Diagnostic Use of IDH1/2 Mutation Analysis in Routine Clinical Testing of Formalin-Fixed, Paraffin-Embedded Glioma Tissues

Transcription

1 J Neuropathol Exp Neurol Copyright Ó 2009 by the American Association of Neuropathologists, Inc. Vol. 68, No. 12 December 2009 pp. 1319Y1325 ORIGINAL ARTICLE Diagnostic Use of IDH1/2 Mutation Analysis in Routine Clinical Testing of Formalin-Fixed, Paraffin-Embedded Glioma Tissues Craig Horbinski, MD, PhD, Julia Kofler, MD, Lindsey M. Kelly, Geoffrey H. Murdoch, MD, PhD, and Marina N. Nikiforova, MD Abstract Mutations in isocitrate dehydrogenase enzyme isoforms 1 (IDH1) and 2 (IDH2) have been identified in many adult astrocytomas and oligodendrogliomas. These mutations are targeted to specific codons (e.g. R132 in IDH1 and R172 in IDH2), making assays to detect them in clinical specimens feasible. We describe a simple and accurate molecular assay for detection of IDH1/2 mutations on routine formalin-fixed paraffin-embedded tissues. Using this polymerase chain reactionybased assay, we tested 75 glial neoplasms and 57 nonneoplastic conditions that can mimic gliomas including radiation changes, viral infections, and infarcts. Of the gliomas, 37 (49%) were positive for IDH1 or IDH2 mutations; the most common mutation was IDH1 (97%). Two of 12 gangliogliomas were positive for IDH1 mutation, and both had unfavorable clinical outcomes (p G 0.03). None of the nonneoplastic cases were positive for IDH mutations. The assay detected IDH mutations in biopsy material containing mostly glioma and in concomitant near-miss stereotactic core biopsies that were otherwise equivocal for the presence of glioma by light microscopy. These results indicate that testing for IDH1/2 mutations can be effectively performed in a clinical setting and can enhance the accuracy of diagnosis of gliomas when traditional diagnostic methods are not definitive. Key Words: Ganglioglioma, Glioma, Isocitrate dehydrogenase, Molecular genetics, Paraffin sections, Pediatric oligodendroglioma, Stereotactic biopsy. INTRODUCTION Mutations of isocitrate dehydrogenase enzyme isoforms 1(IDH1) and 2(IDH2) have recently been found in a large proportion of diffuse astrocytic and oligodendroglial neoplasms (1Y4). Isocitrate dehydrogenase enzyme isoform 1 and IDH2 catalyze the conversion of isocitrate to >-ketoglutarate while reducing nicotinamide adenine dinucleotide phosphate (NADP + ). Isocitrate dehydrogenase enzyme isoform 1 is located in peroxisomes, whereas IDH2 is present in mitochondria (5). Approximately 90% of IDH1 mutations occur in exon From the Department of Pathology, University of Kentucky, Kentucky (CH); and Department of Pathology, University of Pittsburgh, Pittsburgh, Pennsylvania (JK, LMK, GHM, MNN). Send correspondence and reprint requests to: Marina N. Nikiforova, MD, C601, 200 Lothrop St, Pittsburgh, PA 15213; nikiforovamn@ upmc.edu Craig Horbinski was supported by a Callie Rohr American Brain Tumor Association Fellowship. Online-only color figures are available at 4 at codon 132, where a CGTYCAT transition changes a single amino acid from arginine to histidine (R132H). Other transitions and transversions have been found but, so far, are restricted to codon 132, which is in the isocitrate-binding pocket of IDH1. Although the mutation is always heterozygous, it exerts a dominant negative inhibition of IDH1 dimer activity (6). Less common IDH2 mutations occur in an analogous codon at position R 172 (4). It is not clear whether depletion of >-ketoglutarate or NADPH reducing equivalents fully explains the actions of these apparently pro-oncogenic mutants. Despite the unresolved mechanism of IDH1/2 mutation effects, they have been shown to be specific for gliomas compared with non-cns neoplasms (2, 4); rarely, however, carcinomas and leukemias harbor IDH1 mutations (7). In addition, IDH1/2 mutations may be prognostic factors associated with longer