Integration of Genetic and Epigenetic Alterations in the Discovery of Molecular Drivers of Malignancy in Glioma By Ashley A. Smith B.S. and B.A, Roger Williams University, Bristol RI, 02809 A dissertation submitted in partial fulfillment of the requirements for degree of Doctor of Philosophy in the Division of Biology and Medicine at Brown University Providence, Rhode Island May 2014 © Copyright 2014 by Ashley A. Smith This dissertation by Ashley A. Smith is accepted in its present form by the Division of Biology and Medicine as satisfying the dissertation requirements for the degree of Doctor of Philosophy Date ________________ _____________________________________ Karl T. Kelsey, M.D., M.O.H., Advisor Recommended to the Graduate Council Date ________________ _________________________________________ Carmen J. Marsit Ph.D. Reader (Chair) Date ________________ _________________________________________ E. Andrés Houseman, Sc.D., Reader Date ________________ _________________________________________ Yen-Tsung Huang, M.D., M.P.H., S.M., Sc.D. Reader Date ________________ _________________________________________ John K. Wiencke (Outside Reader) Approved by the Graduate Council Date ________________ _________________________________________ Peter M. Weber, Ph.D., Dean of the Graduate School iii CURRICULUM VITAE Ashley A. Smith Department of Pathology and Laboratory Medicine, Box G-E3, Brown University, Providence, RI 02912 Phone: (704) 813-9269; E-mail: Ashley_Smith@brown.edu; Born: June 6, 1984 in Reading PA EDUCATION 2013 Ph.D. (Pathobiology), Brown University, Providence, RI, USA Qualification Exam Project: DNA methylation profiles in brain tumors associate with histology, predict outcome and reflect immune response 2006 B.S. (Biology), Roger Williams University, Bristol, RI, USA B.A. (Chemistry), Roger Williams University, Bristol, RI, USA PUBLICATIONS 1. Smith AA, Huang, YT, Eliot M, Houseman EA, Marsit CJ, Wiencke JK, Kelsey KT. “A novel approach to the discovery of survival biomarkers in glioma using a joint analysis of DNA methylation and gene expression”. Epigenetics. 2013: In Review 2. Christensen BC*, Smith AA*, Zheng S, Koestler DC, Houseman EA, Marsit CJ, Wiemels JL, Nelson HH, Karagas MR, Wrensch MR, Kelsey KT, Wiencke JK. “DNA methylation, isocitrate dehydrogenase mutation, and survival in glioma”. JNCI. 2010: 103(2):143-53. 3. Pietruska J, Liu X, Smith A, McNeil K, Weston P, Zhitkovich A, Hurt R, Kane A, “ Bioavailability, intracellular mobilization of nickel, and HIF-1 activation in human lung epithelial cells exposed to metallic nickel and nickel oxide nanoparticles ”. Toxicol Sci., 2011: 124(1):138-48. * Co-authorship MEETING ABSTRACTS 1. Smith AA, Accomando WP, Wiencke JK, Houseman EA, Marsit CJ, Kelsey KT. “Utility of DMRs that characterize monocytes in non-diseased and diseased brain”. In: Society of NeuroOncology: Proceedings; 2012 Nov. 15-18; Washington, DC, USA. Abstract ME-10. 2. Christensen BC*, Smith AA*, Zheng S, Koestler DC, Houseman EA, Marsit CJ, Wiemels JL, Nelson HH, Karagas MR, Wrensch MR, Kelsey KT, Wiencke JK. “IDH mutation defines methylation class and survival in human glioma”. In: Society of NeuroOncology ; 2010 Nov. 19-21; Montreal, Quebec. Canada. Abstract OM-33 3. Sorger T, Ammon N, Smith A, Ronayne R. “The growth and adhesion of two mouse mesothelial cell lines are related to their rates of tumor formation”. In: American Society for Cell Biology; 2005 Dec. 10:14; San Francisco, CA, USA. Abstract L462  Co-authorship iv INVITED LECTURES 1. Brown University Pathobiology Retreat, “Joint analysis of DNA methylation and gene expression on survival in glioma”. August 2013. Warren, RI, USA 2. Society of NeuroOncology Conference, “IDH mutation defines methylation class and survival in human glioma”. November 2010. Montreal, Quebec, Canada. 3. International Perspectives on Environmental Nanotechnology.”Bioavailability and Toxicity of Nickel in Metallic Nanoparticles”. October 2008. Chicoago, IL, USA. TEACHING AND MENTORING EXPERIENCE Graduate Courses 2012 BIOL 2860 (Molecular Mechanisms of Disease); 1.5 lecture hours on epigenetics and cancer; Brown University Pathobiology Graduate Program, Providence, RI Undergraduate Courses 2010 BIOL 1290 (Cancer Biology); Teachers Assistant; Brown University, Providence, RI, USA 2008 BIOL 2860 (Molecular Mechanisms of Disease); 1.5 lecture hours on Microscopy; Brown University Department of Pathobiology Graduate Program, Providence, RI, USA 2005 BIO 325L (Cell Biology Lab); Teachers Assistant; Roger Williams University, Bristol ,RI, USA Undergraduate Mentoring 2011 Emily Doyle (Intern, Brown University) “CpG Island methylator phenotype associated with Kras and Braf mutations in colon cancer”, Brown University, Providence, RI RESEARCH EXPERIENCE Brown University, Department of Pathobiology, Providence, RI, USA Project: Integration of genetic and epigenetic alterations in the discovery of molecular drives of malignancy in glioma Doctoral Research with Advisor Dr. Karl T. Kelsey, MD, MPH (2008-2013) Brown University, Department of Pathobiology, Providence, RI, USA Project: Bioavailability, intracellular mobilization of nickel, and HIF-1 activation in human lung epithelial cells exposed to metallic nickel and nickel oxide nanoparticles Research Assistance working under Principle Investigator Dr.Agnes Kane (2006-2008) v Roger Williams University, Department of Biology, Bristol, RI, USA Project: The growth and adhesion of two mouse mesothelial cell lines are related to their rates of tumor formation Undergraduate Research with Advisor Dr. Thomas Sorger, PhD (2003-2006) Brown University, Rhode Island Hospital Department of Surgical Research, Bristol, RI, US Project: Designed constructs for containing superior FRET proteins Summer internship with principle investigator Dr. Minosoo Kim, PhD (2005) Roger Williams University, Department of Marine Biology, Bristol, RI, USA Project: Aquaculture Research assistant under supervision of Brad Bourque, MS (2002-2005) CERTIFICATIONS Teaching Certificate I, Sheridan Teaching Center, Brown University AWARDS/HONORS Graduate Student and Post-doctoral Travel Award, Brown University (2012) Rhoda Simper Travel Award, Brown University, Pathobiology Program (2011) Graduated “magna cum laude”, Roger Williams University (2006) Who’s Who among students (Inducted May 2006) Tri-Beta Biological Honor Society (Inducted 2005) Roger Williams University All Academic Team (2003-2006) Roger Williams University Dean’s List (2003-2006) Annual Roger Williams University Achievement Scholarship (2002-2006) Annual Unilever Bestfoods Academic Scholarship (2002-2006) vi Acknowledgements First and foremost, I would like to express my gratitude to my advisor, Dr. Karl T. Kelsey. Without your guidance, mentorship, and constant patience, I would not be the scientist I am today. Thank you for allowing me to explore my own theories even if you did not always agree, and most importantly, thank you for teaching me that it is ok to fail as long as you keep going. I would also like to thank Dr. John K. Weincke, who I consider a second mentor. Your input and guidance is reflected throughout this thesis. Additionally, I would like to thank Dr. Carmen J. Marsit, who has acted as a mentor to me since the time at which I was applying to grad school, who continued to teach me as my professor, and who continues to support me as the chair of my committee. I would also like to thank Dr. Andrés Houseman and Dr. Yen-Tsung Huang, whose statistical advice and constant patience was critical to this thesis. Thank you to all of my collaborators and friends in the laboratory (past and present) particularly, Graham, Haley, Billy, Scott, Rondi (lab mom), and Liz, for the unwavering support (and libation) both inside and outside of lab. In addition, I would like to express a particular thanks to Michele, who has been there for me in every aspect that one person could, I am eternally grateful. I would also like to thank my friends (human and animal) from Windswept Farms, OSCF, and Providence Rugby, for giving me an outlet so I could maintain my sanity. Thank you to all of my friends, family, and extended family who supported me throughout this entire process including: Pop and Nonni for your support not just these past 5 years, but vii throughout my life; Dad, for always helping me find the humor in unfortunate situations; my Sisterface, for constantly smacking the sense back into me #YOLO, and my Mentor and DT, both of whom had to talk me off the ledge countless times but never let me quit. Finally, I would like to thank my mother, Allison A. Smith: you are my one true inspiration and a constant reminder of why I do what I do. ILUITU <3 viii Preface The sum of the work presented in this Ph.D. thesis has been executed by me in collaboration with internal and external investigators, who have been acknowledged appropriately in Chapters 2 and 3. My effort was critical in the planning, execution, analysis, and discussion as presented herein. ix Table of Contents ABSTRACT ……...……….....…………………………………………………………...1 CHAPTER 1: INTRODUCTION ……..……………………………………………….3 THESIS OVERVIEW…....………………………………………………………………..4 GLIOMA: PRESENTATION, DIAGNOSIS, AND TREATMENT……………………..5 GLIOMA: EPIDEMIOLOGY, RISK, AND SURVIVAL………………………………10 GLIOMA: GENETICS…………………………………………………………………..11 GLIOMA: EPIGENETICS………………………………………………………………14 GLIOMA: INTEGRATION OF GENETICS AND EPIGENETICS……………………17 CONCLUSION…………………………………………………………………………..18 REFERENCES…………………………………………………………………………..20 CHAPTER 2: DNA METHYLATION, ISOCITRATIE DEHYDROGENASE MUTATION AND SURVIVAL IN GLIOMA………………………………………..32 CONTEXT AND CAVEATS……………………………………………………………35 ABSTRACT……………………………………………………………………………...37 INTRODUCTION……………………………………………………………………….39 PATIENTS, MATERIALS, AND METHODS……………………………………….....41 STATISTICAL ANALYSIS………………………………………………………….…46 RESULTS……………………………………………………………………………..…49 DISCUSSION……………………………………………………………………………54 REFERENCES…………………………………………………………………………..60 x CHAPTER 3: A NOVEL APPROACH TO THE DISCOVERY OF SURVIVAL BIOMARKERS IN GLIOMA USING A JOINT ANALYSIS OF DNA METHYLATION AND GENE EXPRESSION……………………………………..102 ABSTRACT…………………………………………………………………………….103 INTRODUCTION……………………………………………………………………...106 RESULTS………………………………………………………………………………108 DISCUSSION…………………………………………………………………………..111 MATERIALS AND METHODS……………………………………………………….117 ACKNOWLEDMENTS………………………………………………………………..122 REFERENCES………………………………………………………………………....123 CHAPTER 4: DISCUSSION ………………………………………………………...153 CONCLUSION……………...………………………………………………………….154 FUTURE DIRECTIONS……………………….………………………………………168 REFERENCES………………………………………………………………………....169 xi List of Tables CHAPTER 2 TABLE 1: Patient demographic and tumor characteristics……………………………..69 TABLE 2: Patient age, grade-specific glioma histology, grade, TP53 mutation, and EGFR amplification stratified by IDH mutation status……………………………….…70 TABLE 3: Survival analysis using multivariable Cox proportional hazards model….…71 SUPPLEMENTARY TABLE 1: Primer sequences for quantitative methylation specific polymerase chain reaction (QMSP), IDH mutation, TP53 mutation, and EGFR amplification experiments……………………………………………………………......73 SUPPLEMENTARY TABLE 2: Recursively partitioned mixture model methylation class by glioma histology and predicted methylation class membership for The Cancer Genome Atlas (TCGA) glioblastoma samples…………………………………………..74 SUPPLEMENTARY TABLE 3: Association between GoldenGate array methylation values and quantitative methylation specific polymerase chain reaction (QMSP)………75 SUPPLEMENTARY TABLE 4: Identification numbers (ID) and RPMM methylation class membership for The Cancer Genome Atlas (TCGA) glioblastoma samples used in validation…………………………………………………………………………………76 SUPPLEMENTARY TABLE 5: Cellular pathways enriched among statistically significantly differentially methylated CpG loci in common among glioblastomas, astrocytomas, oligoastrocytomas, and oligodendrogliomas…………………………..…79 SUPPLEMENTARY TABLE 6: Statistically significantly differentially hypomethylated CpG loci in human gliomas……………………………………………………………...80 SUPPLEMENTARY TABLE 7: Statistically significantly differentially hypermethylated CpG loci in human gliomas…………………………………………………………..….85 SUPPLEMENTARY TABLE 8: Cellular pathways enriched among statistically significantly differentially methylated CpG loci in gliomas with an IDH mutation compared to gliomas without IDH mutation…………………………………………….99 SUPPLEMENTARY TABLE 9: Recursively partitioned mixture model (RPMM) methylation class membership and glioma tumor grade and histology………………...100 SUPPLEMENTARY TABLE REFERENCES………………………………………...101 xii CHAPTER 3 TABLE 1: Patient demographic and tumor characteristics…………………………....133 TABLE 2: Final 35 DNA methylation/gene expression pairs that are significantly associated with survival……………………………...…………………………………134 TABLE 3: Functions of significant genes and potential mechanisms in glioma.………135 SUPPLEMENTARY TABLE 1: DNA methylation/ expression pairs that are significantly associated with survival (q-value<0.1)……………….………………………………...151 SUPPLEMENTARY TABLE 2: Expression-based and alternative associations of DNA methylation on gene expression and survival……………………………………152 xiii List of Figures CHAPTER 2 FIGURE 1: Association between glioma histologic subtypes and DNA methylation pattern…………………………………………………………………………………....66 FIGURE 2: Differential methylation and the ratio of hyper- to hypomethylated loci in gliomas ………………………………………………………………………………..…67 FIGURE 3: Association between IDH mutation and methylation phenotype in gliomas …....…………………………………………………………………………..…68 SUPPLEMENTARY FIGURE 1. Association between IDH mutation and increased MGMT methylation……………………………………………………………………....72 CHAPTER 3 FIGURE 1: Significant expression-based and alternative associations of DNA methylation on gene expression and survival…………………………………………..131 FIGURE 2: Model for mediation analysis……………………………………………...132 SUPPLEMENTARY FIGURE 1: Removal of IDH1 mutants…………………………139 SUPPLEMENTARY FIGURE 2: Directionality of significant pairs……………….….140 SUPPLEMENTARY FIGURE 3: Map of methylation loci locations from significant methylation/expression pairs ……………………………………………………….….141 xiv Abstract of Integration of Genetic and Epigenetic Alterations in the Discovery of Molecular Drivers of Malignancy in Glioma, by Ashley A. Smith, Ph.D. Brown University, May 2014 Gliomas are a family of extremely aggressive brain cancers, which, despite current treatment options, have poor prognoses. There are distinct subtypes of gliomas, and accurately identifying these is critical for diagnosis and management. Often, the pathologic diagnosis of these subtypes is difficult, and research is underway to discover novel biomarkers that aid in accurate subtype identification and prognostication. This thesis focuses on the joint analysis of DNA methylation profiles with somatic mutation and gene expression data in glioma, assessing the nature of their association with each other and, subsequently, with histology and disease outcome. The ultimate goal is to develop potential prognostic biomarkers of the disease. DNA methylation was determined for several different grades and histologies of glioma in addition to non-brain-tumor controls. The same samples were sequenced for IDH1/2 mutations. We, and others, discovered an IDH hypermethylator phenotype, showing a tight association between the occurrence of IDH mutation and hypermethylation. This phenotype had a higher prevalence in low-grade and secondary gliomas. Besides mutation, DNA methylation is also associated with other somatic alterations, which can alter gene expression. To better understand how DNA methylation and gene expression drive glioma, we used an integrative bioinformatics approach; our goal was to investigate DNA methylation that modulates gene expression as well as 1 independent DNA methylation (methylation that may exert its phenotypic effects through alternative mechanisms), assessing the nature of their association with disease survival. Our model supports the existing theory that DNA methylation can work through gene expression to influence survival outcome but also suggests that DNA methylation can work alone or through alternative mechanisms to influence glioma outcome. In addition, our approach offers an alternative method of biomarker discovery, which could potentially be used for diagnostic and therapeutic purposes. Overall, this work supports the hypothesis that somatic mutations are not solely responsible for the glioma phenotype. Epigenetics, particularly DNA methylation, is also important in both the genesis and outcome of the disease. Furthermore, our model provides an alternative approach for biomarker discovery that may also be applicable to cancers other than glioma. 2 Chapter 1 Thesis Overview and Introduction 3 Thesis Overview Gliomas are a family of extremely aggressive brain cancers, which, despite currently available treatments, have poor prognoses, with high-grade glioblastoma multiforme (GBM) having a median survival time of 15 months. There exist many individual subtypes of glioma, which are both histologically and molecularly distinct, and accurately identifying these subtypes is critical for diagnosis, prognosis, and treatment. Often, the pathologic diagnosis of these subtypes can be difficult, and research is underway to define novel biomarkers of the disease that can assist in accurate subtype identification. There is an array of somatic alterations that can contribute to tumorigenesis, although it is now recognized that genetic alterations alone cannot explain the phenotypes of all human tumors. Currently, increasing attention is being focused on the potential for epigenetic alterations to drive these tumors. The integration of both epigenetic and genetic alterations is critical to more fully understand tumorigenesis. Using an integrated approach could be particularly valuable for studying cancers with poorly understood etiologies as well as for largely incurable cancers, such as glioma. The aims of this thesis were to focus upon the joint analysis of DNA methylation profiles with mutation and expression data in glioma, assessing their associations with histology and outcome, and evaluating their potential utility as biomarkers of the disease. This thesis begins with a broad introduction to glioma and its histological subtypes, as well as the biology of DNA methylation alterations, gene expression changes, and mutations associated with these phenotypes. Chapter 2 provides details on the 4 integration of glioma DNA methylation and IDH mutation, resulting in the discovery of an IDH-driven hypermethylator phenotype that is associated with the survival outcome of specific glioma subtypes. Chapter 3 describes the results of a two-part bioinformatics- based analysis integrating DNA methylation and gene expression. The first part focuses on methylation-mediated changes in gene expression, which result in differential glioma survival, and the second focuses on DNA methylation mediating survival directly or through mechanisms other than direct changes in gene expression. Additionally, this analysis highlights potential biomarkers of the disease. Finally, Chapter 4 summarizes the conclusions of the previous chapters, discussing the importance of this work and provides potential future directions for this research. Glioma: presentation, diagnosis, and treatment Gliomas are malignant brain tumors thought to arise from glial cells or their precursors 1 and account for almost 80% of all primary malignant brain tumors2. Clinical presentation of the disease includes headaches, seizures, focal neurologic deficits, confusion, memory loss, and personality changes3 . However, many patients, particularly with low-grade glioma, remain asymptomatic4 . Patients suspected of having glioma undergo imaging for initial lesion conformation and grading 3. Though magnetic resonance imaging (MRI) is the gold standard for investigation of suspected glioma, confirmatory diagnosis is still based on stereotactic biopsies4,5 . New imagery methods such as diffusion and perfusion-weighted imaging, proton MR spectroscopy, and 5 susceptibility-weighted imaging provide even more insight into tumor grade and can influence therapeutic decisions5. Upon glioma conformation, a stereotactic biopsy is taken, or if placement is conducive to surgery, tumors are resected and biopsied, with the ladder method being preferable for better histological diagnosis, reduction of symptoms from mass effect, and increased efficacy of therapies6,7. Biopsies are classified based on guidelines set forth by the World Health Organization (WHO), which divides gliomas into several different subtypes and grades1. Subtypes are graded using a I-IV numerical grading system where higher numbers are associated with increased malignancy. Numerical grade is based on the presence or absence of several characteristics, including mitosis, necrosis, nuclear atypia, and endothelial cell proliferation. In addition, tumors are divided into several histological types based on their morphology and predominate cell type. The major histological types include astrocytomas, oligodendrogliomas, mixed oligoastrocytomas, and ependymomas1. Several subtypes can be found within each major type of glioma. The most common subtypes of astrocytic tumors include diffuse and pylocitic astrocytomas. Diffuse astrocytomas (predominately of astrocytic origin), account for almost 80% of adult gliomas and are most frequently found in the cerebral hemispheres 1,8. Diffuse astrocytomas (well-differentiated, anaplastic, and glioblastoma) range from grade II-IV respectively, with glioblastoma multiforme (GBM) being the most malignant of all gliomas. Pilocytic astrocytomas are generally a benign tumor with a WHO grade of I and usually arise in the cerebellum. The second major glioma type, oligodendroglioma (predominantly oligodendrocytic in origin), accounts for 5-15% of gliomas and is usually found in the cerebral hemispheres, specifically the frontal or temporal lobes. 6 Oligodendrogliomas are further divided into well-differentiated (grade II) and anaplastic (grade III) 1,8. In addition, mixed oligoastrocytomas consist of a mix of both astrocytes and oligodendrocytes with both well-differentiated (grade II) and anaplastic (grade III) histologies1. Finally, in adults, ependymomas (predominantly of ependymal origin) are most commonly found in the spinal cord8. Ependymal tumors consist of 4 different subtypes subependymoma, myxopapillary, well-differentiated, and anaplastic, ranging from grade I-III1. Due to the heterogeneity of each of the individual subtypes and varying locations of each, glioma management and treatment can vary accordingly. The general treatment scheme for glioma consists of resection (if applicable), radiation, and/or chemotherapy4,9. Due to the location and infiltrative nature of gliomas, many cannot be resected completely or remain inoperable, and tumor resection is closely associated with patient survival 9. However, advances in surgical techniques have enhanced the ability of surgeons to preform more complete glioma resection 10. Preoperative techniques such as MRI can work together with intraoperative techniques such as neuronavigation to aid in determining the borders of the brain lesion 10. This technique is particularly helpful in locating small deep-seated lesions with an accuracy of about 2 mm11 . Fluorescence-guided resection is another intraoperative imaging technique where fluorescence is used to contrast normal vs. tumor tissue, allowing for more accurate and complete resection10. Techniques such as functional MRI (fMRI) aid in the visualization of active parts of the brain and can be beneficial in obtaining a gross impression of the lesion preoperatively10 . Additional techniques include CT, 3D planning, fiber tracking, and transcranial magnetic stimulation10. If the nature or placement of the tumor does not allow for resection, then a stereotactic biopsy is taken for diagnostic 7 purposes3. Immediately after surgery/biopsy, the main course of treatment is radiotherapy3. Radiotherapy is used for both low- (WHO grade II) and high-grade (WHO grade III, IV) gliomas, typically at a maximum dose of 60 Gy, as higher doses have not been associated with improved outcome and can lead to increased toxicity 4,9 . In addition to radiotherapy, chemotherapy may be used, mostly for high-grade tumors3,9 , as it is controversial whether chemotherapy should be offered to low-grade glioma patients before treatment with radiotherapy4. Concomitant and adjuvant temozolomide (TMZ) is the most commonly used chemotherapeutic drug for glioma treatment with advantages including oral dosing, ability to cross the blood brain barrier (BBB), preferable toxicity profile compared to other drugs, increased effectiveness, and improved quality-of-life 6,12. Other chemotherapeutics include carmustine wafers (Gliadel) and PCV (combination of Procarbazine, CCNU, and Vincristine) 3,4,9 . Depending on the tumor grade and type and patient age, a combination of both radiotherapy and chemotherapy is often used 3,4,9. Additionally, increased knowledge of the pathogenesis of glioma has spurred discussion and trials for targeted molecular-based13, epigenetic-based14 and antiangiogenic-based12,15 therapies. Unfortunately, the initial brain lesion is not the only concern for treatment. Another major issue with glioma patients is the management of comorbidities associated with the primary tumor. These conditions include seizures, peritumoral edema, venous thromboembolism, cognitive dysfunction, and fatigue 16. Seizures are a common symptom of glioma, with approximately 20-62% of patients experiencing tumor-related epilepsy during the course of their disease 16. General treatment for seizures includes a variety of antiepileptic drugs. Unfortunately, antiepileptic drugs can have unwanted interactions 8 with other glioma-related treatments including induction of the cytochrome P-450 system (as seen with the drug phenytoin), which increases the metabolism of many chemotherapeutic agents. For this reason, antiepileptic drugs that do not induce these enzymes (such as clonazepam) are preferred 16. Edema is another side effect of the tumor and if not controlled can lead to serious complications and morbidity. Excess fluid build- up is caused by a disruption in the blood-brain barrier, allowing fluid into the extracellular space of the brain parenchyma. Corticosteroids are usually used to manage peritumoral edema by decreasing endothelial permeability. Unfortunately, there are several complications associated with corticosteroids, including gastrointestinal problems, steroid myopathy, and osteoporosis. Using lower doses can reduce side effects, and most subside after treatment has stopped. Venous thromboembolism (VTE) is another complication experienced by glioma patients and can be treated mechanically using elastic compression stockings as well as with anticoagulation therapies such as low molecular weight heparins. Lastly, disruption in cognitive functions and increased fatigue, though not necessarily associated with morbidity, can significantly reduce quality of life in glioma patients. Medications such as methylphenidate have been shown to improve neurobehavioral functioning, reducing fatigue and depression, while increasing cognition16. Finally recurrence of the primary tumor is often seen. Recurrence of low- grade glioma has been associated with increased malignancy due to transformation 17. However, recurrence is more frequent in higher-grade tumors with a median time-to- tumor progression of ~6.9 months18 . Unfortunately, treatment options for recurrent gliomas are limited due to difficulty of resection and drug resistance19. 9 Glioma: epidemiology, risk, and survival During the years 2005-2009 the incidence (age adjusted) of primary brain and central nervous system (CNS) tumors in the United States was approximately 20.6 per 100,000 people, with the average incidence of malignant tumors in adults (20+ years of age) ranging from 5.80-11.70 per 100,000 people2. Of these, gliomas accounted for 29% of all adult tumors and approximately 80% of all adult malignant tumors, with an incidence rate of 6.03 per 100,000 people. GBM and astrocytomas accounted for approximately 76% of all gliomas, with GBM having the highest incidence rate among malignant tumors. Gliomas are most commonly found in patients between the 4th and 6th decades of life, with lower grades often found at the younger end of the age range 4,7. In addition, malignant glioma incidence is statistically significantly higher in males than in females and in caucasians compared to blacks 2. There are few risk factors associated with glioma, with environmental/behavioral risk factors being the most attractive to study, since they are modifiable 20,21. Of these, ionizing radiation is the only known environmental risk factor. However, it has been suggested that non-ionizing radiation could be associated with gliomagenesis. Specifically mentioned is the use of cell phones, which emit low-radiofrequency in close proximity to the head and brain. Though it is possible cell phone use could cause an increase in glioma risk, no substantial evidence for this has been provided21. Allergies and immunologic changes; specifically, reduced immunoglobulin E (IgE) have been inversely associated with glioma risk 22. Genetic risk factors involved in gliomagenesis include single nucleotide polymorphisms (SNPs), which affect detoxification, DNA repair, and cell cycle regulation 3 . 