Purpose:

Most World Health Organization (WHO) grade I meningiomas carry a favorable prognosis. Some become clinically aggressive with recurrence, invasion, and resistance to conventional therapies (grade 1.5; recurrent/progressive WHO grade I tumors requiring further treatment within 10 years). We aimed to identify biomarker signatures in grade 1.5 meningiomas where histopathology and genetic evaluation has fallen short.

Experimental Design:

Mass spectrometry (MS)–based phosphoproteomics and peptide chip array kinomics were used to compare grade I and 1.5 tumors. Ingenuity Pathway Analysis (IPA) identified alterations in signaling pathways with validation by Western blot analysis. The selected biomarker was evaluated in an independent cohort of 140 samples (79/140 genotyped for meningioma mutations) by tissue microarray and correlated with clinical variables.

Results:

The MS-based phosphoproteomics revealed differential Ser/Thr phosphorylation in 32 phosphopeptides. The kinomic profiling by peptide chip array identified 10 phosphopeptides, including a 360% increase in phosphorylation of RB1, in the 1.5 group. IPA of the combined datasets and Western blot validation revealed regulation of AKT and cell-cycle checkpoint cascades. RB1 hyperphosphorylation at the S780 site distinguished grade 1.5 meningiomas in an independent cohort of 140 samples and was associated with decreased progression/recurrence-free survival. Mutations in NF2, TRAF7, SMO, KLF4, and AKT1 E17K did not predict RB1 S780 staining or progression in grade 1.5 meningiomas.

Conclusions:

RB1 S780 staining distinguishes grade 1.5 meningiomas, independent of histology, subtype, WHO grade, or genotype. This promising biomarker for risk stratification of histologically bland WHO grade I meningiomas provides insight into the pathways of oncogenesis driving these outlying clinically aggressive tumors.

Translational Relevance

Cranial meningiomas are heterogeneous. World Health Organization (WHO) grade using histopathologic criteria falls short in predicting progression-free survival in some benign grade I tumors. Molecular diagnostics have become important in accurately grading other tumors of the brain and thus better predicting the natural history. Using proteomic techniques in a group of genetically defined clinically aggressive grade I meningiomas (grade 1.5 meningiomas; recurrent, progressive WHO grade I tumors requiring further treatment within 10 years), RB1 phosphorylation at the S780 site proved to be a biomarker and mediator of this group. Although mutations in NF2, SMO, AKT, KLF4, TRAF7, or wild-type genotypes did not identify grade 1.5 meningiomas, RB1 phosphorylation at S780 defined the group. Utilizing a phosphoproteomic approach identified phosphorylation at S780 in RB1, a long-appreciated tumor suppressor gene. Validation of this biomarker in larger cohorts from independent medical centers is warranted. Furthermore, additional molecular study may shed light onto how RB1 S780 is involved with the behavior of clinically aggressive grade 1.5 meningiomas.

Meningiomas account for one-fourth of all primary brain neoplasms (1). The World Health Organization (WHO) provides a link between histopathology and risk of progression/recurrence by classifying meningiomas as grade I (benign), grade II (atypical, clear cell, and chordoid subtypes), or grade III (anaplastic; ref. 2). There are nine different histologic subtypes of WHO I, with the majority cured by surgery. Some WHO grade I tumors metastasize (3), rapidly grow (4), and invade brain, blood vessels, and cranial nerves (5). Benign meningiomas can reoccur after gross total resection (6) with rapid progression (7). Radiotherapy or radiosurgery are reserved for when surgery is no longer an option, although some tumors prove refractory (8). These observations support the concept of a group of histologically benign meningiomas, which unexpectedly behave clinically aggressive.

WHO I meningiomas lack atypical or anaplastic features and are managed as benign tumors. Brain invasion is enough to upgrade otherwise bland tumors to grade II (atypical), but is dependent on capturing brain tissue on the pathology section. These tumors have higher rates of recurrence, progression, and altered mortality rates (9). Identification of these tumors is imperative for treatment.

Molecular markers are needed to identify clinically aggressive grade I meningiomas (grade 1.5 meningiomas; recurrent, progressive WHO grade I tumors requiring further treatment within 10 years; ref. 10), where histology falls short. The grade 1.5 meningioma follows a more clinically aggressive natural history as compared with its WHO grade I counterpart. Deletion of 1p and/or 14q is thought to predict recurrence and progression, correlating with tumor grade (6, 11). Recent studies show that benign tumors with similar alterations in their genome can become clinically aggressive and reoccur (12). These results have not translated into biomarkers for diagnosis, risk stratification nor offer chemotherapeutic targets. Recently, methylation status was shown to segregate WHO I tumors and lower risk WHO II meningiomas (13), suggesting that epigenetics may play a larger role.

Pathways in cancer cells lead to increased proliferation, differentiation, migration, and survival. Posttranslational modifications include phosphorylation by kinases. Advances in proteomic techniques have led to a new understanding of the phosphoproteome in human tissue (14) and cell lines (15). Peptide arrays (16), reverse-phase protein arrays (17), antibody arrays (18), and mass spectrometry (MS; ref. 16) have been used to characterize cancer. Phosphoproteomics have resulted in advances in oncology; phosphorylation of oncoproteins in their native environment, molecular mechanisms that drive tumorigenesis (17), radioresistance (19), cellular networks of drug response (20). From these data, pharmacologic targets have led to the development of kinase inhibitors.

Few studies have investigated the proteome of benign meningiomas (21, 22). Most of them lack clinical information and compare different grades. Our group used phosphoproteomics to establish signatures of histologically well-defined WHO grade I–III meningiomas (16). Using low-resolution 2D gel followed by MS identification, we also described unique proteins in clinically aggressive WHO I meningiomas (10). Utilizing higher-resolution phosphoproteomics, we aimed to further characterize grade 1.5 meningiomas for risk stratification and evaluation of patterns of oncogenesis.

Tissue specimens, patients, and clinical data

Studies were conducted in accordance with U.S. Common Rule ethical guidelines. Data and specimen collection were reviewed and approved by the University of Washington Institutional Review Board and Human Subjects Division. Written-informed consent was obtained from all subjects. Methods were carried out in accordance with relevant guidelines and regulations. Patients underwent surgery at the University of Washington Hospitals between January 1, 1998 and December 31, 2012. Samples were collected and stored in −80°C. Data were gathered regarding history, demographics, imaging, neuropathology reports, operative information, and outcomes. Resected tumors were regraded according to revised criteria (2). Histologic subtype, mitoses, Ki-67/MIB, sheeting, macronuclei, hypercellularity, and necrosis were recorded. Invasion was recorded from operative or pathology reads. Specimens were reviewed by three neuropathologists and neurosurgeons. Total resection was defined as absence of residual enhancement on postoperative MRI within 48 hours of surgery. Recurrence was defined as at least 1 cm of enhancement on subsequent MRI. Progression was considered to be at least 1 cm of growth of residual tumor on MRI after surgery. Patients were initially divided according to tumor grade (WHO I, II, and III). Benign meningiomas were grouped into: clinically “nonaggressive” WHO grade I meningioma, designated as WHO I, and clinically “aggressive” WHO grade I meningioma designated as grade 1.5. WHO I meningiomas underwent complete resection, with no evidence of progression/recurrence on imaging at a follow-up period of 120 months. Grade 1.5 included (i) patients undergoing gross resection for a WHO I meningioma requiring repeated resection for recurrence within 120 months (second surgery confirming WHO I); (ii) patients undergoing gross resection of a WHO I meningioma and requiring stereotactic radiosurgery (SRS) for a recurrence within 120 months, and demonstrating progression despite SRS; (iii) patients undergoing more than two operations for a recurrent WHO I tumor. The discovery set included five individual meningioma fresh-frozen tissue samples classified as WHO I and five classified as grade 1.5. Clinical and histologic features of the discovery set are described in Supplementary Tables S1 and S2, respectively. The Cohort 01 tissue microarray (TMA) included 81 meningioma samples (31 grade I, 19 grade 1.5, 27 grade II, and four grade III; Supplementary Table S3). Five samples (sample ID 2, sample ID 24, sample ID 46, sample ID 51, and sample ID 76) were excluded from the analysis due to tissue loss during sectioning, transfer, or staining. The values of staining obtained for the sample ID 13 and 14 were averaged because they are from the same patient and resected at same surgery. Cohort 02 included target resequencing of NF2, TRAF7, SMO, KLF4, and AKT1 E17K of 20 tumors (from Cohort 01, Supplementary Table S3). Target resequencing was performed in 59 additional cases in Cohort 03 and stained for RB1 S780 in this TMA (Supplementary Table S4).

