SURVIVAL PREDICTION FOR BRAIN TUMOR PATIENTS USING GENE EXPRESSION DATA

Brain tumor is one of the most aggressive types of cancer in humans, with an estimated median survival time of 12 months and only 4% of the patients surviving more than 5 years after disease diagnosis. Until recently, brain tumor prognosis has been based only on clinical information such as tumor grade and patient age, but there are reports indicating that molecular profiling of gliomas can reveal subgroups of patients with distinct survival rates. We hypothesize that coupling molecular profiling of brain tumors with clinical information might improve predictions of patient survival time and, consequently, better guide future treatment decisions. In order to evaluate this hypothesis, the general goal of this research is to build models for survival prediction of glioma patients using DNA molecular profiles (U133 Affymetrix gene expression microarrays) along with clinical information. First, a predictive Random Forest model is built for binary outcomes (i.e. short vs. long-term survival) and a small subset of genes whose expression values can be used to predict survival time is selected. Following, a new statistical methodology is developed for predicting time-to-death outcomes using Bayesian ensemble trees. Due to a large heterogeneity observed within prognostic classes obtained by the Random Forest model, prediction can be improved by relating time-to-death with gene expression profile directly. We propose a Bayesian ensemble model for survival prediction which is appropriate for high-dimensional data such as gene expression data. Our approach is based on the ensemble "sum-of-trees" model which is flexible to incorporate additive and interaction effects between genes. We specify a fully Bayesian hierarchical approach and illustrate our methodology for the CPH, Weibull, and AFT survival models. We overcome the lack of conjugacy using a latent variable formulation to model the covariate effects which decreases computation time for model fitting. Also, our proposed models provides a model-free way to select important predictive prognostic markers based on controlling false discovery rates. We compare the performance of our methods with baseline reference survival methods and apply our methodology to an unpublished data set of brain tumor survival times and gene expression data, selecting genes potentially related to the development of the disease under study. A closing discussion compares results obtained by Random Forest and Bayesian ensemble methods under the biological/clinical perspectives and highlights the statistical advantages and disadvantages of the new methodology in the context of DNA microarray data analysis.

Development and validation of the Neuro Symptom Inventory in the primary brain tumor patient population

Symptoms has been shown to predict quality of life, treatment course and survival in solid tumor patients. Currently, no instrument exists that measures both cancer-related symptoms and the neurologic symptoms that are unique to persons with primary brain tumors (PBT). The aim of this study was to develop and validate an instrument to measure symptoms in patients who have PBT. A conceptual analysis of symptoms and symptom theories led to defining the symptoms experience as the perception of the frequency, intensity, distress, and meaning that occurs as symptoms are produced, perceived, and expressed. The M.D. Anderson Symptom Inventory (MDASI) measures both symptoms and how they interfere with daily functioning in patients with cancer, which is similar to the situational meaning defined in the analysis. A list of symptoms pertinent to the PBT population was added to the core MDASI and reviewed by a group of experts for validity. As a result, 18 items were added to the core MDASI (the MDASI-BT) for the next phase of instrument development, establishing validity and reliability through a descriptive, cross-sectional approach with PBT patients. Data were collected with a patient completed demographic data sheet, an investigator completed clinician checklist, and the MDASI-BT. Analysis evaluated the reliability and validity of the MDASI-BT in PBT patients. Data were obtained from 201 patients. The number of items was reduced to 22 by evaluation of symptom severity as well as cluster analysis. Regression analysis showed more than half (56%) of the variability in symptom severity was explained by the brain tumor module items. Factor analysis confirmed that the 22 item MDASI-BT measured six underlying constructs: (a) affective; (b) cognitive; (c) focal neurologic deficits; (d) constitutional symptoms; (e) treatment-related symptoms; and (f) gastrointestinal symptoms. The MDASI-BT was sensitive to disease severity and if the patient was hospitalized. The MDASI-BT is the first instrument to measure symptoms in PBT patients that has demonstrated reliability and validity. It is the first step in a program of research to evaluate the occurrence of symptoms and plan and evaluate interventions for PBT patients. ^

