Relatively little is known about the molecular changes that promote the formation or growth of pilocytic astrocytomas. We investigated genomic alterations in 25 pilocytic astrocytomas, including 5 supratentorial and 20 posterior fossa tumors, using oligonucleotide array comparative genomic hybridization. Large changes were identified in 7 tumors and included gains of chromosomes 5, 6, and 7 and losses of chromosomes 16, 17, 19, and 22. The most common alteration was a 1.9-MB region of low-level gain at chromosome 7q34 identified in 17 of 20 posterior fossa tumors. In most tumors, the region of gain ended within the BRAF locus and encompassed only exons that encode the BRAF kinase domain. We confirmed copy number increase at the 7q34 locus using quantitative polymerase chain reaction with primers adjacent to the HIPK2, RAB19B, and BRAF genes. Western blot analysis revealed that 3 of 6 pilocytic astrocytomas with 7q34 gain contained high levels of phosphorylated extracellular signal-related kinase (ERK) and nitrogen-activated protein kinase/ERK kinase (MEK), while 1 tumor lacking 7q34 gain and 2 normal brain specimens did not. Immunohistochemical stains of a tissue microarray containing 43 pilocytic astrocytoma identified ERK phosphorylation in 35 (81%). These data indicate that focal gains at chromosome 7q34 and increased BRAF-MEK-ERK signaling are common findings in sporadic pilocytic astrocytomas.
Comparative genomic hybridization
Pilocytic astrocytomas are World Health Organization grade I gliomas that can arise throughout the central nervous system (1). They are diagnosed most commonly in the first 2 decades of life but are also sometimes found in young or middle-aged adults. In one single-institution study examining 1,195 pediatric brain tumors diagnosed between 1974 and 2003, pilocytic astrocytomas made up 18% of the intracranial lesions and represented the single most common tumor type (2). Although progression-free survival for pilocytic astrocytomas is very good overall, many children suffer significant morbidity from their disease, and some die, particularly those whose tumors are of the pilomyxoid variant (1, 3, 4).
It is hoped that a better understanding of the molecular changes that drive the initiation and growth of pilocytic astrocytoma will lead to more effective therapies. Relatively little is currently known, however, regarding such molecular events. Gene expression studies indicate that the overall expression profile of pilocytic tumors is distinct from that of infiltrating astrocytomas (5, 6). In addition, molecular alterations such as epidermal growth factor receptor amplification, p53 mutation, and PTEN loss, which are common in infiltrating astrocytomas, particularly those in adults, are almost never identified in pilocytic astrocytomas (1). Pilocytic astrocytomas, particularly in the optic nerve, are sometimes associated with neurofibromatosis type 1 in which loss of the NF1 gene product results in activation of the Ras and cyclic adenosine monophosphate signaling pathways (7, 8). The more common sporadic pilocytic astrocytomas, however, do not seem to mutate or lose NF1 (9-11). Prior cytogenetic and comparative genomic hybridization (CGH) studies of pilocytic astrocytoma did not detect chromosomal alterations in most cases (12-17). We therefore used a more sensitive oligonucleotide array CGH to identify smaller chromosomal gains and losses in a group of 25 pilocytic astrocytomas. The most frequent change identified was gain of a 1.9-MB sequence located at chromosome 7q34, which includes a portion of the BRAF locus. We were also able to document activation of MEK and ERK proteins, which function downstream of BRAF, in most pilocytic astrocytomas.
Materials and Methods
Pilocytic astrocytoma and control tissues were obtained from the Department of Pathology, Johns Hopkins University School of Medicine, with institutional review board approval. All tumors resected from 1990 to 2007 with sufficient excess frozen tissue were used in the study. Tumor classification and cellularity were determined by 2 neuropathologists (CE and PB), and only samples with more than 70% viable tumor present were used in molecular analyses. Only 1 tumor (T62) came from a patient with neurofibromatosis type 1. No tumors were of the pilomyxoid variant. All of the pilocytic astrocytomas represented primary resections, except for T242, which was a recurrent lesion removed 13 years after initial diagnosis. The tissue array was constructed as previously described (18), with between 4 and 12 cores 0.6 mm in diameter taken from each tumor. Statistical analyses were performed using GraphPad Prism4 (GraphPad Software, San Diego, CA).