Targeted resequencing with molecular inversion probes

DNA was extracted using QIAamp DNA Mini Kit (Qiagen) following quantification on Qubit (Life Technologies). Probes were synthesized by Integrated DNA Technologies, Inc., pooled, and phosphorylated. Sequencing and analysis were performed as described (16). Reads were mapped to GRCh37 using Burrows-Wheeler Aligner (BWA)-MEM after the molecular inversion probes (MIPs) targeting sequence was hard clipped from the read. Indel realignment and base quality recalibration was performed according to GATK Best Practice's documentation. Variants were called using GATK's HaplotypeCaller in gvcf format and subsequently jointly called using GATK's GenotypeGVCF. A cutoff allele frequency > 4% was applied.

iTRAQ labeling and phosphopeptide enrichment

We utilized techniques similar to our prior work (16). Briefly, proteins were extracted from fresh-frozen meningioma specimens with T-PER Buffer (Thermo Scientific Pierce) supplemented with 1:100 Phosphatase Inhibitor Cocktail and 1:100 Halt Protease Inhibitor Cocktail EDTA free (Thermo Scientific Pierce). Protein lysates were quantified by Quibit. For the Isobaric Tags for Relative and Absolute Quantitation (iTRAQ)-based quantitative phosphoproteomic experiment, equal amounts of protein from five nonirradiated individual samples of WHO I and five nonirradiated individual samples from grade 1.5 meningioma were pooled to create two pools. Pooled lysates were precipitated, digested, labeled with tags (Life Technologies): WHO I—iTRAQ114 and grade 1.5—iTRAQ115, combined, and desalted. Phosphopeptides were allowed to bind to TiO2 spin tips using Phosphopeptide Enrichment and Clean-up Kit (Thermo Scientific Pierce), eluted, and cleaned using graphite columns (Thermo Scientific Pierce). Samples were dried and resuspended in TFA. LC/MS-MS analysis was performed as described previously (16). The spotted sample plates were analyzed using 4800 Plus MALDI TOF/TOF TM (AB SCIEX) with mass range of 800–3,500 m/z and S/N > 50. MS/MS spectral data were analyzed using ProteinPilot 4.0 (AB SCIEX) referencing International Protein Index and UniProtKB/Swiss-Prot database using Proteome Discoverer 1.3 (Thermo Scientific) with the parameters: MMTS for cysteine alkylation, up to two trypsin missed cleavages, biological modification, amino acid, and substitutions were set for ID focus; phosphorylation emphasis, FDR <5%, and confidence >95%. Data were normalized, and quantification expressed as ratio with WHO I levels (iTRAQ 114) as the denominator. A protein was considered differentially expressed when the iTRAQ ratio (grade 1.5:WHO I) was >1.20 or <0.83 or a fold change >20%. The fold change cutoff for up- or downregulation was determined on the basis of pilot studies evaluating the label-specific experimental variation between two replicates for the same experimental group. Similar approaches have been employed by our group (16, 23, 24, 25) and others (26, 27) to identify relevant candidates in iTRAQ studies.

Serine/threonine kinase profiling

The PamStation12 and STK PamChip peptide arrays (PamGene International BV) were used (16). The fluorescent platform measures the ability of active kinases in a specimen to phosphorylate specific peptides imprinted on multiplex chip arrays. Each chip contains four arrays. Each array displays 140 Ser/Thr and four positive control immobilized peptides. Each peptide represents a 15 amino-acid sequence from putative phosphorylation sites in human proteins derived from the literature and correlated with one or multiple upstream kinases. Kinase(s) in the sample actively phosphorylate substrates on the PamChip, in the presence of ATP. An antibody is used to detect phosphorylation, and a second FITC-conjugated antibody is used to quantify the signal. Three temperature-controlled peptide chips ran in parallel. Chips were blocked with 2% BSA (Sigma-Aldrich). Proteins were extracted from fresh-frozen meningiomas with T-PER Buffer (Thermo Scientific Pierce) supplemented with 1:100 Phosphatase Inhibitor Cocktail and 1:100 Halt Protease Inhibitor Cocktail EDTA free (Thermo Scientific Pierce). Protein lysates were quantified by Quibit. Equal amounts of protein from five nonirradiated individual samples of WHO I and five nonirradiated individual samples from grade 1.5 meningiomas were pooled to create two pools (WHO I and 1.5). One microgram of protein from each pool was applied to individual arrays with kinase buffer, 400 μmol/L ATP, and FITC-conjugated antibodies. Signal intensities were quantitated by BioNavigator 6.1.42 (PamGene), expressed per 100-ms exposure, and log transformed. Mean value <20% for peptides with a signal >2,000 was considered to ensure quality standards. Normalization was applied. Three replicated quantitations were combined using FDR < 1%. A P value <0.05 and a >10% fold change were considered significant.

Western blot analysis

Three WHO I (1,174, 1,289, and 1,149) and three grade 1.5 samples (1,432, 1,494, and 1,893), which were part of our discovery set (Supplementary Tables S1 and S2), were submitted to Western blot for validation. The other samples (WHO I: 1,480, 6, and grade 1.5: 2,002, 1,379) were not included as these tissues were exhausted in performing MIP, iTRAQ MS, and serine/threonine kinase (STK) profiling. Proteins were extracted from meningioma fresh-frozen tissue with T-PER Buffer (Thermo Scientific Pierce) supplemented with 1:100 Phosphatase Inhibitor Cocktail and 1:100 Halt Protease Inhibitor Cocktail EDTA free (Thermo Scientific Pierce). Proteins were quantified using Quibit. Twenty micrograms of protein extract was combined with reducing Laemmli buffer, boiled for 5 minutes, and then resolved by polyacrylamide gels. Polyacrylamide gels were transferred to polyvinylidene difluoride membranes, incubated with primary antibody, overnight, and then with horseradish peroxidase–conjugated secondary antibody. Membranes were developed with Clarity Western ECL Substrate (Bio-Rad) and visualized in GelDoc XR+ System (Bio-Rad). For loading control, membranes were stripped and reprobed with anti-β-actin. Band intensity was quantified using Image Studio Lite Version 5.0 (LI-COR Biosciences). Relative band densitometry is presented as mean ± SEM of the three samples of each meningioma grade. Antibodies are listed in Supplementary Table S5.

