Despite these genetic differences, no significant differentiation in patient survival has been noted between de novo and secondary glioblastomas when controlled for age. In fact, there have been no widely validated prognostic genetic markers for glioblastoma patients. Rather, several genetic changes, including PTEN and EGFR mutations, have been linked to poor prognosis in patients with anaplastic astrocytomas (10), suggesting that these are markers of transformation to glioblastomas.
Molecular profiles of glioma patient specimens have suggested that gene expression may predict patient outcome more accurately than pathologic measures (11–14). These analyses have provided large sets of genes which may be expected to regulate the process of tumor progression. To explore genome-scale expression information for potential value in defining contributors to the malignancy of gliomas with the worst prognosis—glioblastoma patients over the age of 50—we examined tumor RNA in relation to patient survival. Affymetrix gene chip analysis of 41 tumor specimens was examined using computational statistical methods to explore the potential for generating gene expression–based markers of survival, and to elucidate expression-based associations among any genes showing such potential. Additional analysis of the full genome-scale gene expression data using statistical graphical models that define empirical association networks over genes leads to the identification of additional genes linked to those arising in the primary predictive models. These results have been contrasted with traditional DNA studies including measurement of EGFR amplification, mutational analysis of EGFR, TP53, and PTEN, and loss of heterozygosity detection at 9p, 10p, 10q, and 17p.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
PCR-based molecular analysis. Normal DNA was extracted from lymphocytes. Tumor DNA and RNA was isolated from sections cut from the frozen block. Exons 5 to 8 of the TP53 gene and all 9 exons of PTEN were resequenced by capillary electrophoresis on ABI 3100. EGFR DNA amplification assay was done by co-PCR amplifying a 3' untranslated region fragment of EGFR gene with a fragment of exon 3 of IFNG gene as internal control, using fluorescent tagged primers. EGFR/IFNG peak area ratios of >5 are considered as indication of EGFR gene amplification. CDKN2A (p16INK4A) deletion assay was carried out by SYBR Green fluorescent assay on ABI 7900HT. A 
Ct CDKN2A-Globin (internal control) value of >1.7 was considered indicative of homozygous deletion of CDKN2A. Loss of heterozygosity analysis of 9p, 10, and 17p was done by comparing allele intensities of PCR amplified loci (three from each arm) from tumor and corresponding patient's lymphocyte DNA. Fragment analysis was done by capillary electrophoresis on ABI 3100. Peak height ratios (tumor/blood) <0.65 or >1.67 were considered indicative of loss of heterozygosity.
To detect the EGFR vIII variant, RNA extracted from tumor tissue was reverse-transcribed using Invitrogen Superscript II kit and PCR-amplified using primers from exons 1 and 8. The PCR products were electrophoresed in a 3% agarose gel. This assay generates a 111-bp product in vIII variants and a 912-bp product in the wild-type. Levels of SPARC and doublecortex (DCX) transcripts were assayed by SYBR Green fluorescent assay on ABI 7900HT. Normalization of input cDNA amount was done by comparing amplification of housekeeping genes glyceraldehyde-3-phosphate dehydrogenase and ß2-microglobulin. Ct values represent average Ct SPARC or DCX minus average Ct B2M or glyceraldehyde-3-phosphate dehydrogenase.
Microarray chip RNA hybridization procedures. Total RNA was extracted from tumor tissue with Qiagen (Valencia, CA) RNEasy kits, and assessed for quality with an Agilent Lab-on-a-Chip 2100 Bioanalyzer. Hybridization target probes were prepared from total RNA according to standard Affymetrix protocols and hybridized to the human U133A GeneChip (see Supplementary Materials for full details).
Data preprocessing prior to the formal statistical analysis involved standard processes of normalization, expression intensity estimation and screening for genes showing reasonable variation across samples. The Affymetrix U133a DNA microarrays provide assay of over 20,000 probe sets. The expression intensities for all genes across the 41 samples were estimated using robust multi-array average (16), with probe-level quantile normalization, as implemented in the Bioconductor software suite (17). The resulting robust multi-array average expression intensity estimates were then screened to identify genes whose robust multi-array average levels probe vary at least 4-fold across the samples, and whose maximum level exceeded seven on the log2 scale, leading to P = 8,408 genes/probe sets whose robust multi-array average expression intensities are the candidate predictors in the regression model analysis and computational search.