10 Low-grade pilocytic astrocytomas and ependymal tumors have the best prognosis, with an approximate 5-10 year survival rate of 91.4% and 77.6% respectively. Grade II oligodendrogliomas or astrocytomas have a survival range of 5-10 years 3 and; generally, anaplastic oligodendrogliomas (3-5 years) have a better prognosis than anaplastic astrocytomas (2-3 years) 3. The poorest survival among gliomas is associated with GBM, where median survival is only 12-15 months8 , with a 5-year survival of only 4.7%2. However, recent literature has reported on the molecular complexity of these tumors in the hopes of improving survival with better diagnosis and more targeted treatments. Glioma: genetics The variability in the etiology, progression, and histologies of gliomas is in part due to their genetic heterogeneity, which includes somatic mutations, deletion/amplifications, copy number variation (CNV) and insertion of repetitive elements. Somatic mutations, particularly in tumor suppressor genes, were some of the first implicated in gliomagenesis. Over 65% of gliomas, predominantly low-grade and secondary GBMs, contain mutated TP533,13,23. Mutations in the RB1 tumor suppressor gene are observed mainly in high-grade gliomas. Additionally, p53 and RB pathways may be affected by mutations/amplifications in MDM1/2/4/ and CDKN2A/b (INK4A and ARF), as well as CDK4/613,23,24. Dysregulation of many tyrosine kinase-signaling pathways is also present in malignant glioma. For instance, PDGFR overexpression/amplification is ubiquitous among malignant gliomas, and EGFR amplification/overexpression/mutation has become a marker of high-grade glioma and 11 primary GBM, both of which can cause oncogenic dysregulation of PI3K-AKT-mTOR and Ras-MAPK signaling pathways13,23. Also associated with these pathways are mutation/deletion of PTEN, which is the primary negative regulator of the PI3K-AKT- mTOR signaling pathway, and mutations in NF1, which is the primary negative regulator of the Ras-MAPK pathway23,24. Loss of heterozygosity (LOH) of 1p19q is the most prevalent loss among oligodendrogliomas and a predictor of better prognosis25 . Most recently implicated in glioma are alterations in isocitrate dehydrogenase 1/2(IDH1/2) 26 and telomerase reverse transcriptase (TERT) 27 . The metabolic enzyme IDH1/2 is mutated at high prevalence in low-grade gliomas and secondary GBMs26 . Interestingly, patients with IDH1 mutations tend to be younger and have a better survival outcome26,28. Novel mutations in the promoter region of TERT have also been discovered 27; they appear to be mutually exclusive with IDH1 mutations and demonstrate poorer outcome29,30. Additionally, mutations in ATRX (α thalassemia/mental retardation syndrome X-linked) have been observed in GBMs wild-type (WT) for TERT31. ATRX is involved in chromatin remodeling that is active in telomere biology31. Both mutations in TERT and ATRX suggest the importance of telomerase activation in the development of glioma29 . There are several recurrent translocations reported in glioma, including the in-frame gene fusion of fibroblast growth factor receptor1/3 (FGFR1/3) and transforming acidic coiled- coil (TACC) to form FGFR1/3-TACC332 and EGFR fusions with septin 14 (SEPT14) 33. The ladder aids in activation of the STAT3 pathway, whose dysregulation has been associated with glioma infiltration and growth34. Finally, genetic risk factors are also involved in glioma etiology. Extensive genome-wide association studies (GWAS) and candidate-gene studies have found associations between glioma risk and single- 12 nucleotide polymorphisms (SNPs) 35. Of these, GWAS studies are the most consistently replicated, revealing 8 SNPs/near 7 different genes that are significantly associated with glioma risk: TERT, EGFR, CCDC26, CDKN2A, PHLDB1, RTEL1, and TP53 35-39. Integration of these genetic events has allowed for increased understanding of the pathogenesis of glioma and yielded distinct genetic profiles that aid in distinguishing different subtypes for better diagnosis and treatment. Efforts put forth by Godard et al, and Nutt et al have demonstrated that gliomas can be classified based on differential gene expression40,41, and expression-based classes correlated better with survival than histological outcome41. Further investigation revealed that gene expression profiles could be used to further distinguish classes within individual subtypes and aided in the discovery of prognostic markers such as FABP7, whose increased expression is associated with poorer outcome in GBM 42. Further studies used gene-expression signatures to classify gliomas based on their resemblance to different stages of neurogenesis, resulting in three subclasses: proneural, proliferative, and mesenchymal43. These classes were further supported and refined by integrating gene-expression with copy number, and mutation data 24,44 . The integration with other genetic events resulted in the aforementioned proneural and mesenchymal classes with the addition of classical and neuronal classes24,43,44 . Proneural classes are strongly associated with high levels of TP53 mutation, PDGF amplification/mutation, IDH1/2 mutation, younger age, and have a trend toward increased survival. The mesenchymal subtype is defined by high expression of CHI3L1, MET, and NF1 deletion/mutation. High levels of EGFR amplification/mutation define the classical subtype, and there is a clear difference in response to treatment observed between classical and mesenchymal subtypes21,35,36,23 . Finally, tumors of the 13 neural subtype appear to be the most molecularly similar to normal brain, this group also contains the oldest patients44 . Though genetic-based classes have aided in both prognosis and therapeutic intervention, it has become increasingly apparent that genetics alone cannot explain the phenotype of this complex disease, highlighting the need for studies focusing on not only the genome, but also the epigenome. Glioma: epigenetics An epigenetic trait is defined by a heritable, stable change in expression and/or cellular phenotype that does not result from change to the DNA sequence45,46. Epigenetic regulators include histone modifications47, microRNA48,49, and DNA methylation50-52 , and are critical in normal development contributing to the vast array of cellular phenotypes52-54. However, dysregulation of these regulators has been associated with the etiology of many human diseases55. Due to its assay accessibility, DNA methylation has been one of the most widely studied epigenetic events56-59. DNA methylation occurs on cytosines found 5’ to guanines in the DNA sequence (CpG dinucleotides) 52. Maintenance/deposition of methylation is controlled mainly by three DNA methyltransferases (DNMT1, 3A, 3B) using S-adenosyl methionine as the methyl donor53,60. In mammals, approximately 60-80% of CpGs are methylated53. CpG dinucleotides are under-represented in the genome, however, they have been found at higher than expected quantities in gene promoter regions61 , and clusters of them are referred to as CpG islands62 . The placement of CpG islands in promoter regions of genes 14 allows for epigenetic regulation of transcriptional activity through structural changes in associated chromatin53,55,60 . For instance, methylation of a CpG island in the promoter region of a gene can work together with histone modifications causing chromatin condensation and inhibition of transcriptional activity, essentially silencing expression of the gene. CpG shores (CpGs that lie ~2kb away from CpG Islands) have also been implicated in transcriptional activity as well as cell programing63,64 . Furthermore, patterns of DNA methylation can be used to distinguish individual cell types/mixtures and tissues52,65-67, including different regions of non-diseased brain68 . DNA methylation is important in many normal processes besides transcriptional regulation and cell programming, including genomic imprinting, silencing of aberrant repetitive elements, and regulation of transcriptional enhancers and splice site variants52 . Disruption of normal DNA methylation events can cause dysregulation of these processes, which has been associated with adverse health affects including diseases such as cancer 55. One of the first epigenetic changes implicated in human cancer was a general loss of methylation in tumors compared with normal tissue69,70. Hypomethylation is primarily associated with aberrant expression of repetitive elements but can also lead to loss of imprinting and activation of oncogenes69,71,72. Furthermore, hypomethylation can promote deletions and translocations by favoring mitotic recombination73 . Overall, hypomethylation is associated with genomic instability, which can aid in tumor progression71,72. Gene-specific hypermethylation is also observed in cancer and is associated with transcriptional inactivation72-75. Most ubiquitously observed in carcinogenesis is methylation-induced silencing of tumor suppressor genes, which can aid in tumorigenesis by altering many cancer-related pathways74. Patterns of methylation 15 can also be important prognostic and diagnostic tools in cancer. Differentially methylated regions (DMRs) are regions of the genome demonstrating variable methylation and can be used not only to distinguish different cell and tissue types; but also to aid in distinguishing normal and tumor tissue as well as individual cancer subtypes64,76. Genes with differential DNA methylation have become ideal candidates for biomarker selection for both the diagnosis and prognostication of disease while simultaneously highlighting potential therapeutic targets77 . Another reason DNA methylation is so attractive to study is because, unlike genetic alterations, epigenetic alterations are potentially reversible. The reversibility of DNA methylation has been harnessed for therapeutic reasons in myelodysplastic syndromes and myelogenous leukemia, for which the Food and Drug Administration has already approved the use of drugs which prevent re-methylation (i.e. 5-azacytidine and 5-aza-2’-deoxicytidine) 78-80 . Significant advances in the field of epigenetics have led to the discovery of several epigenetically altered genes/pathways in glioma. Genome-wide hypomethylation is seen in approximately 80% of GBMs, and this loss of methylation is correlated with increased proliferation and aberrant transcriptional activity81. The promoter region of putative oncogene MAGEA1 is hypomethylated in GBM and is associated with increased expression of this cancer-testis antigen 81,82. Increased activation of MAGE proteins have been implicated in multiple cancers and are associated with T-cell recognition, p53 inhibition, and response to chemotherapy 81,82. More commonly seen in glioma is locus- specific hypermethylation81,83,84. Promoter hypermethylation has been observed in many cancer-related gene pathways, including DNA repair, cell cycle progression, apoptosis, angiogenesis, and cell growth85-89. Disruption of any of these pathways can ultimately 16 lead to variable effects on survival. One example of this phenomenon is the epigenetic silencing of the DNA repair gene MGMT, which has become a strong predictor of glioma outcome and response to treatment90,91. MGMT normally functions by removing aberrant alkyl groups from the O6 position of guanine90,91 . In cancer treatment, MGMT expression can decrease the therapeutic efficacy of radiation and alkylating agents such as temozolomide by repairing therapy-induced damage to the tumor cells. DNA gene promoter methylation silencing of MGMT is, then, associated with significantly better survival following chemotherapeutic treatments90,91. Promoter methylation of SOCS3 has been implicated in secondary and low-grade gliomagenesis via the STAT3 and MAPK- pathways92,93. Methylation of SOCS3 is significantly associated with poorer survival outcome92,93. These examples demonstrate the impact that the epigenome can have on tumorigenesis as well as its importance for diagnosis and survival outcome and as a biomarker of the disease. Glioma: Integration of genetics and epigenetics The genetic landscape of glioma is fairly well studied; however, its relationship with the glioma epigenome is poorly understood. Previous literature suggests that complex somatic alterations are involved in gliomagenesis that aid not only in distinguishing glioma from other diseases but also in distinguishing different glioma subtypes. These alterations include both genetic events, such as amplifications/deletions and mutations, as well as epigenetic events such as hyper- and hypo-methylation, all of which can dysregulate cancer-related signaling pathways promoting tumorigenesis and 17 modulating outcome. The importance of analyses integrating the cancer genome and epigenome has been observed with the identification of a CpG island methylator phenotype in colorectal cancer94-96. The integration of both methylation profiles and mutation data demonstrated distinct classes of colorectal cancer, with CIMP-high tumors showing extensive promoter methylation and mutations in the BRAF oncogene95. In contrast, a CIMP-low phenotype is associated with promoter methylation of a more limited set of genes, particularly age-related genes, and is also associated with mutation in the KRAS oncogene95. CIMP-negative tumors display rare methylation as well as TP53 mutation. The prognosis associated with these subgroups also varies, with CIMP-high tumors having the best outcome96. In glioma, the link between promoter methylation and gene expression has been established on a single-locus level. However, large-scale integration approaches of methylation patterns and genetic alterations in glioma have not been attempted to date. Conclusion This thesis aims to carefully assess the epidemiology of DNA methylation in glioma. Novel high-throughput DNA methylation arrays (Illumina), which interrogate approximately 1,500 cancer-related CpG loci, were used to identify the epigenetic determinants of methylation in glioma and how they associate with genetic alterations such as mutations. The initial results suggested the correlation of a hypermethylator phenotype and IDH1 mutations with tumor histology and increased prognosis. To further demonstrate the importance of integrative analysis in gliomagenesis and improved 18 prognosis, data obtained from The Cancer Genome Atlas (TCGA) were used to determine the joint effect of DNA methylation and gene expression on survival outcome in glioma using a novel bioinformatics-based approach. 19 References 1. Louis DN, Ohgaki H, Wiestler OD, et al. The 2007 WHO classification of tumours of the central nervous system. Acta Neuropathol. Aug 2007;114(2):97-109. 2. Dolecek TA, Propp JM, Stroup NE, Kruchko C. CBTRUS statistical report: primary brain and central nervous system tumors diagnosed in the United States in 2005-2009. Neuro Oncol. Nov 2012;14 Suppl 5:v1-49. 3. Wen PY, Kesari S. Malignant gliomas in adults. N Engl J Med. Jul 2008;359(5):492-507. 4. Sanai N, Chang S, Berger MS. Low-grade gliomas in adults. J Neurosurg. Nov 2011;115(5):948-965. 5. Mohammed W, Xunning H, Haibin S, Jingzhi M. Clinical applications of susceptibility-weighted imaging in detecting and grading intracranial gliomas: a review. Cancer Imaging. 2013;13:186-195. 6. Pouratian N, Schiff D. Management of low-grade glioma. Curr Neurol Neurosci Rep. May 2010;10(3):224-231. 7. Wang Y, Jiang T. Understanding high grade glioma: molecular mechanism, therapy and comprehensive management. Cancer Lett. May 2013;331(2):139-146. 8. V. K, A.K. A, J.C. A. Robbins Basic Pathology. 9 ed. Canada: Elsevier Saunders; 2013. 9. Stupp R, Tonn JC, Brada M, Pentheroudakis G, Group EGW. High-grade malignant glioma: ESMO Clinical Practice Guidelines for diagnosis, treatment 20 and follow-up. Annals of oncology : official journal of the European Society for Medical Oncology / ESMO. May 2010;21 Suppl 5:v190-193. 10. Vranic A. New developments in surgery of malignant gliomas. Radiol Oncol. Sep 2011;45(3):159-165. 11. Ganslandt O, Behari S, Gralla J, Fahlbusch R, Nimsky C. Neuronavigation: concept, techniques and applications. Neurol India. Sep 2002;50(3):244-255. 12. Bradley D, Rees J. Updates in the management of high-grade glioma. J Neurol. Jul 2013. 13. Masui K, Cloughesy TF, Mischel PS. Review: molecular pathology in adult high-grade gliomas: from molecular diagnostics to target therapies. Neuropathol Appl Neurobiol. Jun 2012;38(3):271-291. 14. Clarke J, Penas C, Pastori C, et al. Epigenetic pathways and glioblastoma treatment. Epigenetics. Jun 2013;8(8). 15. Plate KH, Scholz A, Dumont DJ. Tumor angiogenesis and anti-angiogenic therapy in malignant gliomas revisited. Acta Neuropathol. Dec 2012;124(6):763-775. 16. Wen PY, Schiff D, Kesari S, Drappatz J, Gigas DC, Doherty L. Medical management of patients with brain tumors. Journal of neuro-oncology. Dec 2006;80(3):313-332. 17. Jaeckle KA, Decker PA, Ballman KV, et al. Transformation of low grade glioma and correlation with outcome: an NCCTG database analysis. Journal of neuro- oncology. Aug 2011;104(1):253-259. 21 18. Stupp R, Mason W, van den Bent M, et al. Radiotherapy plus concomitant and adjuvant temozolomide for glioblastoma. N Engl J Med. Mar 2005;352(10):987-996. 19. Haar CP, Hebbar P, Wallace GC, et al. Drug resistance in glioblastoma: a mini review. Neurochem Res. Jun 2012;37(6):1192-1200. 20. Fisher JL, Schwartzbaum JA, Wrensch M, Wiemels JL. Epidemiology of brain tumors. Neurol Clin. Nov 2007;25(4):867-890, vii. 21. Bondy ML, Scheurer ME, Malmer B, et al. Brain tumor epidemiology: consensus from the Brain Tumor Epidemiology Consortium. Cancer. Oct 2008;113(7 Suppl):1953-1968. 22. Wiemels JL, Wiencke JK, Patoka J, et al. Reduced immunoglobulin E and allergy among adults with glioma compared with controls. Cancer Res. Nov 2004;64(22):8468-8473. 23. Huse J, Holland E. Targeting brain cancer: advances in the molecular pathology of malignant glioma and medulloblastoma. Nat Rev Cancer. May 2010;10(5):319-331. 24. Brennan CW, Verhaak RG, McKenna A, et al. The Somatic Genomic Landscape of Glioblastoma. Cell. Oct 2013;155(2):462-477. 25. Aldape K, Burger PC, Perry A. Clinicopathologic aspects of 1p/19q loss and the diagnosis of oligodendroglioma. Arch Pathol Lab Med. Feb 2007;131(2):242-251. 26. Parsons DW, Jones S, Zhang X, et al. An integrated genomic analysis of human glioblastoma multiforme. Science. Sep 2008;321(5897):1807-1812. 22 27. Killela PJ, Reitman ZJ, Jiao Y, et al. TERT promoter mutations occur frequently in gliomas and a subset of tumors derived from cells with low rates of self- renewal. Proceedings of the National Academy of Sciences of the United States of America. Apr 2013;110(15):6021-6026. 28. Hartmann C, Meyer J, Balss J, et al. Type and frequency of IDH1 and IDH2 mutations are related to astrocytic and oligodendroglial differentiation and age: a study of 1,010 diffuse gliomas. Acta Neuropathol. Oct 2009;118(4):469-474. 29. Arita H, Narita Y, Fukushima S, et al. Upregulating mutations in the TERT promoter commonly occur in adult malignant gliomas and are strongly associated with total 1p19q loss. Acta Neuropathol. Aug 2013;126(2):267- 276. 30. Nonoguchi N, Ohta T, Oh JE, Kim YH, Kleihues P, Ohgaki H. TERT promoter mutations in primary and secondary glioblastomas. Acta Neuropathol. Aug 2013. 31. Kannan K, Inagaki A, Silber J, et al. Whole-exome sequencing identifies ATRX mutation as a key molecular determinant in lower-grade glioma. Oncotarget. Oct 2012;3(10):1194-1203. 32. Singh D, Chan JM, Zoppoli P, et al. Transforming fusions of FGFR and TACC genes in human glioblastoma. Science. Sep 2012;337(6099):1231-1235. 33. Frattini V, Trifonov V, Chan JM, et al. The integrated landscape of driver genomic alterations in glioblastoma. Nat Genet. Aug 2013. 23 34. Luwor RB, Stylli SS, Kaye AH. The role of Stat3 in glioblastoma multiforme. J Clin Neurosci. Jul 2013;20(7):907-911. 35. Walsh KM, Anderson E, Hansen HM, et al. Analysis of 60 reported glioma risk SNPs replicates published GWAS findings but fails to replicate associations from published candidate-gene studies. Genet Epidemiol. Feb 2013;37(2):222-228. 36. Sanson M, Hosking FJ, Shete S, et al. Chromosome 7p11.2 (EGFR) variation influences glioma risk. Human molecular genetics. Jul 2011;20(14):2897- 2904. 37. Shete S, Hosking FJ, Robertson LB, et al. Genome-wide association study identifies five susceptibility loci for glioma. Nat Genet. Aug 2009;41(8):899- 904. 38. Stacey SN, Sulem P, Jonasdottir A, et al. A germline variant in the TP53 polyadenylation signal confers cancer susceptibility. Nat Genet. Nov 2011;43(11):1098-1103. 39. Wrensch M, Jenkins RB, Chang JS, et al. Variants in the CDKN2B and RTEL1 regions are associated with high-grade glioma susceptibility. Nat Genet. Aug 2009;41(8):905-908. 40. Godard S, Getz G, Delorenzi M, et al. Classification of human astrocytic gliomas on the basis of gene expression: a correlated group of genes with angiogenic activity emerges as a strong predictor of subtypes. Cancer Res. Oct 2003;63(20):6613-6625. 24 41. Nutt CL, Mani DR, Betensky RA, et al. Gene expression-based classification of malignant gliomas correlates better with survival than histological classification. Cancer Res. Apr 2003;63(7):1602-1607. 42. Liang Y, Diehn M, Watson N, et al. Gene expression profiling reveals molecularly and clinically distinct subtypes of glioblastoma multiforme. Proceedings of the National Academy of Sciences of the United States of America. Apr 2005;102(16):5814-5819. 43. 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. Mar 2006;9(3):157-173. 44. Verhaak RG, Hoadley KA, Purdom E, et al. Integrated genomic analysis identifies clinically relevant subtypes of glioblastoma characterized by abnormalities in PDGFRA, IDH1, EGFR, and NF1. Cancer Cell. Jan 2010;17(1):98-110. 45. Berger SL, Kouzarides T, Shiekhattar R, Shilatifard A. An operational definition of epigenetics. Genes Dev. Apr 2009;23(7):781-783. 46. Bird A. Perceptions of epigenetics. Nature. May 2007;447(7143):396-398. 47. Turner BM. Cellular memory and the histone code. Cell. Nov 2002;111(3):285-291. 48. Chuang JC, Jones PA. Epigenetics and microRNAs. Pediatr Res. May 2007;61(5 Pt 2):24R-29R. 49. Costa FF. Non-coding RNAs, epigenetics and complexity. Gene. Feb 2008;410(1):9-17. 25 50. Riggs AD. X inactivation, differentiation, and DNA methylation. Cytogenet Cell Genet. 1975;14(1):9-25. 51. Holliday R. The inheritance of epigenetic defects. Science. Oct 1987;238(4824):163-170. 52. Cedar H, Bergman Y. Programming of DNA methylation patterns. Annu Rev Biochem. 2012;81:97-117. 53. Smith ZD, Meissner A. DNA methylation: roles in mammalian development. Nat Rev Genet. Mar 2013;14(3):204-220. 54. Weinhold B. Epigenetics: the science of change. Environ Health Perspect. Mar 2006;114(3):A160-167. 55. Portela A, Esteller M. Epigenetic modifications and human disease. Nat Biotechnol. Oct 2010;28(10):1057-1068. 56. Bibikova M, Barnes B, Tsan C, et al. High density DNA methylation array with single CpG site resolution. Genomics. Oct 2011;98(4):288-295. 57. Bibikova M, Le J, Barnes B, et al. Genome-wide DNA methylation profiling using Infinium® assay. Epigenomics. Oct 2009;1(1):177-200. 58. Herman JG, Graff JR, Myöhänen S, Nelkin BD, Baylin SB. Methylation-specific PCR: a novel PCR assay for methylation status of CpG islands. Proceedings of the National Academy of Sciences of the United States of America. Sep 1996;93(18):9821-9826. 59. Tost J, Gut IG. DNA methylation analysis by pyrosequencing. Nat Protoc. 2007;2(9):2265-2275. 26 60. Jaenisch R, Bird A. Epigenetic regulation of gene expression: how the genome integrates intrinsic and environmental signals. Nat Genet. Mar 2003;33 Suppl:245-254. 61. Bird A. DNA methylation patterns and epigenetic memory. Genes Dev. Jan 2002;16(1):6-21. 62. Bird AP. CpG-rich islands and the function of DNA methylation. Nature. 1986 May 15-21 1986;321(6067):209-213. 63. Aran D, Sabato S, Hellman A. DNA methylation of distal regulatory sites characterizes dysregulation of cancer genes. Genome Biol. Mar 2013;14(3):R21. 64. Doi A, Park IH, Wen B, et al. Differential methylation of tissue- and cancer- specific CpG island shores distinguishes human induced pluripotent stem cells, embryonic stem cells and fibroblasts. Nat Genet. Dec 2009;41(12):1350-1353. 65. Ehrlich M, Gama-Sosa MA, Huang LH, et al. Amount and distribution of 5- methylcytosine in human DNA from different types of tissues of cells. Nucleic acids research. Apr 1982;10(8):2709-2721. 66. Rakyan VK, Down TA, Thorne NP, et al. An integrated resource for genome- wide identification and analysis of human tissue-specific differentially methylated regions (tDMRs). Genome Res. Sep 2008;18(9):1518-1529. 67. Houseman EA, Accomando WP, Koestler DC, et al. DNA methylation arrays as surrogate measures of cell mixture distribution. BMC Bioinformatics. 2012;13:86. 27 68. Ladd-Acosta C, Pevsner J, Sabunciyan S, et al. DNA methylation signatures within the human brain. Am J Hum Genet. Dec 2007;81(6):1304-1315. 69. Feinberg AP, Vogelstein B. Hypomethylation of ras oncogenes in primary human cancers. Biochemical and biophysical research communications. Feb 1983;111(1):47-54. 70. Gama-Sosa MA, Slagel VA, Trewyn RW, et al. The 5-methylcytosine content of DNA from human tumors. Nucleic acids research. Oct 1983;11(19):6883- 6894. 71. Ehrlich M. DNA hypomethylation in cancer cells. Epigenomics. Dec 2009;1(2):239-259. 72. Baylin SB, Ohm JE. Epigenetic gene silencing in cancer - a mechanism for early oncogenic pathway addiction? Nat Rev Cancer. Feb 2006;6(2):107-116. 73. Esteller M. Epigenetics in cancer. N Engl J Med. Mar 2008;358(11):1148-1159. 74. Esteller M. CpG island hypermethylation and tumor suppressor genes: a booming present, a brighter future. Oncogene. Aug 2002;21(35):5427-5440. 75. Jones PA, Baylin SB. The fundamental role of epigenetic events in cancer. Nat Rev Genet. Jun 2002;3(6):415-428. 76. Rakyan VK, Down TA, Balding DJ, Beck S. Epigenome-wide association studies for common human diseases. Nat Rev Genet. Aug 2011;12(8):529-541. 77. Mulero-Navarro S, Esteller M. Epigenetic biomarkers for human cancer: the time is now. Crit Rev Oncol Hematol. Oct 2008;68(1):1-11. 78. Issa JP, Gharibyan V, Cortes J, et al. Phase II study of low-dose decitabine in patients with chronic myelogenous leukemia resistant to imatinib mesylate. 28 Journal of clinical oncology : official journal of the American Society of Clinical Oncology. Jun 2005;23(17):3948-3956. 79. Kaminskas E, Farrell A, Abraham S, et al. Approval summary: azacitidine for treatment of myelodysplastic syndrome subtypes. Clinical cancer research : an official journal of the American Association for Cancer Research. May 2005;11(10):3604-3608. 80. Garcia-Manero G, Kantarjian HM, Sanchez-Gonzalez B, et al. Phase 1/2 study of the combination of 5-aza-2'-deoxycytidine with valproic acid in patients with leukemia. Blood. Nov 2006;108(10):3271-3279. 81. Nagarajan RP, Costello JF. Epigenetic mechanisms in glioblastoma multiforme. Semin Cancer Biol. Jun 2009;19(3):188-197. 82. Cadieux B, Ching TT, VandenBerg SR, Costello JF. Genome-wide hypomethylation in human glioblastomas associated with specific copy number alteration, methylenetetrahydrofolate reductase allele status, and increased proliferation. Cancer Res. Sep 2006;66(17):8469-8476. 83. Natsume A, Kondo Y, Ito M, Motomura K, Wakabayashi T, Yoshida J. Epigenetic aberrations and therapeutic implications in gliomas. Cancer Sci. Jun 2010;101(6):1331-1336. 84. Martinez R, Esteller M. The DNA methylome of glioblastoma multiforme. Neurobiol Dis. Jul 2010;39(1):40-46. 85. Amatya VJ, Naumann U, Weller M, Ohgaki H. TP53 promoter methylation in human gliomas. Acta Neuropathol. Aug 2005;110(2):178-184. 29 86. Baeza N, Weller M, Yonekawa Y, Kleihues P, Ohgaki H. PTEN methylation and expression in glioblastomas. Acta Neuropathol. Nov 2003;106(5):479-485. 87. Bello MJ, Rey JA. The p53/Mdm2/p14ARF cell cycle control pathway genes may be inactivated by genetic and epigenetic mechanisms in gliomas. Cancer genetics and cytogenetics. Jan 2006;164(2):172-173. 88. Costello JF, Berger MS, Huang HS, Cavenee WK. Silencing of p16/CDKN2 expression in human gliomas by methylation and chromatin condensation. Cancer Res. May 1996;56(10):2405-2410. 89. Watanabe T, Yokoo H, Yokoo M, Yonekawa Y, Kleihues P, Ohgaki H. Concurrent inactivation of RB1 and TP53 pathways in anaplastic oligodendrogliomas. J Neuropathol Exp Neurol. Dec 2001;60(12):1181-1189. 90. Hegi ME, Diserens AC, Gorlia T, et al. MGMT gene silencing and benefit from temozolomide in glioblastoma. N Engl J Med. Mar 2005;352(10):997-1003. 91. Esteller M, Garcia-Foncillas J, Andion E, et al. Inactivation of the DNA-repair gene MGMT and the clinical response of gliomas to alkylating agents. N Engl J Med. Nov 2000;343(19):1350-1354. 92. Martini M, Pallini R, Luongo G, Cenci T, Lucantoni C, Larocca LM. Prognostic relevance of SOCS3 hypermethylation in patients with glioblastoma multiforme. International journal of cancer. Journal international du cancer. Dec 2008;123(12):2955-2960. 93. Lindemann C, Hackmann O, Delic S, Schmidt N, Reifenberger G, Riemenschneider MJ. SOCS3 promoter methylation is mutually exclusive to 30 EGFR amplification in gliomas and promotes glioma cell invasion through STAT3 and FAK activation. Acta Neuropathol. Aug 2011;122(2):241-251. 94. Issa J. CpG island methylator phenotype in cancer. Nat Rev Cancer. Dec 2004;4(12):988-993. 95. Ogino S, Kawasaki T, Kirkner G, Loda M, Fuchs C. CpG island methylator phenotype-low (CIMP-low) in colorectal cancer: possible associations with male sex and KRAS mutations. J Mol Diagn. Nov 2006;8(5):582-588. 96. Shen L, Toyota M, Kondo Y, et al. Integrated genetic and epigenetic analysis identifies three different subclasses of colon cancer. Proceedings of the National Academy of Sciences of the United States of America. Nov 2007;104(47):18654-18659. 31 Chapter 2 DNA Methylation, Isocitrate Dehydrogenase Mutation, and Survival in Glioma Brock C. Christensen, Ashley A. Smith, Shichun Zheng, Devin C. Koestler, E. Andres Houseman, Carmen J. Marsit, Joseph L. Wiemels, Heather H. Nelson, Margaret R. Karagas, Margaret R. Wrensch, Karl T. Kelsey, John K. Wiencke BCC and AAS contributed equally to the work. MRW, KTK, and JKW are joint lead investigators. Journal of National Cancer Institute; January 19, 2011; 103:143-153 32 DNA Methylation, Isocitrate Dehydrogenase Mutation, and Survival in Glioma Brock C. Christensen, Ashley A. Smith, Shichun Zheng, Devin C. Koestler, E. Andres Houseman, Carmen J. Marsit, Joseph L. Wiemels, Heather H. Nelson, Margaret R. Karagas, Margaret R. Wrensch, Karl T. Kelsey, John K. Wiencke Affiliations of authors: Department of Pathology and Laboratory Medicine (BCC, AAS, CJM, KTK); Department of Community Health, Brown University, Providence, RI 02912, USA (BCC, DCK, EAH, KTK); Department of Neurological Surgery, Helen Diller Family Cancer Center (SZ, MRW, JKW); Department of Epidemiology and Biostatistics, University of California San Francisco, San Francisco, CA (JLW); Department of Biostatistics, Harvard School of Public Health, Boston, MA (EAH); Masonic Cancer Center, Division of Epidemiology and Community Health, University of Minnesota, Minneapolis, MN (HHN); Section of Biostatistics and Epidemiology, Department of Community and Family Medicine, Dartmouth Medical School, Lebanon, NH (MRK). BCC and AAS contributed equally to the work. MRW, KTK, and JKW are joint lead investigators. Correspondence to: John K. Wiencke, PhD, Department of Neurological Surgery, Helen Diller Family Cancer Center, University of California San Francisco, San Francisco, CA 91458 (e-mail: John.Wiencke@UCSF.edu) Funding 33 This study was funded by the National Institute of Health, grant numbers R01CA52689 (to MRW) and P50CA097257 (to MRW and JKW); R01CA078609, R01CA121147, R01CA126939, and R01CA100679 (to KTK); R01ES06717 and R01CA126831 (to JKW); P30CA077598 (to HHN); and Tobacco-Related Diseases Research Program, grant number 18CA-0127 (to JLW) Notes The funders did not have any role in the study design, collection of data, interpretation of the results, preparation of the manuscript, or the decision to submit the manuscript for publication. 34 Context and Caveats Prior knowledge Human gliomas often have mutations in the isocitrate dehydrogenase genes (IDH1 and IDH2). IDH mutation is associated with improved survival in glioma patients. Epigenetic alterations like DNA methylation at CpG dinucleotides play an important role in gene regulation. Integration of genetic and epigenetic data is important for a better understanding of glioma development. Study design DNA methylation profile of CpG loci and methylation class of 131 glioma and seven non-glioma brain tissues were determined. The association between IDH mutation and methylation class was analyzed. Survival analysis of patients carrying IDH mutation vs. wild-type IDH was also performed. Contribution CpG loci showed differential methylation between glioma and non-glioma tissues. Statistically significant associations were found between DNA methylation class and histologic subtypes, and between DNA methylation class and IDH mutation of gliomas. Patients carrying IDH mutation in gliomas showed improved survival compared with patients carrying IDH wild-type after adjustment for age and grade-specific tumor histology. Implications A distinct methylation pattern in glioma tissues is associated with IDH mutation. 35 Limitations Mutation data was not available for all tissue samples, which may have limited the statistical power of the analyses. 36 Abstract Background: Although much is known about molecular and chromosomal characteristics that distinguish glioma histologic subtypes, DNA methylation patterns of gliomas and their association with other tumor features such as mutation of isocitrate dehydrogenase (IDH) genes, has only recently begun to be investigated. Methods: DNA methylation of glioblastomas, astrocytomas, oligodendrogliomas, oligoastrocytomas, ependymomas, and pilocytic astrocytomas (n = 131) from the Brain Tumor Research Center at the University of California San Francisco, as well as non- tumor brain tissues (n = 7), was assessed with the Illumina GoldenGate methylation array. Methylation data were subjected to recursively partitioned mixture modeling (RPMM) to derive methylation classes. Differential DNA methylation between tumor and non-tumor was also assessed. The association between methylation class and IDH mutation (IDH1 and IDH2 isoforms) was tested using univariate and multivariable analysis for tumors with available substrate for sequencing (n = 95). Survival of glioma patients carrying mutant IDH (n = 56) was compared with patients carrying wild-type IDH (n = 39) by using a multivariable Cox proportional hazards model and Kaplan-Meier analysis. All statistical tests were two-sided. Results: We observed a statistically significant association between RPMM methylation class and glioma histologic subtype (P < 2.2  10-16). Compared with non-tumor brain tissues, across glioma tumor histologic subtypes, the differential methylation ratios of CpG loci were statistically significantly different (Permutation P < .0001). Methylation class was strongly associated with IDH mutation in gliomas (P = 3.0  10-16 ). Compared with glioma patients whose tumors harbored wild-type IDH, patients whose tumors 37 harbored mutant IDH showed statistically significantly improved survival (HR of death = 0.27, 95% confidence interval [CI] = 0.10 to 0.72). Conclusion: The homogeneity of methylation classes for gliomas with IDH mutation, despite their histologic diversity, suggests that IDH mutation is associated with a distinct DNA methylation phenotype and an altered metabolic profile in glioma. 