TMA

Samples were fixed in formalin, processed in tissue processor, and embedded in paraffin to produce formalin-fixed, paraffin-embedded blocks. Regions suitable for TMA were selected in duplicate. A blinded scientist created a randomized TMA. All TMA slides for cohorts 01 and 03 were sectioned producing 4-μm sections, placed on slides, and stained with hematoxylin and eosin (H&E). Antibody optimization was performed using Leica Bond III Fully Automated IHC and ISH Staining System (Leica Bio-Systems). Anti-phospho-RB1 S780 was used. The final step included the Bond Polymer Define Detection System (Leica Bio-Systems), containing endogenous peroxidase blocking, secondary antibody, and a streptavidin–biotin detection system. Slides were counterstained with Gill hematoxylin. Controls were included with antibody run. The slides were scanned using NanoZoomer Digital Pathology System (Hamamatsu Photonics, K. K.). Images were analyzed by Visiopharm which converted the initial digital imaging into grayscale values using HDAB–DAB with the Chromaticity Red feature subtracted, an H&E with filter of 3 × 3 pixels, and an RGB - G feature. Analyses were conducted in a blinded fashion. Quantification of RB1 was performed by automated image analysis of regions of interest. The ratio of RB1 S780 per total tissue area was determined. The patterns of staining were divided into quartile groups (0, low, medium, and high) and only the highest group used as high. The most stringent of criteria for labeling RB1 as high was used, and this objective cutoff was 0.088. RB1 staining was considered low when ratio < 0.088 and considered high when ratio > 0.088.

Bioinformatics tools

Computational prediction of kinase phosphorylation was performed by GPS 2.1 (28) and PhosphoNet Kinexus (www.phosphonet.ca). Kinome trees were annotated using Kinome Render (29). Prediction of phosphorylation consensus motifs was performed by NetPhos server (30). IPA (Ingenuity Systems) was used to identify mechanisms and functions and predict upstream regulators (16).

Statistical analysis

Comparison of the levels of the peptides/proteins in meningioma groups was performed using the nonparametric Mann–Whitney test and Kruskal–Wallis, when appropriate. Statistical significance of RB1 S780 phosphorylation by grade was analyzed by mixed-effects regression using grade ranks to test for an ordinal relationship. Correlation between RB1 S780 phosphorylation and clinical variables was assessed by Spearman and Kruskal–Wallis, as appropriate. Mutation prevalence by meningioma grade was evaluated using exact logistic regression. Analyses were adjusted for multiple comparisons using Holm–Bonferroni when appropriate. Time-to-recurrence was evaluated using log-rank tests and Cox regression. For the statistical analyses of band densitometry for Western blotting, the data are presented as mean ± SEM of the three samples of each meningioma grade (P values were calculated using both Student t test and nonparametric testing). All differences were significant when P < 0.05.

Data availability

Datasets generated and analyzed are available from the corresponding author on request.

MS-based phosphoproteomic profiling of meningioma grade 1.5

Although alterations in SMO, KLF4, TRAF7, NF2, and AKT E17K have been found in meningiomas, the downstream protein signature underlying clinically aggressive WHO grade I meningiomas is unknown. In an attempt to understand protein-specific alterations underlying a clinically aggressive phenotype of WHO grade I meningioma, we utilized proteomic profiling. Phosphorylation and kinase activity could lead to novel mechanistic insight and potential biomarker identification and thus iTRAQ phosphopeptide enrichment followed by quantitative MS analysis of the ten meningioma specimens (discovery set, Supplementary Tables S1 and S2) was performed. WHO grade I and grade 1.5 pools were labeled iTRAQ114 and iTRAQ115, respectively, following MS analysis (Supplementary Fig. S1A). A total of 649 unique phosphopeptides corresponding to 165 proteins were identified (Supplementary Table S6). From the 649 candidates, we selected 32 unique phosphopeptides with an individual composite score of ≥95% confidence, top-ranked matching sequence for that spectrum, and iTRAQ ratios from meningioma grade 1.5 compared with benign WHO I tissue greater than 1.20 or less than 0.83 (fold change>20%) at an FDR < 5% (Supplementary Fig. S1B). Fifteen proteins (46.9%, n = 15/32) showed hyperphosphorylation at Ser/Thr residues, among them the ABCF1, ABLIM1, ADD1, AKAP12, CTTN, PBDC1, DBNL, EIF4B, GLCE, HNRNPD, HSPB1, LMNA, Septin-2, VCAN, and VCL. Seventeen proteins (53.1%, n = 17/32) showed hypophosphorylation at serine or threonine residues, including ADD3, CANX, DPYSL3, EML4, EPB41L2, G3BP1, HSP90AB1, MAP1B, MARCKS, PGRMC1, PLEC, PPP1R2, SDC2, SEC22B, TGOLN2, TP53I1, and VIM. Western blot validation of the iTRAQ analysis was performed (Supplementary Fig. S1C). Phosphoproteins for Western blot validation were selected according to antibody commercial availability. The levels of phosphorylation or protein expression were analyzed in three individual fresh-frozen tissue specimens of each meningioma grade, which were also part of the discovery set. We observed that phosphorylation of DPYSL3 S522 (P = 0.003), G3BP1 S232 (P = 0.004), and CANX S583 (P = 0.006) correlated with the phosphoproteome dataset (Supplementary Figs. S1C and S2).

We used IPA to explore alterations in phosphorylation of 32 phosphoproteins (Supplementary Figs. S3 and S4). PKC, cAMP/PKA, PI3K-MTOR-AKT, MAPK-ERK, and RHO GTPases are common themes in these cascades. The MKNK1 and mTOR were predicted as upstream regulators of the cascade (Supplementary Fig. S4B).

Serine–threonine kinome profiling of grade 1.5 meningiomas

We performed kinome profiling of WHO grade I and grade 1.5 meningiomas. Pooled lysates from fresh-frozen specimens were characterized using PamChip peptide array Ser/Thr kinase platform (STK). The experimental workflow is shown in Fig. 1A. Only peptides above the limit of detection were included (n = 101/140; Supplementary Table S7). The 2Log transformed signal intensities of the 101 peptides above limit of detection were clustered and represented as a heatmap (Fig. 1B; Supplementary Table S7). Fold change analysis of grade 1.5 versus WHO I meningiomas showed 10 of 101 peptides with significantly increased phosphorylation changes, among them RB_774_786. RB1 demonstrated over a 360% increase in phosphorylation levels in the more clinically aggressive, 1.5 group (Fig. 1C). This phosphopeptide contains three phosphosites (T774, T778, and S780). Further, analysis of the phosphorylation consensus motif indicates that S780 in the motif T-R-P-P-T-L-S-P-I-P-H-I-P of RB1 is the predicted phosphorylation site. This is a consensus site for CDKs kinases (Supplementary Fig. S5C). Downstream activation of related signaling pathways (Fig. 1D) was predicted in this group of tumors.

Figure 1.