Oligonucleotide Array CGH Studies
Human genomic DNA was isolated from 25 mg of snap-frozen tissue using the DNeasy tissue kit and treated with RNase A according to the manufacturer's instructions (Qiagen, Valencia, CA). Concentration and quality of isolated genomic DNA were assessed using a spectrophotometer (NanoDrop Technologies, Wilmington, DE). Only DNA with a 260/A280 ratio greater than or equal to 1.8 and a 260/A230 ratio greater than or equal to 2.0 was used. Integrity of genomic DNA was further confirmed by low-voltage 0.6% agarose gel electrophoresis, with a mean band size of approximately 50 kb. Sample labeling was performed following the recommended Agilent protocol (Agilent Technologies, Santa Clara, CA). Briefly, 3 μg of genomic DNA was digested with 5 units of AluI and RsaI (Promega, Madison, WI) for 2 hours at 37°C. Labeling reactions were carried out with the total amount of digested DNA for 3 hours at 37°C using a BioPrime Array CGH Genomic Labeling Module (Invitrogen, Carlsbad, CA). The resulting labeled samples were then purified, concentrated on a Centricon YM-30 column, and then mixed with 10× blocking agent and 2× hybridization buffer (Agilent Technologies). Before array hybridization, probes were denatured at 95°C for 3 minutes and then immediately transferred to 37°C for 30 minutes. To remove any precipitates, the mixtures were centrifuged at 14,000 × g for 5 minutes. These mixtures were then hybridized to microarrays for 40 hours at 65°C in a rotating oven (Robbins Scientific, Mountain View, CA) at 20 rpm. Hybridized microarrays were washed and dried according to the manufacturer's protocols and imaged with Agilent G2565BA microarray scanner using default settings. Data were extracted using Feature Extraction Software v9.1 (Agilent Technologies). Data were analyzed using Agilent CGH Analytics v3.5. Aberrant regions (gains or losses) were evaluated using the Aberration Detection Method 2 algorithm with the threshold set at 12 (19).
Immunohistochemistry was performed on deparaffinized sections of a tissue microarray containing cores from 43 pilocytic astrocytomas. In brief, after antigen retrieval using Antigen Unmasking Solution (Vector Laboratories, Burlingame, CA), sections were incubated for 2 hours at room temperature with an antibody specific for phospho-ERK1/2 diluted 1:100 (clone 20G11; Cell Signaling Technologies, Danvers, MA). Primary antibody was detected using the Vectastain Elite kit (Vector Laboratories). Stained arrays were scored by a neuropathologist (CE) blinded with respect to the genomic changes in each tumor. At least 3 intact cores were required for a case to be scored.
Real-Time Polymerase Chain Reaction Analysis
Human genomic RNA-free DNA was isolated from snap-frozen tissue as described above. The sequences of the primers used in this study are as follows: HIPK2 forward 5′ TAGCTGGCGCTGAGTCACTC 3′; HIPK2 reverse 5′ TTACCGGCTCCCTCAGAGTG 3′; Rab19B forward 5′ TAACTTTTGTTTACTCTCCTCTGAG 3′; Rab19B reverse 5′ GCTGCATCACTTACACTTTTAAGAT 3′; BRAF forward 5′ GGCACATCACTGAACATAATTATC 3′; BRAF reverse 5′ AGCATGATATCACAAAGGTACT 3′. Annealing sites are adjacent to the gene indicated in the primer name. The DNA extracted from the tissue samples was adjusted to a concentration of 5 ng/μL. In each experiment, 1 μL (5 ng) of DNA was used. The standard amplification protocol consists of an initial denaturation step at 95°C for 8 minutes, followed by 35 amplification cycles at 95°C for 10 seconds, 58.1°C for 30 seconds, and 65°C for 30 seconds. Fluorescence measurements were taken at the end of the annealing phase at 58.1°C. During the evaluation phase of the assay, each amplification reaction was checked for the absence of nonspecific polymerase chain reaction (PCR) products by performing melt-curve analysis at the end of each run. In addition, PCR products amplified using each primer set were evaluated on 1% agarose gels and stained with ethidium bromide to confirm the amplification of a single product. Standard curves were prepared for each run using normal female genomic DNA (Promega). For each clinical sample, the absolute copy number of the target genes (RAB19b, HIPK2, and BRAF) and of the reference genes (RNaseP and Actin-γ) were measured in tumor tissue and in healthy control tissue. Finally, the relative copy number of target gene as compared with reference genes was calculated as previously described (20).