Statistical analysis. The predictive analysis evaluated linear regression models of the form y = a0 + a1x1 + ... + akxk + e, where y represents log survival time, each xi represents the expression level of gene i, k is a small integer, and e represents an unexplained, random component. The challenge of statistical analysis is to search for subsets of genes that together define significant predictive regressions—that is, to select both the number k of genes, or variables, and then the specific set of genes x1, ..., xk by searching over subsets. This includes the possibility of no association with any genes, i.e., k = 0. Technically, with many genes available, this requires some form of stochastic search. The analysis is based on a so-called shotgun stochastic search (18), which in a distributed computer environment, allows the rapid evaluation of many such models so long as the search is constrained to values of k that are reasonably small. The parallel computational strategies implemented are very efficient and the search over models generally focuses quickly on subsets of relevant models with higher probability (if such a model exists).
Analysis here with n = 41 samples confirms that a number of models with three to four genes are of some interest. The analysis heavily penalizes more complex models, initially very strongly favoring the null hypothesis of no significant predictors in this model context among the thousands of genes in a manner that naturally counters the false discovery propensity of purely likelihood-based model search analyses. In addition, routine calculations confirm that the false-positive rate for discovery of single variable regressions as significant as those identified among the top candidates here is tiny. Of a number of regression models involving between three and five genes that are identified, many rely on overlapping sets of genes with two of the three "key" genes—SPARC, Doublecortex, and Semaphorin3B—appearing in a larger number of most highly scoring models. This reflects inherent collinearities among gene subsets, some of which is naturally induced by coregulation of genes within common pathways, so that models based on distinct although overlapping sets of predictive genes may well reflect a single or small number of relevant biological pathways rather than distinct explanations.
The overall practical relevance of the set of regressions identified (as opposed to nominal statistical significance of any one model) is evaluated by cross-validation prediction. That is, the analysis is repeatedly done in a leave-one-out context, with the tumor left out then being predicted based on the set of models defined and weight by the analysis of the remaining n–1 samples, as is (or should be) standard predictive evaluation in problems where predictive value is of primary interest (19–21). Predictions are based on standard weighted model averaging: models identified are evaluated according to their relative data-based probabilities of model fit, and these probabilities provide weights to use in averaging predictions for the hold-out (or future) tumor samples.
Further statistical analysis of the gene expression data aimed to explore a number of genes implicated in the survival regressions to identify additional, statistically associated genes that would then be candidates for potential biological interpretation. A gene showing up as a marker of survival may be a statistical surrogate of other, potentially mechanistic genes. This component of the statistical analysis applied the regression model search repeatedly; now, rather than treating logged survival times as the variable to predict, we used expression of each of a selected small set of genes as the outcome variable. Genes selected as responses for this analysis are the three key genes already discussed, including each of the two probe sets representing DCX, and an additional gene, KIAA0831. These genes represent the four (really five, with the two versions of DCX) most highly scored genes, in terms of posterior probability of appearing in regression models for survival of the full set of over 8,000 genes. Exploring regression models separately for each of these genes as response generates, in each case, a set of models and ranks the genes appearing as predictors in those models according to posterior probabilities, just as in regressions for survival. The four most highly scoring genes in each case are identified in Supplementary Table 1 along with the primary genes already mentioned. Note that, for doublecortex, where two probe sets appeared as predictors of survival, each probe set was considered separately as a response, but both probe sets were removed from the set of predictors for these procedures.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
|
|
Expression levels of Osteonectin, Doublecortex, and Semaphorin3B together associate with patient survival. Dominant regression models involve probe sets for Osteonectin, Doublecortex, and Semaphorin3B. The overall most likely model is in fact the regression on these three genes, and other models with appreciable posterior probability involve subsets of two of these three together with one other gene. Together, these three genes provide explanatory markers of survival (Table 2). Poorer survival is associated with higher levels of each of these three genes; none of them serves as a useful predictive marker alone, but the concordance of higher values together seems to associate with poorer survival (Table 3). Of note, the expression levels of individual genes were not highly correlated with one another, except for very high correlation between the two Doublecortex isoforms (Supplementary Table 2). One informative plot that summarizes the roles of these three genes as markers of survival is given in Fig. 2A. The metagene plotted is simply the dominant singular factor (principal component) of the expression levels of these three genes across samples, and is plotted here in a three-dimensional scatter plot together with the expression levels of two of the three—SPARC and Doublecortex (see ref. 16 for discussion of the use of metagenes defined as singular factors from groups of statistically associated genes in related contexts). The points are color-coded according to the predicted mean of log survival corresponding to the expression levels, running from blue (lower risk) to red (higher risk).