Kinome profiling (STK PamChip) of grade 1.5 meningioma. A, STK PamChip peptide chip array experimental design and workflow. The gray circle is the theoretical epitope. B, Heatmap of peptide signal intensities. The bar on the top right shows the relation between 2Log-transformed signal intensities and color. Red: hyperphosphorylation. Green: hypophosphorylation. (+) highest phosphorylation levels; (-) lowest phosphorylation levels. Each row represents clustered peptides detected on the STK PamChip with 2Log signal intensity above the limit of detection (101). Each column represents a group of pooled samples submitted to the kinome profiling (grades I and 1.5). C, Fold change analysis (cutoff 20% and P < 0.05) reveals 10 hyperphosphorylated peptides. RED, upregulation; GREEN, downregulation. The order of phosphorylated peptides is according to fold changes, from largest to smallest. The gray row indicates RB1 chosen for further validation. D, The IPA core analysis of hyperphosphorylated peptides in meningioma grade 1.5 predicts activation of Fcγ receptor–mediated phagocytosis in macrophages and monocytes, endothelin-1, type II diabetes mellitus, and glioma signaling pathways. The x-axis shows –log (P). The numerical value on the top represents the percentage of genes in the dataset. Numerical values on the right show the number of genes in the canonical pathway.

Figure 1.

Kinome profiling (STK PamChip) of grade 1.5 meningioma. A, STK PamChip peptide chip array experimental design and workflow. The gray circle is the theoretical epitope. B, Heatmap of peptide signal intensities. The bar on the top right shows the relation between 2Log-transformed signal intensities and color. Red: hyperphosphorylation. Green: hypophosphorylation. (+) highest phosphorylation levels; (-) lowest phosphorylation levels. Each row represents clustered peptides detected on the STK PamChip with 2Log signal intensity above the limit of detection (101). Each column represents a group of pooled samples submitted to the kinome profiling (grades I and 1.5). C, Fold change analysis (cutoff 20% and P < 0.05) reveals 10 hyperphosphorylated peptides. RED, upregulation; GREEN, downregulation. The order of phosphorylated peptides is according to fold changes, from largest to smallest. The gray row indicates RB1 chosen for further validation. D, The IPA core analysis of hyperphosphorylated peptides in meningioma grade 1.5 predicts activation of Fcγ receptor–mediated phagocytosis in macrophages and monocytes, endothelin-1, type II diabetes mellitus, and glioma signaling pathways. The x-axis shows –log (P). The numerical value on the top represents the percentage of genes in the dataset. Numerical values on the right show the number of genes in the canonical pathway.

Close modal

Combined analysis of the iTRAQ/MS-based global phosphoproteome and kinome for interpretation of signaling networks in grade 1.5 meningiomas

We combined datasets (significantly altered 32 phosphopeptides in the iTRAQ and 10 proteins from the STK chip) to perform computational analysis, to gain insight into the mechanisms of grade 1.5 meningiomas. First, we utilized computational prediction of phosphorylation sites by their cognate Ser/Thr kinases. The analysis predicted 85 potential upstream kinase regulators mainly included in the AGC and CMGC groups (Supplementary Fig. S3A). CDKs, CDC42, and MAPKs are three of the largest CMGC groups and mediate the function of a variety of tumor suppressors in the ERK-MAPK transduction pathway (31). The AGC group includes AKT, PKA, PKG, and PKC protein kinases (32). The IPA analysis predicted mTOR as a regulator of grade 1.5 meningiomas (Supplementary Table S8, mTOR highlighted in yellow) and showed a direct relationship with hyperphosphorylated RB1 as well as with FRAP1-Hd, SEC22B, HSPB1, G3BP1, SEPT2, PGRMC1, and HSP90AB1 (Supplementary Fig. S5B). The RB1 S780 site is a consensus site for CDKs kinases. Taken together, the computational analyses suggest that PI3K–AKT–mTOR, ERK–MAPK, PKA, CDC42–RAC1–RHOA, and RB1/E2F signaling pathways may be altered in grade 1.5 meningiomas.

Canonical pathways and upstream kinases validation by Western blotting

Western blot analysis of PI3K-AKT-MTOR, ERK-MAPK, PKA, CDC42-RAC1-RHOA, and RB1/E2F signaling pathways (Fig. 2; Supplementary Fig. S7) was performed. Significant decreases in AKT S473, AKT T308, and downstream targets IKKα T23 and RAF1 were found (PI3K-AKT-MTOR; Fig. 2A; Supplementary Fig. S7A). In the ERK-MAPK pathway, the increased expression of p38 MAPK distinguishes the 1.5 group (Fig. 2B; Supplementary Fig. S7B). Phosphorylated cAMP/PKA targets, CAMKII and PKAC, were seen (Fig. 2C). Proteins intimately involved with the CDC42–RAC1–RHOA pathway played a smaller role and were not significantly altered in the grade 1.5 group of tumors (Fig. 2D). Significantly increased levels of CDK4, CDK6, and RB1 S780 were seen (Fig. 2E; Supplementary Fig. S7E), together implying a defining signature of cell-cycle G1–S checkpoint signaling in grade 1.5 meningiomas. Figure 2F is a table illustrating significantly altered proteins and respective P values.

Figure 2.

Western blotting validation of aggressiveness-related canonical pathways and kinases in grade 1.5 meningioma. Western blotting assay was performed in three grade I and three grade 1.5 fresh-frozen tissue lysates. Cropped lanes. A, PI3K/AKT/MTOR targets. B, ERK-MAPK targets. C, cAMP/PKA targets. D, CDC42-RAC1-RHOA targets. E, Cell-cycle G1–S checkpoint targets. F, The table demonstrates a list of significantly altered proteins and kinases in 1.5 meningiomas as well as their respective P values. Bold letter and ‘*' indicate proteins and kinases with P < 0.05. See also Supplementary Fig. S7 for band intensity and statistics.

Figure 2.

Western blotting validation of aggressiveness-related canonical pathways and kinases in grade 1.5 meningioma. Western blotting assay was performed in three grade I and three grade 1.5 fresh-frozen tissue lysates. Cropped lanes. A, PI3K/AKT/MTOR targets. B, ERK-MAPK targets. C, cAMP/PKA targets. D, CDC42-RAC1-RHOA targets. E, Cell-cycle G1–S checkpoint targets. F, The table demonstrates a list of significantly altered proteins and kinases in 1.5 meningiomas as well as their respective P values. Bold letter and ‘*' indicate proteins and kinases with P < 0.05. See also Supplementary Fig. S7 for band intensity and statistics.

Close modal

Hyperphosphorylation of RB1 S780 predicts recurrence in WHO I meningiomas, identifying the 1.5 subgroup

We used TMA methods to explore whether RB1 S780 staining was found in meningiomas of varying grades, specifically grade 1.5 tumors. RB1 S780 staining was analyzed in multiple TMA cohorts (Supplementary Table S9) which included samples prior to and after radiotherapy treatment (Supplementary Table S10). The levels of RB1 S780 staining were quantified (Supplementary Fig. S8). Meningioma specimens with no prior radiotherapy treatment (107 total tumors; 47 grade I, 28 grade 1.5, 31 grade II, and 1 grade III) and specimens with prior radiotherapy treatment (33 total tumors; 0 grade I, 16 grade 1.5, 14 grade II, and 3 grade III; Supplementary Tables S3 and S4) were included. We performed univariate and multivariate analyses of RB1 S780 phosphorylation to evaluate the ability to distinguish 1.5 tumors (Supplementary Tables S3A and S11). Potential confounders revealed by the univariate analysis (Supplementary Table S11) were considered in the multivariate model (Table 1A). RB1 S780 phosphorylation distinguishes 1.5 tumors. RB1 S780 phosphorylation by grade is statistically significant, and it is increased in grade 1.5 tumors in all samples (Fig. 3C) as well as in samples with no prior radiotherapy treatment (Fig. 3D; Table 1A). Irradiated samples did not show differences in RB1 S780 phosphorylation (Fig. 3E; Table 1A). It is noteworthy that the discovery set of tumors were nonirradiated samples.