38 Introduction Malignant glioma is the most common form of primary malignant brain tumor and the glioma histologic subtypes include glioblastomas, grades 2 and 3 astrocytomas, grades 2 and 3 oligodendrogliomas, grades 2 and 3 oligoastrocytomas, ependymomas, and pilocytic astrocytomas (1). Presently, there are limited treatment options for glioma; glioblastoma, the most common glioma subtype, remains an incurable disease with a median survival of 15 months, even with radiation and temozolomide therapy (2). A comprehensive appreciation of the integrated genomics and epigenomics of glioma is needed to better understand the multiple cellular pathways involved in their development, establish markers of resistance to traditional therapies, and contribute to the development of targeted therapies. Epigenetic alterations can alter gene expression, gene expression potential, or the regulation of gene function, and thereby contribute to gliomagenesis. Arguably, the most widely studied epigenetic mark is DNA methylation that occurs at cytosine residues in the context of CpG dinucleotides. Approximately half of human genes have concentrations of CpGs in their promoter regions and the methylation state of these and other gene-associated CpGs are widely regarded as critical indicators of gene regulation. Since 2008, sequencing of gliomas has identified mutations in the isocitrate dehydrogenase 1 and 2 (IDH1 and IDH2) genes (3-5). The IDH1 and IDH2 enzymes convert isocitrate to alpha (α)–ketoglutarate producing NADPH and participate in cellular metabolic processes such as glucose sensing, lipid metabolism, and oxidative respiration (reviewed in [6]). Mutations in IDH1 are consistently found in codon 132 for arginine (R132), and mutations in IDH2 consistently occur at the analogous amino acid R172 (3, 39 7). Mutations in IDH1 and IDH2 (IDH when referring to both) are unlike most cancer- associated enzyme mutations because they confer neomorphic enzyme activity rather than inactivating, or constitutively activating, the enzyme. The mutant form of IDH enzymes convert α-ketoglutarate to 2-hydroxyglutarate in an NADPH-dependent manner, and via an unknown mechanism contribute to the pathophysiology of gliomas and leukemias (5, 7, 8). IDH mutations occur in approximately 80% of grades 2-3 gliomas and secondary glioblastomas, but less than 10% of primary glioblastomas (4, 5). In gliomas, IDH mutation has been associated with genetic alterations in other genes including the tumor suppressors and oncogenes (5). IDH mutation also has been associated with younger age and improved survival in glioma patients (5, 9). The somatic genetic signature of any individual tumor is critical to assessing its clinical and etiologic character. Similarly, the profile of somatic epigenetic alterations is central to forming a complete understanding of the pattern of disrupted cellular functioning responsible for the deadly behavior of gliomas. Major advances in the clinical role of epigenetics in gliomas include the findings that promoter methylation silencing of the O-6-methylguanine-DNA methyltransferase (MGMT) gene is associated with response to temozolomide treatment (10). Epigenetic silencing of MGMT gene is found in approximately 80% of gliomas with mutant IDH1, compared with approximately 60% of gliomas with wild-type IDH1 (9). Other common alterations in gliomas are mutations in tumor protein p53 (TP53) (11) and amplification of the epidermal growth factor receptor (EGFR) oncogene (12). Better definitions of the somatic nature of gliomas should integrate both their genetic and epigenetic alterations. In this study, we assessed CpG methylation patterns, IDH mutation, TP53 mutation, and 40 EGFR amplification in histologically diverse gliomas to define epigenetic subgroups of potential clinical and etiologic relevance. Patients, Materials, and Methods Patients and Tissue Samples Fresh frozen tumor tissues of patients (n = 131) diagnosed with glioma between 1990 and 2003 were obtained from the University of California San Francisco (UCSF) Brain Tumor Research Center Tissue Bank. Tumors were previously reviewed by UCSF neuropathologists to assign histologic subtypes and grades according to the World Health Organization classification for patients operated on at the UCSF Medical Center (1). Tumor samples were defined as secondary glioblastoma if the patients had previous histological diagnosis of a lower-grade glioma. Non-tumor brain tissue samples were obtained from cancer-free patients (n = 7) who underwent temporal lobe resection for treatment of epilepsy at the UCSF Medical Center. Patient ages were documented at the time of initial diagnosis. Other demographic and survival data were obtained from UCSF patient records and the California Cancer Registry. The Institutional Review Board approval certification was obtained from the UCSF Committee on Human Research, and subjects provided written, informed consent for tissue collection. Cell lines, Cell Culture, and Reagents A431 cells (a human epidermoid cancer cell line that is known to have EGFR amplification and overexpression) and HT29 cells (a human colon adenocarcinoma cell line without EGFR amplification) were obtained from American Type Culture Collection 41 (ATCC, Manassas, VA). Cell lines were maintained in Dulbecco's Modified Eagle's Medium and RPMI 1640 medium (both from Invitrogen, Carlsbad, CA), respectively, with 10% fetal bovine serum (Hyclone, Logan UT), at 37ºC in 5% CO 2. When cultures reached 80% confluency, cells were harvested for DNA extraction. DNA Extraction, Bisulfite Modification, and Methylation Analysis Genomic DNA from 131 glioma tissue samples and seven non-tumor brain tissue samples was isolated from approximately 25 mg wet weight of each frozen tissue sample using QIAamp DNA mini kit (Qiagen Inc., Valencia, CA), according to the manufacturer's instructions. DNA was eluted twice in a total of 100 µl of elution buffer. The same DNA extraction method was applied to A431 and HT29 cell lines that served as EGFR amplification controls. For DNA methylation analysis, 1 g of genomic DNA was first subjected to bisulfite modification using the EZ DNA Methylation Kit (Zymo Research Corporation, Orange, CA), according to the manufacturer's instructions. Bisulfite modification converts unmethylated cytosine residues to uracil and preserves methylated cytosine residues as cytosines. GoldenGate DNA methylation bead arrays (Illumina Inc., San Diego, CA) were used to interrogate methylation of 1505 CpG loci associated with 803 cancer-related genes, according to the manufacturer's instructions. GoldenGate methylation arrays were used to analyze bisulfite-modified DNA from 131 glioma and 7 non-tumor samples for methylation, and processed at the UCSF Institute for Human Genetics, Genomics Core Facility. The GoldenGate array methylation data were deposited in the Gene Expression Omnibus and are publicly available (accession GSE20395). The Cancer Genome Atlas 42 (TCGA), a public data portal, was used to obtain GoldenGate methylation array data for validation of methylation classes. Quantitative methylation-specific polymerase chain reaction (PCR) (QMSP) was used to confirm methylation data from the GoldenGate array. Candidate genes were selected based on previous studies (13-16) that reported aberrant methylation in astrocytic glioma and included MGMT, Ras association domain family member 1 (RASSF1), PYD and CARD domain containing (PYCARD), homeobox A9 (HOXA9), paternally expressed 3 (PEG3), and slit homolog 2 (SLIT2). CpGenome Universal Methylated DNA (Millipore, Billerica, MA) was bisulfite modified and used as a positive control for QMSP analysis. QMSP was performed using Applied Biosystems 7900HT Fast Real-Time PCR System (Applied Biosystems, Carlsbad, CA). The reaction plate was prepared using the Beckman Coulter automated liquid handler-Biomex 3000 (Beckman Coulter, Fullerton, CA). Each reaction contained 10.0 µL 2× Power SYBR Green PCR Master Mix (Applied Biosystems), 100-400 nM of forward and reverse primers (Supplementary Table 1, available online) and 25 ng of DNA template in a total reaction volume of 20 µL. For the amplification of RASSF1, 2–3% dimethyl sulfoxide (DMSO) was added to the reaction mix. PCR conditions are modified by different primer concentrations and DMSO was added to ensure that primer dimers and non-specific amplification products were not included in the threshold cycle (Ct) calculation. To confirm specificity of amplicons from QMSP, we performed dissociation curve analysis. The PCR conditions were: 95ºC for 10 minutes, and 40 cycles of 95ºC for 15 seconds, 60ºC for 30 seconds, and 72ºC for 30 seconds. SYBR Green fluorescence data was collected only during the extension reaction at 72ºC. Ct values were calculated by the 7900HT system software, and average relative quantification (RQ) values were obtained 43 for each sample using actin, beta (ACTB) amplification as the referent, where RQ = (target gene / ACTB) / (Universal methylation calibrator / ACTB). Spearman rank correlation coefficients (rho) and P values were calculated to assess the correlation between GoldenGate array data and QMSP results. Mutation analysis IDH mutation. The region spanning R132 codon of IDH1, and the region spanning R172 codon of IDH2 were amplified by PCR with primers designed with Primer 3 sofware (v.0.4.0) with the exception of the forward sequencing primer, which was selected from Balss et al. (4). PCR reaction mixtures contained 10–25 ng DNA, 1 buffer, 0.2 mM dNTP mix, 0.2 µM forward and reverse primers, 0.04 units of HotStarTaq, and 1 mM MgCl2 (Qiagen Inc.), in a 25 L volume. The PCR conditions were: 95ºC for 10 minutes, 40 cycles of 94ºC for 30 seconds, 60ºC for 30 seconds, and 72ºC for 1 minute. The resulting products were analyzed on a 1.5% agarose gel. DNA was purified using the QIAquick® PCR Purification Kit (Qiagen Inc.) and sent to Rhode Island Genomics and Sequencing Center at the University of Rhode Island, where it was sequenced in both directions using the BigDyeTerminator v3.1 Cycle Sequencing Kit (Applied Biosystems, Foster City, CA). Sequences were analyzed with the help of Applied Biosystems Sequence Scanner Software v1.0. All primers for IDH1 mutation analysis are listed in Supplementary Table 1, available online. TP53 mutation. For TP53 mutation analysis, PCR–single-strand conformation polymorphism technique was used, and DNA sequencing was done as previously 44 described (8). Primers for PCR amplification of fragments of exons 5 to 8 of TP53 are listed in Supplementary Table 1, available online. PCR reaction mixtures contained 50 ng DNA, 20 µmol/L dNTP, 10 mmol/L Tris-HCl (pH 9.0), 1.5 mmol/L MgCl2, 0.1% Triton X-100, 10 pmol of forward and reverse primers, 1 unit Taq (Perkin-Elmer Cetus, Norwalk, CT), and 0.2 µCi [33 P] dCTP (DuPont New England Nuclear, Boston, MA), in a 30 µL volume. DNA with TP53 mutation confirmed by sequencing was included as positive control. The PCR reaction was carried out using 35 cycles: 94ºC for 30 seconds, annealed for 30 seconds at 58 ºC for exons 5 and 8, and 60 ºC for exons 6 and 7 (primers listed in Supplementary Table 1, available online) and 72ºC for 1 minute. Three microliters of PCR product were mixed with 2 µL of 0.1 N NaOH and then mixed with 5 µL of gel loading buffer solution (United States Biochemical Corp. Cleveland, OH) and heated at 94ºC for 4 minutes. DNA was analyzed on 6% nondenatured polyacrylamide gel, supplemented with 10% glycerol. Electrophoresis was performed at room temperature for 20 hours and exposed to autoradiography films for 16 hours for detection of bands. Direct sequencing of PCR fragments for both DNA strands was done on all tumor DNAs that showed aberrant migration patterns on single-strand conformation polymorphism gel to determine the corresponding DNA sequences using dsDNA cycle sequencing system (Life Technologies, Gaithersburg, MD), as described in Wiencke et al. (17). EGFR amplification. EGFR amplification was measured by a quantitative PCR method using the ABI 7900 Real-Time PCR system (Applied Biosystems) and the commonly used DNA-binding dye, SYBR Green I, which has been shown to be equivalent to 45 TaqMan PCR assay for the assessment of gene copy number (18). Quality control measures for the real-time SYBR green assay included running triplicate determinations for both EGFR and control gene, glyceraldehyde-3-phosphate dehydrogenase (GAPDH). DNA from A431 and HT29 cell lines, with known copy number states for EGFR, served as positive and negative controls, respectively, for amplification. Statistical Analysis Data assembly. Methylation data were assembled with BeadStudio methylation software from Illumina. All GoldenGate methylation array data points are represented by fluorescent signals (Cy dyes) from both methylated (Cy5) and unmethylated (Cy3) alleles. The methylation level, designated as beta (β) is calculated as β = (max[Cy5, 0])/(|Cy3| + |Cy5| + 100), in which the average β value is derived from the approximately 30 replicate methylation measurements, because each CpG probe set is present on the array and measured in each sample approximately 30 times. Raw average β values were analyzed without normalization as recommended by Illumina. At each CpG locus, for each tissue DNA sample, the detection P value was used to determine sample performance; all samples had detection P values less than 1  105 at more than 75% of CpG loci and passed performance criteria. There were 8 CpG loci that had a median detection P value of greater than .05, and these 8 CpGs were excluded from the analysis. All CpG loci on the X chromosome were excluded from analysis. The final dataset contained 1413 autosomal CpG loci associated with 773 genes. For each CpG locus, the differential methylation values (delta-beta [Δβ]) were calculated by subtracting the average β value 46 of tumors from the mean β value of the seven non-tumor brain samples. Subsequent analyses were carried out using the R software (19). All statistical tests were two-sided. Unsupervised Clustering, Recursively Partitioned Mixture Modeling, and Survival. Hierarchical clustering of the DNA methylation data was performed using the R function hclust with Euclidean distance metric and Ward linkage. To discern and describe the relationships between CpG methylation data and patient and tumor covariates, a modified model-based form of unsupervised clustering known as recursively partitioned mixture modeling (RPMM), was used as described in Houseman et al. (20) and as used in Christensen et al. (21). The analysis of associations between methylation class (categorical) and individual categorical covariates was performed using the Fisher exact test. To test for association between methylation class and continuous covariates, a permutation test was run with the Kruskal-Wallis test statistic, and a likelihood ratio test was used for comparing the association between methylation class and IDH mutation to a model including age and histology. To test for associations between IDH mutation and grade-specific tumor histology, and IDH mutation and tumor grade, Fisher’s exact tests were used. To test for associations between IDH mutation and primary vs. secondary glioblastoma, IDH mutation and TP53 mutation, and IDH mutation and EGFR amplification, Chi-square tests were used. The assumption of proportionality for Cox proportional hazards modeling was verified by calculating Pearson correlation coefficients for the corresponding set of Schoenfeld residuals with a transformation of time based on the Kaplan-Meier estimate of the survival function (22), and graphically by plotting log(survival time) vs. log(-log[survival as a function of time, t]). 47 Locus-by-locus analysis. To examine differential methylation between tumor and non- tumor tissues, gliomas were stratified by grade-specific histologic subtypes, and individual CpG loci were compared between subtypes of glioma and non-tumor samples using a Wilcoxon rank-sum test. Because this results in the simultaneous comparison of all CpG loci between glioma subtypes and non-tumor sample types, false discovery rate estimation and Q-values computed by the qvalue package in R (23) were used to adjust for multiple testing. Differentially methylated CpGs were counted as hyper- or hypomethylated if both the tumor vs. non-tumor Q less than .05 and the median methylation value |Δβ| greater than 0.2. An equivalent approach was used in the analysis of differential methylation for gliomas with mutant or wild-type IDH, compared with non-tumor tissues. Pathway Analysis. A canonical pathway analysis was conducted with the use of Ingenuity Pathway Analysis software (Ingenuity Systems, Redwood City, CA). CpG gene-loci associated with the Illumina GoldenGate methylation array were used as reference and loci from differential methylation analysis, as described later in the article, were compared. The statistical significance of gene-locus enrichment within canonical pathways was measured with a Fisher’s exact test. 48 Results Unsupervised Clustering and Modeling of Glioma and Non-Tumor DNA Methylation Data Histological grade and patient demographic data for the 131 gliomas and patient demographic data for the seven non-tumor brain tissues are presented in Table 1. To characterize DNA methylation of gliomas and non-tumor brain tissues, the bisulfite- modified DNA samples were hybridized to the GoldenGate DNA methylation array. Unsupervised clustering of DNA methylation data from 1413 autosomal CpG loci showed that non-tumor brain tissues cluster with each other and are distinct from tumor tissues (Figure 1, A). Furthermore, we observed that oligodendrogliomas and astrocytomas generally clustered together and demonstrated a greater number of methylated loci relative to ependymomas, pilocytic astrocytomas, as well as non-tumor brain tissues. Concomitantly, glioblastomas (also known as grade IV astrocytoma), predominantly clustered together at the bottom of the heatmap (Figure 1, A) and displayed more hypermethylated CpG loci than ependymomas. In order to further investigate the DNA methylation patterns of gliomas and non- tumor brain tissue, we implemented an agnostic approach by applying a modified model- based form of unsupervised clustering known as recursively partitioned mixture modeling (RPMM) (20). RPMM allows for precise inference regarding the potential covariates associated with intrinsic similarities and differences in CpG methylation by generating distinct classes of DNA methylation for the modeled samples based upon the DNA methylation array data. We applied RPMM clustering to all 131 tumors, which generated 11 methylation classes (Figure 1, B). Methylation classes contain samples with DNA 49 methylation patterns that are most similar to each other, and samples with different DNA methylation patterns are distinguished by their membership in a different methylation class. Methylation class was statistically significantly associated with both tumor histologic subtype (P < 2.2  10-16) and grade (P < 2.2  10-16) (Supplementary Table 2, available online). Methylation Array and Methylation Class Validation Methylation data from GoldenGate arrays have been extensively validated by our group and others using a variety of methods (24-28). The methylation array data presented in this study were validated by correlating CpG methylation array data to quantitative methylation-specific PCR (QMSP) data for genes commonly methylated in gliomas— MGMT, RASSF1, PYCARD, HOXA9, PEG3, and SLIT2 (Supplementary Table 3, available online). To determine the validity of association between histology and methylation class we utilized publicly available GoldenGate methylation array data for 71 glioblastoma samples from The Cancer Genome Atlas (TCGA). Using the RPMM classification (Figure 1, B), we predicted the methylation class for each glioblastoma sample of TCGA and confirmed that 70 of 71 (99%) TCGA glioblastoma samples were classified in RPMM methylation classes that contained glioblastoma samples (Supplementary Table 2, available online). The identification numbers and the predicted RPMM methylation classes of TCGA tumors are listed in Supplementary Table 4, available online. 50 Ratios of Hypermethylated to Hypomethylated CpG Loci and Tumor Histology We examined the differential methylation (Δβ) between tumor and non-tumor brain tissues and observed a striking pattern of the number of hyper- and hypomethylated CpG loci among different tumor subtypes (Figure 2, A). Glioblastomas showed a low ratio of hyper- to hypomethylated loci (ratio = 1.3), compared with the ratio for grades 2 and 3 astrocytomas, grades 2 and 3 oligoastrocytomas, and grade 2 oligodendrogliomas (ratios = 3.7, 7.6, and 9.7, respectively). Conversely, ependymomas showed increased hypomethylation (ratio = 0.3). The ratios of hyper- to hypomethylated CpG loci were statistically significantly different across glioma tumor histologic subtypes (Permutation P<. 0001). Histology-related hyper- and hypomethylation patterns were also evident in unsupervised hierarchical clustering of Δβ methylation values for all 1413 autosomal CpG loci (Figure 2, B). We next assessed the cellular pathways associated with statistically significantly differentially hypomethylated and (separately) hypermethylated CpG loci that were common among glioblastomas, astrocytomas, oligoastrocytomas, and oligodendrogliomas. There were 18 CpG loci with statistically significant differential hypomethylation (Q<. 05) and common among glioblastomas, astrocytomas, oligoastrocytomas, and oligodendrogliomas. An analysis of cellular pathways enriched among these 18 CpG loci, compared with all genes represented on the methylation array, revealed statistically significant enrichment of metabolism and biosynthesis pathways (Supplementary Table 5, available online). In addition, there were 35 statistically significantly differentially hypermethylated (Q<0.05) CpG loci common among glioblastomas, astrocytomas, oligoastrocytomas, and oligodendrogliomas. An analysis of 51 cellular pathways enriched among these 35 CpG loci showed that oxidative stress response and retinoic acid mediated apoptosis signaling pathways were statistically significantly enriched (Supplementary Table 5, available online). For each grade-specific tumor histology, all statistically significant differentially hypomethylated and hypermethylated CpG loci are detailed in Supplementary Tables 6 and 7, respectively, available online. Glioma Methylation Classes, IDH Mutation, and Survival The analysis of differentially methylated CpG loci in cellular pathways suggested that metabolic pathways as a group were commonly hypomethylated in gliomas. We hypothesized that genetic mutations in the metabolic pathways were associated with the observed DNA methylation phenotype. To test this hypothesis, we sequenced a subset of 95 tumors with available DNA for IDH1 and IDH2 mutations. IDH2 mutation was detected in only two tumors, and IDH1 mutation was detected in 55 tumors (total IDH mutation prevalence = 58.9%). IDH mutations were more common in oligoastrocytoma, oligodendroglioma, or astrocytoma histologic subtypes than in glioblastomas, pilocytic astrocytomas, or ependymomas (P = 6.4  10-9); in lower-grade than higher-grade tumors (P = .01); in tumors with TP53 mutation compared with wild-type TP53 (P = .06); and in younger patients (mean age = 36.6 years vs. 47.4 years, P = .0009) (Table 2). However, IDH mutation was not associated with EGFR amplification (P = .10) (Table 2). Additionally, tumors with IDH mutation showed statistically significantly higher MGMT methylation (P = 3.6  10-4) (Supplementary Figure 1, available online). 52 Next we investigated the number of statistically significantly differentially methylated CpG loci between tumor and non-tumor samples stratified by IDH mutation status. Tumors with IDH mutation revealed a striking contrast between the number of statistically significantly differentially hypermethylated loci, as well as the ratio of hyper- to hypomethylated loci in IDH mutant tumors vs. IDH wild-type tumors (mutant = 7.8 vs. wild-type = 0.22) (Figure 3, A). We utilized the statistically significantly differentially hypermethylated and hypomethylated CpG loci in IDH mutant tumors to conduct an enrichment analysis of cellular pathways. We found that cellular signaling pathways were hypermethylated, whereas metabolism and biosynthesis pathways that included starch and sucrose metabolism and pentose and glucuronate interconversion pathways, were hypomethylated in IDH mutant tumors (Supplementary Table 8, available online). Methylation profiling with RPMM of the 95 gliomas with both methylation data and IDH mutation status resulted in nine methylation classes (Figure 3, B). Methylation classes were statistically significantly associated with patient age (Permutation P = 3.0  10-4), histology (P<2.2  10-16), and grade (P = 6.0  10-9) (Supplementary Table 9, available online). IDH mutation was also strongly associated with methylation class (P = 3.0  10-16) (Figure 3, C), and this association remained statistically significant when controlling for age and histology (likelihood ratio P<.0001). Only two methylation classes had IDH mutant tumors (class L and class RLLR), and greater than 98% of the tumors (all but one) in these two classes had an IDH mutation (Figure 3C). Furthermore, methylation classes L and RLLR were both more highly methylated than the other methylation classes (Figure 3, B). 53 Last, we examined the potential association between IDH mutation and patient survival among cases with available mutation data (n = 95) because previous studies reported increased survival among glioma patients with IDH mutation (3, 5). In a multivariate Cox proportional hazards model controlling for age at diagnosis, sex, and grade-specific histology, we observed that patients whose tumors harbored IDH mutation showed statistically significantly better survival, compared with patients (n = 39) whose tumors harbored wild-type IDH (HR of death = 0.27, 95% confidence interval [CI] = 0.10 to 0.72) (Figure 3, D, and Table 3). Discussion In this study, we demonstrate a distinct pattern of methylation across histological subtypes of glioma that is associated with genetic mutation in IDH gene loci. The two methylation classes associated with mutant IDH tumors had a homogeneous, hypermethylation-rich character compared to the methylation classes for tumors with wild-type IDH. Additionally, the tumors with wild-type IDH belonged to several distinct methylation classes. The contrast between a single homogenous hypermethylated profile and several heterogeneous hypomethylated profiles (associated with distinct histologic types) strongly suggests that IDH mutation “drives” the observed hypermethylated phenotype, irrespective of tumor histology. In support of this, we note that IDH1 mutation is more robustly associated with methylation class, compared with the classical glioma tumor genetic markers like TP53 mutation and EGFR amplification. IDH mutations are heterozygous and allow the enzyme normally responsible for conversion of isocitrate to α-ketoglutarate to convert α-ketoglutarate to 2- 54 hydroxyglutarate in an NADPH–dependent manner and results in accumulation of 2- hydroxyglutarate (7, 8). Despite the observed hypermethylated profile of IDH mutant tumors, analysis of cellular pathways showed hypomethylation of several metabolic pathways, potentially to compensate for mutation-related metabolic stress. Because the methylation profile of IDH mutant tumors is generally homogenous, it is possible that the hypermethylation phenotype is either selected for, or driven by, the hypomethylation of compensatory metabolic pathways, thus directly linking and temporally situating these events. The level of α-ketoglutarate has been shown to be slightly lower in IDH1 mutant gliomas, though this decrease was not statistically significant (8). However, IDH1 localizes to the cytosol and peroxisomes, whereas IDH2 is localizes to mitochondria; and because most IDH mutations in gliomas are in IDH1, pan-cellular α-ketoglutarate levels may not represent available cytosolic α-ketoglutarate levels. Furthermore, IDH1 R132 mutation has been shown to favor an active conformation of the enzyme, increase its affinity for NADPH, and favor reduction of α-ketoglutarate to 2-hydroxyglutarate over the conversion of isocitrate to α-ketoglutarate, which may reduce the availability of cytosolic α-ketoglutarate and NADPH (8). Hence, a potential mechanism responsible for the strong association between epigenetic profile and IDH mutation is related to potentially altered availability of α-ketoglutarate in these tumors. The Jumonji-domain- containing histone demethylases require α-ketoglutarate as a substrate for their enzymatic activity (29) and altered activity of these histone demethylases could lead to aberrantly remodeled chromatin, potentially resulting in epigenetic alterations at the DNA-level as well. However, studies that are beyond the scope of this manuscript would be necessary to disentangle the complex networks of chromatin remodeling enzymes, their targets, and 55 their responses to altered levels of enzymatic substrate. Alternatively, (or perhaps in conjunction) lower concentrations of NADPH associated with mutant IDH1 (30) may result in a decreased capacity for reductive processes in defense against reactive oxygen species. Furthermore, α-ketoglutarate itself is a potent antioxidant (6) and its decreased availability in IDH mutant cells alone, or together with lower NADPH levels could drive the selection of cells with compensatory metabolic gene expression profiles mediated by altered epigenetic patterns including chromatin configuration and DNA methylation. Consistent with the suggestion that gene expression profiles are altered in association with DNA methylation related to IDH mutation, an analysis of glioblastoma gene expression subtypes showed that IDH mutation occurred almost exclusively proneural glioblastomas (31). More broadly, and similar to the hypermethylation phenotype we describe here, hypermethylator phenotypes have previously been associated with other cancers. This phenotype was first described in colon cancer, and is commonly referred to as CpG Island Methylator Phenotype (CIMP) (32). Specifically, colorectal cancers can be divided in CIMP-high, CIMP-low, and non-CIMP based on the methylation of 5-8 specific gene promoters (33, 34). Similar to IDH in glioma, CIMP status in colon tumors has been associated with specific mutations; CIMP-High with BRAF and CIMP-Low and non- CIMP with KRAS (35). Recently, Noushmehr et al. described a CIMP in glioblastomas, termed G-CIMP, which they found to be tightly associated with IDH1 mutation (36). In a number of lower-grade gliomas Noushmehr et al. performed methylation profiling of eight markers of G-CIMP and confirmed that IDH1 mutation is associated with G-CIMP in low-grade tumors, which is consistent with our array-based findings. Furthermore, 56 over 83% of G-CIMP positive glioblastomas with IDH1 mutation were of the proneural glioblastoma gene expression subtype (36), additional evidence supporting an association between distinct, IDH-related methylation in our data (from diverse glioma histologic subtypes), and a specific gene expression phenotype. In addition, MGMT methylation is often investigated in glioma since it has been associated with increased sensitivity to alkylating agents such as Temozolomide and can impact response to therapy (37). In fact, increased MGMT methylation can also distinguish CIMP-High and CIMP- Low from non-CIMP in colon cancer (38). Our results, consistent with previous work (9), demonstrate an association between increased MGMT methylation and IDH mutation. Finally, some studies have reported CIMP positive colon cancers to have a relatively better prognosis (39), and from both the work of Noushmehr et al. and ours, this appears to be consistent with the pattern of survival observed in CIMP gliomas. The association between IDH mutation and a single methylation profile across several histologic subtypes suggests that genetic and epigenetic alterations are not independent. This observation also has profound implications for the development of new therapies for glioma. Although pharmacological inhibition of 2-hydroxyglutarate has been suggested as a possible approach to treating IDH mutant gliomas (40) such drugs do not yet exist. However, DNA methylation is a modifiable therapeutic target; DNA methyltransferase inhibitors and histone deacetylase inhibitors are in clinical trials and showing some promise for the treatment of hematopoietic malignancies (41-43). Our work suggests that a simple diagnostic test for DNA methylation (or mutation) can identify a class of tumors for which the modification of DNA methylation may have therapeutic efficacy. This class of tumors is not discernable by any of the classic 57 histopathologic or tumor markers for glioma. The recognition that IDH mutation has value as a clinical prognostic marker and is associated with a broad DNA methylation phenotype suggests that glioma therapeutic protocols that reverse DNA methylation should be pursued. Our study has a few limitations. Although we studied 131 histologically diverse tumors, we did not have IDH mutation, TP53 mutation, and EGFR amplification data on all subjects and had somewhat limited statistical power to explore the relationships between IDH mutation and these alterations. Future investigations that include larger numbers of histologically diverse samples and higher-resolution methylation array techniques, along with measurements of other somatic alterations (IDH mutation, mRNA expression, and copy number) will afford a more comprehensive understanding of the molecular and chromosomal characteristics that distinguish glioma subtypes. Understanding whether these glioma molecular and chromosomal subtypes are differentially associated with glioma risk loci (44) also will help to understand the etiology and possibly outcomes of this often-catastrophic disease. In summary, our work demonstrates a clear relationship between genetic and epigenetic events in human gliomas by associating IDH mutations with a homogenous methylation profile, and demonstrates that profiles of methylation differ by histologic subtype of disease. Additionally, and consistent with previous work, we also showed that patients with IDH mutation have a significantly improved survival. Advances in therapy for glioma may be realized by targeting DNA methylation. Much attention has recently been given to the utility of MGMT methylation in predicting response to therapy, and our 58 data further suggest that other DNA methylation markers may improve clinical assessment, guide therapies, and potentially uncover novel therapeutic avenues altogether. 