Occupation and glioma: The Harris County Brain Tumor Study, 2001--2010

The relationship between occupational exposures and glioma has not been adequately assessed due to the lack of studies in current scientific literature. To address this disparity, the Harris County Brain Tumor Study, an ongoing population-based case-control study, began in January 2001. Longest-held occupation for 382 cases and 629 controls were frequency matched on age (within 5 years), sex, and race and placed into 14 predetermined occupational categories. Adjusted odds ratios and 95% confidence intervals were calculated for each category using multiple logistic regression. Potential confounders assessed included sex, age, smoking status, education and income. For all subjects, significantly elevated adjusted odds ratios were found in health-related (aOR=1.66; 95%CI=1.03, 2.68), teaching (aOR=1.84; 95%CI=1.17, 2.88), and protective service (aOR=3.6; 95%CI=1.05, 12.31) occupational categories after controlling for sex and education. A significantly lowered odds ratio was seen in the writers, artists, and entertainers category (aOR=0.14; 95%CI=0.03, 0.58). In the stratified analyses, which controlled for education, males had a significantly elevated odds ratio for protective service workers (aOR=4.83; 95%CI=1.24, 18.83) while a significantly lower odds ratio was found in mechanics and machine operators (aOR=0.33; 95%CI=0.12,0.87). In females, we observed a significantly elevated odds ratio in teachers (aOR=1.99; 95%CI=1.20,3.31) and a significantly lower odds ratio in clerical workers (aOR=0.63; 95%CI=0.45,0.90). These analyses revealed several significant associations and allowed for separate analyses by gender, distinguishing this study from many glioma studies. Further analyses should provide a large enough sample size to stratify by gender as well as histological subtype.^

Identification of genetic risk factors for cerebellar mutism in pediatric brain tumor patients

Following posterior fossa surgery for resection of childhood medulloblastoma and primitive neuroectodermal tumor (M/PNET), cerebellar mutism (CM) may develop. This is a condition of absent or diminished speech in a conscious patient with no evidence of oral apraxia, which can be accompanied by other symptoms of the posterior fossa syndrome complex, which includes ataxia and hypotonia. Little is known about the etiology. Therefore, we conducted a SNP, gene, and pathway-level analysis to assess the role of host genetic variation on the risk of CM in M/PNET subjects following treatment. Cases (n= 20) and controls (n= 53) were recruited from the Childhood Cancer Epidemiology and Prevention Center, in Houston, TX. DNA samples were genotyped using the Illumina Human 1M Quad SNP chip. Ten pathways were identified from logistic regression used to identify the marginal effect of each SNP on CM risk. The minP test was conducted to identify associations between SNPs categorized to genes and CM risk. Pathways were assessed to determine if there was a significant enrichment of genes in the pathway compared to all other pathways. There were 78 genes that reached the threshold of min P ≤0.05 in 948 genes. The Neurotoxicity pathway was the most significant pathway after adjusting for multiple comparisons (q=0.040 and q=0.005, using Fisher's exact test and a test of proportions, respectively). Most genes within the Neurotoxicity pathway that reached a threshold of minP ≤0.05 were known to have an apoptosis function, possibly inducing neuronal apoptosis in the dentatothalamocortical pathway, and may be important in CM etiology in this population. This is the first study to assess the potential role of genetic risk factors on CM. As an exploratory study, these results should be replicated in a larger sample. ^

Targeting the Blood-brain Barrier with a Non-canonical Iron-mimicry Mechanism

Treatment of central nervous system (CNS) diseases is limited by the blood-brain barrier (BBB), a selective vascular interface restricting passage of most molecules from blood into brain. Specific transport systems have evolved allowing circulating polar molecules to cross the BBB and gain access to the brain parenchyma. However, to date, few ligands exploiting such systems have proven clinically viable in the setting of CNS diseases. We reasoned that combinatorial phage-display screenings in vivo would yield peptides capable of crossing the BBB and allow for the development of ligand-directed targeting strategies of the brain. Here we show the identification of a peptide mediating systemic targeting to the normal brain and to an orthotopic human glioma model. We demonstrate that this peptide functionally mimics iron through an allosteric mechanism and that a non-canonical association of (i) transferrin, (ii) the iron-mimic ligand motif, and (iii) transferrin receptor mediates binding and transport of particles across the BBB. We also show that in orthotopic human glioma xenografts, a combination of transferrin receptor over-expression plus extended vascular permeability and ligand retention result in remarkable brain tumor targeting. Moreover, such tumor targeting attributes enables Herpes simplex virus thymidine kinase-mediated gene therapy of intracranial tumors for molecular genetic imaging and suicide gene delivery with ganciclovir. Finally, we expand our data by analyzing a large panel of primary CNS tumors through comprehensive tissue microarrays. Together, our approach and results provide a translational avenue for the detection and treatment of brain tumors.