Table 1A.

RB1 S780 phosphorylation by grade.

SampleGradeN (%)pRB1 meanLog pRB1 meanChange from grade IChange from grade I (modeled)Sig. (vs. grade I)Sig. (ordinal)
Full samplea Grade I 30 (40%) 0.03 −5.07 — — — 0.036 
 Grade 1.5 15 (20%) 0.11 −2.98 2.09 1.75 0.006  
 Grade II 26 (35%) 0.06 −4.41 0.66 0.19 0.720  
 Grade III 4 (5%) 0.04 −3.35 1.72 0.88 0.410  
No radiotherapy Txb Grade I 29 (52%) 0.03 −5.01 — — — 0.0003 
 Grade 1.5 10 (18%) 0.15 −2.00 3.01 3.26 <0.0001  
 Grade II 16 (29%) 0.05 −4.60 0.41 0.38 0.524  
 Grade III 1 (2%) 0.02 −3.91 1.10 1.11 0.566  
Radiotherapy Txc Grade I 1 (5%) 0.00 −6.63 — — — 0.911 
 Grade 1.5 5 (26%) 0.02 −4.93 1.70 1.20 0.526  
 Grade II 10 (53%) 0.09 −4.09 2.54 0.63 0.748  
 Grade III 3 (16%) 0.05 −3.16 3.48 0.80 0.724  
SampleGradeN (%)pRB1 meanLog pRB1 meanChange from grade IChange from grade I (modeled)Sig. (vs. grade I)Sig. (ordinal)
Full samplea Grade I 30 (40%) 0.03 −5.07 — — — 0.036 
 Grade 1.5 15 (20%) 0.11 −2.98 2.09 1.75 0.006  
 Grade II 26 (35%) 0.06 −4.41 0.66 0.19 0.720  
 Grade III 4 (5%) 0.04 −3.35 1.72 0.88 0.410  
No radiotherapy Txb Grade I 29 (52%) 0.03 −5.01 — — — 0.0003 
 Grade 1.5 10 (18%) 0.15 −2.00 3.01 3.26 <0.0001  
 Grade II 16 (29%) 0.05 −4.60 0.41 0.38 0.524  
 Grade III 1 (2%) 0.02 −3.91 1.10 1.11 0.566  
Radiotherapy Txc Grade I 1 (5%) 0.00 −6.63 — — — 0.911 
 Grade 1.5 5 (26%) 0.02 −4.93 1.70 1.20 0.526  
 Grade II 10 (53%) 0.09 −4.09 2.54 0.63 0.748  
 Grade III 3 (16%) 0.05 −3.16 3.48 0.80 0.724  

Note: Statistical significance by mixed-effects regression (log-transformed), using grade ranks to test for an ordinal relationship.

aAnalyses adjust for invasion (P = 0.06), location on array (P = 0.02), and sex (P = 0.29).

bAnalyses adjust for sex (P = 0.91).

cAnalyses adjust for macronuclei (P = 0.04) and sex (P = 0.19).

Figure 3.

RB1 S780 validation in a TMA cohort of clinical specimens. A, Representative WHO I and 1.5 meningioma H&E section from two cases. B, The same representative cases stained with RB1 S780. Note the increased cytoplasmic and nuclear staining in the 1.5 meningioma. C, Ratio of RB1 S780 staining per total tissue area versus tumor grade – all samples (nonirradiated samples + samples with prior radiotherapy). D, Ratio of RB1 S780 staining per total tissue area versus tumor grade–nonirradiated samples only. E, Ratio of RB1 S780 staining per total tissue area versus tumor grade (only samples with prior radiotherapy). In C, D, and E, x-axis indicates RB1 S780 intensity of staining. y-axis shows meningioma grades. “n” indicates number of samples. Statistical significance by mixed-effects regression using grade ranks to test for an ordinal relationship. See also Table 1A and 1B. F, Progression/recurrence-free survival versus ratio of RB1 S780 staining per total tissue area independently of tumor grade—all samples. G, Progression/recurrence-free survival versus ratio of RB1 S780 staining per total tissue area independent of tumor grade in nonirradiated samples. H, Progression/recurrence-free survival in irradiated tumors, ratio of RB1 S780 staining per total tissue area independent of tumor grade. I, Progression/recurrence-free survival in all histologically benign meningiomas (grades I, n = 47, and 1.5, n = 44). J, Progression/recurrence-free survival in all nonirradiated histologically benign meningiomas (grades I, n = 47, and 1.5, n = 28). K, Progression/recurrence-free survival in all irradiated histologically benign meningiomas (grades I, n = 0, and 1.5, n = 16). L, Progression/recurrence-free survival in nonirradiated histologically benign meningiomas. Grade I tumors were samples collected at the first and only surgery (n = 47). Grade 1.5 tumors were collected from either initial (first surgery, n = 16) or recurrent (second–fifth surgery, n = 12) specimens. M, Progression/recurrence-free survival of grade II and III tumor patients–nonirradiated samples (n = 32). Statistical significance by Cox regression. See also Table 3A and 3B.

Figure 3.

RB1 S780 validation in a TMA cohort of clinical specimens. A, Representative WHO I and 1.5 meningioma H&E section from two cases. B, The same representative cases stained with RB1 S780. Note the increased cytoplasmic and nuclear staining in the 1.5 meningioma. C, Ratio of RB1 S780 staining per total tissue area versus tumor grade – all samples (nonirradiated samples + samples with prior radiotherapy). D, Ratio of RB1 S780 staining per total tissue area versus tumor grade–nonirradiated samples only. E, Ratio of RB1 S780 staining per total tissue area versus tumor grade (only samples with prior radiotherapy). In C, D, and E, x-axis indicates RB1 S780 intensity of staining. y-axis shows meningioma grades. “n” indicates number of samples. Statistical significance by mixed-effects regression using grade ranks to test for an ordinal relationship. See also Table 1A and 1B. F, Progression/recurrence-free survival versus ratio of RB1 S780 staining per total tissue area independently of tumor grade—all samples. G, Progression/recurrence-free survival versus ratio of RB1 S780 staining per total tissue area independent of tumor grade in nonirradiated samples. H, Progression/recurrence-free survival in irradiated tumors, ratio of RB1 S780 staining per total tissue area independent of tumor grade. I, Progression/recurrence-free survival in all histologically benign meningiomas (grades I, n = 47, and 1.5, n = 44). J, Progression/recurrence-free survival in all nonirradiated histologically benign meningiomas (grades I, n = 47, and 1.5, n = 28). K, Progression/recurrence-free survival in all irradiated histologically benign meningiomas (grades I, n = 0, and 1.5, n = 16). L, Progression/recurrence-free survival in nonirradiated histologically benign meningiomas. Grade I tumors were samples collected at the first and only surgery (n = 47). Grade 1.5 tumors were collected from either initial (first surgery, n = 16) or recurrent (second–fifth surgery, n = 12) specimens. M, Progression/recurrence-free survival of grade II and III tumor patients–nonirradiated samples (n = 32). Statistical significance by Cox regression. See also Table 3A and 3B.