59 References 1. Kleihues P, Burger PC, Scheithauer BW. The new WHO classification of brain tumours. Brain Pathol 1993;3(3):255-68. 2. Stupp R, Mason WP, van den Bent MJ, Weller M, Fisher B, Taphoorn MJ, et al. Radiotherapy plus concomitant and adjuvant temozolomide for glioblastoma. N Engl J Med 2005;352(10):987-96. 3. Parsons DW, Jones S, Zhang X, Lin JC, Leary RJ, Angenendt P, et al. An integrated genomic analysis of human glioblastoma multiforme. Science 2008;321(5897):1807-12. 4. Balss J, Meyer J, Mueller W, Korshunov A, Hartmann C, von Deimling A. Analysis of the IDH1 codon 132 mutation in brain tumors. Acta Neuropathol 2008;116(6):597-602. 5. Yan H, Parsons DW, Jin G, McLendon R, Rasheed BA, Yuan W, et al. IDH1 and IDH2 mutations in gliomas. N Engl J Med 2009;360(8):765-73. 6. Reitman ZJ, Yan H. Isocitrate dehydrogenase 1 and 2 mutations in cancer: alterations at a crossroads of cellular metabolism. J Natl Cancer Inst 2010;102(13):932- 41. 7. Ward PS, Patel J, Wise DR, Abdel-Wahab O, Bennett BD, Coller HA, et al. The common feature of leukemia-associated IDH1 and IDH2 mutations is a neomorphic enzyme activity converting alpha-ketoglutarate to 2-hydroxyglutarate. Cancer Cell 2010;17(3):225-34. 60 8. Dang L, White DW, Gross S, Bennett BD, Bittinger MA, Driggers EM, et al. Cancer-associated IDH1 mutations produce 2-hydroxyglutarate. Nature 2009;462(7274):739-44. 9. Sanson M, Marie Y, Paris S, Idbaih A, Laffaire J, Ducray F, et al. Isocitrate dehydrogenase 1 codon 132 mutation is an important prognostic biomarker in gliomas. J Clin Oncol 2009;27(25):4150-4. 10. Hegi ME, Diserens AC, Gorlia T, Hamou MF, de Tribolet N, Weller M, et al. MGMT gene silencing and benefit from temozolomide in glioblastoma. N Engl J Med 2005;352(10):997-1003. 11. Mashiyama S, Murakami Y, Yoshimoto T, Sekiya T, Hayashi K. Detection of p53 gene mutations in human brain tumors by single-strand conformation polymorphism analysis of polymerase chain reaction products. Oncogene 1991;6(8):1313-8. 12. Wong AJ, Bigner SH, Bigner DD, Kinzler KW, Hamilton SR, Vogelstein B. Increased expression of the epidermal growth factor receptor gene in malignant gliomas is invariably associated with gene amplification. Proc Natl Acad Sci U S A 1987;84(19):6899-903. 13. Dallol A, Krex D, Hesson L, Eng C, Maher ER, Latif F. Frequent epigenetic inactivation of the SLIT2 gene in gliomas. Oncogene 2003;22(29):4611-6. 14. Maegawa S, Itaba N, Otsuka S, Kamitani H, Watanabe T, Tahimic CG, et al. Coordinate downregulation of a novel imprinted transcript ITUP1 with PEG3 in glioma cell lines. DNA Res 2004;11(1):37-49. 61 15. Stone AR, Bobo W, Brat DJ, Devi NS, Van Meir EG, Vertino PM. Aberrant methylation and down-regulation of TMS1/ASC in human glioblastoma. Am J Pathol 2004;165(4):1151-61. 16. Yu J, Zhang H, Gu J, Lin S, Li J, Lu W, et al. Methylation profiles of thirty four promoter-CpG islands and concordant methylation behaviours of sixteen genes that may contribute to carcinogenesis of astrocytoma. BMC Cancer 2004;4:65. 17. Wiencke J, Aldape K, McMillan A, Wiemels J, Moghadassi M, Miike R, et al. Molecular features of adult glioma associated with patient race/ethnicity, age, and a polymorphism in O6-methylguanine-DNA-methyltransferase. Cancer Epidemiol Biomarkers Prev 2005;14(7):1774-83. 18. De Preter K, Speleman F, Combaret V, Lunec J, Laureys G, Eussen B, et al. Quantification of MYCN, DDX1, and NAG gene copy number in neuroblastoma using a real-time quantitative PCR assay. Mod Pathol 2002;15(2):159-66. 19. R Development CT. R: A Language and Environment for Statistical Computing. In. Vienna, Austria: R Foundation for Statistical Computing; 2007. 20. Houseman EA, Christensen BC, Marsit CJ, Karagas MR, Wrensch MR, Yeh RF, et al. Model-based clustering of DNA methylation array data: a recursive-partitioning algorithm for high-dimensional data arising as a mixture of beta distributions. BMC Bioinformatics 2008;9(365). 21. Christensen BC, Houseman EA, Godleski JJ, Marsit CJ, Longacker JL, Roelofs CR, et al. Epigenetic profiles distinguish pleural mesothelioma from normal pleura and predict lung asbestos burden and clinical outcome. Cancer Res 2009;69(1):227-34. 62 22. Grambsch P, Therneau T. Proportional hazards tests and diagnostics based on weighted residuals. Biometrika 1994;81(3):515-26. 23. Storey J, Taylor J, Siegmund D. Strong control, conservative point estimation, and simultaneous conservative consistency of false discovery rates: A unified approach. J Royal Stat Soc 2004;Series B(66):187-205. 24. Bibikova M, Lin Z, Zhou L, Chudin E, Garcia EW, Wu B, et al. High-throughput DNA methylation profiling using universal bead arrays. Genome Res 2006;16(3):383-93. 25. Byun HM, Siegmund KD, Pan F, Weisenberger DJ, Kanel G, Laird PW, et al. Epigenetic profiling of somatic tissues from human autopsy specimens identifies tissue- and individual-specific DNA methylation patterns. Hum Mol Genet 2009;18(24):4808-17. 26. Christensen BC, Houseman EA, Marsit CJ, Zheng S, Wrensch MR, Wiemels JL, et al. Aging and environmental exposures alter tissue-specific DNA methylation dependent upon CpG island context. PLoS Genet 2009;5(8):e1000602. 27. Irizarry RA, Ladd-Acosta C, Carvalho B, Wu H, Brandenburg SA, Jeddeloh JA, et al. Comprehensive high-throughput arrays for relative methylation (CHARM). Genome Res 2008;18(5):780-90. 28. Ladd-Acosta C, Pevsner J, Sabunciyan S, Yolken RH, Webster MJ, Dinkins T, et al. DNA methylation signatures within the human brain. Am J Hum Genet 2007;81(6):1304-15. 29. Krieg AJ, Rankin EB, Chan D, Razorenova O, Fernandez S, Giaccia AJ. Regulation of the histone demethylase JMJD1A by hypoxia-inducible factor 1 alpha enhances hypoxic gene expression and tumor growth. Mol Cell Biol 2010;30(1):344-53. 63 30. Winkler BS, DeSantis N, Solomon F. Multiple NADPH-producing pathways control glutathione (GSH) content in retina. Exp Eye Res 1986;43(5):829-47. 31. Verhaak RGW, Hoadley KA, Purdom E, Wang V, Qi Y, Wilkerson MD, et al. Integrated Genomic Analysis Identifies Clinically Relevant Subtypes of Glioblastoma Characterized by Abnormalities in PDGFRA, IDH1, EGFR, and NF1. Cancer Cell 2010;17(1):98-110. 32. Toyota M, Ahuja N, Ohe-Toyota M, Herman JG, Baylin SB, Issa JP. CpG island methylator phenotype in colorectal cancer. Proc Natl Acad Sci U S A 1999;96(15):8681- 6. 33. Ogino S, Nosho K, Kirkner G, Kawasaki T, Chan A, Schernhammer E, et al. A cohort study of tumoral LINE-1 hypomethylation and prognosis in colon cancer. J Natl Cancer Inst 2008;100(23):1734-8. 34. Issa J. CpG island methylator phenotype in cancer. Nat Rev Cancer 2004;4(12):988-93. 35. Ogino S, Kawasaki T, Kirkner G, Loda M, Fuchs C. CpG island methylator phenotype-low (CIMP-low) in colorectal cancer: possible associations with male sex and KRAS mutations. J Mol Diagn 2006;8(5):582-8. 36. Noushmehr H, Weisenberger DJ, Diefes K, Phillips HS, Pujara K, Berman BP, et al. Identification of a CpG island methylator phenotype that defines a distinct subgroup of glioma. Cancer Cell 2010;17(5):510-22. 37. Chin L, Meyerson M, Aldape K, Bigner D, Mikkelsen T, VandenBerg S, et al. Comprehensive genomic characterization defines human glioblastoma genes and core pathways. Nature 2008;455(7216):1061-1068. 64 38. Ogino S, Kawasaki T, Kirkner G, Suemoto Y, Meyerhardt J, Fuchs C. Molecular correlates with MGMT promoter methylation and silencing support CpG island methylator phenotype-low (CIMP-low) in colorectal cancer. Gut 2007;56(11):1564-71. 39. Ogino S, Goel A. Molecular classification and correlates in colorectal cancer. J Mol Diagn 2008;10(1):13-27. 40. Frezza C, Tennant DA, Gottlieb E. IDH1 mutations in gliomas: when an enzyme loses its grip. Cancer Cell 2010;17(1):7-9. 41. Borthakur G, Huang X, Kantarjian H, Faderl S, Ravandi F, Ferrajoli A, et al. Report of a phase 1/2 study of a combination of azacitidine and cytarabine in acute myelogenous leukemia and high-risk myelodysplastic syndromes. Leuk Lymphoma;51(1):73-8. 42. Santos FP, Kantarjian H, Garcia-Manero G, Issa JP, Ravandi F. Decitabine in the treatment of myelodysplastic syndromes. Expert Rev Anticancer Ther;10(1):9-22. 43. Mercurio C, Minucci S, Pelicci PG. Histone deacetylases and epigenetic therapies of hematological malignancies. Pharmacol Res. 44. Wrensch M, Jenkins RB, Chang JS, Yeh RF, Xiao Y, Decker PA, et al. Variants in the CDKN2B and RTEL1 regions are associated with high-grade glioma susceptibility. Nat Genet 2009;41(8):905-8. 65 Figure 1. Association between glioma histologic subtypes and DNA methylation pattern. A) The average methylation beta (β) values of both gliomas (n = 131) and non-tumor tissue samples (n =7) were subjected to unsupervised hierarchical clustering based on Euclidean distance metric and Ward linkage and are shown in the heatmap. Each row represents a sample and each column represents a CpG locus (all 1413 autosomal loci). The scale bar at the bottom shows the range of β values (0 to 1). Tissue histology and grade are defined in color keys next to the heatmap, on the left. GBM2 = secondary glioblastoma multiforme; GBM = primary glioblastoma multiforme; AS3 = grade 3 astrocytoma; AS2 = grade 2 astrocytoma; OA3 = grade 3 oligoastrocytoma; OA2 = grade 2 oligoastrocytoma; OD2 = grade 2 oligodendroglioma; EP = ependymoma; PA = pilocytic astrocytoma. B) Recursively partitioned mixture model (RPMM) of glioma and non-tumor brain tissue samples (n = 138). Methylation profile classes are stacked in rows separated by red lines and class height corresponds to the number of samples in each class. Class methylation at each CpG locus (columns) is the mean methylation for all samples in a class. To the left of the RPMM is the clustering dendrogram. In the heatmap and RPMM, blue designates methylated CpG loci (average β = 1), and yellow designates unmethylated CpG loci (average β = 0). 66 Figure 2. Differential methylation and the ratio of hyper- to hypomethylated loci in gliomas. Differential the non-tumor brain samples (n = 7) for each CpG locus. A) The number of statistically significantly differentially hyper- and hypomethylated loci (Q<0.05 and |Δβ|>0.2), are plotted by grade-specific glioma histology. GBM = primary glioblastoma multiforme; GBM2 = secondary glioblastoma multiforme; AS3 = grade 3 astrocytoma; AS2 = grade 2 astrocytoma; OA3 = grade 3 oligoastrocytoma; OA2 = grade 2 oligoastrocytoma; OD2 = grade 2 oligodendroglioma; EP = ependymoma; PA = pilocytic astrocytoma. B) Δβ values for all tumors (n = 131) were subjected to unsupervised hierarchical clustering based on Euclidean distance metric and Ward linkage. Each row represents a sample and each column represents a CpG locus (all 1413 autosomal loci). The scale bar at the top shows the range of Δβ values (-1 to 1). Tissue histology and grade are defined in color keys next to the heatmap on the left. In the heatmap blue designates differentially hypermethylated CpG loci in tumors (Δβ = 1), and yellow designates differentially hypomethylated CpG loci in tumors (Δβ = -1). 67 Figure 3. Association between IDH mutation and methylation phenotype in gliomas. A) The number of statistically significantly differentially hyper- and hypomethylated loci (Q<0.05 and |Δβ|>0.2), are plotted by tumor IDH mutation status. B) Recursively partitioned mixture model (RPMM) of glioma samples with both methylation and mutation data (n = 95). Methylation profile classes are stacked in rows separated by red lines, class height corresponds to the number of samples in each class. Class methylation at each CpG locus (columns) is the mean methylation for all samples in a class where blue designates methylated CpG loci (average β = 1), and yellow designates unmethylated CpG loci (average β = 0). To the right of the RPMM is the clustering dendrogram. C) Methylation-class-specific IDH mutation status (Fisher’s P = 3.0E-16). D) Kaplan-Meier survival probability strata for IDH mutant (red, n = 56) and IDH wild-type (black, n = 39) tumors, tick marks are censored observations and banding patterns represent 95% confidence intervals (CIs). 68 Table 1. Patient demographic and tumor characteristics* Tumor histology and grade of glioma tissues (n=131) Non- Primary Secondary tumor Glioblasto Glioblasto Grade 3 Grade 2 Grade 3 Grade 2 Grade 2 Pilocytic Eppendymo brain ma ma Aastrocyto Astrocyto Oligoastrocyto Oligoastrocyto Oligodendroglio Astrocyto ma (n=15) tissue multiforme multiforme ma (n=9) ma (n=20) ma (n=9) ma (n=22) ma (n=20) ma (n=4) Characteristic (n=7) (n=20) (n=12) Age at diagnosis, y Median 33 55 34.5 40 40 40 33 35.5 41 28.5 Range 23 - 42 21 - 78 18 - 49 23 - 57 21 - 64 26 - 52 19 - 48 20 - 59 19 - 70 22 - 39 Sex, No. (%) Female 3 (43) 7 (35) 4 (33) 6 (67) 10 (50) 4 (44) 9 (41) 10 (50) 4 (27) 2 (50) Male 4 (57) 13 (65) 8 (67) 3 (33) 10 (50) 5 (56) 13 (59) 10 (50) 11 (73) 2 (50) Race, No. (%) White - 18 (90) 11 (92) 5 (56) 17 (85) 7 (78) 16 (73) 18 (90) 12 (80) 4 (100) Hispanic - 1 (5) 1 (8) 2 (22) 1 (5) 0 1 (4) 1 (5) 2 (13) 0 American Indian - 0 0 0 1 (5) 0 0 0 0 0 Asian - 0 0 1 (11) 1 (5) 0 2 (9) 1 (5) 0 0 Unknown 7 (100) 1 (5) 0 1 (11) 0 2 (22) 3 (14) 0 1 (7) 0 Survival, d Median NA 759 1244 1933 1584 3007 2937 2532 2498 2789 Range NA 108-2477 466-4973 516-4494 305-4043 603-6459 612-5843 4-5988 478-5983 948-3279 *Non-tumor brain tissues (n=7) were obtained from cancer-free patients who underwent temporal lobe resection for treatment of epilepsy at the UCSF Medical Center. Glioma tissues (n=131) were obtained between 1990 and 2003 from the University of California San Francisco Brain Tumor Research Center Tissue Bank. 69 Table 2. Patient age, grade-specific glioma histology, grade, TP53 mutation, and EGFR amplification stratified by IDH mutation status* IDH Mutation† Patient age and tumor characteristics No Yes Age at diagnosis, y P = 9.0E-04‡ Median age (range) 49 (17–78) 35 (20–59) Mean age (SD) 47.4 (17.5) 36.6 (8.7) Tumor histology§, No. (%) P = 6.4E-09|| Grade 2 Astrocytoma 5 (26) 14 (74) Grade 3 Astrocytoma 0 (0) 4 (100) Ependymoma 14 (100) 0 (0) Primary Glioblastoma 15 (79) 4 (21) Secondary Glioblastoma (P=.005)¶ 1 (14) 6 (86) Grade 2 Oligoastrocytoma 2 (13) 13 (87) Grade 2 Oligodendroglioma 1 (6) 15 (94) Tumor grade, No. (%) P = .01# 1 - - 2 22 (34) 42 (66) 3 0 (0) 5 (100) 4 16 (62) 10 (38) TP53 mutation, No. (%) P = .06** No 27 (63) 16 (37) Yes 5 (31) 11 (69) EGFR amplification, No. (%) P = .10†† No 28 (51) 27 (49) Yes 5 (100) 0 (0) * Analysis of patient age and tumor characteristics vs isocitrate dehydrogenase (IDH) gene mutation status. TP53 = tumor protein 53. EGFR = epidermal growth factor receptor. † IDH gene mutation was assessed by sequencing tumor DNA. ‡ Association between age and IDH mutation was assessed using two-sided Student’s t-test. § Tumors were previously reviewed by neuropathologists at the University of California San Francisco to assign histologic subtypes and grades according to the World Health Organization classification. || Association between grade-specific histology and IDH mutation was assessed using two-sided Fisher’s exact test. ¶ Association between primary vs. secondary glioblastoma and IDH mutation was assessed using two-sided χ2 test. # Association between tumor grade and IDH mutation was assessed using two-sided Fisher’s exact test 2 ** Association between TP53 mutation and IDH mutation was assessed using two-sided χ test. †† Association between EGFR amplification and IDH mutation was assessed using two-sided χ2 test. 70 Table 3. Survival analysis using multivariable Cox proportional hazards model* Variable HR† (95% CI) Age 1.03 (1.00 to 1.06) Sex Female 1.0 (Referent) Male 0.73 (0.34 to 1.55) IDH mutation‡ No 1.0 (Referent) Yes 0.27 (0.10 to 0.72) Histology§ Grade 2 astrocytoma 1.0 (Referent) Grade 3 astrocytoma 1.79 (0.35 to 9.13) Ependymoma 0.25 (0.06 to 1.06) Primary glioblastoma 1.77 (0.60 to 5.22) Secondary glioblastoma 3.94 (1.20 to 12.9) Grade 2 oligoastrocytoma 2.8 (0.06 to 1.39) Grade 3 oligoastrocytoma|| - Grade 2 oligodendroglioma 0.75 (0.21 to 2.69) * Cox proportional hazards model of survival included age, sex, IDH mutation, and grade- specific histology. HR = hazards ratio, CI = confidence interval, IDH = isocitrate dehydrogenase gene. † Adjusted HR values. ‡ IDH gene mutation was assessed by sequencing tumor DNA. §Tumors were previously reviewed by neuropathologists at the University of California San Francisco to assign histologic subtypes and grades according to the World Health Organization classification. || n =1, HR = 1.4E-07, standard error = 4,910, confidence interval indeterminable. 71 0.8 P=3.6E-04 0.6 P = 3.6E-04 Relative MGMT methyaltion 0.4 0.2 - - 0.0 IDH wild-type IDH mutant Supplementary Figure 1. Association between IDH mutation and increased MGMT methylation. IDH mutation status vs. relative MGMT methylation from quantitative methylation specific PCR demonstrates statistically significantly increased MGMT methylation among tumors with IDH mutation (P = 3.6 × 10-4). Black bars indicate mean relative MGMT methylation in IDH wild-type (0.04) and IDH mutant tumors (0.17). 72 Supplementary Table 1. Primer sequences for quantitative methylation specific polymerase chain reaction (QMSP), IDH mutation, TP53 mutation, and EGFR amplification experiments* Experiment Forward 5'-3' Reverse 5'-3' Amplicon Size Reference QMSP RASSF1A-M GTGTTAACGCGTTGCGTATC AACCCCGCGAACTAAAAACG 94 Yu et al. 2004 (1) MGMT3-M GATTTGGTGAGTGTTTGGGTC ACCACTCGAAACTACCACCG 79 This study HOXA9-M GAATTTAAGGGTTGTTCGGGC GACCGCTCAAAAAATACCGCG 81 This study PYCARD-M GGTTGTAGCGGGGTGAGC CGACGATCAAATTCTCCAACG 96 Stone et al. 2004 (2) PEG3-M TCGTCGTATTTGTCGTTAATTAATTC GCAAACGCTATCCTAATTAATTAAACG 123 Maegawa et al. 2004 (3) SLIT2-M TTTAGGTTGCGGCGGAGTC CAACGAACCCGTAACAAAACG 147 Dallol et al. 2003 (4) ACTB TGGTGATGGAGGAGGTTTAGTAAGT AACCAATAAAACCTACTCCTCCCTTAA 133 Harden et al. 2003 (5) IDH1 Amplification ATATTCTGGGTGGCACGGTCTT CCTTGCTTAATGGGTGTAGATACCA 227 This study Sequencing F CGGTCTTCAGAGAAGCCATT This study Sequencing R CATGCAAAATCACATTATTGCCAAC This study IDH2 Amplification TTCTGGTTGAAAGATGGCG CAGGTCAGTGGATCCCCTC 251 This study Sequencing F ATGGCGGCTGCAGTGGG This study Sequencing R CAGGTCAGTGGATCCCCTC This study TP53 Exon 5 GTTCACTTGTGCCCTGA AGCCCTGTCGTCTCT Wiencke et al. 2005 (6) Exon 6 CTCTGATTCCTCACTG CCAGAGACCCCAGTTGCAAACC Wiencke et al. 2005 (6) Exon 7 TGCTTGCCACAGGTCT ACAGCAGGCCAGTGT Wiencke et al. 2005 (6) Exon 8 AGGACCTGATTTCCTTAC TCTGAGGCATAACTGC Wiencke et al. 2005 (6) Gene Amplification EGFR CCGCATTAGCTCTTAGACCCA GAATGCAACTTCCCAAAATGTGC 98 This study GAPDH CTCCCCACACACATGCACTTA CCTAGTCCCAGGGCTTTGATT 99 This study * RASSF1=Ras association domain family member 1, MGMT=O-6-methylguanine-DNA methyltransferase, HOXA9=homeobox A9, PYCARD=PYD and CARD domain containing, PEG3=paternally expressed 3, SLIT2=slit homolog 2, ACTB=actin, beta, IDH=isocitrate dehydrogenase, TP53=tumor protein 53, EGFR=epidermal growth factor receptor, GAPDH=glyceraldehyde-3-phosphate dehydrogenase. 73 Supplementary Table 2. Recursively partitioned mixture model methylation class by glioma histology and predicted methylation class membership for The Cancer Genome Atlas (TCGA) glioblastoma samples* Predicted TCGA GBM Class Methylation Class AS2 AS3 EP GBM GBM2 OA2 OA3 OD2 PA LLLLL 0 0 3 0 0 1 0 0 0 0 LLLLR 0 0 3 0 0 0 0 0 0 0 LLLRL 0 0 5 0 0 0 0 0 0 0 LLLRR 1 0 3 0 0 0 0 0 0 0 LLRLL 1 0 0 0 0 0 0 1 1 0 LLRLR 0 0 0 0 0 1 0 0 3 0 LLRR 4 0 0 0 0 1 0 0 0 1 LRL 0 1 0 14 0 0 0 0 0 56 LRR 0 0 1 3 8 0 0 0 0 10 RL 12 8 0 2 4 10 4 3 0 4 RR 2 0 0 1 0 9 5 16 0 0 * AS2=grade 2 Astrocytoma, AS3=grade 3 astrocytoma, EP=ependymoma, GBM=primary glioblastoma multiforme, GBM2=secondary glioblastoma multiforme, OA2=grade 2 oligoastrocytoma, OA3=grade 3 oligoastrocytoma, OD2=grade 2 oligodendroglioma, PA=pilocytic astrocytoma, TCGA=The Cancer Genome Atlas. Tumors were previously reviewed by UCSF neuropathologists to assign histologic subtypes and grades according to the World Health Organization classification. 74 Supplementary Table 3. Association between GoldenGate array methylation values and quantitative methylation specific polymerase chain reaction (QMSP)* Spearman GENE_CpG † No. ‡ (rho) § P || PEG3_E496 110 0.32 5.90E-04 HOXA9_E252 117 0.52 1.50E-09 HOXA9_E252 117 0.53 6.90E-10 MGMT_P272 110 0.45 7.40E-07 MGMT_P281 110 0.47 2.30E-07 PYCARD_E87 107 0.81 < 2.2E-16 PYCARD_P150 107 0.26 6.70E-03 PYCARD_P393 107 0.43 5.00E-06 RASSF1A_E116 118 0.7 < 2.2E-16 RASSF1A_P244 118 0.59 3.50E-12 SLIT2_E111 106 0.4 2.30E-05 SLIT2_P208 106 0.4 2.40E-05 * PEG3 = paternally expressed 3, HOXA9 = homeobox A9, MGMT = O-6-methylguanine-DNA methyltransferase, PYCARD = PYD and CARD domain containing, RASSF1 = Ras association domain family member 1, SLIT2 = slit homolog 2. † This column lists the Illumina GoldenGate methylation array annotation for CpGs where the gene name is listed first in all capital letters and italics followed by an E for exon or P for promoter to indicate the location of the CpG relative to the transcription start site, and the number indicates the distance of the CpG from the transcription start site. ‡ Number of samples with both GoldenGate array and QMSP methylation data. § Spearman correlation coefficient (rho) || Two-sided Spearman’s rank correlation test for association between GoldenGate array methylation value and QMSP methylation value. 75 Supplementary Table 4. Identification numbers (ID) and RPMM methylation class membership for The Cancer Genome Atlas (TCGA) glioblastoma samples used in validation. RPMM methylation TCGA ID class TCGA-02-0001-01C-01D-0186-05 LRR TCGA-02-0002-01A-01D-0186-05 LRL TCGA-02-0003-01A-01D-0186-05 LRL TCGA-02-0006-01B-01D-0186-05 LRL TCGA-02-0007-01A-01D-0186-05 LRL TCGA-02-0009-01A-01D-0186-05 LRL TCGA-02-0010-01A-01D-0186-05 LRR TCGA-02-0011-01B-01D-0186-05 LRR TCGA-02-0014-01A-01D-0186-05 LRR TCGA-02-0021-01A-01D-0186-05 LRL TCGA-02-0024-01B-01D-0186-05 LRR TCGA-02-0027-01A-01D-0186-05 LRL TCGA-02-0028-01A-01D-0186-05 RL TCGA-02-0033-01A-01D-0186-05 LRL TCGA-02-0034-01A-01D-0186-05 LRL TCGA-02-0037-01A-01D-0186-05 LRL TCGA-02-0038-01A-01D-0186-05 LRL TCGA-02-0043-01A-01D-0186-05 LRL TCGA-02-0046-01A-01D-0186-05 LRL TCGA-02-0047-01A-01D-0186-05 LRR TCGA-02-0052-01A-01D-0186-05 LRL TCGA-02-0054-01A-01D-0186-05 LRL TCGA-02-0055-01A-01D-0186-05 LRL TCGA-02-0057-01A-01D-0186-05 LRL TCGA-02-0058-01A-01D-0186-05 RL TCGA-02-0060-01A-01D-0186-05 LRL TCGA-02-0064-01A-01D-0199-05 LRL TCGA-02-0069-01A-01D-0199-05 LRR TCGA-02-0071-01A-01D-0199-05 LRL TCGA-02-0074-01A-01D-0199-05 LRL TCGA-02-0075-01A-01D-0199-05 LRL 76 TCGA-02-0080-01A-01D-0199-05 RL TCGA-02-0083-01A-01D-0199-05 LRL TCGA-02-0085-01A-01D-0199-05 LRL TCGA-02-0086-01A-01D-0199-05 LRL TCGA-02-0089-01A-01D-0199-05 LRL TCGA-02-0099-01A-01D-0199-05 LRL TCGA-02-0102-01A-01D-0199-05 LRL TCGA-02-0107-01A-01D-0199-05 LRL TCGA-02-0113-01A-01D-0199-05 LRL TCGA-02-0114-01A-01D-0199-05 LRR TCGA-02-0115-01A-01D-0199-05 LRL TCGA-02-0116-01A-01D-0199-05 LRL TCGA-06-0119-01A-08D-0218-05 LRL TCGA-06-0121-01A-04D-0218-05 LRL TCGA-06-0122-01A-01D-0218-05 LRL TCGA-06-0124-01A-01D-0218-05 LRL TCGA-06-0125-01A-01D-0218-05 LRL TCGA-06-0126-01A-01D-0218-05 LRL TCGA-06-0128-01A-01D-0218-05 RL TCGA-06-0129-01A-01D-0218-05 LRR TCGA-06-0130-01A-01D-0218-05 LRL TCGA-06-0133-01A-02D-0218-05 LRL TCGA-06-0137-01A-01D-0218-05 LRL TCGA-06-0137-01A-02D-0218-05 LRL TCGA-06-0137-01A-03D-0218-05 LRL TCGA-06-0137-01B-02D-0218-05 LRL TCGA-06-0139-01A-01D-0218-05 LLRR TCGA-06-0140-01A-01D-0218-05 LRL TCGA-06-0141-01A-01D-0218-05 LRR TCGA-06-0142-01A-01D-0218-05 LRL TCGA-06-0143-01A-01D-0218-05 LRL TCGA-06-0145-01A-01D-0218-05 LRL TCGA-06-0145-01A-02D-0218-05 LRL TCGA-06-0145-01A-03D-0218-05 LRL TCGA-06-0145-01A-04D-0218-05 LRL TCGA-06-0145-01A-05D-0218-05 LRL 77 TCGA-06-0145-01A-06D-0218-05 LRL TCGA-06-0147-01A-01D-0218-05 LRL TCGA-06-0148-01A-01D-0218-05 LRL TCGA-06-0169-01A-01D-0218-05 LRL 78 Supplementary Table 5. Cellular pathways enriched among statistically significantly differentially methylated CpG loci in common among glioblastomas, astrocytomas, oligoastrocytomas, and oligodendrogliomas*. Cellular Pathway P† Hypomethylated Methane Metabolism .02 Stilbene, Coumarine and Lignin Biosynthesis .02 Metabolism of Xenobiotics by Cytochrome P450 .02 PXR/RXR Activation .02 Retinol Metabolism .04 TREM1 Signaling .04 Phenylalanine Metabolism .05 Hypermethylated Retinoic acid Mediated Apoptosis Signaling .005 Primary Immunodeficiency Signaling .01 RAN Signaling .02 NRF2-mediated Oxidative Stress Response .03 EGF Signaling .05 * CpG loci with statistically significantly differential methylation (Q<0.05 and |Δβ|>0.2) between tumor and non-tumor tissue were examined for cellular pathway enrichment with Ingenuity pathways analysis software. PXR=nuclear receptor subfamily 1, group I, member 2; RXR=retinoid X receptor, gamma; TREM1=triggering receptor expressed on myeloid cells 1; RAN=RAN, member RAS oncogene family; NRF2=nuclear factor (erythroid-derived 2)-like 2; EGF=epidermal growth factor. † Two-sided Fisher’s exact test for enrichment of genes whose CpG loci are represented among the genes in the listed pathways. 79 Supplementary Table 6. Statistically significantly differentially hypomethylated CpG loci in human gliomas. Median Median Q- Δβ Q- Δβ GENE_CpG* value Value GENE_CpG* value Value Primary glioblastoma Grade 3 Astrocytoma CASP10_P334_F 0.002 -0.587 ACVR1_P983_F 0.007 -0.363 CD82_P557_R 0.002 -0.289 CASP10_P334_F 0.007 -0.412 CDK2_P330_R 0.002 -0.234 CD82_P557_R 0.007 -0.289 DDR1_P332_R 0.002 -0.299 DDR1_P332_R 0.007 -0.349 DSG1_P159_R 0.002 -0.249 GFAP_P1214_F 0.007 -0.324 GFAP_P1214_F 0.002 -0.293 GSTM2_P109_R 0.007 -0.255 GSTM2_P109_R 0.002 -0.261 IL16_P93_R 0.007 -0.394 LEFTY2_P561_F 0.002 -0.452 IL8_E118_R 0.007 -0.578 MPO_E302_R 0.002 -0.227 LEFTY2_P561_F 0.007 -0.426 MPO_P883_R 0.002 -0.637 MKRN3_E144_F 0.007 -0.480 PSCA_P135_F 0.002 -0.205 MKRN3_P108_F 0.007 -0.509 PTHR1_P258_F 0.002 -0.509 MPO_P883_R 0.007 -0.514 TRIP6_P1274_R 0.002 -0.500 PADI4_P1011_R 0.007 -0.246 TRPM5_P721_F 0.002 -0.265 PTHR1_P258_F 0.007 -0.446 UGT1A1_P315_R 0.002 -0.276 TRIP6_P1090_F 0.007 -0.497 IFNG_E293_F 0.002 -0.274 TRIP6_P1274_R 0.007 -0.489 MKRN3_P108_F 0.002 -0.355 UGT1A1_P315_R 0.007 -0.335 NOTCH4_E4_F 0.002 -0.561 CCL3_E53_R 0.008 -0.261 WNT8B_E487_F 0.002 -0.226 CDK2_P330_R 0.008 -0.235 IFNG_P188_F 0.002 -0.205 HBII_52_E142_F 0.008 -0.253 IL8_E118_R 0.002 -0.666 SERPINE1_P519_F 0.010 -0.454 TMPRSS4_P552_F 0.002 -0.356 P2RX7_P597_F 0.012 -0.491 TRIP6_P1090_F 0.002 -0.504 PLA2G2A_P528_F 0.012 -0.210 GFAP_P56_R 0.002 -0.375 ACVR1_E328_R 0.015 -0.353 SERPINE1_E189_R 0.002 -0.251 CXCL9_E268_R 0.015 -0.205 CCL3_E53_R 0.002 -0.302 JAK3_P1075_R 0.015 -0.455 PADI4_P1011_R 0.002 -0.361 MMP2_P303_R 0.015 -0.291 CASP10_P186_F 0.002 -0.630 PDGFRA_E125_F 0.015 -0.266 SH3BP2_E18_F 0.002 -0.267 GFAP_P56_R 0.019 -0.312 ACVR1_P983_F 0.002 -0.321 EPM2A_P113_F 0.023 -0.205 BLK_P14_F 0.002 -0.273 PRSS1_E45_R 0.023 -0.329 SLC14A1_P369_R 0.002 -0.318 PRSS1_P1249_R 0.023 -0.389 SPP1_E140_R 0.002 -0.382 TJP2_P518_F 0.023 -0.335 MMP10_E136_R 0.003 -0.290 TRPM5_P721_F 0.026 -0.213 PRSS1_P1249_R 0.003 -0.393 BLK_P14_F 0.033 -0.217 TNK1_P221_F 0.003 -0.207 PTK7_E317_F 0.033 -0.219 MKRN3_E144_F 0.003 -0.230 NOTCH4_E4_F 0.049 -0.304 MPL_P657_F 0.003 -0.376 Grade 2 Astrocytoma 80 TNFSF10_E53_F 0.003 -0.458 CD82_P557_R 0.012 -0.298 S100A2_E36_R 0.003 -0.228 DDR1_P332_R 0.012 -0.310 SERPINE1_P519_F 0.003 -0.507 GFAP_P1214_F 0.012 -0.371 TJP2_P518_F 0.003 -0.347 LEFTY2_P561_F 0.012 -0.402 JAK3_P1075_R 0.004 -0.493 TRIP6_P1090_F 0.012 -0.458 KLK11_P103_R 0.004 -0.383 TRIP6_P1274_R 0.012 -0.476 ACVR1_E328_R 0.005 -0.304 UGT1A1_P315_R 0.012 -0.316 HBII_52_E142_F 0.005 -0.261 GFAP_P56_R 0.013 -0.450 IL16_P93_R 0.005 -0.312 GSTM2_P109_R 0.013 -0.260 S100A2_P1186_F 0.005 -0.454 MKRN3_E144_F 0.013 -0.347 ZNFN1A1_E102_F 0.005 -0.252 PTHR1_P258_F 0.014 -0.233 IL8_P83_F 0.005 -0.585 CDK2_P330_R 0.017 -0.232 NAT2_P11_F 0.005 -0.203 MPO_P883_R 0.018 -0.246 PI3_P1394_R 0.005 -0.367 MKRN3_P108_F 0.019 -0.499 SHB_P691_R 0.005 -0.254 ACVR1_E328_R 0.019 -0.321 EMR3_P39_R 0.006 -0.481 IL12B_P392_R 0.021 -0.301 KLK10_P268_R 0.006 -0.208 ACVR1_P983_F 0.025 -0.356 CD86_P3_F 0.008 -0.436 IL16_P93_R 0.025 -0.258 CSF3R_P8_F 0.009 -0.370 IL8_E118_R 0.026 -0.416 GSTM2_P453_R 0.010 -0.263 CASP10_P334_F 0.032 -0.269 NOS2A_P288_R 0.010 -0.264 SLC14A1_P369_R 0.037 -0.274 MBD2_P233_F 0.011 -0.254 P2RX7_P597_F 0.047 -0.230 Grade 3 MMP2_P303_R 0.011 -0.332 Oligoastrocytoma MMP9_P189_F 0.011 -0.413 ACVR1_P983_F 0.003 -0.428 PDGFRA_E125_F 0.011 -0.250 CD82_P557_R 0.003 -0.246 FGFR2_P460_R 0.013 -0.327 DDR1_P332_R 0.003 -0.349 PRSS1_E45_R 0.013 -0.304 GFAP_P1214_F 0.003 -0.377 ALPL_P433_F 0.016 -0.227 GSTM2_P109_R 0.003 -0.255 FGF1_E5_F 0.016 -0.213 LEFTY2_P561_F 0.003 -0.427 PADI4_P1158_R 0.016 -0.243 MKRN3_E144_F 0.003 -0.384 VAV1_P317_F 0.016 -0.214 MKRN3_P108_F 0.003 -0.615 MC2R_P1025_F 0.018 -0.243 MPO_P883_R 0.003 -0.346 CASP10_E139_F 0.022 -0.305 TRIP6_P1090_F 0.003 -0.540 NGFR_P355_F 0.022 -0.263 TRIP6_P1274_R 0.003 -0.528 PIK3R1_P307_F 0.022 -0.234 UGT1A1_P315_R 0.003 -0.451 STAT5A_E42_F 0.024 -0.298 CASP10_P334_F 0.004 -0.313 C4B_E171_F 0.027 -0.205 CDK2_P330_R 0.004 -0.247 CPA4_E20_F 0.027 -0.258 ACVR1_E328_R 0.005 -0.332 DDR2_E331_F 0.032 -0.246 IL16_P93_R 0.005 -0.381 HDAC1_P414_R 0.032 -0.522 GFAP_P56_R 0.007 -0.452 TNFSF10_P2_R 0.032 -0.369 PDGFRA_E125_F 0.007 -0.236 Secondary Glioblastoma PWCR1_P357_F 0.007 -0.204 81 CD82_P557_R 0.002 -0.270 IL8_E118_R 0.009 -0.521 CXCL9_E268_R 0.002 -0.441 TRPM5_P721_F 0.014 -0.211 DSG1_P159_R 0.002 -0.406 BLK_P14_F 0.040 -0.205 EMR3_P1297_R 0.002 -0.406 RUNX1T1_E145_R 0.040 -0.229 Grade 2 GABRA5_P1016_F 0.002 -0.339 Oligoastrocytoma GFAP_P1214_F 0.002 -0.408 ACVR1_E328_R 0.006 -0.403 GSTM2_P109_R 0.002 -0.251 CDK2_P330_R 0.006 -0.251 IFNG_E293_F 0.002 -0.326 DDR1_P332_R 0.006 -0.334 IFNG_P188_F 0.002 -0.447 GFAP_P1214_F 0.006 -0.275 IL8_P83_F 0.002 -0.615 GSTM2_P109_R 0.006 -0.237 ITK_P114_F 0.002 -0.312 IL16_P93_R 0.006 -0.360 JAK3_P1075_R 0.002 -0.562 IL8_E118_R 0.006 -0.497 KLK10_P268_R 0.002 -0.211 LEFTY2_P561_F 0.006 -0.369 KLK11_P103_R 0.002 -0.355 MKRN3_E144_F 0.006 -0.267 KRT1_P798_R 0.002 -0.297 TRIP6_P1090_F 0.006 -0.456 LEFTY2_P561_F 0.002 -0.489 TRIP6_P1274_R 0.006 -0.509 MPO_E302_R 0.002 -0.427 UGT1A1_P315_R 0.006 -0.385 MPO_P883_R 0.002 -0.633 MKRN3_P108_F 0.007 -0.488 PADI4_P1011_R 0.002 -0.647 CD82_P557_R 0.007 -0.295 PI3_P1394_R 0.002 -0.330 GFAP_P56_R 0.007 -0.444 PRSS1_E45_R 0.002 -0.498 CASP10_P334_F 0.007 -0.277 PRSS1_P1249_R 0.002 -0.664 PDGFRA_E125_F 0.007 -0.250 PSCA_P135_F 0.002 -0.468 ACVR1_P983_F 0.007 -0.341 PTHR1_P258_F 0.002 -0.560 P2RX7_P597_F 0.007 -0.284 PWCR1_P357_F 0.002 -0.238 MPO_P883_R 0.008 -0.258 SPI1_P929_F 0.002 -0.206 HTR2A_E10_R 0.010 -0.248 SPP1_P647_F 0.002 -0.291 PTHR1_P258_F 0.024 -0.235 Grade 2 TMPRSS4_P552_F 0.002 -0.520 Oligodendroglioma TRIP6_P1274_R 0.002 -0.539 ACVR1_E328_R 0.002 -0.420 UGT1A1_P315_R 0.002 -0.419 DDR1_P332_R 0.002 -0.302 WNT8B_E487_F 0.002 -0.369 GFAP_P1214_F 0.002 -0.243 CDK2_P330_R 0.002 -0.241 GSTM2_P109_R 0.002 -0.254 CSF3R_P8_F 0.002 -0.498 IL16_P93_R 0.002 -0.401 DDR1_P332_R 0.002 -0.328 MKRN3_E144_F 0.002 -0.207 IL16_P93_R 0.002 -0.421 MKRN3_P108_F 0.002 -0.515 MKRN3_P108_F 0.002 -0.496 TRIP6_P1274_R 0.002 -0.511 P2RX7_P597_F 0.002 -0.562 UGT1A1_P315_R 0.002 -0.426 PSCA_E359_F 0.002 -0.434 MPO_P883_R 0.002 -0.204 TNK1_P221_F 0.002 -0.254 CASP10_P334_F 0.003 -0.316 TRPM5_P721_F 0.002 -0.448 TRIP6_P1090_F 0.003 -0.429 HBII_52_E142_F 0.002 -0.468 LEFTY2_P561_F 0.004 -0.308 NOTCH4_E4_F 0.002 -0.604 ACVR1_P983_F 0.004 -0.259 82 CASP10_P334_F 0.003 -0.554 PDGFRA_E125_F 0.004 -0.256 HLA_DQA2_E93_F 0.003 -0.226 CDK2_P330_R 0.005 -0.233 ACVR1_E328_R 0.003 -0.370 SPP1_E140_R 0.005 -0.292 ACVR1_P983_F 0.003 -0.442 CD82_P557_R 0.007 -0.235 BLK_P14_F 0.003 -0.360 IL8_E118_R 0.008 -0.354 GFAP_P56_R 0.003 -0.471 PEG3_E496_F 0.008 -0.276 GLI2_E90_F 0.003 -0.320 GFAP_P56_R 0.009 -0.374 MKRN3_E144_F 0.003 -0.531 Ependymoma MPL_P657_F 0.003 -0.449 ACVR1_E328_R 0.007 -0.387 PLA2G2A_P528_F 0.003 -0.399 ACVR1_P983_F 0.007 -0.525 ZNFN1A1_E102_F 0.003 -0.267 CD82_P557_R 0.007 -0.291 PLA2G2A_E268_F 0.004 -0.277 CDK2_P330_R 0.007 -0.236 TNFSF8_P184_F 0.004 -0.219 DDR1_P332_R 0.007 -0.335 CCR5_P630_R 0.004 -0.271 FGF1_E5_F 0.007 -0.393 EMR3_P39_R 0.004 -0.288 FGF1_P357_R 0.007 -0.288 FGF7_P44_F 0.004 -0.308 FGFR2_P460_R 0.007 -0.413 CCL3_E53_R 0.005 -0.589 GFAP_P1214_F 0.007 -0.456 CD1A_P6_F 0.005 -0.215 GSTM2_P109_R 0.007 -0.263 EMR3_E61_F 0.005 -0.458 GSTM2_P453_R 0.007 -0.399 PTPRH_E173_F 0.005 -0.489 LEFTY2_P561_F 0.007 -0.466 SERPINE1_P519_F 0.005 -0.492 MMP14_P13_F 0.007 -0.561 TEK_P526_F 0.005 -0.228 MPL_P657_F 0.007 -0.453 TGFB1_P833_R 0.005 -0.334 RIPK1_P744_R 0.007 -0.446 TRIP6_P1090_F 0.005 -0.521 RIPK1_P868_F 0.007 -0.414 IL8_E118_R 0.006 -0.563 SERPINE1_P519_F 0.007 -0.504 ALPL_P433_F 0.007 -0.278 SH3BP2_E18_F 0.007 -0.330 HLA_DPA1_P28_R 0.007 -0.225 SPARC_P195_F 0.007 -0.209 NOS2A_P288_R 0.007 -0.501 TNK1_P221_F 0.007 -0.291 PDGFRA_E125_F 0.007 -0.288 TRIP6_P1274_R 0.007 -0.511 SH3BP2_E18_F 0.007 -0.300 UGT1A1_P315_R 0.007 -0.421 CSF3_P309_R 0.010 -0.302 FASTK_P598_R 0.008 -0.292 PEG3_E496_F 0.010 -0.245 MATK_P190_R 0.008 -0.294 CASP10_P186_F 0.012 -0.548 SERPINE1_E189_R 0.008 -0.262 DDR2_E331_F 0.012 -0.410 SHB_P691_R 0.008 -0.299 HLA_DPA1_P205_R 0.012 -0.351 SLC14A1_P369_R 0.008 -0.487 SLC14A1_P369_R 0.012 -0.361 FAS_P65_F 0.009 -0.237 IL6_P213_R 0.017 -0.306 CHI3L2_E10_F 0.013 -0.253 KIAA0125_E29_F 0.017 -0.245 GFAP_P56_R 0.013 -0.508 MMP2_P303_R 0.017 -0.329 HPN_P374_R 0.013 -0.527 SIN3B_P514_R 0.017 -0.201 IL1RN_P93_R 0.013 -0.335 SPP1_E140_R 0.017 -0.356 NAT2_P11_F 0.013 -0.335 AOC3_P890_R 0.021 -0.237 HPN_P823_F 0.015 -0.573 LEFTY2_P719_F 0.021 -0.268 ALPL_P433_F 0.020 -0.334 83 SLC22A18_P216_R 0.021 -0.201 TGFBI_P173_F 0.020 -0.295 TJP2_P518_F 0.021 -0.347 TJP2_P518_F 0.020 -0.375 VAMP8_P114_F 0.021 -0.337 TRIP6_P1090_F 0.020 -0.523 NAT2_P11_F 0.024 -0.322 IL8_E118_R 0.022 -0.560 PADI4_P1158_R 0.033 -0.364 MUC1_P191_F 0.022 -0.262 PDGFB_P719_F 0.033 -0.318 ZNFN1A1_E102_F 0.022 -0.203 CCKAR_P270_F 0.038 -0.453 IGF2AS_P203_F 0.026 -0.241 ZNF264_P397_F 0.038 -0.272 HTR2A_E10_R 0.029 -0.292 MBD2_P233_F 0.043 -0.300 RBP1_P426_R 0.029 -0.400 SHB_P691_R 0.043 -0.284 S100A2_P1186_F 0.029 -0.255 MMP10_E136_R 0.032 -0.367 TNFSF10_E53_F 0.032 -0.342 EPHA2_P203_F 0.035 -0.318 LCN2_P86_R 0.035 -0.332 SEPT5_P464_R 0.035 -0.243 * This column lists the Illumina GoldenGate methylation array annotation for CpGs where the gene name is listed first in all capital letters followed by an E for exon or P for promoter to indicate the location of the CpG relative to the transcription start site, and the number indicates the distance of the CpG from the transcription start site, and F indicates forward strand and R indicates reverse strand. 