The role of deregulated E2F1 expression in brain tumors

Mutations disabling the retinoblastoma (Rb) pathway are among the most common in human cancers, including brain cancer. These mutations promote tumor development through deregulated control of the E2F family of transcription factors. E2F1 belongs to a class of E2F's identified as transcriptional activators and involved in the G1/S phase transition of the cell. However, E2F-1 presents with a paradox as it is considered to have membership in two gene classes, functioning as both an oncogene and a tumor suppressor. This unusual trait generates a degree of uncertainty on the role that E2F1 plays in the development or maintenance of any given tumor. Here we show that E2F1 functions as an oncogene in brain tumors through the generation of mice engineered to overexpress E2F1 specifically within glial cells and neuronal progenitors as directed by the GFAP promoter. Mice carrying the transgene develop with high penetrance a phenotype characterized by neurological deficits including paresia, ataxia, head tilt and seizures. MRI imagining of the tgE2F1 mice reveals a low incidence of mild hydrocephalus, and most notably, histological analysis demonstrates that 25% of tgE2F1 mice present with the spontaneous formation of malignant brain tumors. Overall these neoplasms show histological features from a wide range of aggressive brain cancers including medulloblastoma, choroid plexus carcinoma, primary neuroectodermic tumor and malignant gliomas. Isolation and characterization of astrocytes from the tgE2F1 animal reveals a highly proliferative population of cells with 55% ± 2.5 of the tgE2F1astrocytes, 35% ± 3.4 normal mouse astrocytes in S-phase and the acquired capacity to grow in anchorage independent conditions. Additionally tgE2F1 astrocytes show an aberrant phenotype with random chromosomal fusions and nearly all cells demonstrating polyploidy. Taken together, this model forces a comparison to human brain tumor formation. Mouse age as related to tumoral mimics the human scenario with juvenile tgE2F1 mice presenting embryonal tumors typically identified in children, and older tgE2F1 mice demonstrating gliomas. In this regard, this study suggests a global role for E2F1 in the formation and maintenance of multilineage brain tumors, irrefutably establishing E2F1 as an oncogene in the brain. ^

A nested case-control study of radiation exposure and brain cancer incidence in the United States Air Force

A nested case-control study design was used to investigate the relationship between radiation exposure and brain cancer risk in the United States Air Force (USAF). The cohort consisted of approximately 880,000 men with at least 1 year of service between 1970 and 1989. Two hundred and thirty cases were identified from hospital discharge records with a diagnosis of primary malignant brain tumor (International Classification of Diseases, 9th revision, code 191). Four controls were exactly matched with each case on year of age and race using incidence density sampling. Potential career summary extremely low frequency (ELF) and microwave-radiofrequency (MWRF) radiation exposures were based upon the duration in each occupation and an intensity score assigned by an expert panel. Ionizing radiation (IR) exposures were obtained from personal dosimetry records.^ Relative to the unexposed, the overall age-race adjusted odds ratio (OR) for ELF exposure was 1.39, 95 percent confidence interval (CI) 1.03-1.88. A dose-response was not evident. The same was true for MWRF, although the OR = 1.59, with 95 percent CI 1.18-2.16. Excess risk was not found for IR exposure (OR = 0.66, 45 percent CI 0.26-1.72).^ Increasing socioeconomic status (SES), as identified by military pay grade, was associated with elevated brain tumor risk (officer vs. enlisted personnel age-race adjusted OR = 2.11, 95 percent CI 1.98-3.01, and senior officers vs. all others age-race adjusted OR = 3.30, 95 percent CI 2.0-5.46). SES proved to be an important confounder of the brain tumor risk associated with ELF and MWRF exposure. For ELF, the age-race-SES adjusted OR = 1.28, 95 percent CI 0.94-1.74, and for MWRF, the age-race-SES adjusted OR = 1.39, 95 percent CI 1.01-1.90.^ These results indicate that employment in Air Force occupations with potential electromagnetic field exposures is weakly, though not significantly, associated with increased risk for brain tumors. SES appeared to be the most consistent brain tumor risk factor in the USAF cohort. Other investigators have suggested that an association between brain tumor risk and SES may arise from differential access to medical care. However, in the USAF cohort health care is universally available. This study suggests that some factor other than access to medical care must underlie the association between SES and brain tumor risk. ^

Medulloblastomas overexpress the p53-inactivating oncogene WIP1/PPM1D.