Close modal

Kaplan–Meier and Cox analyses of RB1 S780 phosphorylation versus progression/recurrence-free survival were performed to identify its potential prognostic value. In all samples, independent of WHO grade, RB1 S780 phosphorylation is associated with lower progression/recurrence-free survival (P = 0.004; Fig. 3F; Table 1B). In nonirradiated samples (P = 0.0001; Fig. 3G;Table 1B), increased RB1 S780 staining was also associated with decreased progression/recurrence-free survival. This was not visualized in the sample cohort with prior radiotherapy treatment (Fig. 3H; Table 1B).

Table 1B.

RB1 S780 phosphorylation progression/recurrence survival plots.

SignificanceHR
FigureSampleFactorCox unadj.Cox adjustedComparisonEstimate95% Lower95% Upper
5Fa All pRB1 0.005 0.004 pRB1 high (vs. low) 2.93 1.41 6.09 
5Gb No radiotherapy pRB1 0.0002 0.0001 pRB1 high (vs. low) 7.77 2.72 22.2 
5H Radiotherapy pRB1 0.894 — pRB1 high (vs. low) 0.90 0.19 4.31 
5Ic No radiotherapy Grade 0.066 — Grade 1.5 (vs. 1) — — — 
     Grade 2 (vs. 1) — — — 
     Grade 3 (vs. 1) — — — 
5Jc No radiotherapy, grade ≤ 1.5 pRB1 0.107 — pRB1 high (vs. low) — — — 
5Kc No radiotherapy, grade ≥ 2 pRB1 0.934 — pRB1 high (vs. low) — — — 
SignificanceHR
FigureSampleFactorCox unadj.Cox adjustedComparisonEstimate95% Lower95% Upper
5Fa All pRB1 0.005 0.004 pRB1 high (vs. low) 2.93 1.41 6.09 
5Gb No radiotherapy pRB1 0.0002 0.0001 pRB1 high (vs. low) 7.77 2.72 22.2 
5H Radiotherapy pRB1 0.894 — pRB1 high (vs. low) 0.90 0.19 4.31 
5Ic No radiotherapy Grade 0.066 — Grade 1.5 (vs. 1) — — — 
     Grade 2 (vs. 1) — — — 
     Grade 3 (vs. 1) — — — 
5Jc No radiotherapy, grade ≤ 1.5 pRB1 0.107 — pRB1 high (vs. low) — — — 
5Kc No radiotherapy, grade ≥ 2 pRB1 0.934 — pRB1 high (vs. low) — — — 

Note: Some HRs could not be estimated due to low cell counts.

aMultivariate Cox model adjusts for grade (P < 0.01) and small cell (P = 0.01).

bMultivariate Cox model adjusts for grade (P = 0.06), small-cell (P = 0.01), and sheet architecture (P = 0.05).

cMultivariate Cox model inconclusive due to low cell counts.

Within the histologically benign samples, grade 1.5 meningiomas had higher levels of phosphorylation of RB1 at S780. Prior radiotherapy tended to correlate with decreased phosphorylated RB1 across groups (Fig. 3E). Considering that radiotherapy may change the characteristics of a tumor and only nonirradiated samples were present in the discovery set, we focused on the significance of progression/recurrence-free survival versus RB1 S780 staining in samples without previous radiotherapy, by grade. Phosphorylation of RB1 at S780 in nonirradiated, histologically benign, tumors correlated with progression/recurrence-free survival (Fig. 3J; Table 1B). This correlation was not seen in this group when radiotherapy was given prior to the surgery (Fig. 3K).

To better understand the role that RB1 S780 staining plays in temporal aggressiveness of the grade 1.5 group of tumors, we separated these tumors based on the timing of the sample collection. That is, whether the sample was obtained at the first operation versus the second to fifth surgery in the patient's course of treatment. There were 140 meningiomas stained for RB1 S780 (of which 79 underwent MIP resequencing). This included 47 grade I tumors and 28 grade 1.5 tumors with no radiotherapy history. Of the 28 grade 1.5 meningiomas, 16 were obtained from the first operation and 12 from the second to fifth operation. To address whether RB1 S780 staining occurs prior to the recurrence or progression, we have separated out the samples obtained at the first and subsequent (second–fifth) operations, for clarity (see Supplementary Tables S9 and S10). Figure 3I shows a Kaplan–Meier curve comparing three groups (without prior radiotherapy): grade I, grade 1.5 at first operation, and grade 1.5 at second to fifth operation. Phosphorylated RB1 S780 identifies the grade 1.5 group regardless of timing of tissue collection (Fig. 3l). The difference in progression-free survival, between grade 1.5 groups (first vs. subsequent operations), was not significant (P = 0.236). To further illustrate the specificity of phosphorylated RB1 S780 for the grade 1.5 group of tumors, an analysis was undertaken in grade 2 and 3 meningiomas. RB1 S780 staining in histologically aggressive meningiomas (grade II + III), without prior radiotherapy, did not correlate with progression/recurrence-free survival (Fig. 3M). RB1 staining at the S780 site is a unique, specific, and robust marker for grade 1.5 meningiomas and may highlight a pathway that can be exploited for therapies.

Screening of the SMO, KLF4, TRAF7, NF2, and AKT E17K mutations by MIP and relationship with RB1 S780 staining

Recurrent mutations in SMO, KLF4, TRAF7, NF2, and AKT genes (33) could potentially underlie the pathogenesis of grade 1.5 meningiomas. We sought to explore the relationship between alterations in these genes and both RB1 S780 staining and recurrence/progression-free survival. Target resequencing of NF2, TRAF7, SMO, KLF4, and AKT1 E17K was performed in an additional cohort of 79 tumors (26 grade I, 25 grade 1.5, 26 grade II, and 2 grade III) with both MIP genotyping and RB1 S780 staining. Tables 2A and B show the analyses for testing a relationship between WHO grade and presence of mutation for each of six genetic factors, as well as RB1 S780 staining, respectively. Exact logistic regression was used in order to avoid the pitfalls of low cell counts and thus correlated effects due to repeated subjects were ignored. The only mutation showing a significant relationship with grade is TRAF7, which remained statistically significant (P = 0.020) after adjusting for the experiment-wise error due to multiple comparisons (Holm–Bonferroni, m = 6). However, the strength of this relationship appears to be driven mainly by the high prevalence in grade I as compared with grades 1.5/II/III. There are no discernable differences in prevalence or odds ratios among the higher grades, thus inferring a monotonic relationship among the higher grades is not necessarily warranted. Mutations in SMO initially showed significance on its own, but the effect (which was in the opposite direction of TRAF7) washed away after adjusting for multiple comparisons (P = 0.108). High RB1 was highly related to grade in the mutation (N = 79, P = 0.008) and nonradiated (N = 107, P = 0.013) samples, but less so in the full (N = 140, P = 0.059) sample. However, note that the relationship is decidedly not monotonic, as contrary to the mutation analyses, the ORs decrease within just the higher grades (1.5/II/III). For this reason, the relationship would be better characterized with grade as dichotomous (i.e., I vs. 1.5/II/III). RB1 S780 staining was most reliable in identifying the grade 1.5 meningiomas, in this group.

Table 2A.

Mutation rate by grade.