84 Supplementary Table 7. Statistically significantly differentially hypermethylated CpG loci in human gliomas. Median Median Δβ Q- Δβ GENE_ CpG* Q-value Value GENE_ CpG* value Value Grade 3 Primary Glioblastoma Oligoastrocytoma AATK_P519_R 0.002 0.214 AATK_P519_R 0.003 0.217 CD40_P372_R 0.002 0.310 ABCG2_P310_R 0.003 0.748 FZD9_E458_F 0.002 0.771 ALOX12_E85_R 0.003 0.647 IRAK3_E130_F 0.002 0.277 ALOX12_P223_R 0.003 0.359 IRAK3_P185_F 0.002 0.538 ATP10A_P147_F 0.003 0.603 MEST_E150_F 0.002 0.521 BMP4_P123_R 0.003 0.668 MEST_P4_F 0.002 0.477 BMP4_P199_R 0.003 0.231 SLIT2_P208_F 0.002 0.281 CCKAR_E79_F 0.003 0.377 TES_P182_F 0.002 0.675 CD40_E58_R 0.003 0.541 TNFRSF10A_P91_F 0.002 0.706 CD40_P372_R 0.003 0.314 TP73_P945_F 0.002 0.291 CD81_P272_R 0.003 0.599 CD81_P272_R 0.002 0.603 CD9_P504_F 0.003 0.582 GFI1_P45_R 0.002 0.455 CDH3_E100_R 0.003 0.720 HOXA9_P1141_R 0.002 0.748 CDH3_P87_R 0.003 0.764 MEST_P62_R 0.002 0.464 CDKN1B_P1161_F 0.003 0.226 TAL1_E122_F 0.002 0.250 CFTR_P115_F 0.003 0.238 TNFRSF10D_E27_F 0.002 0.664 COL18A1_P494_R 0.003 0.802 HTR1B_P222_F 0.002 0.570 CRIP1_P874_R 0.003 0.597 RAB32_P493_R 0.002 0.370 CTNNA1_P382_R 0.003 0.809 TNFRSF10A_P171_F 0.002 0.550 DDIT3_P1313_R 0.003 0.444 DIO3_P674_F 0.002 0.512 DES_E228_R 0.003 0.752 FLT3_E326_R 0.002 0.669 DNAJC15_P65_F 0.003 0.343 FLT4_E206_F 0.002 0.278 DSC2_E90_F 0.003 0.829 HOXA11_E35_F 0.002 0.472 EIF2AK2_P313_F 0.003 0.604 F2R_P839_F 0.002 0.416 ELK3_P514_F 0.003 0.585 HOXA5_P1324_F 0.002 0.329 EPHA2_P203_F 0.003 0.282 IRAK3_P13_F 0.002 0.252 EPHA2_P340_R 0.003 0.476 KIAA1804_P689_R 0.002 0.370 ERBB3_E331_F 0.003 0.814 MT1A_P600_F 0.002 0.245 ERCC6_P698_R 0.003 0.435 CD40_E58_R 0.002 0.318 ERN1_P809_R 0.003 0.252 PRKCDBP_E206_F 0.002 0.634 ESR2_E66_F 0.003 0.790 DNAJC15_P65_F 0.003 0.324 ESR2_P162_F 0.003 0.796 DSC2_E90_F 0.003 0.596 EYA4_P508_F 0.003 0.441 MAPK10_E26_F 0.003 0.204 EYA4_P794_F 0.003 0.369 ALOX12_P223_R 0.003 0.296 F2R_P839_F 0.003 0.438 DES_E228_R 0.003 0.221 FABP3_E113_F 0.003 0.592 HOXA9_E252_R 0.003 0.676 FAS_P65_F 0.003 0.618 85 ISL1_E87_R 0.003 0.236 FES_E34_R 0.003 0.842 MOS_E60_R 0.003 0.595 FES_P223_R 0.003 0.770 ALOX12_E85_R 0.004 0.587 FGFR3_P1152_R 0.003 0.365 BMP4_P199_R 0.004 0.222 FRZB_E186_R 0.003 0.805 HOXA11_P698_F 0.004 0.777 FZD9_E458_F 0.003 0.762 HTR1B_E232_R 0.004 0.412 GFI1_P45_R 0.003 0.535 RARRES1_P426_R 0.004 0.340 GLI3_P453_R 0.003 0.788 GATA6_P21_R 0.005 0.579 GNMT_E126_F 0.003 0.806 GFI1_E136_F 0.005 0.425 GNMT_P197_F 0.003 0.682 LY6G6E_P45_R 0.005 0.230 GP1BB_P278_R 0.003 0.261 SLC22A3_P528_F 0.005 0.203 GRB7_P160_R 0.003 0.389 TES_E172_F 0.005 0.412 GUCY2D_P48_R 0.003 0.507 ZNF215_P129_R 0.005 0.233 HFE_E273_R 0.003 0.796 HOXB2_P488_R 0.006 0.351 HHIP_P578_R 0.003 0.273 HIC_1_seq_48_S103 RARA_P176_R 0.006 0.349 _R 0.003 0.228 TUSC3_E29_R 0.006 0.676 HIC2_P498_F 0.003 0.432 DIO3_P90_F 0.006 0.286 HOXA9_E252_R 0.003 0.280 HIC_1_seq_48_S103_ R 0.006 0.210 HRASLS_P353_R 0.003 0.359 HOXA5_P479_F 0.006 0.213 HS3ST2_E145_R 0.003 0.596 TAL1_P594_F 0.006 0.794 ICA1_P61_F 0.003 0.330 GATA6_P726_F 0.008 0.565 ICA1_P72_R 0.003 0.759 GJB2_P931_R 0.008 0.249 IFNGR2_P377_R 0.003 0.217 TDGF1_P428_R 0.008 0.259 IGF1_E394_F 0.003 0.225 ISL1_P379_F 0.009 0.222 IGFBP2_P306_F 0.003 0.574 DIO3_E230_R 0.010 0.471 IL17RB_E164_R 0.003 0.813 RAN_P581_R 0.010 0.327 INSR_P1063_R 0.003 0.428 ZP3_E90_F 0.010 0.424 IRAK3_E130_F 0.003 0.260 DDIT3_P1313_R 0.011 0.370 IRAK3_P13_F 0.003 0.279 TWIST1_E117_R 0.011 0.265 IRF5_E101_F 0.003 0.435 MYOD1_E156_F 0.014 0.317 ITPR2_P804_F 0.003 0.754 NEFL_P209_R 0.014 0.616 JAK3_E64_F 0.003 0.625 IPF1_P750_F 0.016 0.439 JAK3_P156_R 0.003 0.425 EYA4_P794_F 0.018 0.214 KIAA1804_P689_R 0.003 0.745 GSTM1_P266_F 0.020 0.316 KIT_P367_R 0.003 0.488 BMP4_P123_R 0.022 0.420 KLK10_P268_R 0.003 0.492 HOXA9_P303_F 0.022 0.394 LOX_P313_R 0.003 0.781 PALM2_AKAP2_P420 _R 0.022 0.252 LY6G6E_P45_R 0.003 0.313 TFAP2C_P765_F 0.022 0.368 LYN_P241_F 0.003 0.786 JAK3_P156_R 0.024 0.340 MAP3K1_E81_F 0.003 0.822 SFN_P248_F 0.024 0.344 MAP3K1_P7_F 0.003 0.796 IPF1_P234_F 0.030 0.389 MATK_P190_R 0.003 0.532 86 MAP3K1_P7_F 0.030 0.450 MEST_E150_F 0.003 0.606 TNFRSF10C_P7_F 0.030 0.398 MEST_P4_F 0.003 0.495 BCR_P346_F 0.032 0.284 MEST_P62_R 0.003 0.525 SCGB3A1_E55_R 0.032 0.287 MET_E333_F 0.003 0.794 CAV2_E33_R 0.036 0.363 MGMT_P272_R 0.003 0.240 FZD9_P175_F 0.036 0.391 MGMT_P281_F 0.003 0.202 HOXA5_E187_F 0.036 0.259 MMP14_P13_F 0.003 0.323 ITPR2_P804_F 0.040 0.203 MOS_E60_R 0.003 0.694 TAL1_P817_F 0.040 0.254 MST1R_P392_F 0.003 0.225 TSP50_P137_F 0.040 0.259 MT1A_E13_R 0.003 0.567 CRIP1_P874_R 0.044 0.234 MT1A_P49_R 0.003 0.397 IGFBP1_P12_R 0.044 0.403 MYCL1_P502_R 0.003 0.753 HOXB2_P99_F 0.049 0.243 MYLK_P469_R 0.003 0.376 PYCARD_E87_F 0.049 0.308 NOTCH3_P198_R 0.003 0.733 WRN_P969_F 0.049 0.287 PAX6_P1121_F 0.003 0.518 Secondary Glioblastoma POMC_P400_R 0.003 0.404 AATK_P519_R 0.002 0.215 PRKCDBP_E206_F 0.003 0.732 ABCG2_P310_R 0.002 0.475 PYCARD_E87_F 0.003 0.710 ALOX12_E85_R 0.002 0.633 PYCARD_P393_F 0.003 0.276 ALOX12_P223_R 0.002 0.354 RAB32_E314_R 0.003 0.542 BMP4_P199_R 0.002 0.234 RAB32_P493_R 0.003 0.476 CD81_P272_R 0.002 0.651 RAN_P581_R 0.003 0.410 CFTR_P115_F 0.002 0.453 RARRES1_P426_R 0.003 0.549 CFTR_P372_R 0.002 0.631 RASSF1_E116_F 0.003 0.894 CRIP1_P274_F 0.002 0.465 RASSF1_P244_F 0.003 0.514 CRIP1_P874_R 0.002 0.522 RBP1_E158_F 0.003 0.889 DES_E228_R 0.002 0.662 RBP1_P150_F 0.003 0.797 DSC2_E90_F 0.002 0.696 RBP1_P426_R 0.003 0.504 EYA4_P794_F 0.002 0.337 SCGB3A1_E55_R 0.003 0.795 FZD9_E458_F 0.002 0.708 SEPT9_P374_F 0.003 0.223 GFI1_E136_F 0.002 0.729 SLC22A3_P528_F 0.003 0.276 GFI1_P45_R 0.002 0.545 TES_P182_F 0.003 0.666 HIC_1_seq_48_S103_ R 0.002 0.222 TGFB2_E226_R 0.003 0.830 HOXA9_E252_R 0.002 0.656 THBS1_E207_R 0.003 0.837 TIMP3_seq_7_S38_ IRAK3_E130_F 0.002 0.612 F 0.003 0.419 IRF5_E101_F 0.002 0.231 TJP1_P390_F 0.003 0.812 TNFRSF10A_P171_ JAK3_P156_R 0.002 0.369 F 0.003 0.566 KIAA1804_P689_R 0.002 0.761 TNFRSF10A_P91_F 0.003 0.727 MAPK10_E26_F 0.002 0.209 TNFRSF10D_E27_F 0.003 0.716 MEST_E150_F 0.002 0.563 TNFRSF10D_P70_F 0.003 0.238 MEST_P4_F 0.002 0.491 TRIP6_E33_F 0.003 0.752 87 MEST_P62_R 0.002 0.500 VAV2_P1182_F 0.003 0.360 MGMT_P272_R 0.002 0.225 ZMYND10_E77_R 0.003 0.696 MGMT_P281_F 0.002 0.394 ZMYND10_P329_F 0.003 0.653 MST1R_P392_F 0.002 0.222 ZP3_E90_F 0.003 0.745 MT1A_E13_R 0.002 0.531 CAV2_E33_R 0.004 0.817 MT1A_P49_R 0.002 0.517 CFTR_P372_R 0.004 0.698 MT1A_P600_F 0.002 0.245 CRIP1_P274_F 0.004 0.459 PRKCDBP_E206_F 0.002 0.697 GSTM1_P266_F 0.004 0.466 RAB32_E314_R 0.002 0.520 MAPK10_E26_F 0.004 0.233 RAB32_P493_R 0.002 0.457 MMP14_P208_R 0.004 0.424 RASSF1_E116_F 0.002 0.888 NGFR_P355_F 0.004 0.207 RASSF1_P244_F 0.002 0.463 PGF_P320_F 0.004 0.661 RBP1_E158_F 0.002 0.894 TAL1_P594_F 0.004 0.574 SYK_E372_F 0.002 0.348 WT1_P853_F 0.004 0.384 TAL1_P594_F 0.002 0.656 CCNA1_E7_F 0.005 0.626 TNFRSF10A_P171_F 0.002 0.558 CTNNB1_P757_F 0.005 0.308 TNFRSF10A_P91_F 0.002 0.716 E2F5_P516_R 0.005 0.361 TNFRSF10C_E109_F 0.002 0.381 ENC1_P484_R 0.005 0.409 TNFRSF10D_E27_F 0.002 0.713 EVI1_P30_R 0.005 0.719 TNFRSF10D_P70_F 0.002 0.240 GRB7_E71_R 0.005 0.226 COL18A1_P494_R 0.002 0.485 HTR1B_E232_R 0.005 0.580 CTNNA1_P382_R 0.002 0.315 IGFBP2_P353_R 0.005 0.227 ERBB3_E331_F 0.002 0.606 IRAK3_P185_F 0.005 0.383 ESR2_P162_F 0.002 0.750 TDGF1_P428_R 0.005 0.257 ZMYND10_E77_R 0.002 0.683 TNFRSF1B_P167_F 0.005 0.318 ERG_E28_F 0.002 0.687 BCR_P346_F 0.007 0.586 RARRES1_P426_R 0.002 0.524 GFI1_E136_F 0.007 0.650 TES_P182_F 0.002 0.608 IGFBP1_P12_R 0.007 0.476 THBS1_E207_R 0.002 0.825 SH3BP2_P771_R 0.007 0.447 CCNA1_E7_F 0.003 0.752 ZNF215_P129_R 0.007 0.674 DNAJC15_P65_F 0.003 0.367 CPA4_E20_F 0.009 0.249 ERN1_P809_R 0.003 0.216 FAS_P322_R 0.009 0.252 GNMT_E126_F 0.003 0.797 GRB10_P260_F 0.009 0.723 GP1BB_P278_R 0.003 0.241 IGFBP1_E48_R 0.009 0.537 HIC2_P498_F 0.003 0.262 TNFRSF10C_P7_F 0.009 0.404 HOXB2_P488_R 0.003 0.365 CDKN1C_P626_F 0.011 0.489 HS3ST2_E145_R 0.003 0.709 FGFR2_P266_R 0.011 0.324 ICA1_P61_F 0.003 0.319 GJB2_P931_R 0.011 0.373 TDGF1_P428_R 0.003 0.229 HDAC1_P414_R 0.011 0.242 GJB2_P931_R 0.003 0.313 HOXB2_P488_R 0.011 0.288 ERCC6_P698_R 0.004 0.289 ITGA6_P298_R 0.014 0.465 RBP1_P150_F 0.004 0.758 NR2F6_E375_R 0.014 0.239 SLC22A3_P528_F 0.004 0.284 PTCH2_E173_F 0.014 0.368 88 FRZB_E186_R 0.004 0.806 STAT5A_E42_F 0.014 0.218 HHIP_P578_R 0.004 0.464 TGFBI_P173_F 0.014 0.352 HOXA9_P303_F 0.004 0.304 CRK_P721_F 0.017 0.344 ICA1_P72_R 0.004 0.498 EVI1_E47_R 0.017 0.555 SCGB3A1_E55_R 0.004 0.783 JUNB_P1149_R 0.017 0.316 SFN_P248_F 0.004 0.288 MC2R_P1025_F 0.017 0.209 TJP1_P390_F 0.004 0.616 MCM2_P260_F 0.017 0.213 AREG_P217_R 0.005 0.411 MOS_P27_R 0.017 0.353 BMP4_P123_R 0.005 0.619 MT1A_P600_F 0.017 0.301 CD40_P372_R 0.005 0.318 PLAUR_E123_F 0.017 0.627 F2R_P839_F 0.005 0.441 PLSCR3_P751_R 0.017 0.377 GNMT_P197_F 0.005 0.666 TMEFF1_P626_R 0.017 0.235 GRB7_E71_R 0.005 0.399 AREG_P217_R 0.021 0.475 GRB7_P160_R 0.005 0.227 EPHB4_E476_R 0.021 0.224 GSTM1_P266_F 0.005 0.453 FGFR2_P460_R 0.021 0.279 HFE_E273_R 0.005 0.819 LAMC1_P808_F 0.021 0.372 HOXA9_P1141_R 0.005 0.733 PTK2_P735_R 0.021 0.311 HOXB2_P99_F 0.005 0.267 TFAP2C_P765_F 0.021 0.265 MAP3K1_E81_F 0.005 0.712 WRN_P969_F 0.026 0.309 MAP3K1_P7_F 0.005 0.691 COL1A2_P407_R 0.032 0.222 PYCARD_E87_F 0.005 0.693 HDAC5_E298_F 0.032 0.286 RAN_P581_R 0.005 0.383 RARRES1_P57_R 0.032 0.346 RBP1_P426_R 0.005 0.501 KCNK4_P171_R 0.040 0.303 TNFRSF10C_P7_F 0.005 0.454 FGF1_P357_R 0.048 0.208 ZMYND10_P329_F 0.005 0.615 HCK_P858_F 0.048 0.314 ITPR2_P804_F 0.006 0.631 MMP9_P237_R 0.048 0.221 RARRES1_P57_R 0.006 0.354 PADI4_P1158_R 0.048 0.208 Grade 2 ST6GAL1_P164_R 0.006 0.629 Oligoastrocytoma FES_E34_R 0.007 0.743 ALOX12_E85_R 0.006 0.614 ZP3_E90_F 0.007 0.385 CRIP1_P274_F 0.006 0.420 EPHA2_P340_R 0.009 0.330 FZD9_E458_F 0.006 0.669 TAL1_E122_F 0.009 0.285 LOX_P313_R 0.006 0.769 CD40_E58_R 0.010 0.458 MT1A_P600_F 0.006 0.275 CDH3_E100_R 0.010 0.648 RBP1_E158_F 0.006 0.725 CTNNB1_P757_F 0.010 0.207 SEPT9_P374_F 0.006 0.208 ELL_P693_F 0.010 0.362 CD40_E58_R 0.007 0.467 FRZB_P406_F 0.010 0.290 MEST_P4_F 0.007 0.473 HTR1B_E232_R 0.010 0.371 PRKCDBP_E206_F 0.007 0.676 JAK3_E64_F 0.010 0.432 AATK_P519_R 0.007 0.210 PAX6_P1121_F 0.010 0.258 CAV2_E33_R 0.007 0.580 WNT10B_P823_R 0.010 0.274 ERBB3_E331_F 0.007 0.698 HOXA11_P698_F 0.015 0.713 FAS_P65_F 0.007 0.593 89 CDH3_P87_R 0.017 0.589 GFI1_P45_R 0.007 0.297 EYA4_P508_F 0.017 0.373 GLI3_P453_R 0.007 0.592 FES_P223_R 0.017 0.667 ICA1_P61_F 0.007 0.380 FGFR3_P1152_R 0.017 0.256 IGFBP1_P12_R 0.007 0.539 HCK_P858_F 0.017 0.403 MATK_P190_R 0.007 0.433 MOS_E60_R 0.017 0.624 MT1A_E13_R 0.007 0.410 DIO3_E230_R 0.021 0.384 TAL1_P594_F 0.007 0.225 LYN_P241_F 0.024 0.626 ZP3_E90_F 0.007 0.371 MET_E333_F 0.024 0.284 ALOX12_P223_R 0.007 0.341 NOTCH3_P198_R 0.024 0.292 BMP4_P199_R 0.007 0.233 POMC_P400_R 0.024 0.341 CCNA1_E7_F 0.007 0.639 TGFB2_E226_R 0.024 0.777 COL18A1_P494_R 0.007 0.600 ATP10A_P147_F 0.028 0.314 DDIT3_P1313_R 0.007 0.434 CCKAR_E79_F 0.028 0.315 ESR2_P162_F 0.007 0.677 FAS_P65_F 0.028 0.589 MAP3K1_E81_F 0.007 0.810 HOXA5_P479_F 0.028 0.273 NOTCH3_P198_R 0.007 0.672 HRASLS_P353_R 0.028 0.312 RASSF1_E116_F 0.007 0.760 IMPACT_P234_R 0.028 0.341 RBP1_P150_F 0.007 0.679 TIMP3_seq_7_S38_F 0.028 0.308 BMP4_P123_R 0.007 0.645 EVI1_E47_R 0.038 0.237 CD81_P272_R 0.007 0.539 IRAK3_P185_F 0.038 0.595 DSC2_E90_F 0.007 0.701 PGF_P320_F 0.049 0.271 ENC1_P484_R 0.007 0.250 SGCE_E149_F 0.049 0.266 EYA4_P794_F 0.007 0.353 ST6GAL1_P528_F 0.049 0.647 F2R_P839_F 0.007 0.411 TAL1_P817_F 0.049 0.201 FES_P223_R 0.007 0.709 Grade 3 Astrocytoma GRB10_P260_F 0.007 0.553 AATK_P519_R 0.007 0.211 GRB7_P160_R 0.007 0.362 HIC_1_seq_48_S103 ABCG2_P310_R 0.007 0.728 _R 0.007 0.204 ALOX12_E85_R 0.007 0.616 JAK3_P156_R 0.007 0.380 ALOX12_P223_R 0.007 0.325 MYCL1_P502_R 0.007 0.646 CD40_P372_R 0.007 0.282 PAX6_P1121_F 0.007 0.322 CD81_P272_R 0.007 0.462 POMC_P400_R 0.007 0.376 CFTR_P372_R 0.007 0.692 RAB32_E314_R 0.007 0.413 COL18A1_P494_R 0.007 0.635 RASSF1_P244_F 0.007 0.436 CRIP1_P874_R 0.007 0.473 TES_P182_F 0.007 0.277 TIMP3_seq_7_S38_ CTNNA1_P382_R 0.007 0.726 F 0.007 0.441 DES_E228_R 0.007 0.517 TJP1_P390_F 0.007 0.756 DSC2_E90_F 0.007 0.804 TNFRSF10D_E27_F 0.007 0.696 EPHA2_P340_R 0.007 0.433 TNFRSF10D_P70_F 0.007 0.226 ERBB3_E331_F 0.007 0.866 ZMYND10_P329_F 0.007 0.654 ERN1_P809_R 0.007 0.217 ABCG2_P310_R 0.007 0.689 ESR2_P162_F 0.007 0.790 AREG_P217_R 0.007 0.446 90 EYA4_P794_F 0.007 0.312 CDKN1B_P1161_F 0.007 0.227 F2R_P839_F 0.007 0.403 CRIP1_P874_R 0.007 0.462 FRZB_P406_F 0.007 0.588 CTNNA1_P382_R 0.007 0.666 FZD9_E458_F 0.007 0.701 DES_E228_R 0.007 0.320 GFI1_E136_F 0.007 0.706 IGFBP2_P306_F 0.007 0.441 GFI1_P45_R 0.007 0.573 IL17RB_E164_R 0.007 0.716 GSTM1_P266_F 0.007 0.518 MEST_E150_F 0.007 0.464 HHIP_P578_R 0.007 0.379 MET_E333_F 0.007 0.691 HIC_1_seq_48_S103_ R 0.007 0.216 EPHA2_P340_R 0.008 0.438 HOXA9_E252_R 0.007 0.549 FES_E34_R 0.008 0.690 ICA1_P61_F 0.007 0.336 FGFR3_P1152_R 0.008 0.336 ICA1_P72_R 0.007 0.695 GFI1_E136_F 0.008 0.614 IFNGR2_P377_R 0.007 0.220 GUCY2D_P48_R 0.008 0.515 IL6_P611_F 0.007 0.336 IRF5_E101_F 0.008 0.222 IRF5_E101_F 0.007 0.202 KIAA1804_P689_R 0.008 0.609 ITPR2_P804_F 0.007 0.648 LY6G6E_P45_R 0.008 0.262 JAK3_P156_R 0.007 0.383 MEST_P62_R 0.008 0.475 KIAA1804_P689_R 0.007 0.700 RARRES1_P426_R 0.008 0.489 MAP3K1_E81_F 0.007 0.791 TNFRSF10A_P91_F 0.008 0.554 MAP3K1_P7_F 0.007 0.735 TNFRSF10C_P7_F 0.008 0.394 MEST_E150_F 0.007 0.466 CD9_P504_F 0.008 0.489 MOS_E60_R 0.007 0.646 CTNNB1_P757_F 0.008 0.264 MT1A_P600_F 0.007 0.249 ERN1_P809_R 0.008 0.216 MYLK_P469_R 0.007 0.355 ESR2_E66_F 0.008 0.411 PGF_P320_F 0.007 0.535 GSTM1_P266_F 0.008 0.375 PRKCDBP_E206_F 0.007 0.617 IRAK3_P185_F 0.008 0.472 TNFRSF10A_P171_ PTPN6_E171_R 0.007 0.248 F 0.008 0.526 RAB32_E314_R 0.007 0.552 BMPR2_P1271_F 0.009 0.371 RAB32_P493_R 0.007 0.422 CCKAR_E79_F 0.009 0.407 RAN_P581_R 0.007 0.378 ERCC6_P698_R 0.009 0.403 RARA_E128_R 0.007 0.313 EVI1_E47_R 0.009 0.416 RARA_P176_R 0.007 0.570 HHIP_P578_R 0.009 0.226 RARRES1_P426_R 0.007 0.514 HRASLS_P353_R 0.009 0.368 RASSF1_E116_F 0.007 0.826 TGFB2_E226_R 0.009 0.731 RBP1_E158_F 0.007 0.859 FABP3_E113_F 0.010 0.516 RBP1_P426_R 0.007 0.495 RAB32_P493_R 0.010 0.424 SCGB3A1_E55_R 0.007 0.733 RAN_P581_R 0.010 0.386 SYK_E372_F 0.007 0.258 RUNX1T1_P103_F 0.010 0.405 TAL1_E122_F 0.007 0.547 SH3BP2_P771_R 0.010 0.306 TES_P182_F 0.007 0.518 ZNF215_P129_R 0.010 0.462 TNFRSF10A_P171_F 0.007 0.542 ATP10A_P147_F 0.011 0.524 TNFRSF10A_P91_F 0.007 0.709 CD40_P372_R 0.011 0.294 91 TNFRSF10D_E27_F 0.007 0.712 HTR1B_E232_R 0.011 0.270 TNFRSF10D_P70_F 0.007 0.223 ICA1_P72_R 0.011 0.640 ZP3_E90_F 0.007 0.557 ITPR2_P804_F 0.011 0.579 BMP4_P199_R 0.008 0.228 MAP3K1_P7_F 0.011 0.760 CD40_E58_R 0.008 0.447 THBS1_E207_R 0.011 0.674 EPHA1_P119_R 0.008 0.509 CDH3_P87_R 0.012 0.515 ESR2_E66_F 0.008 0.767 FRZB_E186_R 0.012 0.684 HOXA11_P698_F 0.008 0.683 GNMT_E126_F 0.012 0.715 HOXA9_P1141_R 0.008 0.536 GNMT_P197_F 0.012 0.645 HS3ST2_E145_R 0.008 0.636 PYCARD_E87_F 0.012 0.652 MAPK10_E26_F 0.008 0.206 RBP1_P426_R 0.012 0.492 MT1A_P49_R 0.008 0.711 SCGB3A1_E55_R 0.012 0.706 POMC_P400_R 0.008 0.362 TRIP6_E33_F 0.012 0.641 PYCARD_E87_F 0.008 0.649 DNAJC15_P65_F 0.013 0.311 RASSF1_P244_F 0.008 0.420 ELK3_P514_F 0.013 0.523 TJP1_P390_F 0.008 0.861 EPHA2_P203_F 0.013 0.258 ZMYND10_E77_R 0.008 0.634 EYA4_P508_F 0.013 0.361 ZMYND10_P329_F 0.008 0.643 KLK10_P268_R 0.013 0.482 ZNF215_P129_R 0.008 0.606 SEMA3B_E96_F 0.013 0.223 CCNA1_E7_F 0.010 0.839 CFTR_P372_R 0.015 0.588 DDIT3_P1313_R 0.010 0.459 PGF_P320_F 0.015 0.413 ERCC6_P698_R 0.010 0.381 CPA4_E20_F 0.016 0.231 ERG_E28_F 0.010 0.743 EVI1_P30_R 0.016 0.269 FABP3_E113_F 0.010 0.547 GP1BB_P278_R 0.016 0.209 FES_E34_R 0.010 0.731 IGF1_E394_F 0.016 0.250 HDAC5_E298_F 0.010 0.222 MMP14_P13_F 0.016 0.222 MYCL1_P502_R 0.010 0.659 ERG_E28_F 0.018 0.394 SOX17_P303_F 0.010 0.235 GJB2_P931_R 0.018 0.243 TAL1_P594_F 0.010 0.624 JAK3_E64_F 0.018 0.564 TIMP3_seq_7_S38_F 0.010 0.543 LYN_P241_F 0.018 0.563 CFTR_P115_F 0.012 0.279 PTCH2_E173_F 0.018 0.297 DNAJC15_P65_F 0.012 0.261 VAV2_P1182_F 0.018 0.328 MEST_P62_R 0.012 0.444 JUNB_P1149_R 0.020 0.252 MOS_P27_R 0.012 0.209 ST6GAL1_P164_R 0.020 0.254 SEPT9_P374_F 0.012 0.205 BCR_P346_F 0.022 0.248 SOX17_P287_R 0.012 0.351 MYLK_P469_R 0.022 0.285 TNFRSF10C_P7_F 0.012 0.472 EIF2AK2_P313_F 0.024 0.562 BMPR2_P1271_F 0.015 0.277 HFE_E273_R 0.024 0.607 CRIP1_P274_F 0.015 0.417 HIC2_P498_F 0.024 0.228 CTNNB1_P757_F 0.015 0.318 IL6_P611_F 0.024 0.384 CTSL_P264_R 0.015 0.445 KIT_P367_R 0.024 0.313 GNMT_E126_F 0.015 0.751 PLAUR_E123_F 0.024 0.527 KIT_P367_R 0.015 0.374 ZNF215_P71_R 0.024 0.261 92 NOTCH3_P198_R 0.015 0.643 ITGA6_P298_R 0.026 0.252 PLSCR3_P751_R 0.015 0.400 MOS_E60_R 0.026 0.570 ACVR1C_P363_F 0.019 0.423 TGFBI_P173_F 0.026 0.226 CDKN1B_P1161_F 0.019 0.366 HCK_P858_F 0.028 0.306 EYA4_P508_F 0.019 0.510 MAPK10_E26_F 0.032 0.208 GLI3_P453_R 0.019 0.668 CDH3_E100_R 0.039 0.541 HOXA9_P303_F 0.019 0.313 TNFRSF1B_P167_F 0.039 0.252 HRASLS_P353_R 0.019 0.342 ACVR1C_P363_F 0.043 0.263 MEST_P4_F 0.019 0.472 TEK_E75_F 0.043 0.200 PTCH2_E173_F 0.019 0.282 MMP14_P208_R 0.046 0.246 RBP1_P150_F 0.019 0.695 MMP7_E59_F 0.046 0.208 SH3BP2_P771_R 0.019 0.378 TMEFF1_P626_R 0.050 0.216 Grade 2 TP73_P945_F 0.019 0.348 Oligodendroglioma CAV2_E33_R 0.023 0.273 AATK_P519_R 0.002 0.208 E2F5_P516_R 0.023 0.271 ABCG2_P310_R 0.002 0.678 FES_P223_R 0.023 0.705 ALOX12_E85_R 0.002 0.617 GJB2_P931_R 0.023 0.314 ALOX12_P223_R 0.002 0.341 GUCY2D_P48_R 0.023 0.261 ATP10A_P147_F 0.002 0.581 LYN_P241_F 0.023 0.508 BMP4_P123_R 0.002 0.646 NRG1_P558_R 0.023 0.324 BMP4_P199_R 0.002 0.230 PAX6_P1121_F 0.023 0.307 BMPR2_P1271_F 0.002 0.431 TNFRSF10C_E109_F 0.023 0.270 CAV2_E33_R 0.002 0.763 WNT10B_P823_R 0.023 0.295 CCKAR_E79_F 0.002 0.381 BMP4_P123_R 0.026 0.609 CCNA1_E7_F 0.002 0.662 CCKAR_E79_F 0.026 0.350 CD40_E58_R 0.002 0.378 CD9_P504_F 0.026 0.545 CD40_P372_R 0.002 0.307 CDH3_E100_R 0.026 0.717 CD81_P272_R 0.002 0.610 CDH3_P87_R 0.026 0.705 CD9_P504_F 0.002 0.527 ELK3_P514_F 0.026 0.339 CDH3_P87_R 0.002 0.706 EPHA2_P203_F 0.026 0.304 CDKN1B_P1161_F 0.002 0.419 FAS_P65_F 0.026 0.623 CFTR_P372_R 0.002 0.500 FGFR3_P1152_R 0.026 0.356 COL18A1_P494_R 0.002 0.725 FRZB_E186_R 0.026 0.782 CRIP1_P274_F 0.002 0.389 GNMT_P197_F 0.026 0.674 CRIP1_P874_R 0.002 0.449 GRB7_E71_R 0.026 0.708 CTNNA1_P382_R 0.002 0.675 GRB7_P160_R 0.026 0.388 CTNNB1_P757_F 0.002 0.319 HCK_P858_F 0.026 0.377 DDIT3_P1313_R 0.002 0.427 HFE_E273_R 0.026 0.776 DES_E228_R 0.002 0.551 HIC2_P498_F 0.026 0.354 DSC2_E90_F 0.002 0.707 HLA_F_E402_F 0.026 0.366 ELK3_P514_F 0.002 0.553 IRAK3_P185_F 0.026 0.490 EPHA2_P203_F 0.002 0.223 JAK3_E64_F 0.026 0.669 EPHB6_E342_F 0.002 0.213 93 LOX_P313_R 0.026 0.632 ERBB3_E331_F 0.002 0.714 MATK_P190_R 0.026 0.362 ERCC6_P698_R 0.002 0.444 MET_E333_F 0.026 0.684 ERN1_P809_R 0.002 0.226 MT1A_E13_R 0.026 0.534 ESR2_E66_F 0.002 0.733 TERT_P360_R 0.026 0.256 ESR2_P162_F 0.002 0.768 TGFB2_E226_R 0.026 0.838 EYA4_P508_F 0.002 0.478 THBS1_E207_R 0.026 0.782 EYA4_P794_F 0.002 0.361 TRIP6_E33_F 0.026 0.701 F2R_P839_F 0.002 0.349 ATP10A_P147_F 0.033 0.528 FES_E34_R 0.002 0.735 BCR_P346_F 0.033 0.273 FES_P223_R 0.002 0.700 IMPACT_P234_R 0.033 0.636 FGFR3_P1152_R 0.002 0.353 RUNX1T1_P103_F 0.033 0.497 FZD9_E458_F 0.002 0.697 AREG_P217_R 0.040 0.491 GFI1_P45_R 0.002 0.314 IGFBP1_P12_R 0.040 0.535 GLI3_P453_R 0.002 0.635 PLAUR_E123_F 0.040 0.391 GNMT_E126_F 0.002 0.739 RARRES1_P57_R 0.040 0.281 GNMT_P197_F 0.002 0.662 CEBPA_P1163_R 0.049 0.453 GP1BB_P278_R 0.002 0.212 COL1A2_E299_F 0.049 0.298 GRB7_E71_R 0.002 0.584 COL1A2_P407_R 0.049 0.245 HHIP_P578_R 0.002 0.312 COL1A2_P48_R 0.049 0.233 HRASLS_P353_R 0.002 0.382 GRB10_P260_F 0.049 0.614 ICA1_P61_F 0.002 0.398 Grade 2 Astrocytoma ICA1_P72_R 0.002 0.717 CD81_P272_R 0.012 0.551 IGFBP2_P306_F 0.002 0.437 ERBB3_E331_F 0.012 0.634 IL17RB_E164_R 0.002 0.794 FZD9_E458_F 0.012 0.641 INSR_P1063_R 0.002 0.507 TNFRSF10A_P91_F 0.012 0.655 IRAK3_E130_F 0.002 0.409 COL18A1_P494_R 0.013 0.625 IRAK3_P185_F 0.002 0.530 MEST_P4_F 0.013 0.447 IRF5_E101_F 0.002 0.301 ESR2_P162_F 0.014 0.778 JAK3_E64_F 0.002 0.512 DSC2_E90_F 0.015 0.716 JAK3_P156_R 0.002 0.379 GFI1_P45_R 0.015 0.384 KIAA1804_P689_R 0.002 0.691 ERCC6_P698_R 0.017 0.203 KIT_P367_R 0.002 0.450 MEST_E150_F 0.017 0.401 KLK10_P268_R 0.002 0.466 MT1A_P600_F 0.017 0.246 LOX_P313_R 0.002 0.775 TGFB2_E226_R 0.017 0.703 LYN_P241_F 0.002 0.634 TJP1_P390_F 0.017 0.726 MAP3K1_E81_F 0.002 0.809 ALOX12_E85_R 0.018 0.607 MAP3K1_P7_F 0.002 0.745 CFTR_P372_R 0.018 0.496 MATK_P190_R 0.002 0.478 CRIP1_P274_F 0.018 0.412 MEST_E150_F 0.002 0.485 TNFRSF10A_P171_F 0.018 0.523 MEST_P4_F 0.002 0.468 DES_E228_R 0.019 0.329 MEST_P62_R 0.002 0.473 ERN1_P809_R 0.019 0.212 MET_E333_F 0.002 0.764 94 RARRES1_P426_R 0.019 0.488 MT1A_P600_F 0.002 0.272 TES_P182_F 0.019 0.478 NGFR_P355_F 0.002 0.261 CAV2_E33_R 0.019 0.385 NOTCH3_P198_R 0.002 0.684 CCNA1_E7_F 0.019 0.617 NTSR1_P318_F 0.002 0.292 FABP3_E113_F 0.019 0.448 PAX6_P1121_F 0.002 0.388 HHIP_P578_R 0.019 0.233 PGF_P320_F 0.002 0.586 JAK3_P156_R 0.019 0.385 PLAUR_E123_F 0.002 0.448 TNFRSF10D_E27_F 0.019 0.689 POMC_P400_R 0.002 0.397 TNFRSF10D_P70_F 0.019 0.201 PRKCDBP_E206_F 0.002 0.669 ZMYND10_P329_F 0.019 0.639 PYCARD_E87_F 0.002 0.658 DDIT3_P1313_R 0.021 0.445 RAB32_E314_R 0.002 0.645 EYA4_P794_F 0.021 0.315 RAN_P581_R 0.002 0.359 PRKCDBP_E206_F 0.021 0.611 RARRES1_P426_R 0.002 0.478 SH3BP2_P771_R 0.021 0.415 RASSF1_E116_F 0.002 0.786 ABCG2_P310_R 0.023 0.686 RASSF1_P244_F 0.002 0.460 CTNNA1_P382_R 0.023 0.609 RBP1_E158_F 0.002 0.792 ICA1_P72_R 0.023 0.527 RBP1_P150_F 0.002 0.668 MATK_P190_R 0.023 0.438 RBP1_P426_R 0.002 0.488 RAB32_E314_R 0.023 0.420 SCGB3A1_E55_R 0.002 0.752 BMP4_P199_R 0.025 0.206 SEPT9_P374_F 0.002 0.225 HIC_1_seq_48_S103_ R 0.025 0.201 SYK_E372_F 0.002 0.227 IGF1_E394_F 0.025 0.254 TES_P182_F 0.002 0.358 LOX_P313_R 0.025 0.771 TGFB2_E226_R 0.002 0.774 MEST_P62_R 0.025 0.426 THBS1_E207_R 0.002 0.757 TIMP3_seq_7_S38_ PTCH2_E173_F 0.025 0.244 F 0.002 0.622 RBP1_E158_F 0.025 0.736 TJP1_P390_F 0.002 0.853 TNFRSF10A_P171_ RBP1_P150_F 0.025 0.635 F 0.002 0.542 RUNX1T1_P103_F 0.025 0.209 TNFRSF10A_P91_F 0.002 0.633 F2R_P839_F 0.026 0.377 TNFRSF10C_P7_F 0.002 0.441 GLI3_P453_R 0.026 0.508 TNFRSF10D_E27_F 0.002 0.711 LY6G6E_P45_R 0.026 0.296 TNFRSF10D_P70_F 0.002 0.229 MYCL1_P502_R 0.026 0.570 TRIP6_E33_F 0.002 0.673 THBS1_E207_R 0.026 0.719 WT1_P853_F 0.002 0.354 DNAJC15_P65_F 0.028 0.274 ZMYND10_E77_R 0.002 0.553 PGF_P320_F 0.028 0.467 ZMYND10_P329_F 0.002 0.664 SCGB3A1_E55_R 0.028 0.617 ZNF215_P129_R 0.002 0.688 ALOX12_P223_R 0.029 0.329 ZP3_E90_F 0.002 0.601 BMP4_P123_R 0.029 0.588 CPA4_E20_F 0.002 0.255 CD40_E58_R 0.029 0.372 FABP3_E113_F 0.002 0.409 HIC_1_seq_48_S103 GSTM1_P266_F 0.029 0.426 _R 0.002 0.209 HRASLS_P353_R 0.029 0.310 IGFBP2_P353_R 0.002 0.213 95 NOTCH3_P198_R 0.029 0.647 ITPR2_P804_F 0.002 0.544 POMC_P400_R 0.029 0.353 MOS_E60_R 0.002 0.665 RAB32_P493_R 0.029 0.371 PODXL_P1341_R 0.002 0.375 ZMYND10_E77_R 0.029 0.446 RAB32_P493_R 0.002 0.436 CTNNB1_P757_F 0.032 0.221 TGFBI_P173_F 0.002 0.279 GRB7_P160_R 0.032 0.311 CDH3_E100_R 0.003 0.720 RBP1_P426_R 0.032 0.477 GRB10_P260_F 0.003 0.390 CCKAR_E79_F 0.035 0.335 GUCY2D_P48_R 0.003 0.389 EPHA2_P340_R 0.035 0.345 HTR1B_E232_R 0.003 0.418 ATP10A_P147_F 0.037 0.453 LY6G6E_P45_R 0.003 0.318 RAN_P581_R 0.037 0.361 MT1A_P49_R 0.003 0.276 TNFRSF10C_E109_ EYA4_P508_F 0.040 0.468 F 0.003 0.292 JAK3_E64_F 0.040 0.556 CTSD_P726_F 0.003 0.321 KIAA1804_P689_R 0.040 0.552 DNAJC15_P65_F 0.003 0.335 MAP3K1_E81_F 0.040 0.769 MYCL1_P502_R 0.003 0.592 RARA_P176_R 0.040 0.503 PADI4_P1158_R 0.003 0.262 EPHA2_P203_F 0.044 0.295 SH3BP2_P771_R 0.003 0.375 GNMT_E126_F 0.044 0.660 TEK_E75_F 0.003 0.222 ITPR2_P804_F 0.044 0.486 TMEFF1_P626_R 0.003 0.288 MYLK_P469_R 0.044 0.249 IFNGR2_P377_R 0.003 0.200 PLAUR_E123_F 0.044 0.442 MMP14_P13_F 0.003 0.280 PYCARD_E87_F 0.044 0.646 SLC22A3_P528_F 0.003 0.231 RASSF1_E116_F 0.044 0.612 CASP6_P201_F 0.004 0.247 TNFRSF10C_P7_F 0.044 0.384 EIF2AK2_P313_F 0.004 0.722 AREG_P217_R 0.047 0.391 GFI1_E136_F 0.004 0.532 CRIP1_P874_R 0.047 0.415 HS3ST2_E145_R 0.004 0.305 IGFBP1_P12_R 0.047 0.515 PTCH2_E173_F 0.004 0.265 LYN_P241_F 0.047 0.410 AREG_P217_R 0.004 0.494 MAP3K1_P7_F 0.047 0.719 EPHA2_P340_R 0.004 0.289 RASSF1_P244_F 0.047 0.393 EVI1_P30_R 0.004 0.427 Ependymoma FAS_P65_F 0.004 0.591 EVI2A_P94_R 0.009 0.237 HFE_E273_R 0.004 0.627 RASSF1_E116_F 0.013 0.574 MMP2_P303_R 0.004 0.383 ERN1_P809_R 0.017 0.210 MYLK_P469_R 0.004 0.461 IFNGR2_P377_R 0.017 0.208 RUNX1T1_P103_F 0.004 0.247 TDGF1_P428_R 0.017 0.314 STAT5A_E42_F 0.004 0.258 PTPRO_P371_F 0.026 0.259 FRZB_E186_R 0.005 0.719 RAB32_P493_R 0.029 0.345 GRB7_P160_R 0.005 0.396 SPP1_E140_R 0.029 0.215 HIC2_P498_F 0.005 0.306 FZD9_E458_F 0.032 0.201 IGFBP1_E48_R 0.005 0.490 KLK10_P268_R 0.032 0.349 IMPACT_P234_R 0.005 0.466 RASSF1_P244_F 0.032 0.406 PYCARD_P393_F 0.005 0.218 96 TES_P182_F 0.032 0.234 VAV2_P1182_F 0.005 0.399 HOXA11_P698_F 0.041 0.252 AHR_P166_R 0.005 0.660 NFKB1_P496_F 0.041 0.236 CDKN1C_P626_F 0.005 0.252 CTSD_P726_F 0.045 0.296 FAS_P322_R 0.005 0.371 HCK_P858_F 0.005 0.323 IGFBP7_P297_F 0.005 0.229 LAMC1_P808_F 0.005 0.363 COL1A2_P407_R 0.006 0.256 HOXA11_P698_F 0.006 0.499 IGF2R_P396_R 0.006 0.272 ZNF215_P71_R 0.006 0.349 CD81_P211_F 0.007 0.360 GSTM1_P266_F 0.007 0.247 MMP14_P208_R 0.007 0.374 PLSCR3_P751_R 0.007 0.218 RARRES1_P57_R 0.007 0.260 MMP2_P197_F 0.008 0.272 TNF_P158_F 0.008 0.235 IGF1_E394_F 0.009 0.310 MLH3_P25_F 0.009 0.250 MT1A_E13_R 0.009 0.489 SEMA3B_E96_F 0.009 0.237 ERG_E28_F 0.010 0.338 EVI1_E47_R 0.012 0.638 JUNB_P1149_R 0.012 0.361 GJB2_P931_R 0.013 0.288 IGFBP1_P12_R 0.013 0.414 TNFRSF1B_P167_F 0.013 0.385 WRN_P969_F 0.013 0.294 ACVR1C_P363_F 0.015 0.378 BCR_P346_F 0.015 0.267 COL1A2_P48_R 0.015 0.275 FGFR2_P460_R 0.015 0.363 ITGB4_P517_F 0.015 0.410 FGFR2_P266_R 0.017 0.375 TNFRSF1B_E5_F 0.017 0.201 E2F5_P516_R 0.019 0.303 FGF1_P357_R 0.019 0.238 HOXA5_E187_F 0.019 0.321 PTPN6_E171_R 0.019 0.238 IL6_P611_F 0.021 0.319 PTK2_P735_R 0.027 0.467 CALCA_E174_R 0.030 0.225 97 TFAP2C_P765_F 0.037 0.250 CREB1_P819_F 0.041 0.248 CRK_P721_F 0.046 0.266 MAPK12_P416_F 0.046 0.214 SH3BP2_E18_F 0.046 0.216 SHB_P691_R 0.046 0.223 * This column lists the Illumina GoldenGate methylation array annotation for CpGs where the gene name is listed first in all capital letters followed by an E for exon or P for promoter to indicate the location of the CpG relative to the transcription start site, the number indicates the distance of the CpG from the transcription start site, and F indicates forward strand and R indicates reverse strand. 98 Supplementary Table 8. Cellular pathways enriched among statistically significantly differentially methylated CpG loci in gliomas with an IDH mutation compared to gliomas without IDH mutation*. Pathways enriched in IDH mutant gliomas P† Hypermethylated Protein Kinase A Signaling .05 Angiopoietin Signaling .06 RAN Signaling .10 Hypomethylated Methane Metabolism .03 Stilbene, Coumarine and Lignin Biosynthesis .03 Metabolism of Xenobiotics by Cytochrome P450 .03 PXR/RXR Activation .04 Retinol Metabolism .05 Phenylalanine Metabolism .06 Starch and Sucrose Metabolism .09 Pentose and Glucuronate Interconversions .09 Androgen and Estrogen Metabolism .10 * CpG loci with statistically significantly differential methylation (Q<0.05 and |Δβ|>0.2) between IDH wild-type and IDH mutant gliomas were examined for cellular pathway enrichment with Ingenuity pathways analysis software. RAN=RAN, member RAS oncogene family; PXR=nuclear receptor subfamily 1, group I, member 2; RXR=retinoid X receptor, gamma. † Two-sided Fisher’s exact test P value for enrichment of genes whose CpG loci are represented in among those in the listed pathways. 99 Supplementary Table 9. Recursively partitioned mixture model (RPMM) methylation class membership and glioma tumor grade and histology*. IDH Mutation Tumor Grade Tumor histology† Methylation Class No Yes 2 3 4 AS2 AS3 EP GBM GBM2 OA2 OA3 OD2 L 1 52 43 5 5 14 4 0 3 2 13 1 16 RLLL 5 0 1 0 4 0 0 1 3 1 0 0 0 RLLR 0 5 0 0 5 0 0 0 1 4 0 0 0 RLR 12 0 0 0 12 0 0 0 12 0 0 0 0 RRLL 5 0 5 0 0 0 0 5 0 0 0 0 0 RRLR 5 0 5 0 0 0 0 5 0 0 0 0 0 RRRLL 2 0 2 0 0 1 0 1 0 0 0 0 0 RRRLR 4 0 4 0 0 4 0 0 0 0 0 0 0 RRRR 4 0 4 0 0 0 0 2 0 0 2 0 0 P = 3.0x10-16‡ P < 2.2x10-16§ P < 2.2x10-16|| * Methylation classes from recursively partitioned mixture model (RPMM) of gliomas with IDH mutation data stratified by IDH mutation status, tumor grade, and grade-specific tumor histology, all statistical tests are two-sided. † AS2=grade 2 Astrocytoma, AS3=grade 3 astrocytoma, EP=ependymoma, GBM=primary glioblastoma multiforme, GBM2=secondary glioblastoma multiforme, OA2=grade 2 oligoastrocytoma, OA3=grade 3 oligoastrocytoma, OD2=grade 2 oligodendroglioma. Tumors were previously reviewed by UCSF neuropathologists to assign histologic subtypes and grades according to the World Health Organization classification. ‡ Fisher’s exact test P value for association between RPMM methylation class and IDH mutation status. § Fisher’s exact test P value for association between RPMM methylation class and tumor grade. || Fisher’s exact test P value for association between RPMM methylation class and grade-specific tumor histology. 100 References 1. Yu J, Zhang H, Gu J, Lin S, Li J, Lu W, et al. Methylation profiles of thirty four promoter-CpG islands and concordant methylation behaviours of sixteen genes that may contribute to carcinogenesis of astrocytoma. BMC Cancer 2004;4:65. 2. Stone AR, Bobo W, Brat DJ, Devi NS, Van Meir EG, Vertino PM. Aberrant methylation and down-regulation of TMS1/ASC in human glioblastoma. Am J Pathol 2004;165(4):1151-61. 3. Maegawa S, Itaba N, Otsuka S, Kamitani H, Watanabe T, Tahimic CG, et al. Coordinate downregulation of a novel imprinted transcript ITUP1 with PEG3 in glioma cell lines. DNA Res 2004;11(1):37-49. 4. Dallol A, Krex D, Hesson L, Eng C, Maher ER, Latif F. Frequent epigenetic inactivation of the SLIT2 gene in gliomas. Oncogene 2003;22(29):4611-6. 5. Harden SV, Tokumaru Y, Westra WH, Goodman S, Ahrendt SA, Yang SC, et al. Gene promoter hypermethylation in tumors and lymph nodes of stage I lung cancer patients. Clin Cancer Res 2003;9(4):1370-5. 6. Wiencke JK, Aldape K, McMillan A, Wiemels J, Moghadassi M, Miike R, et al. Molecular features of adult glioma associated with patient race/ethnicity, age, and a polymorphism in O6-methylguanine-DNA-methyltransferase. Cancer Epidemiol Biomarkers Prev 2005;14(7):1774-83. 