|
Expression profiles can derive additional relationships between genes expressed in patient specimens related to survival. Additional statistical analysis explored statistical associations in gene expression data among a few of the key genes implicated in the survival regressions and other genes that, in a regression context, showed up as predictive of expression fluctuations of this initial set of genes (see Materials and Methods). Figure 3 displays a graph summarizing the predictive relationships identified in this analysis, presented as a statistical graphical association model—a subgraph of the much larger graph relating expression levels across all genes (23, 24). The set of genes here are listed in Supplementary Table 1. Arrows are directed from a gene A to a gene B to represent the appearance of gene A as a predictor of gene B in one of the three most highly weighted regressions for expression of gene B. A dashed edge indicates that gene A had a negative regression coefficient in the highest probability model in which it was involved in predicting gene B. The number labeling an edge from gene A to gene B indicates the aggregate posterior probability of all regression models for gene B that contain gene A as an explanatory variable—an overall measure of the relevance/weight of gene A as predictive of gene B.
|
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Of the genes detected in our expression studies, Osteonectin has been most clearly linked to glioma pathophysiology in prior studies. Osteonectin, also known as secreted protein acidic and rich in cysteine (SPARC) or BM-40, is an extracellular protein that plays an important role in development, tissue healing and remodeling, and angiogenesis (reviewed in ref. 27). Osteonectin/SPARC was originally discovered as an important component of bone (28) but is also expressed in epithelia exhibiting high rates of turnover (gut, skin, and glandular tissue), as well as vascular smooth muscle cells and endothelial cells (29). In addition to its normal physiologic role, Osteonectin/SPARC is abnormally expressed in cancers. Many cancers, including cancers of the gastrointestinal tract, breast, lung, kidney, adrenal cortex, prostate, bladder, and meninges (29–33), express increased SPARC levels that are associated with a conversion to invasive and metastatic tumors. Gene expression analysis of potential tumor markers in malignant gliomas by sequential analysis of gene expression found a 10-fold overexpression of SPARC (34). Immunohistochemical analysis of human gliomas reveals that SPARC is expressed specifically at sites of tumor invasion—in tumor cells at the tumor-brain interface, endothelial cells of tumor-associated vessels, and reactive astrocytes (30). In vitro assays of gliomas have further implicated SPARC expression in increased glioma invasion (35) and angiogenesis (36). We and others have shown that forced expression of SPARC in human glioma cell lines promotes tumor cell invasion in both cell culture and animal models associated with matrix metalloproteinase expression (37, 38). More recently, we have shown that SPARC expression in human glioma cell lines induces increased activation of AKT and promotes cell survival with serum survival (39). Thus, the prominent role of SPARC in influencing patient survival suggests that regulation of the cellular microenvironment may significantly contribute to tumor behavior beyond the progression to high-grade malignancy.