survival compared with grade-matched gliomas that do not harbor such alterations (3, 4, 8). These features make IDH1 and IDH2 attractive targets for ancillary molecular testing in tissues for both diagnostic and prognostic purposes. Prior studies describing IDH mutations predominantly used archival snap-frozen tissue of biopsies. We report a simple and accurate molecular assay for detection of IDH1/2 mutation in routine formalin-fixed paraffin-embedded (FFPE) sections and demonstrate diagnostic use and reliability of this assay for clinical assessment of gliomas. MATERIALS AND METHODS Study Design and Histological Processing Formalin-fixed paraffin-embedded tissue from archival surgical specimens was selected via electronic searching of the institutional pathology laboratory information system during the past 5 years, in keeping with University of Pittsburgh Institutional Review Board policies. We assessed 132 cases including 75 neoplasms and 57 nonneoplastic lesions. All gliomas were re-evaluated to be certain that grading followed World Health Organization 2007 criteria (9). Blocks that contained adequate tissue were sectioned at a thickness of 5 Km. Sections were stained with hematoxylin and eosin to confirm that diagnostic tissue had been retained; adjacent sections were evaluated with the IDH1/2 mutation assay. All molecular analyses and interpretations were performed blinded to diagnosis. IDH1/2 Assays Tumor targets were manually microdissected from 5-Km unstained histological sections. DNA was isolated from each J Neuropathol Exp Neurol Volume 68, Number 12, December

2 Horbinski et al J Neuropathol Exp Neurol Volume 68, Number 12, December 2009 TABLE 1. IDH Mutations in Formalin-Fixed Paraffin-Embedded Tumor Biopsy Samples Neoplasm Type No. Examined IDH1 Mutation % IDH2 Mutation % Fibrillary astrocytoma grade Anaplastic astrocytoma grade Glioblastoma grade Oligodendroglioma grade Anaplastic oligodendroglioma grade 3 Ganglioglioma Pilocytic astrocytoma grade Pediatric oligodendroglioma grade 2 Oligoastrocytoma grade Chordoid glioma Diffuse large B-cell lymphoma Hemangioblastoma Hemangiopericytoma Myxopapillary ependymoma Total Cases were selected and analyzed for IDH1 and IDH2 point mutations at codons 132 and 172, respectively. The 2 cases of ganglioglioma with IDH1 mutations both had atypical aggressive clinical courses; the pediatric oligodendroglioma case was also positive for IDH1 mutation. target using the DNeasy Blood and Tissue kit (Qiagen, Valencia, CA), according to the manufacturer s instructions. The quantity of isolated DNA was assessed using a NanoDrop 1000 spectrophotometer (Thermo Scientific, Wilmington, DE). For the detection of IDH mutations, forward and reverse primers were designed to amplify exon 4 (codon R132) of the IDH1 gene and exon 4 (codon R172) of the IDH2 gene using Primer 3 software ( The IDH1 forward primer (5 -ACC AAA TGG CAC CAT ACG A-3 ) and reverse primer (5 -GCA AAA TCA CAT TAT TGC CAA C-3 ) generated a 130-bp polymerase chain reaction (PCR) product; IDH2 forward primer 5 -GCT GCA GTG GGA CCA CTA TT-3 ) and reverse primer (5 -TGT GGC CTT GTA CTG CAG AG-3 ) generated a 293-bp PCR product. Both primers were acceptable for amplification of DNA from FFPE tissue samples. Polymerase chain reaction amplification was performed using 5 to 50 ng of DNA, 0.2 Hmol of each primer, and AmpliTaq Gold PCR Master Mix (Applied Biosystems, Inc, Foster City, CA). The reaction mixture was subjected to an initial denaturation of 95-C for 10 minutes, followed by 35 cycles of amplification, consisting of denaturation at 95-C for 30 seconds, annealing at 55-C for 30 seconds, and extension 72-C for 60 seconds in a total volume of 50 KL. The PCR products were purified using the MinElute PCR Purification kit (Qiagen). The PCR products were then sequenced in both sense and antisense directions with the primers listed above using the