101 Chapter 3 A novel approach to the discovery of survival biomarkers in glioma using a joint analysis of DNA methylation and gene expression Ashley A. Smith, Yen-Tsung Huang, Melissa Eliot, E. Andres Houseman, Carmen J. Marsit, John K. Wiencke, and Karl T. Kelsey In Review: Epigenetics; November 18, 2013 102 A novel approach to the discovery of survival biomarkers in glioblastoma using a joint analysis of DNA methylation and gene expression Ashley A. Smith1 , Yen-Tsung Huang2, Melissa Eliot2, E. Andres Houseman3, Carmen J. Marsit4,5, John K. Wiencke6, and Karl T. Kelsey1, 2 1 Department of Pathology and Laboratory Medicine, Brown University, Providence, Rhode Island, 02912 2 Department of Epidemiology, Brown University, Providence, Rhode Island, 02903 3 Department of Public Health, Oregon State University, Corvallis, Oregon, 97331 4 Department of Pharmacology and Toxicology, Geisel School of Medicine at Dartmouth, Hanover, New Hampshire, 03755 5 Department of Community and Family Medicine and Section of Biostatistics and Epidemiology, Geisel School of Medicine at Dartmouth, Dartmouth, New Hampshire, 03755 6 Department of Neurological Surgery, University of California San Francisco, San Francisco, California, 94143 Corresponding Author: Karl T. Kelsey, Department of Pathology and Laboratory Medicine, Brown University, Box G-E3, Providence, RI 02912. Phone: 401-863-6420 Fax: 401-863-9008 E-mail: Karl_Kelsey@brown.edu Running Title: Methylation and expression predict glioma survival Conflict of interest: The authors have no potential conflicts of interest to report. Funding: This work was funded by grants from the National Cancer Institute [CA100679, K.T.K.]; and the National Institutes of Health [CA126831, J.K.W] Abbreviations: GBM: glioblastoma multiforme; CNV: copy number variation; G-CIMP: Glioma CpG Island Methylator Phenotype; AFT: accelerated failure time; iBag: integrative Bayesian analysis; BH: Benjamini-Hochberg; FDR: false discovery rate; DF: degree-of-freedom 103 Abstract Glioblastoma multiforme (GBM) is the most aggressive of all brain tumors with a median survival under 1.5 years. Recently, epigenetic alterations have been found to play key roles in both glioma genesis and clinical outcome, demonstrating the need to integrate genetic and epigenetic data into predictive models. To enhance current models through discovery of novel predictive biomarkers, we employed a genome wide, agnostic strategy to specifically capture both expression-based (methylation-directed changes in gene expression) and alternative associations of DNA methylation with disease survival in glioma. Human GBM-associated DNA methylation, gene expression, IDH1 mutation status, and survival data were obtained from The Cancer Genome Atlas. DNA methylation loci and expression probes were paired by gene, and their subsequent association with survival was determined by applying an accelerated failure time model to previously published alternative and expression-based association equations. Significant associations were seen in 27 unique methylation/expression pairs with expression-based, alternative, and combinatorial associations observed (10, 13, and 4 pairs, respectively). The majority of the DNA methylation loci that were predictive were located within CpG islands, and all but three of the locus pairs showed negative correlations, suggesting that for most loci, the methylation/expression pairs were inversely related, consistent with methylation-associated gene regulatory action. Our results indicate that changes in DNA methylation are associated with altered survival outcome through both coordinate changes in gene expression and alternative mechanisms. Furthermore, our approach offers an alternative method of biomarker discovery using a 104 priori gene pairing and precise targeting to identify novel sites for loci-specific therapeutic intervention. Keywords: glioma, DNA methylation, gene expression, biomarker, mediation analysis 105 Introduction Glioblastoma multiforme (GBM) is the most aggressive of all brain tumors, and accounts for approximately 70% of all malignant gliomas. 1 Despite current treatments, patients with GBMs have a median survival of only 12-15 months.1 This disease is thought to result from the outgrowth of clonal populations that harbor a combination of somatic gene alterations that are likely complex.1 Genetic alterations include dysregulation of many angiogenic and proliferative pathways including amplification of EGFR and overexpression of VEGF.1 In addition, dysregulation in many members of the PI(3)K /Akt/RAS signaling pathway have also been implicated in the disease. 1 In 2006, Phillips et al used these genetic alterations, as well as copy number variations (CNV), to distinguish subclasses of GBM, which had prognostic implications.2 These analyses were further supported by several studies that assessed known, prevalent mutations in GBMs (EGFR, PTEN, IDH1, TP53, and NF1), copy number alterations, and expression changes in an integrative approach in order to more precisely define of GBM subtypes important for survival prediction. These data and approaches strongly support the hypothesis that GBMs harbor a complex combination of somatic alterations that determine their phenotype. 3, 4 Recently, Frattini et al (2013), used a novel statistical approach in an attempt to identify drivers of gliomagenesis through integration of somatic mutations and CNV. 5 They classified three types of GBM: 1) GBM having deletions at sites containing mutations, 2) GBM having amplifications at sites containing mutations, and 3) GBM with recurrent mutations and no alteration in CNV.5 They also identified fusion products involving the EGFR-SEPT14 loci. Their integrative analysis further added to the genetic understanding 106 of GBM pathogenesis as well as marked specific targets for possible therapeutic intervention. 5 Epigenetics (particularly DNA methylation) also plays an important role in gliomagenesis and glioma survival. Gene promoter DNA methylation has long been associated with gene silencing and research has now identified a role for methylation in selecting alternate transcripts and gene promoters, giving rise to somatic events that can impact disease survival. 6-10 11, 12 Our group and others have reported an association between isocitrate dehydrogenase 1 and 2 (IDH1/2) mutations and a hypermethylator phenotype in gliomas that is associated with early age of onset and increased patient survival, specifically in lower grade gliomas and secondary GBM.6, 15 Our data, which looked at a TCGA independent population, also demonstrated and association between TP53 and G- CIMP and a lack of association between EGFR and G-CIMP, and an overall increase in methylation genome-wide. 16 DNA methylation does not act solely through the mediation of gene expression (the mechanism that we designate as an expression-based association). DNA methylation has also been found to associate with chromosomal instability, the induction of splice variants, alterations in enhancer regions, changes in microRNA binding regions and expression control regions, and mutations. These somatic changes (which we designate as an alternative association) could also greatly impact survival, but are much less well studied.6-10 These reports have highlighted the crosstalk between various types of carcinogenic somatic alterations and the need for a better understanding of the complex nature of the 107 pattern of somatic gene inactivation, involving genetic and epigenetic alterations that impact upon both the genesis of and survival from glioma. Although there has been a call for these integrative biomarkers that can sharpen predictive tools, most research has focused on the integration of genetic alterations (e.g. mutations) and their association with survival.5, 18, 19 Here, we have made use of The Cancer Genome Atlas (TCGA) data sets to test our bioinformatics-based approach for identifying novel biomarkers of phenotypically important relationships among DNA methylation, gene expression, and survival in GBM. Results DNA methylation and gene expression are significantly associated in GBM samples After removal of all IDH1 mutant samples and replicates to prevent survival bias, the final phase 1 and phase 2 datasets contained n=73 and n=168 samples, respectively. Patient demographic data for all 241 GBM samples are presented in Table 1. Expression and methylation loci were paired by gene symbol for all 241 samples, resulting in a total of 66,202 unique methylation and expression pairs, which were used for the following analysis. In order to ensure functionality of methylation loci in the following analysis an initial screen was conducted to determine the association of methylation and expression with in the same gene. To identify the methylation loci that regulate gene expression level, a linear model, as specified in Equation 2 (see Materials and Methods), was performed using the combined phase 1 and phase 2 datasets (n=241). Pairs were chosen as significant if they had a q-value <0.05. Out of all 66,202 corresponding loci for both expression and methylation, 9821 were found to be significantly associated with each 108 other (84.3% negatively correlated, 15.7% positively correlated). Samples were then separated back into the original phase 1 (n=73) and phase 2 (n=168) sets for survival analysis. DNA methylation and gene expression pairs are significantly associated with patient survival in GBM samples To determine DNA methylation and gene expression pairs that are not only significantly associated with each other, but also significantly associated with survival, a Cox proportional hazards model was run on phase 1, phase 2, and pooled datasets. We used the Cox model to investigate the effect of gene expression, DNA methylation, and their interaction term on survival, adjusting for age, gender, and study. ‘Study’ was included in as a model variable as a precautionary measure due to the inherent difference in how the presence of IDH1 mutation was determined for each of the two datasets. As previously mentioned, tumors with a G-CIMP phenotype or IDH mutation were removed from this analysis due to their association with increased survival in GBM patients. Analysis of the phase 1 data set (n=73) yielded 878 pairs (from the original 9821) that were significantly associated with survival (p<0.05). Those 878 pairs were re-run using the phase 2 data set (n=168) using the same model, which reveals 100 pairs with p<0.05. Finally, we assessed effects of the 100 pairs on overall survival using the pooled dataset (n=241) (Supplementary Material, Table S1). Pairs significantly correlated with survival were chosen based on the q-value (BH) of the pooled model (cutoff: q<0.10). A total of 36 unique methylation/expression pairs from 29 genes were significantly associated with 109 survival. Of these 36 unique pairs, CpG locus cg23134520 was found to contain a SNP (rs6032566) and was removed from further analysis. This yielded 35 unique methylation/expression pairs from 28 different genes, which were used for the final mediation analysis (Table 2). Association of methylated loci with survival can be decomposed into i) those whose action is mediated through expression and ii) those whose association with survival is not mediated in this fashion. We first estimated the association of DNA methylation with survival mediated through presumptive effect on gene expression (expression-based association) and then assessed the association not directly mediated through gene expression (alternative association). The expression-based and alternative associations of paired loci with survival were estimated for the top 35 unique methylation/expression pairs (chosen from the linear model and Cox proportional hazards model) by using an accelerated failure time (AFT) model (see Supplementary Material, Table S2). This yielded 10 unique methylation/expression pairs where expression-mediated methylation was associated with survival outcome (or significant expression-based associations) (Fig. 1A), 13 methylation/expression pairs where methylation did not work through expression of the same gene to effect survival (significant alternative association) (Fig. 1B), and 4 methylation/expression pairs where methylation exerted its effect on survival outcome directly and through gene expression (both significant alternative and expression-based associations) (Fig. 1C). Of the 27 significant methylation and expression pairs, 22 DNA 110 methylation loci were located within a CpG Island and, in general, pairs within the same gene had similar effects on survival (Fig. 1 A-C). In addition, all but three of the locus pairs (associated with CACNB1, RFXANK, and RAB21), had negative correlations, suggesting that the majority of the methylation/expression pairs were inversely related (see Supplementary Material, Fig. S2). Additionally, exon locations of methylation loci from significant pairs can be seen in supplementary material, Fig. S3 Discussion The association of alterations in DNA methylation and gene expression in GBM with disease survival has been a major focus of recent studies, as it is apparent that outcome is not solely driven by somatic mutation. These previous studies generally identified loci whose methylation was inversely correlated with expression and examined that impact of those loci on patient outcome. Uniquely, in our study, we focused upon methylation and attempted to classify the effects of methylation on survival into those mediated by expression and those not mediated by expression, thereby expanding the potential biomarker pool. In 2013, Wang et al used an integrative Bayesian analysis (iBAG) approach to analyze the association of DNA methylation with changes in gene expression and subsequently evaluated the association of changes in gene expression on GBM survival.21 This linear approach was able to identify several genes with significant associations of gene expression modulated by methylation. Consistent with this data, several genes that we identified to be significantly modulated by DNA methylation, including OSMR, STEAP1, 111 21 and GRB10, were also reported by Wang et al in their findings. However, methylation not only exerts its effects on survival through expression of its associated gene, but also can operate through a variety of other mechanisms, including chromosomal fragility/instability, splicing variants, enhancer regions, and dysregulation of microRNA. 6-10 Etcheverry et al (2010) investigated the impact of DNA methylation on gene expression and outcome in GBM.22 Their analysis focused on the relationship between DNA methylation and gene expression and the association of methylation on survival. They identified 421 CpG sites that were significantly inversely correlated between methylation and expression, 291 of these CpG sites matched what we found to be correlated in our analysis. They also identified 13 genes, that appeared to have consistent differential methylation and expression (between GBM and control brain) but were negatively correlated, suggesting that the regulation of these genes may be epigenetically modulated.22 However, Wang et al did not consider the joint effect of methylation and expression on outcome. In addition, IDH1 mutant-associated samples were removed from our study to ensure that the final results would not reflect a bias toward the IDH1 hypermethylator phenotype due to its association with increased survival.6 Our final model focuses not only on how methylation acts through expression to associate with survival; but also assesses how methylation can associate with survival directly or as a proxy for alternative mechanisms (Fig. 2). The final 27 significant methylation/expression pairs (contain genes associated with invasion, angiogenesis, and metabolism, and many have been previously linked to brain/glioma (Table 3). Of the 20 genes that contained the significant pairs, to our knowledge none are associated with common amplifications and deletions found in GBM.23 Ten pairs (from seven genes) had 112 a significant expression-based association with survival, suggesting that DNA methylation in these genes affects survival outcome via gene expression of the associated gene. Interestingly, two genes contained multiple significant methylation/expression pairs. One of these genes, oncostatin M receptor (OSMR), contained two significant pairs, both with the same gene expression probe, but paired with different DNA methylation loci. The DNA methylation loci for these pairs fall in a CpG island within 550 bp of the transcription start site of the OSMR gene and the pairs showed a negative correlation, suggesting that methylation of these loci could inhibit gene expression. The locus pairs (cg03138091_A_24_P388860 and cg26475085_A_24_P388860) were associated with a significant expression-based association for each CpG. It is known that OSMR beta associates with Interleukin 31 Receptor alpha (IL31RA) to form the Interleukin 31 receptor (IL31) complex which activates signal transducer and activator of transcription 3 (STAT3).24 Priester et al (2013) recently demonstrated that silencing of STAT3 inhibits glioma single cell infiltration and tumor growth, suggesting that STAT3 plays an important role in the invasiveness of gliomas.25 If OSMR is silenced via DNA methylation of its promoter, this could lead to a decrease in OSMR gene expression and its association with IL31RA, inhibiting the subsequent activation of STAT3. Without activated STAT3, GBM growth and infiltration could be attenuated, potentially causing an increase in survival. This proposed mechanism supports the expression-based association of OSMR methylation on survival in the present study. In addition to the 10 pairs with significant expression-based associations, there were also 14 methylation/expression pairs (in 12 genes) with significant alternative associations. This suggests that in these genes, DNA methylation is associated with survival either 113 directly or through mechanisms other than direct changes in gene expression. For instance, aquaporin 1 (AQP1) contained one methylation/expression pair, which is located in a CpG island within 300 bp of the transcription start site of the AQP1 gene, and the pair showed a negative correlation, suggesting that methylation of this locus could inhibit gene expression. The major function of aquaporins (AQPs) is transportation of water across cell membranes, the disruption of which has been shown to disturb the blood-brain barrier and lead to cerebral edema. 26-28 AQP1 and AQP4 are most abundantly expressed in the nervous system, and though AQP4 has been more heavily studied, the expression of both has been observed in GBM and found to correlate with malignancy, specifically with cytotoxic cerebral edema, angiogenesis, and migration/invasion.26, 29, 30 Recently, it has been shown that both AQP1 and AQP4 are direct targets of microRNA 320a (miR-320a) and that increased miR-320a is associated with a reduction in AQP1/4 expression.31 Therefore, a possible mechanistic explanation for the alternative association we observe involves methylation of the microRNA target region on AQP1 inhibiting the binding of miR-320a and ultimately allowing transcription of AQP1. Interestingly, there were four methylation/expression pairs (three genes) that had both significant alternative and expression-based associations. Of interest is the gene growth factor receptor-bound protein 10 (GRB10), which contained two significant pairs, both with the same DNA methylation locus but paired with different gene expression probes. The DNA methylation locus for these pairs fall in a CpG island of the GRB10 gene, and the pairs showed a negative correlation. The loci pairs (cg24302095_A_24_P235266 and cg24302095_A_24_P235268) have significant alternative associations that suggest that with a 5% increase in methylation, a decrease in survival may be observed; but the pairs 114 also have significant expression-based associations. GRB10 is an imprinted gene that is differentially expressed from two promoters. In the brain, it is paternally expressed.32 GRB10 interacts with receptor tyrosine kinases and signaling molecules, most commonly insulin receptors and insulin-like growth factor receptors. 32, 33 In addition, monoallelic expression appears to be limited to fetal brain, skeletal muscle, and, most recently, placenta.32, 33 Not only is expression of GRB10 tissue specific, but it is also isoform specific.32 Currently, 13 different splice variants of GRB10 have been identified, with all but one being expressed in the brain.33 Overexpression of some isoforms has been shown to suppress growth.32 Yonghao et al (2011) found decreased expression of GRB10 in many human tumor types, including gliomas, compared to corresponding normal tissue. 34 These tumor samples demonstrated a negative correlation between GRB10 and PTEN expression. Furthermore, in a murine cell line, stabilization of Grb10 due to mTORC1 - mediated phosphorylation resulted in inhibition of PI3K and ERK-MAPK pathways, suggesting a role for Grb10 as a tumor suppressor. 34 Conversely, Nord et al (2009), using a 32K bacterial artificial chromosomes array, found human GRB10 to be a putative novel oncogene in glioblastoma.35 Mechanistic differences might be attributed to inherent imprinting differences in GRB10 between mice and humans. Nonetheless, DNA methylation of this CpG locus has the potential to cause alternative splice sites and may be responsible for the different isoforms of GRB10. Therefore, it is plausible that both the alternative and expression-based associations of this gene have a significant outcome on survival. Further potential mechanisms for genes that contained significant pairs can be found in Table 3. 115 There were several limitations to our work. First, we relied upon publically available data, which did not have complete mutation and survival data. We used a previously validated approach to control for this, but this remains a limitation 9 7 . To address the issue of missing survival data we used an accelerated failure time model to predict the survival time of censored values. In order to ensure functionality of methylation loci in our analysis, an initial screen was conducted, and only methylation and expression pairs that were significantly correlated within the same gene were used. Due to limited patient data, our study consisted only of primary GBM; however, promoter methylation of many GBM associated genes is more common in secondary GBM (ie. 11% promoter methylation for MGMT 36 ), which may explain the lack of detection of previously described genes associated with promoter methylation in glioma. Additionally, there was one pediatric patient out of the 241 samples (age 10) that was not removed from the study prior to analyses. Our approach focuses on methylations that regulate expression of the same gene, as mentioned above, and would miss methylation loci that do not regulate gene expression and are associated with survival through the alternative mechanism. To establish no association with gene expression, difficulties such as distinguishing null findings due to severe multiple comparisons from those with true biology will be an issue. Our approach circumvents the difficulty and is driven by biology: methylation that regulates gene expression is more likely to be functional and thus affects cancer survival. Overall, our findings are consistent with the well-accepted concept that DNA methylation can associate with survival outcome via alterations in gene expression (e.g., OSMR). Our findings also suggest that methylation can associate with survival outcome through 116 mechanisms other than dysregulation of gene transcription, including disruption of microRNA function, as possible in the case of AQP1. Additionally, some methylation/expression pairs have both significant alternative and expression-based associations, suggesting that different tumors are using discrete mechanisms, yielding different survival outcomes, as described for the proposed alternative and expression- based associations of GRB10. It should be noted that promoter methylation of MGMT, which is frequent in low-grade and secondary GBM11, 12,was observed to be significantly correlated with MGMT gene expression (data not shown), but was not observed in our final list of significant pairs. This may be attributable to the data quality (e.g. treatment data), or the relatively large number of subjects required to detect an interaction between treatment and methylation at this locus. Importantly, our data suggest that this approach might profitably be applied to cancers other than GBM. Our method also brings to light pathways for future study as potential mechanisms in the pathogenesis of glioma. Though additional validation is needed, our work supports the concept that DNA methylation can function both through gene expression, and more directly or through alternative mechanisms, to modulate survival outcomes among glioblastoma patients. Materials and Methods: External Data Sets Methylation, expression, and mutation data for glioblastoma multiforme (GBM) were downloaded from The Cancer Genome Atlas (TCGA) for two different sample sets. 117 Level 1 HumanMethylation27 (Illumina) DNA methylation data and level 2 AgilentG4502A_07_1 and 2 gene expression data were downloaded for all available GBM batches. GBM batches 1, 2, 3, and 10 were used as the phase 1 set and GBM batches 16, 20, 26, 38, and 62 were used as the phase 2 data set. Patient samples lacking covariate data were removed; samples were further restricted to patients diagnosed with glioblastoma who were alive 30 days after their date of diagnosis. Data sets were not combined in further analyses due to the fact that phase 2 data did not have definitive IDH mutation status. Since IDH mutations are associated with survival we were hesitant to combine the two datasets as mis-identification of IDH mutations could grossly affect findings. Recursively partitioned mixture model to determine IDH1 mutation status Patient survival, DNA methylation, gene expression, and IDH1 mutation data (phase 1 set only), was obtained for primary glioblastoma multiforme (GBM) samples. It has been widely acknowledged that IDH1 mutants are almost exclusively associated with a hypermethylater (G-CIMP) phenotype, and this phenotype is associated with increased survival in glioma.10,11 Therefore, we wanted to remove IDH mutant samples from our study so results would not be biased due to increased survival associated with this mutation. Since IDH mutation data was not available for the phase 2 sample set, we employed a recursively partitioned mixture model (RPMM) as described by Houseman et al 20 and used in Christensen and Smith et al.6 The RPMM successfully divided the phase 1 set into seven classes (see Supplementary Material, Fig. S1), and the samples in the top two most highly methylated classes, along with the samples having IDH mutations in the 118 phase 1 set, were removed (TCGA.14.1458, TCGA.16.1460, TCGA.19.1788, TCGA.14.1456, TCGA.28.1756, TCGA.14.4157, TCGA.32.4208). Methylation Data Methylation beta values were extracted from raw idat files using GenomeStudio software (Illumina), which calculates beta values using M/(M+U+100), where M is the methylated signal, U is the unmethylated signal, and 100 is an arbitrary offset. Replicates that did not correlate were removed (TCGA.06.0137, TCGA.06.0145). For methylation loci, all loci that contained a detection p-value > 0.05 for any sample were removed from further analysis. Since approximately 25% of the survival data is censored, censored survival times were estimated using an accelerated failure time (AFT) model based on the equation below. Equation 1. ( ) ( ) ( ) Where T follows a Weibull distribution 37 (μ is a scale parameter and follows an extreme value distribution). Next, methylation values were normalized for bead chip to control potential batch effect using the ComBat method 38 with adjustment of age, gender, survival, censored data, and survival-censored interaction. Expression Data TCGA expression and methylation subject identification numbers were matched; all non-matching samples were removed from the datasets. Replicates in expression samples were either averaged or chosen based on the closest mean and standard deviation to the 119 methylation distribution across all samples. The final data sets consist of a phase 1 dataset (n=73) and a phase 2 dataset (n=168) that contain complete data on overall survival, DNA methylation and gene expression with samples considered G-CIMP removed. Final methylation/expression locus pairs Methylation and expression loci were merged based on gene of origin. Annotation files for both platforms (HumanMethylation27 and AgilentG4502A_07_1 and 2) were downloaded from TCGA and matched by gene symbol, (using the manufacturer’s annotation) yielding 66,202 methylation/expression pairs. It should be noted that there are usually several methylation loci and/or expression probes found within each gene, so while each pair is unique upon merging, an individual methylation or expression locus may be repeated among several pairs. Statistical Analysis To choose statistically significant methylation and expression pairs, expression was regressed on methylation in the pooled (n=241) dataset. The associated p-values were adjusted for false discovery rate (FDR) using the Benjamini-Hochberg (BH) procedure.39 All methylation/expression pairs that had a q-value <0.05 were identified as being significantly associated with each other (n=9821 pairs). To further siphon out statistically significant pairings, pairs were then assessed using a Cox proportional hazards model for the effect of expression, methylation, and their interaction on survival, controlling for age, gender, and study (when applicable). A three degree-of-freedom (DF) Chi-square test was performed to test for significance of 120 expression, methylation, and their cross-product interaction. The three-DF models were repeated for both phase 1 (n=73) and phase 2 datasets (n=168) separately and the pooled dataset (n=241). In order to reduce false positives, final statistically significant pairs were selected for having p-values <0.05 in both phase 1 and phase 2 datasets and q-values of <0.1 in the pooled dataset. The associations of methylation and expression on survival were determined by a mediation analysis adopted from VanderWeele 37 using the following equations for the expression-based and alternative associations of methylation on survival: Equation 2. [ | ] Equation 3. ( ) ( ) Equation 4. ( )( ) Equation 5. { ( ))( ) ( ), where T is survival time, E is expression, M is methylation, c is study, σ2 is the variance of the error term in Equation 2, is a random error in Equation 3 following the extreme value distribution, and is a scale parameter. For our purposes, m* is median methylation and (m-m*) is the change in methylation we are interested in observing. For example, we would set m-m* to 0.05 if we wanted to look at the change in survival for a 5% increase in methylation. Equation 2 represents the linear model for the association between expression and methylation, and Equation 3 represents the accelerated failure 121 time model with interaction between methylation and expression. 0 -2 are the regression parameters for the linear model, and θ 0 -θ4 are the regression parameters for the accelerated failure time model. We used a stepwise mediation analysis that considers the relationships between methylation and expression (Equation 2) and their joint effect on survival (Equation 3). In our case, an alternative association is the effect that methylation alone (or as a proxy for alternative mechanisms) has on survival, and expression-based association is the effect of methylation on survival mediated through gene expression. Equation 4 represents the expression-based association, and Equation 5 represents the alternative association of methylation on survival, 37 both of which can be estimated by fitting the models in Equations 2 and 3. We used bootstrap to find the variances and confidence intervals of the expression-based and alternative associations. To determine directionality of the association of methylation on expression, we looked at the coefficient in the linear model regressing expression on methylation (Equation 2). A negative coefficient suggests that methylation and expression are inversely related (i.e., increased methylation is associated with decreased expression and vise versa). A positive correlation demonstrates that methylation and expression are directly related (i.e., increased methylation is associated with increased expression). Acknowledgments The authors would like to thank all individuals involved in the TCGA, particularly the patients who donated samples for use in this research. 122 References 1. Wen PY, Kesari S. Malignant gliomas in adults. N Engl J Med 2008; 359:492- 507. 2. Phillips HS, Kharbanda S, Chen R, Forrest WF, Soriano RH, Wu TD, Misra A, Nigro JM, Colman H, Soroceanu L, et al. Molecular subclasses of high-grade glioma predict prognosis, delineate a pattern of disease progression, and resemble stages in neurogenesis. Cancer Cell 2006; 9:157-73. 3. TCGA. Comprehensive genomic characterization defines human glioblastoma genes and core pathways. Nature 2008; 455:1061-8. 