Doublecortin and doublecortex are gene products encoded by the locus linked to X-linked lissencephaly (40). Whereas males with X-linked lissencephaly display abnormally smooth brains, female patients develop brains with abnormal heterotopic gray matter regions reminiscent of a second cortical region. During brain development, neuronal precursors are generated at periventricular regions then migrate outwards to populate the cortical layers. Heterotopic gray matter regions have been linked to impaired migration of neuronal cell bodies. Doublecortin encodes a microtubule-associated protein with two actin-binding domains that regulate neuronal migration (41–43). The activity of doublecortin is suppressed by phosphorylation at specific residues (42, 43). The kinases regulating these phosphorylation events include Cdk5, MARKS, and protein kinase A (42, 43). These kinases may be abnormally expressed or regulated in some cancers, including gliomas. The expression levels and contributions of doublecortin in cancer have been previously unrecognized. Likewise, the regulation of doublecortin expression is poorly understood. A recent report linked targeted disruption of PTEN expression to changes in doublecortin expression (44). The striking relationship that we have detected between the expression levels of SOX4 and doublecortin strongly suggest co-regulation of these genes. SOX4 is a SRY box containing transcription factor that is expressed during brain development in the cerebellar external granule layer (45). The functions of SOX4 have been dissected through targeted disruption in mice, resulting in early vascular death with defects in cardiac outflow tract formation and pro-B lymphocyte generation (46). SOX4 has not been widely studied in cancer, but SOX4 has been shown to be overexpressed in another central nervous system cancer, medulloblastoma (47). The link between SOX4 expression and that of both doublecortin isoforms strongly suggests that SOX4 may regulate doublecortin expression. In a strong validation of the potential biological relationships that may be derived by the statistical analyses used in our studies, we have detected a similar relationship between SOX11 and one of the expressed transcripts of doublecortin. SOX11 is the SOX family member most closely related to SOX4 with potential overlapping functions.
Semaphorin3B (SEMA3B) is a class III, secreted semaphorin with SEMA, immunoglobulin, and short basic domains. In parallel to other semaphorins, SEMA3B regulates neuronal migration. SEMA3B antagonizes SEMA3A neuronal growth cone repulsion at neuropilin-1 homodimers but acts as an agonist at neuropilin-1/2 heterodimers or neuropilin-2 homodimers (48). The neuropilins are transmembrane receptors without clear independent signaling functions that may act as accessory receptors for vascular endothelial growth factor (VEGF). Although VEGF has been most closely linked to endothelial cell proliferation and increased vascular permeability, evidence of the role of VEGF in cellular migration and brain development are apparent. The SEMA3B locus is located at 3p21.3, which is a homozygous deletion region in small cell lung cancer, suggesting that SEMA3B may act as a tumor suppressor gene in some cancers (49). Reintroduction of p53 into the p53-null U373MG human malignant glioma cell line induced SEMA3B expression (50). The dichotomous role of SEMA3B parallels that of SPARC, which can also restrict tumor cell proliferation and exhibit tumor suppressive roles in cancers as well. The putative SEMA3B receptors, the neuropilins, are expressed in human gliomas and may serve biological roles in tumor malignancy.
In summary, this gene expression study provides evidence that three genes which regulate cellular motility may contribute to the poor prognosis of patients with glioblastomas. No previous studies, of which we are aware of, have elucidated the conclusive links between expression of specific genes and survival of older glioblastoma patients. Although cellular mitogenesis and resistance to apoptosis have been the targets of many biological therapies, our regression analyses using gene expression to explain the survival outcomes revealed that genes whose primary cellular effects may be the regulation of cellular migration appear as candidate markers of poor survival. Together, these results suggest that tumor migration may represent an important effector of glioblastoma malignancy and may warrant accelerated development of specific therapies. Current targeted therapies for glioblastomas have focused on cellular pathways that primarily regulate proliferation and apoptosis. Clinical experience suggests that tumor invasion is a severe challenge in the management of glioblastoma patients. Elegant studies by Berens, Bjerkvig, Rao and others have shown that glioma invasion can be the target of directed therapies and that these approaches may augment the efficacy of traditional therapies (reviewed in ref. 3). Our studies may lend further weight to these approaches and suggest that the gene products whose expression is now linked to poor survival may be useful therapeutic targets. Future studies will prospectively determine the link between the expression of SPARC, Doublecortex, and SEMA3B in gliomas of all grades and patient outcome. Additional studies under way will further dissect the contributions of these gene products to the biology of gliomas, including tumor cell invasion, proliferation, apoptosis, and secretion of angiogenic factors.
|
Acknowledgments |
|---|
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
|
Footnotes |
|---|
Received 11/ 2/04. Revised 2/ 4/05. Accepted 3/ 9/05.