Whole-cell protein extracts were prepared by mincing 25 mg of snap-frozen tissue samples in 50 μL of TNE buffer, supplemented with 1× protease inhibitor cocktail (Roche Diagnostics, Mannheim, Germany) and an inhibitor of protein tyrosine phosphatases, 1 mM Na3VO4 (Sigma, St. Louis, MO). Lysates were cleared by centrifugation at 4°C for 30 minutes. Fifty micrograms of the cleared lysates was run on sodium dodecyl sulfate-polyacrylamide gel electrophoresis gels in MOPS-sodium dodecyl sulfate running buffer (Invitrogen), electrophoretically transferred to nytran membranes, and blocked overnight at 4°C in PBST with 5% nonfat milk. Primary antibodies used in this study were anti-glyceraldehyde-3-phosphate dehydrogenase (clone 6C5, 1:500,000; Research Diagnostics, Concord, MA) and antibodies (all from Cell Signaling Technologies) to ERK1/2 (1:1000), phospho-ERK1/2 (clone 20G11, 1:1000), phospho-MEK1/2 (Clone 166F8, 1:1000), and MEK1/2 (Clone L38C12). Secondary antibodies were anti-mouse or anti-rabbit 20 (KPL, Inc., Gaithersburg, MD; 1:5000), and visualization was by chemiluminescence (Western Lightning; Perkin Elmer, Foster City, CA).
Oligonucleotide Array CGH Analysis of Pilocytic Astrocytoma Genomic DNA
Genomic DNA extracted from 25 snap-frozen pilocytic astrocytomas was analyzed using Agilent oligonucleotide arrays containing 60-mer probes representing more than 236,000 coding and noncoding human sequences with a 8.9-KB overall median spacing. Patient demographic data and tumor location are shown in Table 1. Statistically significant gains and losses were determined using Agilent CGH Analytics v3.5 software. Among the 25 tumors analyzed, 3 showed gains and 4 showed losses of entire chromosomes (Table 1). The genome-wide changes in 1 tumor (T1098), including gains of chromosomes 5 and 6, are depicted in Figure 1A. A more detailed view of the chromosome 6 gain is shown in Figure 1B, in which there is a significant rightward deflection in the trendline representing the mean hybridization ratio along the chromosome. In contrast, most chromosomes showed a roughly equal hybridization of labeled tumor and normal DNA probes, as depicted in the trace for chromosome 6 in tumor T125 (Fig. 1C). The relatively normal karyotypes detected in most pilocytic tumors were not due to low sensitivity of the array because analysis of malignant brain tumors such as medulloblastoma identified numerous gains and losses, including many we have previously documented (21) such as amplification of the c-myc oncogene (data not shown).
Representative chromosomal alterations. (A) Gains of chromosomes 5 and 6, as well as a smaller region (7q34), were identified in the T1098 genome. (B) The rightward shift of the hybridization trendline indicates gain of chromosome 6 in T1098. (C) No such gain is present in T125, with the trendline centered at 0. (D) A 1.9-MB region located at cytoband 7q34 represented the most common focal change. In all panels, red and green points represent single oligonucleotide probes showing either a gain or loss in the tested sample as compared with the reference. Regions of gain considered significant by the Agilent CGH Analytics software are marked by red bars on the right side of the panel.