Medulloblastoma is the most common malignant brain tumor of childhood. Despite numerous advances, clinical challenges range from recurrent and progressive disease to long-term toxicities in survivors. The lack of more effective, less toxic therapies results from our limited understanding of medulloblastoma growth. Although TP53 is the most commonly altered gene in cancers, it is rarely mutated in medulloblastoma. Accumulating evidence, however, indicates that TP53 pathways are disrupted in medulloblastoma. Wild-type p53-induced phosphatase 1 (WIP1 or PPM1D) encodes a negative regulator of p53. WIP1 amplification (17q22-q23) and its overexpression have been reported in diverse cancer types. We examined primary medulloblastoma specimens and cell lines, and detected WIP1 copy gain and amplification prevalent among but not exclusively in the tumors with 17q gain and isochromosome 17q (i17q), which are among the most common cytogenetic lesions in medulloblastoma. WIP1 RNA levels were significantly higher in the tumors with 17q gain or i17q. Immunoblots confirmed significant WIP1 protein in primary tumors, generally higher in those with 17q gain or i17q. Under basal growth conditions and in response to the chemotherapeutic agent, etoposide, WIP1 antagonized p53-mediated apoptosis in medulloblastoma cell lines. These results indicate that medulloblastoma express significant levels of WIP1 that modulate genotoxic responsiveness by negatively regulating p53.

Abnormal expression of REST/NRSF and Myc in neural stem/progenitor cells causes cerebellar tumors by blocking neuronal differentiation.

Medulloblastoma, one of the most malignant brain tumors in children, is thought to arise from undifferentiated neural stem/progenitor cells (NSCs) present in the external granule layer of the cerebellum. However, the mechanism of tumorigenesis remains unknown for the majority of medulloblastomas. In this study, we found that many human medulloblastomas express significantly elevated levels of both myc oncogenes, regulators of neural progenitor proliferation, and REST/NRSF, a transcriptional repressor of neuronal differentiation genes. Previous studies have shown that neither c-Myc nor REST/NRSF alone could cause tumor formation. To determine whether c-Myc and REST/NRSF act together to cause medulloblastomas, we used a previously established cell line derived from external granule layer stem cells transduced with activated c-myc (NSC-M). These immortalized NSCs were able to differentiate into neurons in vitro. In contrast, when the cells were engineered to express a doxycycline-regulated REST/NRSF transgene (NSC-M-R), they no longer underwent terminal neuronal differentiation in vitro. When injected into intracranial locations in mice, the NSC-M cells did not form tumors either in the cerebellum or in the cerebral cortex. In contrast, the NSC-M-R cells did produce tumors in the cerebellum, the site of human medulloblastoma formation, but not when injected into the cerebral cortex. Furthermore, the NSC-M-R tumors were blocked from terminal neuronal differentiation. In addition, countering REST/NRSF function blocked the tumorigenic potential of NSC-M-R cells. To our knowledge, this is the first study in which abnormal expression of a sequence-specific DNA-binding transcriptional repressor has been shown to contribute directly to brain tumor formation. Our findings indicate that abnormal expression of REST/NRSF and Myc in NSCs causes cerebellum-specific tumors by blocking neuronal differentiation and thus maintaining the "stemness" of these cells. Furthermore, these results suggest that such a mechanism plays a role in the formation of human medulloblastoma.

The Development and Implementation of an Anthropomorphic Head Phantom for the Assessment of Proton Therapy Treatment Procedures

Proton therapy has become an increasingly more common method of radiation therapy, with the dose sparing to distal tissue making it an appealing option, particularly for treatment of brain tumors. This study sought to develop a head phantom for the Radiological Physics Center (RPC), the first to be used for credentialing of institutions wishing to participate in clinical trials involving brain tumor treatment of proton therapy. It was hypothesized that a head phantom could be created for the evaluation of proton therapy treatment procedures (treatment simulation, planning, and delivery) to assure agreement between the measured dose and calculated dose within ±5%/3mm with a reproducibility of ±3%. The relative stopping power (RSP) and Hounsfield Units (HU) were measured for potential phantom materials and a human skull was cast in tissue-equivalent Alderson material (RLSP 1.00, HU 16) with anatomical airways and a cylindrical hole for imaging and dosimetry inserts drilled into the phantom material. Two treatment plans, proton passive scattering and proton spot scanning, were created. Thermoluminescent dosimeters (TLDs) and film were loaded into the phantom dosimetry insert. Each treatment plan was delivered three separate times. Each treatment plan passed our 5%/3mm criteria, with a reproducibility of ±3%. The hypothesis was accepted and the phantom was found to be suitable for remote audits of proton therapy treatment facilities.