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
|
|
 |
|
  H.-Q. Qu, K. Jacob, S. Fatet, B. Ge, D. Barnett, O. Delattre, D. Faury, A. Montpetit, L. Solomon, P. Hauser, et al. Genome-wide profiling using single-nucleotide polymorphism arrays identifies novel chromosomal imbalances in pediatric glioblastomas Neuro Oncology, October 15, 2009; (2009) nop001v1. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. Bredel, D. M. Scholtens, G. R. Harsh, C. Bredel, J. P. Chandler, J. J. Renfrow, A. K. Yadav, H. Vogel, A. C. Scheck, R. Tibshirani, et al. A Network Model of a Cooperative Genetic Landscape in Brain Tumors JAMA, July 15, 2009; 302(3): 261 - 275. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 L. Capparuccia and L. Tamagnone Semaphorin signaling in cancer cells and in cells of the tumor microenvironment - two sides of a coin J. Cell Sci., June 1, 2009; 122(11): 1723 - 1736. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 P. L. Nghiemphu, W. Liu, Y. Lee, T. Than, C. Graham, A. Lai, R. M. Green, W. B. Pope, L. M. Liau, P. S. Mischel, et al. Bevacizumab and chemotherapy for recurrent glioblastoma: A single-institution experience Neurology, April 7, 2009; 72(14): 1217 - 1222. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 V Martin, L Mazzucchelli, and M Frattini An overview of the epidermal growth factor receptor fluorescence in situ hybridisation challenge in tumour pathology J. Clin. Pathol., April 1, 2009; 62(4): 314 - 324. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. Li, J. Walling, S. Ahn, Y. Kotliarov, Q. Su, M. Quezado, J. C. Oberholtzer, J. Park, J. C. Zenklusen, and H. A. Fine Unsupervised Analysis of Transcriptomic Profiles Reveals Six Glioma Subtypes Cancer Res., March 1, 2009; 69(5): 2091 - 2099. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
Y. Zhang, N. Zhang, B. Dai, M. Liu, R. Sawaya, K. Xie, and S. Huang FoxM1B Transcriptionally Regulates Vascular Endothelial Growth Factor Expression and Promotes the Angiogenesis and Growth of Glioma Cells Cancer Res., November 1, 2008; 68(21): 8733 - 8742. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 H. Colman and K. Aldape Molecular Predictors in Glioblastoma: Toward Personalized Therapy Arch Neurol, July 1, 2008; 65(7): 877 - 883. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 D. Hagerstrand, A. Smits, A. Eriksson, S. Sigurdardottir, T. Olofsson, M. Hartman, M. Nister, H. Kalimo, and A. Ostman Gene expression analyses of grade II gliomas and identification of rPTP{beta}/{zeta} as a candidate oligodendroglioma marker Neuro-oncol, February 1, 2008; 10(1): 2 - 9. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 T. Haque, D. Faury, S. Albrecht, E. Lopez-Aguilar, P. Hauser, M. Garami, Z. Hanzely, L. Bognar, R. F. Del Maestro, J. Atkinson, et al. Gene Expression Profiling from Formalin-Fixed Paraffin-Embedded Tumors of Pediatric Glioblastoma Clin. Cancer Res., November 1, 2007; 13(21): 6284 - 6292. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 D. Krex, B. Klink, C. Hartmann, A. von Deimling, T. Pietsch, M. Simon, M. Sabel, J. P. Steinbach, O. Heese, G. Reifenberger, et al. Long-term survival with glioblastoma multiforme Brain, October 1, 2007; 130(10): 2596 - 2606. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
R. J. Gilbertson and D. H. Gutmann Tumorigenesis in the Brain: Location, Location, Location Cancer Res., June 15, 2007; 67(12): 5579 - 5582. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 D. Faury, A. Nantel, S. E. Dunn, M.-C. Guiot, T. Haque, P. Hauser, M. Garami, L. Bognar, Z. Hanzely, P. P. Liberski, et al. Molecular Profiling Identifies Prognostic Subgroups of Pediatric Glioblastoma and Shows Increased YB-1 Expression in Tumors J. Clin. Oncol., April 1, 2007; 25(10): 1196 - 1208. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