Modeled
GradeN (%)N (%) mutationsOR (vs. I)OR (vs. I)Sig. (vs. I)Sig. ordinalSig. adj. multiple comparisons
NF2 Grade I 26 (33%) 9 (35%) 1.00 1.00 — 0.204 0.612 
 Grade 1.5 25 (32%) 14 (56%) 2.40 2.36 0.210   
 Grade II 26 (33%) 14 (54%) 2.20 2.17 0.264   
 Grade III 2 (3%) 1 (50%) 1.89 1.84 1.000   
TRAF7 Grade I 26 (33%) 12 (46%) 1.00 1.00 — 0.003 0.020 
 Grade 1.5 25 (32%) 4 (16%) 0.22 0.23 0.042   
 Grade II 26 (33%) 3 (12%) 0.15 0.16 0.013   
 Grade III 2 (3%) 0 (0%) 0.00 0.53 0.635   
AKT1 Grade I 26 (33%) 3 (12%) 1.00 1.00 — 0.106 0.424 
 Grade 1.5 25 (32%) 1 (4%) 0.32 0.33 0.640   
 Grade II 26 (33%) 0 (0%) 0.00 0.24 0.235   
 Grade III 2 (3%) 0 (0%) 0.00 3.51 1.000   
KLF4 Grade I 26 (33%) 0 (0%) 1.00 1.00 — 0.700 1.000 
 Grade 1.5 25 (32%) 1 (4%) — 1.04 0.980   
 Grade II 26 (33%) 1 (4%) — 1.00 1.000   
 Grade III 2 (3%) 0 (0%) — — —   
SMO Grade I 26 (33%) 0 (0%) 1.00 1.00 — 0.022 0.108 
 Grade 1.5 25 (32%) 1 (4%) — 1.04 0.980   
 Grade II 26 (33%) 5 (19%) — 7.78 0.051   
 Grade III 2 (3%) 0 (0%) — — —   
Wild-type Grade I 26 (33%) 9 (35%) 1.00 1.00 — 0.949 1.000 
 Grade 1.5 25 (32%) 7 (28%) 0.73 0.74 0.837   
 Grade II 26 (33%) 8 (31%) 0.84 0.84 1.000   
 Grade III 2 (3%) 1 (50%) 1.89 1.84 1.000   
Modeled
GradeN (%)N (%) mutationsOR (vs. I)OR (vs. I)Sig. (vs. I)Sig. ordinalSig. adj. multiple comparisons
NF2 Grade I 26 (33%) 9 (35%) 1.00 1.00 — 0.204 0.612 
 Grade 1.5 25 (32%) 14 (56%) 2.40 2.36 0.210   
 Grade II 26 (33%) 14 (54%) 2.20 2.17 0.264   
 Grade III 2 (3%) 1 (50%) 1.89 1.84 1.000   
TRAF7 Grade I 26 (33%) 12 (46%) 1.00 1.00 — 0.003 0.020 
 Grade 1.5 25 (32%) 4 (16%) 0.22 0.23 0.042   
 Grade II 26 (33%) 3 (12%) 0.15 0.16 0.013   
 Grade III 2 (3%) 0 (0%) 0.00 0.53 0.635   
AKT1 Grade I 26 (33%) 3 (12%) 1.00 1.00 — 0.106 0.424 
 Grade 1.5 25 (32%) 1 (4%) 0.32 0.33 0.640   
 Grade II 26 (33%) 0 (0%) 0.00 0.24 0.235   
 Grade III 2 (3%) 0 (0%) 0.00 3.51 1.000   
KLF4 Grade I 26 (33%) 0 (0%) 1.00 1.00 — 0.700 1.000 
 Grade 1.5 25 (32%) 1 (4%) — 1.04 0.980   
 Grade II 26 (33%) 1 (4%) — 1.00 1.000   
 Grade III 2 (3%) 0 (0%) — — —   
SMO Grade I 26 (33%) 0 (0%) 1.00 1.00 — 0.022 0.108 
 Grade 1.5 25 (32%) 1 (4%) — 1.04 0.980   
 Grade II 26 (33%) 5 (19%) — 7.78 0.051   
 Grade III 2 (3%) 0 (0%) — — —   
Wild-type Grade I 26 (33%) 9 (35%) 1.00 1.00 — 0.949 1.000 
 Grade 1.5 25 (32%) 7 (28%) 0.73 0.74 0.837   
 Grade II 26 (33%) 8 (31%) 0.84 0.84 1.000   
 Grade III 2 (3%) 1 (50%) 1.89 1.84 1.000   
Table 2B.

Rate of high RB1 by grade.

Modeled
SampleGradeN (%)N (%) HighOR (vs. I)OR (vs. I)Sig. (vs. I)Sig. ordinalSig. adj. multiple comparisons
N = 79 (Table 2A) Grade I 26 (33%) 5 (19%) 1.00 1.00 — 0.001 0.008 
 Grade 1.5 25 (32%) 24 (96%) 100.80 86.88 <0.001   
 Grade II 26 (33%) 19 (83%) 11.40 10.75 <0.001   
 Grade III 2 (3%) 0 (0%) 0.00 1.89 1.000   
N = 140 (full sample) Grade I 47 (34%) 8 (17%) 1.00 1.00 — 0.010 0.059 
 Grade 1.5 44 (31%) 34 (77%) 16.58 15.89 <0.001   
 Grade II 45 (32%) 21 (47%) 4.27 4.20 0.004   
 Grade III 4 (3%) 1 (25%) 1.63 1.61 1.000   
N = 107 (full sample, nonradiated) Grade I 47 (44%) 8 (17%) 1.00 1.00 — 0.002 0.013 
 Grade 1.5 28 (26%) 27 (96%) 131.63 118.93 <0.001   
 Grade II 31 (29%) 15 (48%) 4.57 4.47 0.007   
 Grade III 1 (1%) 0 (0%) 0.00 5.00 1.000   
Modeled
SampleGradeN (%)N (%) HighOR (vs. I)OR (vs. I)Sig. (vs. I)Sig. ordinalSig. adj. multiple comparisons
N = 79 (Table 2A) Grade I 26 (33%) 5 (19%) 1.00 1.00 — 0.001 0.008 
 Grade 1.5 25 (32%) 24 (96%) 100.80 86.88 <0.001   
 Grade II 26 (33%) 19 (83%) 11.40 10.75 <0.001   
 Grade III 2 (3%) 0 (0%) 0.00 1.89 1.000   
N = 140 (full sample) Grade I 47 (34%) 8 (17%) 1.00 1.00 — 0.010 0.059 
 Grade 1.5 44 (31%) 34 (77%) 16.58 15.89 <0.001   
 Grade II 45 (32%) 21 (47%) 4.27 4.20 0.004   
 Grade III 4 (3%) 1 (25%) 1.63 1.61 1.000   
N = 107 (full sample, nonradiated) Grade I 47 (44%) 8 (17%) 1.00 1.00 — 0.002 0.013 
 Grade 1.5 28 (26%) 27 (96%) 131.63 118.93 <0.001   
 Grade II 31 (29%) 15 (48%) 4.57 4.47 0.007   
 Grade III 1 (1%) 0 (0%) 0.00 5.00 1.000   

Note: ORs and statistical significance by exact logistic regression, without adjusting for other factors. Eight subjects are represented twice in this sample, the effects of which this analysis ignores due to the small sample size (which requires exact logistic regression). Adjustment for multiple comparisons by Holm–Bonferroni (m = 7).