4. Verhaak RG, Hoadley KA, Purdom E, Wang V, Qi Y, Wilkerson MD, Miller CR, Ding L, Golub T, Mesirov JP, et al. Integrated genomic analysis identifies clinically relevant subtypes of glioblastoma characterized by abnormalities in PDGFRA, IDH1, EGFR, and NF1. Cancer Cell 2010; 17:98-110. 5. Frattini V, Trifonov V, Chan JM, Castano A, Lia M, Abate F, Keir ST, Ji AX, Zoppoli P, Niola F, et al. The integrated landscape of driver genomic alterations in glioblastoma. Nat Genet 2013. 6. Christensen BC, Smith AA, Zheng S, Koestler DC, Houseman EA, Marsit CJ, Wiemels JL, Nelson HH, Karagas MR, Wrensch MR, et al. DNA methylation, isocitrate dehydrogenase mutation, and survival in glioma. Journal of the National Cancer Institute 2011; 103:143-53. 7. Ehrlich M. DNA hypomethylation in cancer cells. Epigenomics 2009; 1:239- 59. 123 8. Lopez-Serra P, Esteller M. DNA methylation-associated silencing of tumor- suppressor microRNAs in cancer. Oncogene 2012; 31:1609-22. 9. Wan J, Oliver VF, Zhu H, Zack DJ, Qian J, Merbs SL. Integrative analysis of tissue-specific methylation and alternative splicing identifies conserved transcription factor binding motifs. Nucleic Acids Res 2013. 10. Aran D, Hellman A. DNA methylation of transcriptional enhancers and cancer predisposition. Cell 2013; 154:11-3. 11. Hegi M, Diserens A, Gorlia T, Hamou M, de Tribolet N, Weller M, Kros J, Hainfellner J, Mason W, Mariani L, et al. MGMT gene silencing and benefit from temozolomide in glioblastoma. N Engl J Med 2005; 352:997-1003. 12. Esteller M, Garcia-Foncillas J, Andion E, Goodman SN, Hidalgo OF, Vanaclocha V, Baylin SB, Herman JG. Inactivation of the DNA-repair gene MGMT and the clinical response of gliomas to alkylating agents. N Engl J Med 2000; 343:1350-4. 13. Killela PJ, Reitman ZJ, Jiao Y, Bettegowda C, Agrawal N, Diaz LA, Friedman AH, Friedman H, Gallia GL, Giovanella BC, et al. TERT promoter mutations occur frequently in gliomas and a subset of tumors derived from cells with low rates of self-renewal. Proc Natl Acad Sci U S A 2013; 110:6021-6. 14. Castelo-Branco P, Choufani S, Mack S, Gallagher D, Zhang C, Lipman T, Zhukova N, Walker EJ, Martin D, Merino D, et al. Methylation of the TERT promoter and risk stratification of childhood brain tumours: an integrative genomic and molecular study. Lancet Oncol 2013; 14:534-42. 124 15. Noushmehr H, Weisenberger D, Diefes K, Phillips H, Pujara K, Berman B, Pan F, Pelloski C, Sulman E, Bhat K, et al. Identification of a CpG island methylator phenotype that defines a distinct subgroup of glioma. Cancer Cell 2010; 17:510-22. 16. Christensen BC, Smith AA, Zheng S, Koestler DC, Houseman EA, Marsit CJ, Wiemels JL, Nelson HH, Karagas MR, Wrensch MR, et al. DNA methylation, isocitrate dehydrogenase mutation, and survival in glioma. J Natl Cancer Inst 2011; 103:143- 53. 17. Godard S, Getz G, Delorenzi M, Farmer P, Kobayashi H, Desbaillets I, Nozaki M, Diserens AC, Hamou MF, Dietrich PY, et al. Classification of human astrocytic gliomas on the basis of gene expression: a correlated group of genes with angiogenic activity emerges as a strong predictor of subtypes. Cancer Res 2003; 63:6613-25. 18. Rivenbark AG, Coleman WB. Dissecting the molecular mechanisms of cancer through bioinformatics-based experimental approaches. J Cell Biochem 2007; 101:1074-86. 19. Ferté C, Trister AD, Huang E, Bot BM, Guinney J, Commo F, Sieberts S, André F, Besse B, Soria JC, et al. Impact of bioinformatic procedures in the development and translation of high-throughput molecular classifiers in oncology. Clin Cancer Res 2013; 19:4315-25. 20. Houseman E, Christensen B, Yeh R, Marsit C, Karagas M, Wrensch M, Nelson H, Wiemels J, Zheng S, Wiencke J, et al. Model-based clustering of DNA methylation array data: a recursive-partitioning algorithm for high-dimensional data arising as a mixture of beta distributions. BMC Bioinformatics 2008; 9:365. 125 21. Wang W, Baladandayuthapani V, Morris JS, Broom BM, Manyam G, Do KA. iBAG: integrative Bayesian analysis of high-dimensional multiplatform genomics data. Bioinformatics 2013; 29:149-59. 22. Etcheverry A, Aubry M, de Tayrac M, Vauleon E, Boniface R, Guenot F, Saikali S, Hamlat A, Riffaud L, Menei P, et al. DNA methylation in glioblastoma: impact on gene expression and clinical outcome. BMC Genomics 2010; 11:701. 23. Rao SK, Edwards J, Joshi AD, Siu IM, Riggins GJ. A survey of glioblastoma genomic amplifications and deletions. J Neurooncol 2010; 96:169-79. 24. Chattopadhyay S, Tracy E, Liang P, Robledo O, Rose-John S, Baumann H. Interleukin-31 and oncostatin-M mediate distinct signaling reactions and response patterns in lung epithelial cells. J Biol Chem 2007; 282:3014-26. 25. Priester M, Copanaki E, Vafaizadeh V, Hensel S, Bernreuther C, Glatzel M, Seifert V, Groner B, Kögel D, Weissenberger J. STAT3 silencing inhibits glioma single cell infiltration and tumor growth. Neuro Oncol 2013; 15:840-52. 26. Papadopoulos MC, Verkman AS. Aquaporin water channels in the nervous system. Nat Rev Neurosci 2013; 14:265-77. 27. Bonomini F, Francesca B, Rezzani R. Aquaporin and blood brain barrier. Curr Neuropharmacol 2010; 8:92-6. 28. Wolburg H, Noell S, Fallier-Becker P, Mack AF, Wolburg-Buchholz K. The disturbed blood-brain barrier in human glioblastoma. Mol Aspects Med 2012; 33:579-89. 29. El Hindy N, Bankfalvi A, Herring A, Adamzik M, Lambertz N, Zhu Y, Siffert W, Sure U, Sandalcioglu IE. Correlation of aquaporin-1 water channel protein 126 expression with tumor angiogenesis in human astrocytoma. Anticancer Res 2013; 33:609-13. 30. Saadoun S, Papadopoulos MC, Hara-Chikuma M, Verkman AS. Impairment of angiogenesis and cell migration by targeted aquaporin-1 gene disruption. Nature 2005; 434:786-92. 31. Sepramaniam S, Armugam A, Lim KY, Karolina DS, Swaminathan P, Tan JR, Jeyaseelan K. MicroRNA 320a functions as a novel endogenous modulator of aquaporins 1 and 4 as well as a potential therapeutic target in cerebral ischemia. J Biol Chem 2010; 285:29223-30. 32. Blagitko N, Mergenthaler S, Schulz U, Wollmann HA, Craigen W, Eggermann T, Ropers HH, Kalscheuer VM. Human GRB10 is imprinted and expressed from the paternal and maternal allele in a highly tissue- and isoform-specific fashion. Hum Mol Genet 2000; 9:1587-95. 33. Monk D, Arnaud P, Frost J, Hills FA, Stanier P, Feil R, Moore GE. Reciprocal imprinting of human GRB10 in placental trophoblast and brain: evolutionary conservation of reversed allelic expression. Hum Mol Genet 2009; 18:3066-74. 34. Yu Y, Yoon SO, Poulogiannis G, Yang Q, Ma XM, Villén J, Kubica N, Hoffman GR, Cantley LC, Gygi SP, et al. Phosphoproteomic analysis identifies Grb10 as an mTORC1 substrate that negatively regulates insulin signaling. Science 2011; 332:1322-6. 35. Nord H, Hartmann C, Andersson R, Menzel U, Pfeifer S, Piotrowski A, Bogdan A, Kloc W, Sandgren J, Olofsson T, et al. Characterization of novel and complex 127 genomic aberrations in glioblastoma using a 32K BAC array. Neuro Oncol 2009; 11:803-18. 36. Cecener G, Tunca B, Egeli U, Bekar A, Tezcan G, Erturk E, Bayram N, Tolunay S. The promoter hypermethylation status of GATA6, MGMT, and FHIT in glioblastoma. Cell Mol Neurobiol 2012; 32:237-44. 37. VanderWeele TJ. Causal mediation analysis with survival data. Epidemiology 2011; 22:582-5. 38. Johnson WE, Li C, Rabinovic A. Adjusting batch effects in microarray expression data using empirical Bayes methods. Biostatistics 2007; 8:118-27. 39. Benjamini Y, Hochberg Y. Controlling the false discovery rate: a practical and powerful approach to multiple testing. Jornal of the Royal Statistical Society Series B (Methodological) 1995; 57:289-300. 40. Ruan B, Pong K, Jow F, Bowlby M, Crozier RA, Liu D, Liang S, Chen Y, Mercado ML, Feng X, et al. Binding of rapamycin analogs to calcium channels and FKBP52 contributes to their neuroprotective activities. Proc Natl Acad Sci U S A 2008; 105:33-8. 41. Hanissian SH, Teng B, Akbar U, Janjetovic Z, Zhou Q, Duntsch C, Robertson JH. Regulation of myeloid leukemia factor-1 interacting protein (MLF1IP) expression in glioblastoma. Brain Res 2005; 1047:56-64. 42. Zagzag D, Salnikow K, Chiriboga L, Yee H, Lan L, Ali MA, Garcia R, Demaria S, Newcomb EW. Downregulation of major histocompatibility complex antigens in invading glioma cells: stealth invasion of the brain. Lab Invest 2005; 85:328-41. 128 43. Halestrap AP. The SLC16 gene family - structure, role and regulation in health and disease. Mol Aspects Med 2013; 34:337-49. 44. Miranda-Gonçalves V, Honavar M, Pinheiro C, Martinho O, Pires MM, Cordeiro M, Bebiano G, Costa P, Palmeirim I, Reis RM, et al. Monocarboxylate transporters (MCTs) in gliomas: expression and exploitation as therapeutic targets. Neuro Oncol 2013; 15:172-88. 45. Colen CB, Shen Y, Ghoddoussi F, Yu P, Francis TB, Koch BJ, Monterey MD, Galloway MP, Sloan AE, Mathupala SP. Metabolic targeting of lactate efflux by malignant glioma inhibits invasiveness and induces necrosis: an in vivo study. Neoplasia 2011; 13:620-32. 46. Atkinson GP, Nozell SE, Benveniste ET. NF-kappaB and STAT3 signaling in glioma: targets for future therapies. Expert Rev Neurother 2010; 10:575-86. 47. Smith SJ, Tilly H, Ward JH, Macarthur DC, Lowe J, Coyle B, Grundy RG. CD105 (Endoglin) exerts prognostic effects via its role in the microvascular niche of paediatric high grade glioma. Acta Neuropathol 2012; 124:99-110. 48. Akhtar S, McIntosh P, Bryan-Sisneros A, Barratt L, Robertson B, Dolly JO. A functional spliced-variant of beta 2 subunit of Kv1 channels in C6 glioma cells and reactive astrocytes from rat lesioned cerebellum. Biochemistry 1999; 38:16984-92. 49. Yang X, Zhang Y, Li S, Liu C, Jin Z, Wang Y, Ren F, Chang Z. Rab21 attenuates EGF-mediated MAPK signaling through enhancing EGFR internalization and degradation. Biochemical and biophysical research communications 2012; 421:651- 7. 129 50. Totong R, Schell T, Lescroart F, Ryckebüsch L, Lin YF, Zygmunt T, Herwig L, Krudewig A, Gershoony D, Belting HG, et al. The novel transmembrane protein Tmem2 is essential for coordination of myocardial and endocardial morphogenesis. Development 2011; 138:4199-205. 51. Smith KA, Lagendijk AK, Courtney AD, Chen H, Paterson S, Hogan BM, Wicking C, Bakkers J. Transmembrane protein 2 (Tmem2) is required to regionally restrict atrioventricular canal boundary and endocardial cushion development. Development 2011; 138:4193-8. 52. Halestrap AP, Meredith D. The SLC16 gene family-from monocarboxylate transporters (MCTs) to aromatic amino acid transporters and beyond. Pflugers Arch 2004; 447:619-28. 130 Figure 1. Significant expression-based and alternative associations of DNA methylation on gene expression and survival. The 35 unique DNA methylation/gene expression pairs were subjected to an Accelerated Failure Time (AFT) survival model and applied to alternative and expression-based equations (2-5 in methods). This yielded a total of 27 significant methylation/expression pairs, 10 had significant expression-based associations (A), 14 had significant alternative associations (B), and 4 had both significant expression-based and alternative associations (C). Grey lines indicate alternative associations, black lines indicate expression-based associations, grey circles indicate that the methylation locus for that gene pair was found in CpG Island, and black circles indicate that the methylation locus for that gene pair was not found in a CpG island. The y-axis indicates the change in survival time per 5% increase in methylation; therefore, effects that fall above the line are associated with an increase in survival and effects that fall below the line are associated with a decrease in survival. 131 Alternative association Methylation-associated mutations changes in splice variants dysregulation of regulatory microRNAs activation of repeat regions alterations in enhancer regions METHYLATION EXPRESSION SURVIVAL Expression-based association loss of imprinting hypermethylation of cancer-related pathways Figure 2. Model for mediation analysis. First a linear model adjusted for study was used to determine significantly correlated methylation/expression pairs. Next, a Cox proportional hazards model was used to find significant association between survival and expression, methylation, and their interaction term (adjusting for age, gender, and study). An accelerated failure time model was used to estimate the association between survival and expression, methylation, and their interaction term (adjusting for age, gender, and study), and a mediation analysis was performed to estimate the alternative and expression-based associations on glioma survival. 132 Table 1. Patient demographic and tumor* characteristics Data Sets Characteristic Training Set (n=73) Testing Set (n=168) Pooled (n=241) Age, years Median 56 60 59 Range 18-86 10-86 10-86 Sex, n(%) Female 31 (42.5) 69 (41.1) 100 (41.5) Male 42 (57.5) 99 (58.9) 141(58.5) **Survival (months) Median 12.58 10.6 11.3 Range 1.37-60.0 1.08-60.0 1.08-60.0 *All tumor data obtained from The Cancer Genome Atlas (TCGA) **Censored at 60 months (5 years) 133 Table. 2. Final 35 DNA methylation/gene expression pairs that are significantly associated with survival TargetID_Reporter.REF SYMBOL cg17942096_A_23_P165180 RFXANK cg18345635_A_23_P147345 SLC16A3 cg23943801_A_23_P128166 RAB21 cg27626424_A_23_P34449 LOR cg05743054_A_23_P419947 MLF1 cg18345635_A_23_P158725 SLC16A3 cg18345635_A_23_P147349 SLC16A3 cg11558474_A_23_P94552 TMEM2 cg01781266_NM_018222_2_3793 PARVA cg05845503_A_24_P141275 GRHPR cg05845503_A_23_P60225 GRHPR cg04551925_A_23_P19894 AQP1 cg00973286_A_23_P139715 TNFRSF1A cg16773028_A_32_P40593 KCNA2 cg03138091_A_24_P388860 OSMR cg26475085_A_24_P388860 OSMR cg24812523_A_23_P14346 AKAP6 cg24302095_A_24_P235266 GRB10 cg24302095_A_24_P235268 GRB10 cg22166290_A_24_P402580 BCL11A cg03764161_A_23_P203330 FAM111A cg17726022_A_24_P261734 SLC38A1 cg17726022_A_23_P326510 SLC38A1 cg07663789_A_23_P327451 NPR3 cg04006554_A_23_P214244 ENPP5 cg04006554_A_23_P214240 ENPP5 cg04006554_NM_021572_2_2378 ENPP5 cg05788437_A_23_P80826 FYTTD1 cg06038049_A_23_P35029 CPSF3L cg20089715_A_23_P405754 CACNB1 cg24219058_A_23_P310921 PCDH7 cg20091959_A_23_P210445 L3MBTL cg18138552_A_23_P67464 PSMD8 cg20161089_A_24_P270460 IFI27 cg18320336_A_24_P406335 STEAP1 134 Table 3. Functions of significant genes and potential mechanisms in glioma Potential expression-based Potential alternative role in SYMBOL NAME FUNCTION (GeneCards®) Ref. # role in glioma survival glioma survival CACNB1 calcium channel, voltage- Involved in modulating G It has been proposed that 37 dependent, beta 1 subunit protein inhibition CACNB1 can protect neurons from Ca(2+)-induced cell death by modulating Ca(2+) channels; therefore, methylation-induced inhibition of CACNB1 could lead to loss of their neuroprotective activities (Ruan B et al 2008). IFI27 interferon, alpha-inducible Promotes cell death through ? protein 27 mediation of IFN-alpha MLF1 myeloid leukemia factor 1 Oncoprotein that may be MLF1 and MLF1-like protein 38 involved in lineage were found to co-localize and commitment be over expressed in GBM tumors suggesting they play a role in glioma pathogenesis and survival. (Hanissian SH et al 2005). Dysregulation in expression of MLH1 via methylation could lead to differential survival outcomes. OSMR oncostatin M receptor Member of the type 1 Dysregulation of STAT3 22-23 cytokine receptor family activation via epigenetic which heterodimerizes with induced silencing interleukin 31, which as a (Chattopadhyay et al 2007; complex can induce Priester et al 2013). signaling events RFXANK regulatory factor X- Forms a complex with Methylation-induced decrease 39 associated ankyrin- regulatory factor X- in RFXANK could inhibit MHC containing protein associated protein and class II activation, which is regulatory factor 5, which associated with glioma tumor can then bind X box motif invasion (Zagzag D et al, regions of some major 2005). histocompatibility (MHC) 135 class II molecules, leading to activation SLC16A3 solute carrier family 16, Part of a family of Differential SLC16A3 40- member 3 monocarboxylate expression causing 42,49 (monocarboxylic acid transporters that catalyze dysregulation of glycolytic transporter 4) lactic acid and pyruvate metabolism via MCTs transport across plasma (Halestrap AP et al 2004 and membranes 2013; Miranda-Gonçalves V et al 2013; Colen CB et al 2011). TNFRSF1A tumor necrosis factor This receptor can activate Methylation induced changes 43 receptor superfamily, NF-kappaB, mediate in gene expression can member 1A apoptosis, and function as a dysregulate NF-kappaB regulator of inflammation pathway, which has been previously associated with glioma tumorigenesis and could be a possible therapeutic target of this disease ( Atkinson GP et al 2010). AQP1 aquaporin 1 (Colton blood Molecular water channel Methylation-mediated 24-29 group) protein dysregulation of microRNA mir-320a binding region (Papadopoulos MC et al 2013; Bonomini F et al 2010; Wolburg H et al 2012; El Hindy Ner et al 2013; Saadoun S et al 2005; Sepramaniam S et al 2010). ENPP5 ectonucleotide It may play a role in Possible dysregulation in 44 pyrophosphatase/ neuronal cell angiogenic signaling (Smith phosphodiesterase 5 communication SJ et al 2012). (putative) 136 FYTTD1 forty-two-three domain Required for mRNA export ? containing 1 from the nucleus to the cytoplasm KCNA2 potassium voltage-gated Voltage-gated ion channel Contains an alternatively 45 channel, shaker-related that has a multitude of spliced product in glioma subfamily, member 2 different functions ranging cells which could contribute from regulation of to the inactivation rate of the neurotransmitter release, k(+) current Akhtar S et al heart rate, insulin secretion, 1999) and neuronal excitability L3MBTL Lethal (3) Malignant Brain Polycomb group gene ? Tumor-like 1 (Drosophila) which functions to regulate gene activity via chromatin modifications NPR3 natriuretic peptide receptor Natriuretic peptide receptor ? C/guanylate cyclase C that regulates blood (atrionatriuretic peptide volume/pressure, receptor C) pulmonary hypertension, cardiac function and some metabolic/growth processes PSMD8 proteasome (prosome, Regulatory subunit of the ? macropain) 26S subunit, 26S multicatalytic non-ATPase, 8 proteinase complex, which is involved in the ATP- dependent degradation of ubiquitinated proteins RAB21 RAB21, member RAS GTP-binding protein Rab21 expression has been 46 oncogene family involved in integrin found to attenuate Epidermal internalization and recycling growth factor (EFG)- mediated mitogen-activated protein kinase (MAPK) by inducing EGF-receptor degradation (Yang X et al 2012). 137 STEAP1 six transmembrane Found to be upregulated in ? epithelial antigen of the multiple cancer cells lines prostate 1 and may be a potential metalloreductase TMEM2 transmembrane protein 2 Involved in coordination of ? 47-48 myocardial and endocardial morphogenesis (Totong R et al 2011, Smith KA et al 2011) CPSF3L cleavage and Catalytic subunit of the ? ? polyadenylation specific integrator complex, which factor 3-like mediates the 3-prime end processing of small nuclear RNAs U1 and U3 GRB10 growth factor receptor- Growth receptor binding Methylation induced loss of Methylation changes in 30-33 bound protein 10 protein that interacts with imprinting (Blagitko N et al splice variants, leading to insulin and insulin-like 2009; Monk D et al 2009; Yu Y expression of alternatively growth-factor receptors et al 2011 ;Nord H et al 2009). functioning isoforms (Blagitko N et al 2009; Monk D et al 2009; Yu Y et al 2011 ;Nord H et al 2009 ). GRHPR glyoxylate Enzyme that plays a role in ? ? reductase/hydroxypyruvate metabolism and reduces reductase hydroxypyruvate to D- glycerate and glyoxylate to glycolate and oxidizes D- glycerate to hydroxypyruvate ? - Possible mechanisms relating to glioma and significant expression-based or alternative association are unknown. 138 A B 0.8 0.6 0.4 0.2 0.0 rLL rLR rRLLL rRLLRL rRLLRR rRLR rRR C Class Hypermethylator Phenotype RLLRR TCGA.14.1458, TCGA.16.1460, TCGA.19.1788, TCGA.14.1456 RLLRL TCGA.28.1756, TCGA.14.4157, TCGA.32.4208 rLL rRLLL rRLLRL rLR rRLLRR rRLR rRR Supplementary Figure 1. Removal of IDH1 mutants. A Recursively partitioned mixture model (RPMM) was run on the top 5000 most variable CpG loci from the HumanMethylation27 (Illumina) DNA methylation array testing set (n=190) in order to determine hypermethylated classes, which have been previously associated with IDH1 mutation and increased survival in glioma(1A). The average methylation of each class was plotted (1B) and tumors in the top two most hypermethylated classes (RLLRR and RLLRL) were removed from the analysis as possible IDH1 mutant containing samples (1C). 139 Supplementary Figure 2. Directionality of significant pairs. Gene expression values were plotted against DNA methylation values. A negative correlation demonstrates that methylation and expression are going in opposite directions (i.e. an increase in methylation is associated with a decrease in expression) and a positive correlation demonstrates that methylation and expression are going in the same directions (i.e. an increase in methylation is associated with an increase in expression). 140 141 142 143 144 145 146 147 148 149 Supplementary Fig. 3 Map of methylation loci locations from significant methylation/expression pairs. Each methylation locus obtained from significant methylation and expression pairs was plotted according to its genome location (Illumina annotation file). Exon locations were obtained from Genome Browser, with variants chosen based on the highest number of exons for which methylation loci fell within an exon as opposed to an intron. If the methylation locus was found within a CpG Island, that CpG island range was plotted in green (Illumina annotation file). CPSF3L is not plotted due to the fact that the accession number for this gene (NM_032179.1) was not available on genome browser. 150 Supplementary Table 1. DNA methylation/ expression pairs that are significantly associated with survival (q-value<0.1) Methylation Loci_Expression Probe Gene Symbol p-value q-value** cg17942096_A_23_P165180 RFXANK 0.0007 0.0124 0.0001 0.007 cg18345635_A_23_P147345 SLC16A3 0.0123 0.0075 0.0001 0.007 cg23943801_A_23_P128166 RAB21 0.0133 0.0063 0.0003 0.0078 cg27626424_A_23_P34449 LOR 0.0238 0.0078 0.0004 0.0078 cg05743054_A_23_P419947 MLF1 0.0205 0.0001 0.0004 0.0078 cg18345635_A_23_P158725 SLC16A3 0.0082 0.0336 0.0005 0.0078 cg18345635_A_23_P147349 SLC16A3 0.0091 0.0179 0.0009 0.0127 cg11558474_A_23_P94552 TMEM2 0.0192 0.0171 0.001 0.0127 cg01781266_NM_018222_2_3793 PARVA 0.0204 0.0052 0.0015 0.0157 cg05845503_A_24_P141275 GRHPR 0.0399 0.0191 0.0017 0.0157 cg05845503_A_23_P60225 GRHPR 0.0305 0.0220 0.0019 0.0157 cg04551925_A_23_P19894 AQP1 0.0486 0.0302 0.002 0.0157 cg00973286_A_23_P139715 TNFRSF1A 0.0451 0.0231 0.0022 0.0157 cg16773028_A_32_P40593 KCNA2 0.0474 0.0032 0.0022 0.0157 cg03138091_A_24_P388860 OSMR 0.0070 0.0261 0.0026 0.0174 cg26475085_A_24_P388860 OSMR 0.0227 0.0104 0.0037 0.0228 cg24812523_A_23_P14346 AKAP6 0.0049 0.0239 0.0039 0.0229 cg24302095_A_24_P235266 GRB10 0.0244 0.0006 0.0042 0.0234 cg24302095_A_24_P235268 GRB10 0.0238 0.0005 0.0047 0.0244 cg22166290_A_24_P402580 BCL11A 0.0179 0.0449 0.0053 0.0244 cg03764161_A_23_P203330 FAM111A 0.0425 0.0082 0.0053 0.0244 cg17726022_A_24_P261734 SLC38A1 0.0346 0.0388 0.0054 0.0244 cg17726022_A_23_P326510 SLC38A1 0.0389 0.0335 0.0063 0.0273 cg07663789_A_23_P327451 NPR3 0.0217 0.0261 0.0068 0.0282 cg04006554_A_23_P214244 ENPP5 0.0113 0.0140 0.0091 0.0362 cg04006554_A_23_P214240 ENPP5 0.0068 0.0107 0.0107 0.0398 cg04006554_NM_021572_2_2378 ENPP5 0.0071 0.0115 0.0107 0.0398 cg05788437_A_23_P80826 FYTTD1 0.0232 0.0000 0.0118 0.0422 cg06038049_A_23_P35029 CPSF3L 0.0031 0.0322 0.0154 0.0515 cg20089715_A_23_P405754 CACNB1 0.0044 0.0123 0.0154 0.0515 cg24219058_A_23_P310921 PCDH7 0.0078 0.0085 0.0224 0.0723 cg20091959_A_23_P210445 L3MBTL 0.0286 0.0136 0.0232 0.0725 cg18138552_A_23_P67464 PSMD8 0.0310 0.0205 0.0259 0.0758 cg20161089_A_24_P270460 IFI27 0.0462 0.0093 0.0261 0.0758 cg23134520_A_23_P143218 ACOT8* 0.0463 0.0374 0.0265 0.0758 cg18320336_A_24_P406335 STEAP1 0.0374 0.0277 0.0345 0.0958 ACOT8 was removed from further analysis as SNP rs6032566 lies in the CpG locus of interest (cg23144520)**q-value based on n=878 151 Supplementary Table 2. Expression-based and alternative associations of DNA methylation on gene expression and survival. Confidence Confidence Expression- Confidence Confidence Methylation Loci_Expression Alternative Interval (lower Interval Based Interval Interval Probe SYMBOL Association bound) (upper bound) Association (lower bound) (upper bound) cg24812523_A_23_P14346 AKAP6 0.999 0.96 1.04 1.007 0.994 1.018 cg04551925_A_23_P19894 AQP1 1.111 1.03 1.21 0.982 0.952 1.012 cg22166290_A_24_P402580 BCL11A 0.998 0.92 1.11 0.989 0.964 1.011 cg20089715_A_23_P405754 CACNB1 1.029 0.97 1.10 0.977 0.950 0.994 cg06038049_A_23_P35029 CPSF3L 2.863 1.19 10.97 7.237 1.129 103.291 cg04006554_A_23_P214240 ENPP5 0.915 0.84 0.97 0.997 0.980 1.015 cg04006554_A_23_P214244 ENPP5 0.920 0.85 0.98 0.994 0.977 1.011 cg04006554_NM_021572_2_2378 ENPP5 0.917 0.84 0.98 0.997 0.979 1.014 cg03764161_A_23_P203330 FAM111A 1.172 0.70 1.80 0.970 0.717 1.441 cg05788437_A_23_P80826 FYTTD1 2.577 1.50 5.09 0.768 0.406 1.287 cg24302095_A_24_P235266 GRB10 0.961 0.90 1.00 1.029 1.012 1.053 cg24302095_A_24_P235268 GRB10 0.958 0.90 0.99 1.028 1.012 1.049 cg05845503_A_23_P60225 GRHPR 2.156 1.06 4.30 4.370 0.953 21.125 cg05845503_A_24_P141275 GRHPR 2.277 1.11 4.79 5.474 1.074 29.351 cg20161089_A_24_P270460 IFI27 0.941 0.85 1.05 0.961 0.928 0.989 cg16773028_A_32_P40593 KCNA2 1.061 1.03 1.10 0.996 0.980 1.010 cg20091959_A_23_P210445 L3MBTL 1.064 1.00 1.12 1.013 0.997 1.034 cg27626424_A_23_P34449 LOR 1.036 1.00 1.07 1.001 0.994 1.009 cg05743054_A_23_P419947 MLF1 1.124 0.73 2.38 0.825 0.596 0.933 cg07663789_A_23_P327451 NPR3 1.033 1.00 1.07 0.998 0.991 1.003 cg03138091_A_24_P388860 OSMR 0.984 0.95 1.02 1.018 1.008 1.030 cg26475085_A_24_P388860 OSMR 1.041 0.95 1.15 1.025 1.003 1.052 cg01781266_NM_018222_2_3793 PARVA 1.059 0.95 1.15 0.990 0.963 1.017 cg24219058_A_23_P310921 PCDH7 1.043 0.78 1.76 0.939 0.810 1.005 cg18138552_A_23_P67464 PSMD8 1.610 1.07 2.39 0.850 0.625 1.080 cg23943801_A_23_P128166 RAB21 2.764 1.36 6.06 1.240 0.729 2.441 cg17942096_A_23_P165180 RFXANK 0.922 0.84 1.01 1.050 1.019 1.086 cg18345635_A_23_P147345 SLC16A3 1.009 0.98 1.06 1.017 1.007 1.031 cg18345635_A_23_P147349 SLC16A3 1.010 0.98 1.05 1.011 1.003 1.024 cg18345635_A_23_P158725 SLC16A3 1.011 0.98 1.06 1.015 1.006 1.029 cg17726022_A_23_P326510 SLC38A1 1.040 0.72 1.51 0.875 0.702 1.032 cg17726022_A_24_P261734 SLC38A1 1.052 0.73 1.55 0.868 0.698 1.021 cg18320336_A_24_P406335 STEAP1 0.939 0.89 0.99 0.995 0.981 1.004 cg11558474_A_23_P94552 TMEM2 1.107 1.01 1.23 1.017 0.994 1.042 cg00973286_A_23_P139715 TNFRSF1A 1.096 0.91 1.28 1.080 1.001 1.181 *Associations in bold type are significant 152 Chapter 4 Discussion 153 Conclusion Malignant adult gliomas are the most common type of brain cancer1. In the past decade, advances in diagnosis and treatment, particularly the use of the alkylating agent temozolomide, have only led to minimal improvement in patient survival 2,3. Glioma survival outcome has been found to be associated with age, adjuvant treatments, giant- cell subtype and oligodendroglia differentiation2. In addition, advances in imaging techniques have allowed for better diagnosis4,5 and more complete resection of malignant tumors, which has also been correlated with improved patient survival5,6 . However, advances in the classification of glioma based on its molecular landscape are the most clinically relevant7-10 . The classification of glioma types/subtypes using both genetic and epigenetic profiles has not only enhanced our knowledge of gliomagenesis but has also highlighted both molecular predictors of survival and possible therapeutic targets of glioma11. Genetic markers of glioma such as somatic alterations in the p53 12-15 , Rb15 , EGFR16 , PI3K, and VEGF signaling pathways have now been well established 2,17 , allowing for treatments targeting specific genes and proteins. More successful targeted therapies include anti-angiogenic drugs, including the commonly used bevacizumab18 . Bevacizumab is a monoclonal antibody against VEGF, which, upon binding, inhibits VEGF activity19. In 2009 bevacizumab was granted accelerated approval by the FDA as a single agent for the treatment of recurrent GBM 19. However, some studies have shown that anti-angiogenic drugs can enhance invasion and metastasis20,21. It has been suggested 154 that anti-angiogenic drugs should be used in combination with drugs that inhibit progression, invasion, and/or metastasis to increase overall survival22,23. EGFR is a tyrosine kinase activated growth factor that is involved in the activation of many signaling pathways, including RAS-MEK-ERK and PI3K-AKT24, and is strongly dysregulated in glioma and consistently amplified or mutated in GBM16. Many targeted therapies, including monoclonal antibodies, vaccines, tyrosine-kinase inhibitors, and RNA-based agents have been under review25. Though research on these drugs seems promising, drug resistance is a common endpoint, re-emphasizing the need for novel drug targets25. Recently, greater attention has been given to epigenetics-based targets, both for prognostic and therapeutic purposes. The role of epigenetics in cancer biology has only recently started to come into focus. Epigenetics encompasses events such as histone modifications26, DNA methylation27,28 , and the targeting of genes by microRNA29,30 , all of which are capable of changing an individual’s gene expression and/or cellular phenotype without directly changing the DNA sequence31,32 . One of the most intensely studied areas of epigenetics is DNA methylation, which entails the addition of methyl groups to CpG dinucleotides33. DNA methylation is catalyzed by DNA methyltransferases and causes condensation of chromatin structure, which can lead to dysregulation of gene transcription 33,34. DNA methylation of gene promoters is strongly implicated in a variety of cancers, including gliomas, and has been associated variable prognosis35 . The best-known example of this association in glioma is the promoter methylation of methyl guanine methyl transferase (MGMT), which is associated with increased survival after treatment36,37 . The fact that epigenetics does not involve actual alterations in the DNA sequence makes it more 155 appealing to study because unlike genetic alterations, epigenetic alterations are potentially reversible. Agents inhibiting re-methylation, such as 5-azacytidine, have already been approved for the treatment of hematopoietic cancers38-40 . However, the use of such drugs on solid tumors has proven less effective41 . Additionally, the lack of specificity of DNA-methylating agents is a prevalent concern, as it can lead to global demethylation and consequent expression of oncogenes and transposable elements, ultimately causing genomic instability41 . DNA methylation has become a reliable source of biomarkers, as methylation profiles can distinguish cell lineages42 , tissues43, and disease subtypes, and contribute to improvements in diagnosis, prognosis, and treatment outcome35. On a single locus level, DNA methylation has aided in the treatment and survival of glioma, as seen with the aforementioned promoter methylation of MGMT44. Loss of methylation on a global level has become a defining tumor characteristic45,46. Progress in the understanding of both the genetic and epigenetic landscapes of glioma has led to advances in both the diagnosis and treatment of the disease. Despite these advancements, disease survival remains low. Some researchers theorize that patients would benefit from targeting the molecular landscape as a whole, not just specific somatic alterations47. This theory would rely heavily on the molecular classification of tumors and how specific profiles or molecular characteristics associate with both treatment and survival. The molecular classification of glioma began with the integration of genetic alterations. At the forefront of this research was Phillips et al, who defined 3 classes of glioma based on the integration of copy number variation (CNV), gene expression, and activation of cell signaling8 . The classes, proneural, proliferative, 156 and mesenchymal, each resemble different stages of neurogenesis and each was differentially associated with outcome, with the proliferative and mesenchymal classes demonstrating the poorest survival8. These analyses were further supported by the addition of mutation data, which helped refine associations seen between classes and survival outcome and led to the identification of novel drivers of gliomagenesis 10,48,49. However, as these studies only exhibit the genetic diversity of glioma and its subtypes, there is a need for further analyses that integrate other important aspects of gliomagenesis, such as epigenetics. The integration of genetics and