Smaller regions of DNA copy number change were also detected. The most common was a focal gain in the 7q34 region, which was present in 17 (68%) of the pilocytic astrocytomas (Fig. 1D). The amplitude of this change was modest, suggesting that only 1 or 2 additional copies of the region were present. In 1 case (T1036), gain of all of chromosome 7 was detected with an additional focal gain at 7q34. All 17 of the pilocytic astrocytomas with the focal 7q34 gain were from the posterior fossa, and none of the 5 tumors that arose in the supratentorial cortex had the change; this difference in location is statistically significant (p = 0.001, 2-tailed Fisher exact test). The span of the region of gain varied slightly among cases but was approximately 1.9 MB in size and included 17 known or predicted genes extending from the MGC14289 on the centromeric side to the oncogene BRAF on the telomeric side (Figs. 2A, B). Agilent CGH Analytics software predicted that the telomeric end of the region of gain terminated within the coding region of BRAF in 14 cases (Figs. 2A, B). The break occurred between exons 13 and 14, and sequences encoding the proteins carboxyl terminal 202 amino acids were included in the gain. In the other 3 cases, the entire BRAF gene was included in the amplified region. Seven additional small regions with statistically significant increases in copy number were also identified. These ranged in size from 0.1 to 50 MB and were found on chromosomes 1, 3, 5, 8, 9, 11, and 15 (Table 2). Eight regions of loss ranging from 0.03 to 187 MB in size were also detected (Table 2). Gains in the 7q34 region reported in 2 other studies are shown in Figures 2C and D and are discussed below.
The 7q34 region of gain. (A) The 1.9-MB region of gain at 7q34 contained 17 known or predicted genes. Polymerase chain reaction primers used for quantification of the gain (arrows) were located adjacent to the HIPK2, RAB19B, and BRAF genes. (B) The predicted telomeric end points of the 7q34 gain were predominantly within the BRAF locus. The sequence number (UCSD genome build May 2004/NCBI Build 35) at the telomeric end of the region is shown on the right for each tumor. (C) Pfister and colleagues (22) recently identified gains in this region defined by 2 bacterial artificial chromosome clones in 28 of 53 pilocytic astrocytomas. (D) In 3 of the 10 cases reported by Deshmukh and colleagues (23), the predicted end point of the region of gain fell within BRAF.
We used quantitative PCR analysis to verify the DNA copy number gain identified at 7q34 in our cases. Primer pairs that amplified sequences adjacent to the HIPK2, RAB19B, and BRAF genes (Fig. 2) were used for this analysis. All reactions were performed in triplicate. Table 1 shows the combined haploid copy number average for all 3 primer pairs compared with the mean of 2 reference genes (actin gamma and RNAse P). In 14 of the 17 cases identified by CGH array as showing focal gains at 7q34, the haploid genome value as measured by quantitative PCR was 1.5 or greater, consistent with the presence of 3 or more copies of the locus. Of the 8 tumors that lacked focal 7q34 gains on array CGH analysis, only 2 had a haploid copy number of 1.5 or greater. One of these (T1032) showed a gain of the entire chromosome 7; the quantitative PCR thus reflects this trisomy. In the second discrepant case (T1053), although quantitative PCR analysis showed an extremely high 7q34 copy number, the cycle threshold at which the both 7q34 and reference gene products were detected was much higher than for any other tumor. This may have affected the analysis, and we do not believe that 7q34 is truly highly amplified in this tumor. Because we had relatively few cases that lacked the 7q34 gain, we also analyzed 4 DNA samples extracted from nonneoplastic autopsy brains; the haploid ratio in these never reached 1.5 for any of the primer pairs used.