BigDye Terminator v3.1 Cycle Sequencing kit (Applied Biosystems). The sequencing reaction mixture was subjected to 25 cycles of amplification consisting of denaturation at 95-C for seconds, annealing at 55-C for 15 seconds, and extension at 60-C for 4 minutes in a total volume of 20 KL. The sequencing products were then purified using Centri-Sep spin columns (Princeton Separations, Freehold, NJ) and analyzed by capillary gel electrophoresis on ABI3130 (Applied Biosystems). The Mutation Surveyor software (SoftGenetics, LLC, State College, PA) was used to assist with the interpretation of the sequence electropherograms. Each case was classified as positive or negative for the IDH mutation based on the sequencing results. Each specimen was evaluated for numbers of viable and atypical cells. The sensitivity of detection with this assay was found to be approximately 100 viable nuclei in a microdissected specimen and at least 20% IDH mutant alleles in a background of normal alleles. The testing was not conducted when the specimen did not meet these test requirements. Statistical Analysis The impact of IDH1 status on risk of progression in gangliogliomas was determined with a contingency table and Fisher exact test using GraphPad software (La Jolla, CA). Results were considered significant when the 2-sided value of p G RESULTS A total of 132 FFPE specimens, including 75 neoplasms and 57 nonneoplastic conditions, were analyzed for the presence of IDH1 and IDH2 mutations (Tables 1 and 2). Mutations were found in 37 (49%) of the tumors, all of which were gliomas; no mutation was identified in nonneoplastic lesions. Of the mutations, 97% were in the IDH1 gene; 1 mutation (3%) was detected in IDH2. Representative positive mutations on sequencing are shown in Figure 1. TABLE 2. IDH1 and IDH2 Mutations Are Not Present in Nonneoplastic Conditions That Mimic Glioma IDH1 Mutation IDH2 Mutation No. Nonneoplastic Diagnosis Examined Ischemia/infarct Reactive gliosis around metastatic tumor Reactive gliosis, idiopathic Abscess Vasculopathy/malformation Viral or Toxoplasma encephalitis Demyelination Inflammation NOS Radiation-induced damage Vasculitis PML Hippocampal sclerosis Shunt-induced reactive gliosis Total Fifty-seven cases representing a variety of nonneoplastic conditions that frequent mimic gliomas were analyzed for IDH1 and IDH2 point mutations at codons 132 and 172, respectively. NOS, not otherwise specified; PML, progressive multifocal leukoencephalopathy. Ó 2009 American Association of Neuropathologists, Inc.

3 J Neuropathol Exp Neurol Volume 68, Number 12, December 2009 IDH1/2 Assay for Clinical Use FIGURE 1. IDH1/2 mutations in formalin-fixed paraffin-embedded samples. Unstained slides containing glioma tissue were sequenced for IDH1 and IDH2 mutations. A total of 85.3% of IDH1-mutated cases had CGTYCAT at codon 132 (R132H). Other mutations included CGTYTGT (R132C), CGTYAGT (R132S), CGTYCTT (R132L), and CGTYGGT (R132G). Only 1 IDH2 mutation was found, an AGGYATG (R172M). IDH1/2 Mutations in Glioma Specimens In the neoplastic cases, 67.3% of the grades 2 and 3 diffuse gliomas were positive for IDH1 mutations and 16.7% of the glioblastomas were positive. Of all IDH1-positive cases, 85.3% (29/34) were CGTYCAT at codon 132 (R132H). The remaining cases included 2 CGTYAGT (R132S) in grade 2 and grade 3 oligodendrogliomas; CGTYTGT (R132C) in a grade 3 oligodendroglioma; CGTYGGT (R132G) in a grade 2 fibrillary astrocytoma; and CGTYCTT (R132L) in a secondary glioblastoma. Only 1 case was positive for an IDH2 mutation, that is, an AGGYATG transversion (R172M) in a grade 2 oligodendroglioma. No case was positive for both IDH1 and IDH2 mutations. These results and detection frequencies are in agreement with prior studies (1Y4, 8, 10), indicating that IDH sequencing reliably detects mutations in FFPE tissues. In addition, a selection of less common CNS neoplasms was