epigenetics has greatly enhanced our knowledge of cancer biology, as seen with CpG island methylator phenotypes (CIMP), which can distinguish different tumor subtypes and are significantly associated with outcome. In glioma, the link between promoter methylation and gene expression has been established on a single-locus level. However, no large-scale attempts integrating methylation patterns and genetic alterations in glioma have been made to date. The goal of this thesis was to integrate genetic and epigenetic profiles to obtain molecular drivers of malignancy and survival in glioma. In chapter 2, we discuss the relationship between the common glioma mutant isocitrate dehydrogenase (IDH) and its association with DNA methylation. First DNA methylation signatures of GBM, astrocytomas (AS), oligodendrogliomas (OD), oligoastrocytomas (OA), ependymomas (EP), and pilocytic astrocytomas (PA) (n=131) and those of non-glioma brain tissues (n=7) were obtained using the Infinium GoldenGate array, which interrogates CpG methylation loci in ~1500 cancer related genes. Tumors and non-glioma tissues were then clustered based on their methylation 157 status (β-value), with samples having the most similar methylation patterns clustering together. As expected, gliomas clustered separately from non-glioma brain tissue. Furthermore, a pathway analysis of differentially methylated loci (based on Δβ between glioma and non-glioma brain tissue) demonstrated that as a whole, metabolic pathways were commonly hypomethylated. Interestingly, the metabolite IDH1/2 has recently been found to be mutated in approximately 80% of low-grade gliomas and secondary GBM, and in <10% of primary GBM50,51. This prompted us to investigate the association between IDH and DNA methylation in glioma. Recursively partitioned mixture modeling was used to cluster only glioma samples, resulting in nine classes that were significantly associated with age, histology, and grade. Not surprisingly, IDH mutants were associated with histological subtypes, with an increased number of mutants found in low-grade gliomas and secondary GBM compared with primary GBM. In addition, IDH mutants were exclusively associated with the two homogenous hypermethylated classes, where non-mutants were heterogeneously distributed among the remaining seven classes. This novel finding suggested IDH as a potential driver of a hypermethylator phenotype. In fact, associations between methylation class and both TP53 and EGFR were less robust than that of mutant IDH, further supporting the role of IDH as a driver of the observed hypermethylator phenotype. A Cox proportional hazards model showed that patients whose tumors harbored IDH mutants had significantly improved outcome compared with patients whose tumors harbored non-mutant IDH, suggesting IDH mutation in association with hypermethylation as a potential prognostic biomarker of glioma. The prognostic value of the hypermethylator phenotype, or CIMP, was first observed in colorectal cancer. CIMP classes are determined based on mutations in BRAF and/or KRAS and promoter 158 methylation levels52 , and CIMP subtypes are associated with differential prognosis. Recently, Noushmehr et al confirmed the relationship of mutant IDH with promoter methylation of CIMP-associated loci and successfully defined a glioma-CIMP (G-CIMP) 7 . They found that G-CIMP-positive tumors are frequently found in younger patients with low-grade gliomas, and these patients often show better survival outcomes, which supports our findings7 . Additionally, gene expression data revealed that G-CIMP tumors are enriched in a portion of the previously identified proneural subtype7, which has also been associated with a better prognosis8,10. Mechanistically, the link between IDH mutants and DNA methylation is still under debate. Mutant IDH is a neomorphic enzyme that, instead of catalyzing the oxidative decarboxylation of isocitrate into α-ketoglutarate (α –KG), actually converts α –KG into oncometabolite 2-hydroxyglutare (2-HG) in an NADPH-dependent manner53 . Accumulation of 2-HG has been seen in diseases such as 2-hydroxyglutaric aciduria, which has been associated with increased risk of glioma 53. In 2011, Xu et al found 2-HG to be a weak antagonist of α–KG, which, at high concentrations, can inhibit α–KG- dependent dioxygenases such as histone demethylases and TET2 5-methylcytosine hydroxylases (5mC). Inhibition of histone demethylases can limit the removal of histone- associated methyl groups causing an increase in normal histone methylation. TET2 normally catalyzes the conversion of 5mC to 5-hydroxymethylcytosine (5hmC), which can lead to demethylation of DNA. Therefore, inhibition of TET2 can lead to an increase in DNA methylation. Consequently, the increased production of 2-HG from mutated IDH can cause dysregulation of the normal methylome 54. In 2012, Turcan et al lent further support to the connection between IDH and the G-CIMP55 . The group found enriched 159 methylation of the histone marks H3K9 and H3K27 in cells expressing mutant IDH, which have been shown to promote DNA methylation through recruitment of DMNTs 56. Furthermore, expression of TET2 was inhibited in IDH mutant samples, decreasing production of 5hmC and further supporting previous findings of a possible mechanistic link between IDH mutations and G-CIMP55 . The success of the integration of genetic and epigenetic alterations in defining a prognostically relevant G-CIMP class further demonstrates the need for analyses that can aid in the discovery of other drivers and potential biomarkers of gliomagenesis. Of particular interest are those gliomas that fall outside of the IDH-driven methylator phenotype. In chapter 3, we employed a genome-wide, agnostic strategy for the discovery of novel predictive biomarkers related to the prognosis of glioma. In the previous chapter, we focused on the associations of IDH mutant gliomas and methylation. Interestingly, though IDH mutants were exclusive to the two hypermethylated classes, wild-type IDH was homogenously distributed among the lower methylated classes, suggesting alternative mechanisms of glioma pathogenesis in these patients. Uniquely, in our study, we focused on IDH wild-type samples and the role methylation plays alone or in conjunction with gene expression in the pathogenesis and survival of primary GBM. Not surprisingly, the 27 genes found to be significantly associated with survival in our study are involved in invasion, angiogenesis, and metastasis, and many were previously found to be associated with brain/glioma. We found 10 methylation/expression pairs that had a significant expression-based association with survival, suggesting that DNA methylation in these genes affects survival outcome via expression of the associated gene, supporting 160 the commonly accepted paradigm that methylation effects survival through gene expression. Interestingly, of these 10 methylation/expression pairs, two were found within the OSMR gene. As mentioned in chapter 3, OSMR is associated with STAT3 activation via the JAK/STAT signaling pathway. In a glioma cell line, inhibition of STAT3 activation was associated with reduced cell migration and invasion and mice with STAT3 knockdown tumors exhibited increased survival compared to controls, suggesting that inhibition of STAT3 is important in gliomagenesis and survival. Therefore, methylation induced silencing of OSMR could inhibit the activation of STAT3, thereby attenuating glioma cell migration and invasion. Evidence of methylation-induced silencing of OSMR has already been shown in colorectal (CR) cancer. Methylation- induced silencing of OSMR expression was associated with increased growth due to inhibition of the OSMR substrate OSM57,58 . Furthermore, CR tumors with increased OSMR promoter methylation were associated with a non-invasive phenotype58, suggesting that OSMR could predict a class of tumors that are associated with improved survival. This data supports a possible link between promoter methylation, gene expression, and survival outcome, and suggests methylation of OSMR as a potential biomarker of a novel prognostic phenotype. In our work, 14 methylation/expression pairs were found to have significant alternative associations, suggesting that methylation can function through alternative mechanisms other than expression, to effect survival. Increased expression of the water channel AQP1, which had a significant alternative association in our analysis, has recently been observed in GBM. Interestingly, AQP1 has been shown to contain targets for regulatory microRNAs. Osmotically regulated microRNAs miR-708 and miR-666 were found to 161 inhibit AQP1 expression in BDL endothelial cells, and low AQP1 levels were associated with reduced angiogenesis and fibrosis in a mouse model of liver cirrhosis59. Additionally, hypoxically activated miR-214 was correlated with decreased AQP1 expression in HUVEC cells60, and miR-320a has been found to directly target AQP1 and is associated with decreased mRNA and protein expression of AQP1 during cerebral ischemia. Importantly, AQP1 is associated with cytotoxic cerebral edema, angiogenesis, and invasion in GBM, suggesting that suppression of AQP1 expression could increase survival outcome61. Therefore, a possible mechanistic explanation for the alternative association we observed in our work involves methylation of the microRNA target region on AQP1, which would inhibit the binding of miR-320a and ultimately result in increased expression of AQP1. This model could explain the alternative mechanism associated with glioma survival in this instance61,62. Unexpectedly, there were four methylation- expression pairs that had both significant alternative and expression-based associations, suggesting that methylation can function simultaneously through both expression-based and alternative mechanisms to significantly impact survival. This phenomenon was observed with locus pairs found within the imprinted GRB10 gene. GRB10 has recently been implicated as both a putative oncogene in glioma 63 and potential tumor suppressor64 . The GRB10 gene is of particular interest because it has been found to contain 13 different splice variants, expression of all but one of which (γ2) has been found in the brain65. GRB10 has been shown to have both an inhibitory and stimulatory effect on IGF-1- related proliferation, though not specifically in brain tissue66 . Though the reason behind its conflicting effects is not yet understood, one theory is that different GRB10 isoforms have different regulatory functions but compete for similar substrates. DNA methylation 162 has the ability to regulate differential isoform production via alternative splicing 67. Uniquely, imprinting of GRB10 is tissue-dependent. Monoallelic expression is seen in skeletal muscle and placenta (maternally expressed in humans) and in the brain (paternally expressed in humans) 65 . Disruption of maternal imprinting in mice leads to overgrowth and insulin sensitivity throughout life, while in adult mice, deletion of GRB10 is associated with increased total body mass and up-regulation of cancer associated genes68 . Unfortunately, ablation of imprinting in the paternal allele has not been shown to affect growth 68. However, it is important to note that mice have been found to only have maternally imprinted GRB10 as opposed to humans, who show biallelic imprinting. Overall, this suggests that associations with survival can occur due to both loss of imprinting (expression-based association) and through expression of differentially functioning alternative isoforms (alternative association). It is plausible that these effects could be seen simultaneously in genes within the same tumor and work synergistically, or these effects could occur separately within different tumors, allowing the gene and its associated effect (alternative or expression-based) to be used as possible markers of tumor type. Overall, these findings corroborate the common idea that methylation operates through expression to affect survival outcome, but they also suggest that methylation can associate with survival outcome through mechanisms other than dysregulation of gene transcription. Though additional validation studies are needed, our method may lead to the identification of novel putative genetic and epigenetic biomarkers of glioma that could potentially be useful as therapeutic targets Importantly, this approach could be applicable to cancers other than glioma, and the model can be adjusted to include other variables of interest. For instance, instead of focusing on DNA 163 methylation and gene expression, one could focus on DNA methylation and microRNA. Thus, our analysis provides a conceivable method of biomarker discovery that may be broadly clinically applicable. Unfortunately, there were several limitations to this study. First, we had to rely on publically available data, which did not have complete mutation and survival data. For the missing mutation data, we used an RPMM to remove the hypermethylated classes that our group9 and others7 have previously found to exclusively contain IDH mutants. To address issue of missing survival data we used an accelerated failure time model to predict the survival time of censored values in order to control for survival in a combat model. In order to ensure functionality of methylation loci in our analysis, an initial screen was conducted, and only methylation and expression pairs that were significantly correlated within the same gene were used. Unfortunately, this approach entailed the exclusion of loci that may affect the expression of genes from distant locations, as seen with the methylation of enhancer regions. Additionally, the Infinium HumanMethylation27 array that was used to determine methylation status in these samples is strongly biased towards proximal promoter regions69. Therefore, in future studies it may be beneficial to look at the correlation of methylation loci and all gene expression probes, without focusing on pairs within a single gene. In 2013, Aran et al explored how DNA methylation of distal regulatory sites in normal and malignant cell lines associates with gene expression levels across the genome70. First, they developed a model, which, at a score of less than or equal to 0.85, successfully determined genes that undergo promoter methylation-dependent expression in variable methylation sites (VMS) of malignant cell lines with 2.63% sensitivity and a 12.8% false discovery rate. This 164 model was then applied to VMS +/- 1 megabase of the transcription start site of over 17,000 human genes, excluding sites that fell within 5 KB of promoters/alternative promoters of the associated genes. This yielded 1,911 pairs (486 genes), 1,041 of which were distal regulatory sites largely within promoter and enhancer regions. Further analyses suggested that high-scoring pairs were enriched in transcriptional enhancers and bound transcription factors in a methylation-dependent manner. Furthermore, analysis of the 1,911 distal methylation sites in normal vs. malignant mammary epithelial cells revealed a methylation-dependent association between high-score enhancer regions and the expression of their associated genes and suggested that methylation levels of these enhancers associate better with transcriptional regulation than promoter methylation. Moreover, both hypomethylation and hypermethylation of enhancers was observed in different malignant cell types, suggesting a role for differential enhancer methylation in cancer70. In addition differential enhancer methylation may be useful in differentiating different glioma subtypes. Due to limited patient data, our study consisted only of primary GBM; however, using our method to look at several different histologies could support pre-existing or aid in the discovery of new subclasses of glioma. This is further supported by our observation of methylation and expression pairs that were significant for both expression-based and alternative associations, demonstrating that the pathogenesis of these tumors involves discrete mechanisms that have differential effects on survival outcome. Of further interest is the prognostic signature of gliomas both before and after treatment. Shukla et al used both treated (radio therapy and concomitant temozolomide) patient samples and a series of cell culture experiments (using 5-Aza-2'- deoxycytidine treatment) to identify a methylation-based prognostic signature in high- 165 grade glioma comprising nine genes71 . Using a methylation-based risk-score, the methylation statuses of these nine genes could identify patients as either low-risk or high- risk, with the latter having significantly lower survival. Unsurprisingly, the low-risk group contained a high percentage of IDH1 mutants, proneural associated genes, and G- CIMP tumors. Additionally, the high-risk group was associated with activated NF-kB signaling. Further studies demonstrated that inhibition of NF-kB lead to enhanced sensitivity to chemotherapeutic agents, Not only does this explain the decreased survival observed in high-risk groups, but it also suggests NF-kB as a probable therapeutic target in cases where normal therapy is not successful71 . Our research has aided in the discovery of putative glioma biomarkers, as observed in chapter 2 with the association of IDH and a hypermethylator phenotype with increased survival. Though more validation is required, we have shown the importance of analyses integrating multiple somatic alterations and their associations with outcome, as shown in chapter 3 with the integration of DNA methylation and gene expression. These analyses supported the common idea that DNA methylation works through gene expression to affect survival. Our analysis also demonstrated a unique method of biomarker discovery that can easily lend itself to diseases other than glioma. Most importantly, our analysis demonstrated significant alternative associations, suggesting that DNA methylation can also operate through alternative or combined mechanisms to affect outcome. An alternative association, as defined in chapter 3, is when DNA methylation affects survival without directly influencing gene expression. In this case, DNA methylation does not directly alter gene transcription via promoter methylation, but may change gene expression and survival via dysregulation of microRNAs and enhancer 166 regions. In addition, DNA methylation could also affect survival by promoting/hindering genomic fragility and instability. Survival in patients with tumors having a hypermethylator phenotype, as seen with the IDH-associated hypermethylator phenotype discussed in chapter 2, has often been associated with promoter methylation-induced silencing of tumor suppresser genes. However, with increased knowledge of the role epigenetics plays in cancer and survival, promoter methylation may not be the only relevant epigenetic mechanism at play in the methylator phenotype. As previously discussed, alternative mechanisms such as methylation inhibition of microRNAs and their target regions and methylation of distal sites associated with enhancer or polycomb regions could impact gene expression and survival. Additionally, tumors with a hypermethylator phenotype are generally associated with increased survival. This could also be explained by the enhanced genomic stability observed in these tumors. In cancer, genomic instability is associated with hypomethylation, which can cause increased expression of aberrant transposons and a potential subsequent decrease in survival. Theoretically, the increased hypermethylation seen in CIMP-positive tumors could manifest in methylation of transposons/repeat elements, thereby increasing stability relative to CIMP-negative tumors. This is another possible explanation for the increased survival associated with CIMP-positive tumors vs CIMP-negative tumors. Collectively, this research has enhanced our knowledge of gliomagenesis and has further demonstrated the molecular complexity of glioma. 167 Future Directions Though our method was successful in demonstrating the importance of integration of molecular phenotypes in the classification of molecular drivers of malignancy in glioma, it will be important to properly validate these analyses with a separate set of GBMs. Additionally, our methods could benefit from the use of new high-throughput DNA methylation arrays, such as the HumanMethylation450 BeadChip array (Illumina), which allows for increased coverage of the genome compared to the HumanMethylation27 array, including the interrogation of CpG shores, whose differential methylation patterns have become increasingly recognized as important biomarkers of disease. This increased coverage would allow us to look at the correlation between DNA methylation and gene expression beyond the transcription start site. Following the example of Aran et al, who looked at DNA methylation of distal regulatory sites70, we could tailor our integrated analysis to focus on the association of CpG shore methylation with both gene expression and survival. It has been shown that altered methylation at distal regulatory sites found within enhancer regions correlates with altered gene expression more strongly than altered methylation at promoter regions70. With this knowledge, it would be expected that our analysis would demonstrate increased expression-based associations with survival. 168 References 1. Wen PY, Kesari S. Malignant gliomas in adults. N Engl J Med. Jul 2008;359(5):492-507. 2. Masui K, Cloughesy TF, Mischel PS. Review: molecular pathology in adult high-grade gliomas: from molecular diagnostics to target therapies. Neuropathol Appl Neurobiol. Jun 2012;38(3):271-291. 3. Weller M, Cloughesy T, Perry JR, Wick W. Standards of care for treatment of recurrent glioblastoma--are we there yet? Neuro Oncol. Jan 2013;15(1):4-27. 4. Mohammed W, Xunning H, Haibin S, Jingzhi M. Clinical applications of susceptibility-weighted imaging in detecting and grading intracranial gliomas: a review. Cancer Imaging. 2013;13:186-195. 5. Clarke JL, Chang SM. Neuroimaging: diagnosis and response assessment in glioblastoma. Cancer J. 2012 Jan-Feb 2012;18(1):26-31. 6. Sanai N, Berger MS. Glioma extent of resection and its impact on patient outcome. Neurosurgery. Apr 2008;62(4):753-764; discussion 264-756. 7. Noushmehr H, Weisenberger D, Diefes K, et al. Identification of a CpG island methylator phenotype that defines a distinct subgroup of glioma. Cancer Cell. May 2010;17(5):510-522. 8. 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. Mar 2006;9(3):157-173. 9. Christensen BC, Smith AA, Zheng S, et al. DNA methylation, isocitrate 169 dehydrogenase mutation, and survival in glioma. Journal of the National Cancer Institute. Jan 19 2011;103(2):143-153. 10. Verhaak RG, Hoadley KA, Purdom E, et al. Integrated genomic analysis identifies clinically relevant subtypes of glioblastoma characterized by abnormalities in PDGFRA, IDH1, EGFR, and NF1. Cancer Cell. Jan 2010;17(1):98-110. 11. von Deimling A, Korshunov A, Hartmann C. The next generation of glioma biomarkers: MGMT methylation, BRAF fusions and IDH1 mutations. Brain Pathol. Jan 2011;21(1):74-87. 12. Watanabe T, Yokoo H, Yokoo M, Yonekawa Y, Kleihues P, Ohgaki H. Concurrent inactivation of RB1 and TP53 pathways in anaplastic oligodendrogliomas. J Neuropathol Exp Neurol. Dec 2001;60(12):1181-1189. 13. Bello MJ, Rey JA. The p53/Mdm2/p14ARF cell cycle control pathway genes may be inactivated by genetic and epigenetic mechanisms in gliomas. Cancer genetics and cytogenetics. Jan 2006;164(2):172-173. 14. Amatya VJ, Naumann U, Weller M, Ohgaki H. TP53 promoter methylation in human gliomas. Acta Neuropathol. Aug 2005;110(2):178-184. 15. Costello JF, Berger MS, Huang HS, Cavenee WK. Silencing of p16/CDKN2 expression in human gliomas by methylation and chromatin condensation. Cancer Res. May 1996;56(10):2405-2410. 16. Heimberger AB, Suki D, Yang D, Shi W, Aldape K. The natural history of EGFR and EGFRvIII in glioblastoma patients. J Transl Med. Oct 2005;3:38. 17. Huse J, Holland E. Targeting brain cancer: advances in the molecular 170 pathology of malignant glioma and medulloblastoma. Nat Rev Cancer. May 2010;10(5):319-331. 18. Norden AD, Drappatz J, Wen PY. Antiangiogenic therapies for high-grade glioma. Nat Rev Neurol. Nov 2009;5(11):610-620. 19. Cohen MH, Shen YL, Keegan P, Pazdur R. FDA drug approval summary: bevacizumab (Avastin) as treatment of recurrent glioblastoma multiforme. Oncologist. Nov 2009;14(11):1131-1138. 20. Ebos JM, Lee CR, Cruz-Munoz W, Bjarnason GA, Christensen JG, Kerbel RS. Accelerated metastasis after short-term treatment with a potent inhibitor of tumor angiogenesis. Cancer Cell. Mar 2009;15(3):232-239. 21. Pàez-Ribes M, Allen E, Hudock J, et al. Antiangiogenic therapy elicits malignant progression of tumors to increased local invasion and distant metastasis. Cancer Cell. Mar 2009;15(3):220-231. 22. You WK, Sennino B, Williamson CW, et al. VEGF and c-Met blockade amplify angiogenesis inhibition in pancreatic islet cancer. Cancer Res. Jul 2011;71(14):4758-4768. 23. Narayana A, Kelly P, Golfinos J, et al. Antiangiogenic therapy using bevacizumab in recurrent high-grade glioma: impact on local control and patient survival. J Neurosurg. Jan 2009;110(1):173-180. 24. Oda K, Matsuoka Y, Funahashi A, Kitano H. A comprehensive pathway map of epidermal growth factor receptor signaling. Mol Syst Biol. 2005;1:2005.0010. 25. Taylor TE, Furnari FB, Cavenee WK. Targeting EGFR for treatment of glioblastoma: molecular basis to overcome resistance. Curr Cancer Drug 171 Targets. Mar 2012;12(3):197-209. 26. Turner BM. Cellular memory and the histone code. Cell. Nov 2002;111(3):285-291. 27. Riggs AD. X inactivation, differentiation, and DNA methylation. Cytogenet Cell Genet. 1975;14(1):9-25. 28. Holliday R. The inheritance of epigenetic defects. Science. Oct 1987;238(4824):163-170. 29. Costa FF. Non-coding RNAs, epigenetics and complexity. Gene. Feb 2008;410(1):9-17. 30. Chuang JC, Jones PA. Epigenetics and microRNAs. Pediatr Res. May 2007;61(5 Pt 2):24R-29R. 31. Berger SL, Kouzarides T, Shiekhattar R, Shilatifard A. An operational definition of epigenetics. Genes Dev. Apr 2009;23(7):781-783. 32. Bird A. Perceptions of epigenetics. Nature. May 2007;447(7143):396-398. 33. Smith ZD, Meissner A. DNA methylation: roles in mammalian development. Nat Rev Genet. Mar 2013;14(3):204-220. 34. Jaenisch R, Bird A. Epigenetic regulation of gene expression: how the genome integrates intrinsic and environmental signals. Nat Genet. Mar 2003;33 Suppl:245-254. 35. Portela A, Esteller M. Epigenetic modifications and human disease. Nat Biotechnol. Oct 2010;28(10):1057-1068. 36. Esteller M, Garcia-Foncillas J, Andion E, et al. Inactivation of the DNA-repair gene MGMT and the clinical response of gliomas to alkylating agents. N Engl J 172 Med. Nov 2000;343(19):1350-1354. 37. Hegi ME, Diserens AC, Gorlia T, et al. MGMT gene silencing and benefit from temozolomide in glioblastoma. N Engl J Med. Mar 2005;352(10):997-1003. 38. Garcia-Manero G, Kantarjian HM, Sanchez-Gonzalez B, et al. Phase 1/2 study of the combination of 5-aza-2'-deoxycytidine with valproic acid in patients with leukemia. Blood. Nov 2006;108(10):3271-3279. 39. Kaminskas E, Farrell A, Abraham S, et al. Approval summary: azacitidine for treatment of myelodysplastic syndrome subtypes. Clinical cancer research : an official journal of the American Association for Cancer Research. May 2005;11(10):3604-3608. 40. Issa JP, Gharibyan V, Cortes J, et al. Phase II study of low-dose decitabine in patients with chronic myelogenous leukemia resistant to imatinib mesylate. Journal of clinical oncology : official journal of the American Society of Clinical Oncology. Jun 2005;23(17):3948-3956. 41. PM H, Z L, H K. Demethylation Agents in the Treatment of Cancer. Pharmaceuticals. July 2 2010 2010;3:2022-2044. 42. Houseman EA, Accomando WP, Koestler DC, et al. DNA methylation arrays as surrogate measures of cell mixture distribution. BMC Bioinformatics. 2012;13:86. 43. Rakyan VK, Down TA, Thorne NP, et al. An integrated resource for genome- wide identification and analysis of human tissue-specific differentially methylated regions (tDMRs). Genome Res. Sep 2008;18(9):1518-1529. 44. Hegi M, Diserens A, Gorlia T, et al. MGMT gene silencing and benefit from 173 temozolomide in glioblastoma. N Engl J Med. Mar 2005;352(10):997-1003. 45. Feinberg AP, Vogelstein B. Hypomethylation of ras oncogenes in primary human cancers. Biochemical and biophysical research communications. Feb 1983;111(1):47-54. 46. Gama-Sosa MA, Slagel VA, Trewyn RW, et al. The 5-methylcytosine content of DNA from human tumors. Nucleic acids research. Oct 1983;11(19):6883- 6894. 47. Olar A, Aldape KD. Using the Molecular Classification of Glioblastoma to Inform Personalized Treatment. J Pathol. Oct 2013. 48. Network CGAR. Comprehensive genomic characterization defines human glioblastoma genes and core pathways. Nature. Oct 2008;455(7216):1061- 1068. 49. Frattini V, Trifonov V, Chan JM, et al. The integrated landscape of driver genomic alterations in glioblastoma. Nat Genet. Aug 2013. 50. Parsons DW, Jones S, Zhang X, et al. An integrated genomic analysis of human glioblastoma multiforme. Science. Sep 2008;321(5897):1807-1812. 51. Hartmann C, Meyer J, Balss J, et al. Type and frequency of IDH1 and IDH2 mutations are related to astrocytic and oligodendroglial differentiation and age: a study of 1,010 diffuse gliomas. Acta Neuropathol. Oct 2009;118(4):469-474. 52. Ogino S, Kawasaki T, Kirkner G, Loda M, Fuchs C. CpG island methylator phenotype-low (CIMP-low) in colorectal cancer: possible associations with male sex and KRAS mutations. J Mol Diagn. Nov 2006;8(5):582-588. 174 53. Dang L, White D, Gross S, et al. Cancer-associated IDH1 mutations produce 2- hydroxyglutarate. Nature. Dec 2009;462(7274):739-744. 54. Xu W, Yang H, Liu Y, et al. Oncometabolite 2-hydroxyglutarate is a competitive inhibitor of α-ketoglutarate-dependent dioxygenases. Cancer Cell. Jan 2011;19(1):17-30. 55. Turcan S, Rohle D, Goenka A, et al. IDH1 mutation is sufficient to establish the glioma hypermethylator phenotype. Nature. Mar 2012;483(7390):479-483. 56. Jin B, Li Y, Robertson KD. DNA methylation: superior or subordinate in the epigenetic hierarchy? Genes & cancer. Jun 2011;2(6):607-617. 57. Friedrich M, Höss N, Stögbauer F, et al. Complete inhibition of in vivo glioma growth by oncostatin M. J Neurochem. Mar 2001;76(5):1589-1592. 58. Hibi K, Goto T, Sakuraba K, et al. Methylation of OSMR gene is frequently observed in non-invasive colorectal cancer. Anticancer Res. Apr 2011;31(4):1293-1295. 59. Huebert RC, Jagavelu K, Hendrickson HI, et al. Aquaporin-1 promotes angiogenesis, fibrosis, and portal hypertension through mechanisms dependent on osmotically sensitive microRNAs. Am J Pathol. Oct 2011;179(4):1851-1860. 60. Rutkovskiy A, Valen G, Vaage J. Cardiac aquaporins. Basic Res Cardiol. Nov 2013;108(6):393. 61. El Hindy N, Bankfalvi A, Herring A, et al. Correlation of aquaporin-1 water channel protein expression with tumor angiogenesis in human astrocytoma. Anticancer Res. Feb 2013;33(2):609-613. 175 62. Sepramaniam S, Armugam A, Lim KY, et al. MicroRNA 320a functions as a novel endogenous modulator of aquaporins 1 and 4 as well as a potential therapeutic target in cerebral ischemia. The Journal of biological chemistry. Sep 2010;285(38):29223-29230. 63. Nord H, Hartmann C, Andersson R, et al. Characterization of novel and complex genomic aberrations in glioblastoma using a 32K BAC array. Neuro Oncol. Dec 2009;11(6):803-818. 64. Yu Y, Yoon SO, Poulogiannis G, et al. Phosphoproteomic analysis identifies Grb10 as an mTORC1 substrate that negatively regulates insulin signaling. Science. Jun 2011;332(6035):1322-1326. 65. Monk D, Arnaud P, Frost J, et al. Reciprocal imprinting of human GRB10 in placental trophoblast and brain: evolutionary conservation of reversed allelic expression. Human molecular genetics. Aug 2009;18(16):3066-3074. 66. Vecchione A, Marchese A, Henry P, Rotin D, Morrione A. The Grb10/Nedd4 complex regulates ligand-induced ubiquitination and stability of the insulin- like growth factor I receptor. Mol Cell Biol. May 2003;23(9):3363-3372. 67. Maunakea AK, Chepelev I, Cui K, Zhao K. Intragenic DNA methylation modulates alternative splicing by recruiting MeCP2 to promote exon recognition. Cell Res. Nov 2013;23(11):1256-1269. 68. Desbuquois B, Carré N, Burnol AF. Regulation of insulin and type 1 insulin- like growth factor signaling and action by the Grb10/14 and SH2B1/B2 adaptor proteins. The FEBS journal. Feb 2013;280(3):794-816. 69. Bibikova M, Le J, Barnes B, et al. Genome-wide DNA methylation profiling 176 using Infinium® assay. Epigenomics. Oct 2009;1(1):177-200. 70. Aran D, Sabato S, Hellman A. DNA methylation of distal regulatory sites characterizes dysregulation of cancer genes. Genome Biol. Mar 2013;14(3):R21. 71. Shukla S, Pia Patric IR, Thinagararjan S, et al. A DNA Methylation Prognostic Signature of Glioblastoma: Identification of NPTX2-PTEN-NF-κB Nexus. Cancer Res. Nov 2013;73(22):6563-6573. 177