MEK/ERK Activation in Pilocytic Astrocytoma
In view of the inclusion of a portion of the BRAF locus in the 7q34 region of gain, we examined the possibility that the signaling program regulated by this oncogene might be active in pilocytic astrocytomas. BRAF has been shown to promote tumor growth by activation of the mitogen-activated protein kinase (MAPK) pathway in several tumor types, including cutaneous melanoma and carcinomas of the thyroid and colon (reviewed in Reference 24). Phosphorylation is associated with functional activation of the MEK and ERK proteins in this cascade. Therefore, we examined the expression and phosphorylation status of these in protein lysates from 7 snap-frozen pilocytic astrocytomas and 2 normal cerebral cortex specimens (Fig. 3). Consistent with the notion that RAS-BRAF-ERK signaling is active in pilocytic astrocytomas, we identified high levels of MEK and ERK phosphorylation in 3 of the 7 tumors, but not in the normal brain specimens. Phosphorylated MEK and ERK were also detected in the remaining 4 tumors on longer exposures of the labeled Western blot and were several-fold more abundant in 3 of these compared with normal brain (data not shown). Only 1 tumor without 7q34 gain (T62) had sufficient frozen tissue remaining for Western analysis; this showed lower levels of MEK and ERK phosphorylation.
MEK and ERK activation in pilocytic astrocytoma. (A) Total MEK1 and phosphorylated, active MEK1 (pMEK) were detected in protein lysates from several pilocytic astrocytomas, but not from normal cerebral cortical samples (NB1 and NB2). (B) Unphosphorylated ERK1 and ERK2 were widely expressed, whereas only a few tumors showed extensive ERK phosphorylation (pERK).
We also examined signaling through the MAPK pathway in 43 formalin-fixed, paraffin-embedded pilocytic astrocytoma samples included on a tissue microarray. Consistent with previous analyses in thyroid carcinoma and melanoma of activated BRAF (25,26), both cytoplasmic and nuclear phosphorylated ERK were observed, with the latter predominating. Because nuclear translocation of phosphorylated ERK is thought to be required for the activation of promitotic transcription factors (27), we scored nuclear immunoreactivity in the pilocytic astrocytomas (Fig. 4). Strong nuclear immunoreactivity was present in 9 cases (21%), moderate immunoreactivity in 14 cases (33%), and weak or very focal immunoreactivity in 12 cases (28%). No nuclear immunoreactivity for phosphorylated ERK was observed in 8 pilocytic astrocytomas (19%). We did not observe phosphorylated ERK in normal cerebellar or cerebral cortex (data not shown). Of the 25 cases analyzed by array CGH, only 13 were included on this tissue microarray. As in the Western analysis described above, the concordance between increased copy number at 7q34 and the amount of ERK phosphorylation was imperfect. Only 2 of the cases analyzed using immunohistochemistry lacked the 7q34 gain, and both showed at least focal nuclear phosphorylated ERK immunoreactivity. It is possible that MAPK signaling was activated by other mechanisms such as BRAF mutation in these cases. Taken together, these immunohistochemical and Western blot data indicate that RAS-BRAF-ERK signaling is active in most pilocytic astrocytomas.
Phosphorylated ERK (pERK) immunostaining in representative pilocytic astrocytomas. (A) No pERK immunoreactivity. (B) Weak pERK immunoreactivity with concentration in the nucleus (arrows and inset). (C) Moderate pERK immunoreactivity. Arrows indicate nuclear staining. (D) Strong nuclear and cytoplasmic pERK immunoreactivity (original magnification: 400× for all images).
The genetic alterations causing pilocytic astrocytomas remain largely unknown, and this lack of molecular understanding has hampered attempts to develop targeted therapies. We used high-density oligonucleotide arrays for CGH analysis of 25 pilocytic astrocytomas to address this issue. Large genomic changes were relatively rare in our cases, present in just 7 of 25 tumors (28%), and were similar to those previously reported. We found gains of chromosomes 5, 6, and 7, as has been shown in pilocytic astrocytomas analyzed using CGH, fluorescent in situ hybridization and classical cytogenetics (12, 13, 16, 17, 28). Similarly, the losses we identify of chromosomes 16, 17, 19, and 22 are consistent with several previous reports showing deletion of some or all of these chromosomes (14, 16, 17, 29), although these losses were not found by others (13).