tested, which included a previously reported chordoid glioma (11) and a grade 2 pediatric oligodendroglioma. The chordoid glioma was negative for mutations at both IDH1 and IDH2 codons; the pediatric oligodendroglioma was positive for the common CGTYCAT transition at codon 132 (R132H) on IDH1 (Table 1). IDH1/2 Mutations in Glioma Nonneoplastic Conditions That Mimic Glioma Although IDH1 and IDH2 mutations have been extensively studied in gliomas, little information is available on the presence of these mutations in nonneoplastic tissue samples. Such data are essential before implementation of IDH testing on a routine clinical basis. Fifty-seven nonneoplastic lesions were assessed for IDH1 and IDH2, including conditions that frequently resemble gliomas by light microscopy. None of the cases, including viral infections, radiation-induced changes, and reactive gliosis around metastatic tumors, showed IDH1 or IDH2 mutations (Table 2). These results are important for clinical applications because the demonstration of an IDH mutation indicates the presence of a glioma and that it cannot be attributed to nonneoplastic diseases. IDH Mutations in Suboptimal Tumor Biopsies To assess whether PCR-based assays can detect IDH mutations at the outer edge of infiltrating tumors (a common clinical problem in the evaluation of some stereotactically obtained biopsies), multipart cases were selected that had nondiagnostic core biopsies from the periphery of the tumor in 1 part and a separate part containing unequivocal glioma that was positive for an IDH mutation. Typically, the diagnostic tissue was obtained by the neurosurgeon after an intraoperative consultation determined that the initial biopsy was nondiagnostic. In 3 of 4 cases meeting those criteria a peripheral biopsy was positive for IDH1 or IDH2 mutation although the diagnosis on tissue sample was indeterminate by light microscopic analysis (Fig. 2). In all 3 of these cases, the mutation exactly matched the central diagnostic tissue result. One oligodendroglioma had 2 peripheral biopsies, but in only 1 of the 2 was the IDH2 mutation detected (Figs. 2MYR). These results demonstrate that IDH mutations can be detected even in equivocal biopsies. IDH1 Mutations in Gangliogliomas Included in the neoplastic part of this validation study were 12 gangliogliomas. Of those cases, 2 were positive for the CGTYCAT transition at codon 132 (R132H) of the IDH1 gene (Fig. 3). Interestingly, of the 10 gangliogliomas for which clinical follow-up data were available, only the 2 cases with IDH1 mutations progressed to high-grade gliomas, each within a relatively short time (Table 3). These differences in outcomes were statistically significant (p G 0.03). Of note, the IDH1-positive ganglioglioma with Batypical features[ (Case 2, Table 3) showed hypercellularity, nuclear atypia, and a focal FIGURE 2. IDH mutations are detectable in equivocal formalin-fixed paraffin-embedded samples. Both central CT stereotactic core biopsies with diagnostic tissue ( central bx ) and equivocal biopsies of tumor periphery ( peripheral bx ) were assayed for IDH1 and IDH2 mutations. Case 1 is a gemistocytic astrocytoma (AYF), Case 2 is a grade 3 anaplastic astrocytoma (GYL), and Case 3 is a grade 2 oligodendroglioma (MYR). In each case, the peripheral biopsy showed reactive changes and hypercellularity suspicious for glioma but were nondiagnostic and not representative of the tumor ([D] and [E], [J] and [K], [P] and [Q]). The indeterminate tissue from Cases 1 and 2 revealed IDH1 mutations identical to their corresponding central biopsies ([C] and [F], [I] and [L]). The IDH2 mutation in Case 3 diagnostic tissue (O) was also identified in the peripheral tissue (R). Ó 2009 American Association of Neuropathologists, Inc. 1321

4 Horbinski et al J Neuropathol Exp Neurol Volume 68, Number 12, December Ó 2009 American Association of Neuropathologists, Inc.