Investigation of arterial spin labeling MRI for quantitative cerebral blood flow measurement

Arterial spin labeling (ASL) is a technique for noninvasively measuring cerebral perfusion using magnetic resonance imaging. Clinical applications of ASL include functional activation studies, evaluation of the effect of pharmaceuticals on perfusion, and assessment of cerebrovascular disease, stroke, and brain tumor. The use of ASL in the clinic has been limited by poor image quality when large anatomic coverage is required and the time required for data acquisition and processing. This research sought to address these difficulties by optimizing the ASL acquisition and processing schemes. To improve data acquisition, optimal acquisition parameters were determined through simulations, phantom studies and in vivo measurements. The scan time for ASL data acquisition was limited to fifteen minutes to reduce potential subject motion. A processing scheme was implemented that rapidly produced regional cerebral blood flow (rCBF) maps with minimal user input. To provide a measure of the precision of the rCBF values produced by ASL, bootstrap analysis was performed on a representative data set. The bootstrap analysis of single gray and white matter voxels yielded a coefficient of variation of 6.7% and 29% respectively, implying that the calculated rCBF value is far more precise for gray matter than white matter. Additionally, bootstrap analysis was performed to investigate the sensitivity of the rCBF data to the input parameters and provide a quantitative comparison of several existing perfusion models. This study guided the selection of the optimum perfusion quantification model for further experiments. The optimized ASL acquisition and processing schemes were evaluated with two ASL acquisitions on each of five normal subjects. The gray-to-white matter rCBF ratios for nine of the ten acquisitions were within ±10% of 2.6 and none were statistically different from 2.6, the typical ratio produced by a variety of quantitative perfusion techniques. Overall, this work produced an ASL data acquisition and processing technique for quantitative perfusion and functional activation studies, while revealing the limitations of the technique through bootstrap analysis. ^

Ligand-directed targeting of malignant gliomas

The dramatic poor survival of patients diagnosed with glioblastoma multiforme (GBM) is a reflection of the struggles that accompany traditional treatments. Thus, the development of molecular-based targeted therapies represents new windows for intervention. In this study, we hypothesized that we could select peptide-ligands that selectively target GBM based on the idea that the glioma microenvironment may induce or modify the expression of cell surface receptors that could be accessed by circulating peptides. To select the peptides we employed two distinct in vivo screenings. First, a random phage-displayed peptide library was injected into mice bearing intracranial tumors. Phage that bound to tumor were recovered and sequenced. We found that the tumor-derived phage CLSYKGRC, CNKVSTKC and CQSSREKC were recovered with the highest frequencies and used for subsequent targeting experiments. Second, the phage peptide library was injected into mice without tumors and phage were recovered from brain and sequenced. A phage-displayed peptide (CRTIGPSVC) with homology to transferrin (Tf) was selected and injected into brain tumor-bearing mice. Results showed that after 6 hours of circulation, the CLSYKGRC, CNKVSTKC and CQSSREKC-phage selectively targeted GBM vasculature. In contrast, Tf-like phage accumulated outside the tumor blood vessels in the cytoplasm of cells located within GBM, suggesting it was internalized in vivo. However, after short periods of circulation this phage was restricted to the tumor vasculature. Importantly, none of the selected phage targeted normal brain cells in animals bearing intracranial tumors. An affinity column coupled to the CNKVSTKC zpeptide was used to identify receptors from GBM. Using mass-spectrometry Vimentin, a marker of glial malignancy, was identified as a potential receptor. Other studies showed that the Tf-like phage bound selectively to Apo-Tf (iron free), with no binding to Holo-Tf (iron loaded) or to Tf receptor (TfR). However, the binding of Tf-like phage to glioma cells that express TfR increased in the presence of Apo-Tf. Thus, the Tf-like phage could indirectly target TfR using the endogenous Tf pathway. We propose that the novel peptides identified in this study could be conjugated to therapeutic or imaging agents for use GBM. ^