The advantage of array CGH is its ability to detect relatively small regions of gain and loss; and several such localized copy number alterations were present in the tumors we examined. Some of these overlap with those reported in previous studies, including gains of 1p31-p36 (16), 3q26 (17), 5q23-q35 (29), 7q34 (29), 9q34 (16), and 11q12-q13 (29) and losses of 1q21 (17) and 2q21-q35 (14). The most frequent change was a small region of gain located at 7q34 that was detected in 17 of our 25 cases (68%). The degree of the CGH hybridization ratio change at 7q34, in addition to confirming quantitative PCR analysis, suggests that 1 or 2 extra copies of this locus are present in most tumors. The extent of the 7q34 DNA gain varied only slightly from case to case. In most tumors, it was 1.9 MB in size and included 17 known or predicted genes. Interestingly, in 15 cases, the predicted telomeric end of the 7q34 gain occurred within the BRAF oncogene between exons 13 and 14. Thus, only sequences encoding the proteins' carboxyl terminus would be included (Fig. 5A). In the remaining 3 tumors, the entire BRAF coding region, as well as its promoter, was included in the region of gain. We used phosphorylation-specific antibodies to investigate the possibility that the 7q34 gain might play an oncogenic role in pilocytic astrocytomas by activating the BRAF kinase cascade. Consistent with the concept that this pathway is active in most pilocytic astrocytomas, we found evidence of phosphorylation of MEK1 and ERK1/2 in most tumors examined by Western blot or immunohistochemistry.
RAS pathway activation in pilocytic astrocytomas. (A) Only exons encoding the kinase domain of the BRAF protein were included in the region of gain in most tumors analyzed. (B) The RAS pathway seems to be activated at multiple points in pilocytic astrocytomas, including NF1 loss of function in neurofibromatosis patients (8), rare oncogenic RAS mutations (30, 31), RASSF1A methylation (32), and gain or mutation of BRAF (22).
Gains in the 7q34 region in pilocytic astrocytoma have recently been reported by 2 other groups as well. Deshmukh and colleagues examined 10 pilocytic astrocytomas using 2 array CGH platforms (23) and found gains at 7q34 in 6 of 10 tumor samples using NimbleGen arrays and in 8 of 10 using Affymetrix arrays. They confirmed these results using quantitative PCR with primers specific for HIPK2 and also found copy number gains in this gene in an additional 26 of 61 (43%) independent tumors. Pfister and colleagues performed CGH analysis of bacterial artificial chromosome array on 66 pilocytic astrocytomas and identified gain of a 0.97-MB region defined by 2 bacterial artificial chromosome probes in 30 tumors (22). Duplication of the 7q34 region was confirmed in 28 (53%) of 53 pilocytic astrocytomas in their study using fluorescent in situ hybridization. As in our study, they found evidence of ERK1/2 phosphorylation. Interestingly, they also identified activating V600E point mutations in 4 (6%) of the 66 pilocytic astrocytomas in which BRAF was sequenced. Neither of these groups found that gains of 7q34 were restricted to posterior fossa tumors.
When combined with these 2 recent studies, our data strongly suggest that gain of a small region of DNA located at 7q34 represents the most common genomic change in pilocytic astrocytoma. The extent of the altered region and how many of the genes contained in it play important roles in the initiation and growth of pilocytic astrocytoma remain to be determined. The minimal region of gain defined by the 2 probes used by Pfister and colleagues (RP5-886O8 and RP4-726 N20) extends from TBXAS1 to BRAF (Fig. 2C). The flanking probes they report without gains lie well outside the region depicted in Figure 2 and do not help to define the extent of the change. The size of the minimal region of gain identified by Deshmukh and colleagues was slightly smaller than in our series and varied in different tumors and between array platforms (23). As in our series, however, it contained HIPK2 and a number of additional more telomeric genes of potential importance. They focused their analysis on the homeobox gene HIPK2 and showed that it is overexpressed at an mRNA and protein level in pilocytic astrocytomas. Consistent with a potential oncogenic role for HIPK2 in gliomas, overexpression of this protein in glioblastoma cells resulted in a growth advantage in vitro (23). Interestingly, whereas Deshmukh and colleagues did not investigate MAPK signaling in their tumors, in 3 of the pilocytic astrocytomas, they analyzed using Nimblegen arrays, the region of gain including exons encoding the C-terminus of the BRAF gene (Fig. 2D).