5 J Neuropathol Exp Neurol Volume 68, Number 12, December 2009 IDH1/2 Assay for Clinical Use Ó 2009 American Association of Neuropathologists, Inc. 1323

6 Horbinski et al J Neuropathol Exp Neurol Volume 68, Number 12, December 2009 TABLE 3. IDH1 Mutations and Outcomes in Gangliogliomas Case Age, Years Sex Location Radiation Therapy IDH1 Status Follow-Up Interval, Years Outcome 1 11 F Right temporal No Negative 4 No evidence of recurrence 2 51 F Left parietal Yes Positive 4 Prior resection was ganglioglioma with Batypical features[; progressed to glioblastoma in 4 years 3 10 M Right temporal No Negative 5 Recurrent ganglioglioma 3 years later; no other recurrence thereafter 4 1 M Right parietal No Negative 3 No evidence of recurrence 5 12 M Bifrontal No Negative 5 No evidence of recurrence 6 45 M Left frontal Yes Positive 4 Progressed to anaplastic oligodendroglioma in 2 years 7 8 M Left temporal No Negative 4 Small increase in size of residual tumor 8 16 M Right parietal No Negative 3 No evidence of recurrence 9 31 M Left frontal NA Negative NA NA F Left occipital No Negative 5 No evidence of recurrence M Right frontal Yes Negative 2 No evidence of recurrence M Cerebellum NA Negative NA NA Twelve cases of ganglioglioma were tested for IDH1 and IDH2 mutations. Both tumors with IDH1 mutations (Cases 2 and 6) showed high-grade progression within 2 to 4 years; none of the IDH1-negative tumors with available outcome data showed progression. No ganglioglioma was positive for an IDH2 mutation. F, female; M, male; NA, not available. Ki67 proliferation index of up to 10%. Postoperative radiotherapy was initiated when the tumor showed radiological progression. The other IDH1-positive ganglioglioma (Case 6, Table 3) had 1p19q codeletion, 1% Ki67 proliferation index, and a history of low-grade glioma (not otherwise specified) from an outside institution; adjuvant radiotherapy had been given 8 years before the ganglioglioma diagnosis was made on a resection. The original tissue was not available for analysis. DISCUSSION We describe a novel diagnostic assay for detection of IDH1 and IDH2 mutations present in a large proportion of diffuse gliomas and demonstrate the feasibility and diagnostic use of IDH testing for clinical assessment of gliomas in routine FFPE biopsies. This assay accurately detects mutations and does not generate false-positive results in samples of nonneoplastic conditions that may mimic gliomas. Moreover, this test allows for detection of the mutations even in biopsies that are equivocal on histological evaluation. The definitive identification of a biopsy sample as neoplastic is one of the key tasks for pathologists. Nowhere is this more apparent than in the practice of neuropathology, in which the diagnosis of an infiltrating glioma may trigger a series of treatments such as radiation therapy, which can result in permanent damage to the CNS. The difficulty of such diagnoses is increased by the need for small tissue biopsies of brain and spinal cord. Therefore, ancillary tests are useful in cases where the definitive diagnosis cannot be rendered on the basis of examination by light microscopy alone. Unfortunately for gliomas, the testing for markers such as Ki67 proliferation index, p53 expression, EGFR amplification, and 1p19q codeletion has limitations and, with the exception of EGFR amplification by fluorescence in situ hybridization, is usually unhelpful when dealing with biopsy samples peripheral to the tumor centers. TP53 mutations are seen in most diffuse low-grade gliomas but are scattered over multiple exons, making routine sequencing impractical. The recent discovery of IDH1/2 mutations was a major step forward in neuro-oncology research. Not only is there only 1 codon per gene to assess, but multiple studies on multiple continents have found highly concordant results (1Y4, 10). Furthermore, these mutations are mostly found in the sorts of cases that are diagnostically most often problematic (i.e. low-grade diffuse gliomas), particularly when the biopsy sample is from the periphery of a tumor. In many such cases, the assay will only be able to deliver a diagnosis of Bglioma,[ but recent work has suggested that certain IDH1 mutations (e.g. R132C) indicate astrocytic differentiation, whereas IDH2 mutations are more commonly found in oligodendrogliomas (10) (Figs. 2MYR). In some instances, the glioma diagnosis might not only be proven but also whether the tumor is more likely astrocytic or oligodendroglial. Despite the advantages of this novel IDH assay, there are certain limitations. Even a peripheral biopsy requires a certain density of glioma cells in the core; for example, this method requires at least 30% to 40% glioma cells for consistently reliable results (see Materials and Methods section). Even if there is adequate sampling of glioma nuclei for the FIGURE 3. Gangliogliomas with IDH mutations may demonstrate atypically aggressive clinical courses. None of the gangliogliomas with follow-up data (Table 3) that were negative for IDH1 underwent anaplastic transformation. (AYC) A representative example of a mutation-negative ganglioglioma. (DYO) Both tumors that were IDH1 positive had aggressive courses (Case 2 [DYI] and Case 6 [JYO]) (Table 3). Arrow in (B) highlights a binucleate cell; arrowheads in (G) and (M) highlight microvascular proliferation; arrowhead in (N) identifies a mitotic figure Ó 2009 American Association of Neuropathologists, Inc.