E. A. Maher, C. Brennan, P. Y. Wen, L. Durso, K. L. Ligon, A. Richardson, D. Khatry, B. Feng, R. Sinha, D. N. Louis, et al. Marked Genomic Differences Characterize Primary and Secondary Glioblastoma Subtypes and Identify Two Distinct Molecular and Clinical Secondary Glioblastoma Entities Cancer Res., December 1, 2006; 66(23): 11502 - 11513. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 P. O. Sakariassen, L. Prestegarden, J. Wang, K.-O. Skaftnesmo, R. Mahesparan, C. Molthoff, P. Sminia, E. Sundlisaeter, A. Misra, B. B. Tysnes, et al. Angiogenesis-independent tumor growth mediated by stem-like cancer cells PNAS, October 31, 2006; 103(44): 16466 - 16471. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
T. Okunaga, Y. Urata, S. Goto, T. Matsuo, S. Mizota, K. Tsutsumi, I. Nagata, T. Kondo, and Y. Ihara Calreticulin, a Molecular Chaperone in the Endoplasmic Reticulum, Modulates Radiosensitivity of Human Glioblastoma U251MG Cells. Cancer Res., September 1, 2006; 66(17): 8662 - 8671. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
Q. Wei, L. Clarke, D. K. Scheidenhelm, B. Qian, A. Tong, N. Sabha, Z. Karim, N. A. Bock, R. Reti, R. Swoboda, et al. High-grade glioma formation results from postnatal pten loss or mutant epidermal growth factor receptor expression in a transgenic mouse glioma model. Cancer Res., August 1, 2006; 66(15): 7429 - 7437. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
Y.-J. Chen, V. Hakin-Smith, M. Teo, G. E. Xinarianos, D. A. Jellinek, T. Carroll, D. McDowell, M. R. MacFarlane, R. Boet, B. C. Baguley, et al. Association of Mutant TP53 with Alternative Lengthening of Telomeres and Favorable Prognosis in Glioma. Cancer Res., July 1, 2006; 66(13): 6473 - 6476. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. Wrensch, J. K. Wiencke, J. Wiemels, R. Miike, J. Patoka, M. Moghadassi, A. McMillan, K. T. Kelsey, K. Aldape, K. R. Lamborn, et al. Serum IgE, Tumor Epidermal Growth Factor Receptor Expression, and Inherited Polymorphisms Associated with Glioma Survival. Cancer Res., April 15, 2006; 66(8): 4531 - 4541. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. Liu, B. Dai, S.-H. Kang, K. Ban, F.-J. Huang, F. F. Lang, K. D. Aldape, T.-x. Xie, C. E. Pelloski, K. Xie, et al. FoxM1B Is Overexpressed in Human Glioblastomas and Critically Regulates the Tumorigenicity of Glioma Cells. Cancer Res., April 1, 2006; 66(7): 3593 - 3602. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
D. A. Reardon, J. N. Rich, H. S. Friedman, and D. D. Bigner Recent Advances in the Treatment of Malignant Astrocytoma J. Clin. Oncol., March 10, 2006; 24(8): 1253 - 1265. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 R. E. Brown and A. Law Morphoproteomic Demonstration of Constitutive Nuclear Factor-kappaB Activation in Glioblastoma Multiforme with Genomic Correlates and Therapeutic Implications Ann. Clin. Lab. Sci., January 1, 2006; 36(4): 421 - 426. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. Aghi, P. Gaviani, J. W. Henson, T. T. Batchelor, D. N. Louis, and F. G. Barker II Magnetic Resonance Imaging Characteristics Predict Epidermal Growth Factor Receptor Amplification Status in Glioblastoma Clin. Cancer Res., December 15, 2005; 11(24): 8600 - 8605. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. DeLong, G. Yao, Q. Wang, A. Dobra, E. P. Black, J. T. Chang, A. Bild, M. West, J. R. Nevins, and H. Dressman DIG--a system for gene annotation and functional discovery Bioinformatics, July 1, 2005; 21(13): 2957 - 2959. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Cancer Research | Clinical Cancer Research |
| Cancer Epidemiology Biomarkers & Prevention | Molecular Cancer Therapeutics |
| Molecular Cancer Research | Cancer Prevention Research |
| Cancer Prevention Journals Portal | Cancer Reviews Online |
| Annual Meeting Education Book | Meeting Abstracts Online |