It is not entirely clear how the copy number alterations we identify at 7q34 activate the RAS-BRAF-ERK pathway. It seems that the entire BRAF gene is duplicated in some tumors, possibly leading to its overexpression. Low-level genomic gains including BRAF, most of which involved all of chromosome 7, have been documented in 17 of 23 World Health Organization grade II to IV gliomas, with 13 of these occurring in glioblastoma (33). These authors found that such higher-grade gliomas had significantly increased levels of nuclear phosphorylated ERK when multiple copy number alterations affecting the RAS-BRAF-ERK pathway were present. In most of our pilocytic astrocytoma cases, however, the focal gain at 7q34 includes only the sequences encoding the carboxyl terminal portion of the protein. This region of BRAF contains the kinase domain required to phosphorylate and activate MEK, whereas the amino terminus encodes the RAS binding domain that serves to inhibit kinase function (reviewed in Reference 34). Expression of the kinase domain alone has been shown to result in a protein with unchecked transforming activity that is independent of Ras activation (35,36). Interestingly, translocations of BRAF that fuse the carboxyl terminal kinase domain of the protein to other proteins have recently been identified in congenital melanocytic nevi (37). Murine tumors induced by the Sleeping Beauty transposon have also been shown to contain insertion events in BRAF introns that result in expression of a truncated, constitutively active kinase domain. These experimental and clinical data suggest that alterations of genomic DNA at 7q34 in pilocytic astrocytoma could result in the expression of a shortened or fused BRAF kinase domain, thereby resulting in activation of downstream MEK-ERK signaling. In support of this latter possibility, Jones and colleagues (38) showed that 7q34 gains in pilocytic astrocytomas can generate an in-frame fusion protein containing an active kinase domain.
The discovery that BRAF activation by chromosomal gain or, more rarely, mutation occurs in many sporadic pilocytic astrocytomas complements previous studies that implicate the RAS-BRAF-ERK cascade in these tumors (Fig. 5B). In patients with neurofibromatosis type 1 who often develop pilocytic astrocytomas, loss of function of the NF1 gene product neurofibromin results in increased RAS activity (7). Whereas NF1 inactivation is not commonly found in sporadic pilocytic tumors, oncogenic mutations in RAS have been reported in a small number of sporadic cases. Sharma and colleagues (30) examined K-RAS, H-RAS, and N-RAS in 21 sporadic pilocytic astrocytomas and identified an activating K-RAS mutation in one of these tumors. Janzarik and colleagues (31) screened these genes as well in 15 pilocytic astrocytomas and also found 1 K-RAS mutation. Finally, epigenetic silencing of the RAS association domain family RA(RASSF1A) gene, which is thought to modulate RAS signaling (38), has been documented in many pilocytic astrocytomas (39).
The fact that RAS-BRAF-ERK signaling seems to play a central role in the growth of pilocytic astrocytoma is particularly exciting because inhibitors affecting several kinases in this cascade have been developed (40-42). Some of these drugs have already been used in human clinical trials and might well inhibit the growth of pilocytic astrocytomas. Indeed, Pfister and colleagues (22) showed that pharmacological inhibition of RAS-BRAF-ERK signaling in low-grade gliomas could decrease proliferation of the tumor cells in culture. Such new therapies could be pivotal in the management of clinically problematic pilocytic or pilomyxoid tumors that arise in locations that make complete surgical excision impossible.
The authors would like to thank Ms Patricia Goldthwaite for assistance in identifying clinical material and Dr Wayne Yu of the DNA Microarray Core at the Johns Hopkins Sidney Kimmel Cancer Center for performing the array comparative genomic hybridization.
This work was supported by The Pilocytic/Pilomyxoid Research Fund, and the Children's Cancer Foundation of Baltimore, MD.