7 J Neuropathol Exp Neurol Volume 68, Number 12, December 2009 IDH1/2 Assay for Clinical Use assay, IDH screening alone would miss the 15% to 30% of grades 2 and 3 diffuse gliomas and secondary glioblastomas that are intrinsically negative. Furthermore, primary glioblastomas (the most common gliomas in the adult population) almost by definition are negative for IDH mutations. Fortunately, the latter group of cases frequently has EGFR amplification, which via fluorescence in situ hybridization can be detected in a few infiltrating cells on a glass slide. Thus, the combination of IDH and EGFR ancillary testing could identify a large proportion of diffuse gliomas of all grades, even with suboptimal tumor sampling. Nevertheless, because negative results do not exclude the diagnosis of glioma, this assay should be used only as an ancillary test and not as a substitute for analysis by light microscopy. The IDH mutation screening has been shown to assist in differentiating between pilocytic and diffuse astrocytomas (12), and was recently described in gangliogliomas. Consistent with their World Health Organization grade 1 status, ganglioglioma patient survival after surgery is high and malignant progression is uncommon (13); there may, however, be a link between adjuvant radiation therapy and malignant transformation (14, 15). Nevertheless, in this small cohort, the only gangliogliomas with high-grade transformation and poor outcomes were those that harbored IDH1 mutations before progression (Table 3, Fig. 3). This raises the possibility that such tumors may be variants of ganglioglioma or perhaps are better classified as diffuse gliomas of either astrocytic or oligodendroglial lineage (e.g. Case 6 in Table 3, Figs. 3JYO). A larger study specifically addressing this question will be necessary to validate and extend these observations. In summary, detection of IDH1/2 mutations is feasible, sufficiently robust on FFPE tissue samples, and can be implemented for routine clinical use. It provides additional diagnostic and prognostic information, particularly on small surgical biopsies. It also may reduce the need for additional surgeries and provide important information in cases with equivocal histological features. ACKNOWLEDGMENT The authors thank Colleen Lovell for her histological work. REFERENCES 1. Balss J, Meyer J, Mueller W, et al. Analysis of the IDH1 codon 132 mutation in brain tumors. Acta Neuropathol 2008;116:597Y Bleeker FE, Lamba S, Leenstra S, et al. IDH1 mutations at residue p.r132 (IDH1(R132)) occur frequently in high-grade gliomas but not in other solid tumors. Hum Mutat 2009;30:7Y11 3. Parsons DW, Jones S, Zhang X, et al. An integrated genomic analysis of human glioblastoma multiforme. Science 2008;321:1807Y12 4. Yan H, Parsons DW, Jin G, et al. IDH1 and IDH2 mutations in gliomas. N Engl J Med 2009;360:765Y73 5. Thompson CB. Metabolic enzymes as oncogenes or tumor suppressors. N Engl J Med 2009;360:813Y15 6. Zhao S, Lin Y, Xu W, et al. Glioma-derived mutations in IDH1 dominantly inhibit IDH1 catalytic activity and induce HIF-1alpha. Science 2009;324:261Y65 7. Kang MR, Kim MS, Oh JE, et al. Mutational analysis of IDH1 codon 132 in glioblastomas and other common cancers. Int J Cancer 2009;125:353Y55 8. Sanson M, Marie Y, Paris S, et al. Isocitrate dehydrogenase 1 codon 132 mutation is an important prognostic biomarker in gliomas. J Clin Oncol 2009;27:4150Y54 9. Louis DN, Ohgaki H, Wiestler OD, et al. WHO Classification of Tumors of the Central Nervous System. 4th ed. Lyon, France: IARC; 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 2009;118:469Y Horbinski C, Dacic S, McLendon RE, et al. Chordoid glioma: A case report and molecular characterization of five cases. Brain Pathol 2009; 19:439Y Korshunov A, Meyer J, Capper D, et al. Combined molecular analysis of BRAF and IDH1 distinguishes pilocytic astrocytoma from diffuse astrocytoma. Acta Neuropathol 2009;118:401Y5 13. Im SH, Chung CK, Cho BK, et al. Intracranial ganglioglioma: Preoperative characteristics and oncologic outcome after surgery. J Neurooncol 2002;59:173Y Liauw SL, Byer JE, Yachnis AT, et al. Radiotherapy after subtotally resected or recurrent ganglioglioma. Int J Radiat Oncol Biol Phys 2007; 67:244Y Rumana CS, Valadka AB. Radiation therapy and malignant degeneration of benign supratentorial gangliogliomas. Neurosurgery 1998;42:1038Y43 Ó 2009 American Association of Neuropathologists, Inc. 1325