Micro RNA expression alterations in patients with glioblastoma

Glioblastoma, also known as glioblastoma multiform or GBM, is the most common and most malignant primary brain tumor. The clinical history of patients with glioblastoma is short, usually less than 3 months in more than 50% of cases after diagnosis. Currently, the methods of glioblastoma treatment are chemotherapy, radiotherapy and surgery. Even with the more effective treatment options, patients with glioblastoma most likely have a median survival time of 10 to 12 months. It is necessary to seek other treatment methods, including gene-targeted treatment. The success of gene-targeted treatment depends critically on the knowledge of genes that may be the cause of, or contribute to disease. To establish a correlate between glioblastoma survival timeline and micro RNA expression alteration, a study of 91 glioblastoma patients was conducted at the University of Texas M. D. Anderson Cancer Center. These 91 glioblastoma patients were newly diagnosed from 2002 to 2007. Statistical analysis was conducted to test the association of miRNA expression alteration between long-term survival and short-term survival glioblastoma. The completion of this proposed study will provide a better understanding of the regulatory role of miRNA in glioblastoma progression.^

Immune and genetic risk factors in glioma

Despite extensive research, the etiology of adult glioma remains largely unknown. We sought to further explore the role of immune and genetic factors in glioma etiology using data from the Harris County Brain Tumor Study and the first U.S. genome-wide association study of glioma. First, using a case-control study design, we examined the association between adult glioma risk and surrogates of the timing and frequency of common early childhood infections, birth order and sibship size, respectively. We found that each one-unit increase in birth order was associated with a 12% decreased risk of glioma development in adulthood (OR=0.88, 95% CI=0.81-0.96); however, sibship size was not associated with adult glioma risk (OR=0.96, 95% CI=0.91-1.02). Second, we used a multi-strategic approach to explore the relationships between glioma risk, history of asthma/allergies, and 23 functional SNPs in 11 inflammation genes. We found three inflammation gene SNPs to be significantly associated with glioma risk (COX2/PTGS2 rs20417 [OR=1.41]; IL10 rs1800896 [OR=1.57]; and IL13 rs20541 [OR=0.39]). Joint effects analysis of the risk-conferring alleles of these three SNPs revealed a trend of increasing risk with increasing number of adverse alleles among those without asthma/allergies (p<0.0001). Finally, we conducted a case-only study to explore pairwise SNP-SNP interactions in immune-related pathways among a population of 1304 non-Hispanic white glioma cases. After correction for multiple comparisons, we found 279 significant SNP-SNP interactions among polymorphisms of immune-related genes, many of which have not been previously examined. Our results, cumulatively, suggest an important role for immune and genetic factors in glioma etiology and provide several new hypotheses for future studies.^

Matrix metalloproteinase-1 as a prognosticator in malignant glioma

The overall purpose of this study was to assess the relationship between the promoter region polymorphism (-2607 1G/2G) of matrix metalloproteinase-1 (MMP-1) polymorphism and outcome in brain tumor patients diagnosed with a primary brain tumor between 1994 and 2000 at The University of Texas M. D. Anderson Cancer Center. The MMP-1 polymorphism was genotyped for all brain tumor patients who participated in the Family Brain Tumor Study and for whom blood samples were available. Relevant covariates were abstracted from medical records for all cases from the original protocol, including information on demographics, tumor histology, therapy and outcome was obtained. The hypothesis was that brain tumor patients with the 2G allele have a poorer prognosis and shorter survival than brain tumor patients with the 1G allele. ^ Experimental Design: Genetic variants for the MMP-1 enzyme were determined by a polymerase chain reaction-restriction fragment length polymorphism assay. Comparison was made between the overall survival for cases with the 2G polymorphism and overall survival for cases with the 1G polymorphism using multivariable Cox Proportional-Hazard analysis, controlling for age, sex, Karnofsky Performance Scale (KPS), extent of surgery, tumor histology and treatment received. Kaplan-Meier and Cox Proportional-Hazard analyses were utilized to assess if the MMP-1 polymorphisms were related to overall survival. Results: Overall survival was not statistically significantly different between the 2G allele brain tumor patients and the 1G allele patients and there was no statistically significant difference between tumor types. ^ Conclusions: No association was found between MMP-1 polymorphisms and survival in patients with malignant gliomas. ^