We next aimed to test for a relationship between recurrence/progression time and presence of mutation for each of six genetic factors. Supplementary Table S12 summarizes the difference between each of the "no-mutation" and "mutation" survival curves (log-rank test) and also provides estimates of HRs and mean time to recurrence (Cox proportional-hazards regression). Presence of the TRAF7 mutation was associated with lower risk for recurrence/progression [HR, 0.18; 95% confidence interval (CI), 0.04–0.77; P = 0.009], even after adjustment for multiple comparisons (P = 0.044). Presence of the KLF4 mutation was associated with higher risk for recurrence/progression (HR, 9.65; 95% CI, 2.12–43.9: P < 0.001), even after adjustment for multiple comparisons (P = 0.002). The KLF4 and SMO HR estimates should be considered unreliable due to the low prevalence of mutation. Although TRAF7 and KLF4 mutations may correlate with decreased and increased risk of recurrence/progression, respectively, neither alteration identified the grade 1.5 meningiomas. RB1 S780 staining was most reliable in identifying the grade 1.5 meningiomas with a decrease in progression-free survival.

Meningiomas are a heterogeneous group of tumors, characterized by WHO grading. Up to 20% of benign meningiomas reoccur even after gross resection (6), and progression of residual disease can be rapid (7). Radiotherapy or radiosurgery is usually reserved for when surgery is not an option, although some tumors prove refractory to this treatment (8). These observations support the concept of a group of histologically benign meningiomas, which clinically behave more aggressively. Identifying these outliers is imperative for treatment planning and surveillance, prior to clinical aggressiveness. When WHO I meningiomas invade brain, they are automatically upgraded to WHO II (2). This diagnosis is dependent on the presence of brain tissue in the histologic specimen, bringing up the possibility of WHO I tumors being under graded because of the lack of brain tissue in the specimen. Because these tumors follow a different natural history and recurrence pattern, it is imperative to identify them. These tumors have similar MIB, ki67, and histology patterns (4). Improved biomarkers for identification of these outlying tumors would aid in better classifying for adjuvant planning.

Numerous studies have focused on DNA alterations in aggressive meningiomas (34, 35). Chromosomal gains and losses (ref. 36; e.g., monosomy 22) or specific genotypes (e.g., NF2 mutations) and next-generation sequencing (33) are promising. The majority of WHO grade I meningiomas (50%–60%) are linked to NF2 mutations. We performed target resequencing of NF2, TRAF7, SMO, KLF4, and AKT1 E17K in an additional cohort of meningioma samples, including the discovery set samples (Supplementary Tables S3 and S4). Table 2A shows the analysis testing for a relationship between grade and presence of mutation for each of six genetic factors (five gene mutations and wild-type). We did not find a correlation between alterations in SMO, KLF4, TRAF7, NF2, and AKT E17K mutations with WHO grade or clinical aggressiveness of grade 1.5 tumors. There was a trend of TRAF7 mutations and more benign natural history. Further, there was a trend of KLF4 mutations and a more rapid recurrence/progression natural history. Although these results must be confirmed in a much larger cohort of samples with excellent clinical outcome data, it is even more impressive that RB1 S780 staining performed well in identifying grade 1.5 tumors in this group of samples.

Gene expression profiling focused on tumor grade–specific transcription has been useful for showing unique profiles. A recent study shed light onto several meningioma subgroups, defined by genome methylation status, including an aggressive subgroup of WHO grade I tumors (13). Utilizing 2D gel electrophoresis and MS, our group detected a unique group of proteins that distinguished clinically aggressive grade I tumors (10). Focusing on phosphorylation events, AKAP12 was also associated with meningioma progression (16). Proteomic studies can add to the increasing volume of genetic and transcriptional studies, toward a goal of better biomarkers for patients with meningioma. These data provide proof of this concept.

In the classification of gliomas, the use of FISH for detection of 1p/19q loss in oligodendrogliomas directs therapy and predicts natural history (37). Further, MGMT hypermethylation and IDH1 mutation status help with prognosticating in high-grade gliomas (38). These examples provide real-time biomarkers shedding light onto the pathogenesis of a solid brain tumor, where WHO grading is insufficient.

Protein modulation is an end result of changes in genetic variants, epigenetic alterations, and gene transcription. Study of an aggressive tumor's proteome has been instrumental in the identification of targets for therapeutic strategies. Emerging proteomic techniques have shown potential to characterize the dynamic regulation or dysregulation of protein expression and function. Cancer proteomic profiling has been used for the characterization of specific tumors, and systematic review of these data using bioinformatics has provided insight into pathophysiology.

Kinase inhibitors have played an important role in the treatment of cancer; thus, we utilized proteomic techniques focusing on phosphorylation events. Our group assessed iTRAQ phosphoproteome and kinome peptide array profiling in WHO grade II and III meningiomas (16). This study was validated in a larger cohort of patients and clearly illustrates how analysis of protein functions and networks can yield unique signature and targets of potential diagnostic and prognostic value, in well-defined histologic specimens (16).

Using a similar experimental design (16), we identified dysregulated proteins and cascades in clinically aggressive 1.5 meningiomas. Although unique and well-defined signatures were seen across meningioma grades (16), dramatic molecular differences were not observed between WHO I versus 1.5. These observations were not unexpected with both having WHO I bland histologic features. iTRAQ and peptide chip identified 32 phosphoproteins and 10 phosphopeptides, respectively. Combined datasets revealed mechanisms of 1.5 tumors, allowing molecular characterization. AKT mutation or activation is not a defining feature of this group. The AHR cascade can affect cellular transduction through interactions with Rb (39).

Evidence for RB1 signaling in the pathogenesis of grade 1.5 meningiomas is as follows: (1) the kinome revealed a significant increase in RB1 phosphorylation levels predicted to occur at the S780 site, (2) the predicted phosphorylation site is a consensus for CDKs kinases, (3) the canonical pathway analysis identified the cell cycle G1–S checkpoint signaling and Western blot analysis revealed significant related increases in CDK4, CDK6, and RB1 S780 (Fig. 2E). RB1 is a commonly affected tumor-suppressor gene in cancers (40), regulating multiple pathways to influence proliferation, migration, invasion, and cell cycle (41). Identification of hyperphosphorylation on RB_774_78 in grade 1.5 meningiomas defined this group. Phosphorylation of many sites on RB1 by the cyclins/CDKs has been described (42, 43). RB1 harbors up to 16 potential S/T-P consensus sites for CDKs (43). The majority of the phosphosites are clustered at the carboxyl-terminus. The RB_774_786 (T-R-P-P-T-L-S-P-I-P-H-I-P) peptide contains three phosphosites: T774, T778, and S780. The sequence of RB_774_786 (T-R-P-P-T-L-S-P-I-P-H-I-P) displays the S/T-P consensus site, which includes T778 and S780. The consensus analysis predicted S780 as being the most likely to be phosphorylated (score = 0.926) when compared with T774 and T778 sites (Supplementary Fig. S3C). RB1 S780 has been described (44), whereas T774 and T778 remain uncharacterized. A commercially available antibody allowed investigation of S780 hyperphosphorylation. Our results demonstrate that inactivation of RB1/EF2 signaling can distinguish the subgroup of 1.5 meningioma from WHO I. The kinome profiling suggested mTOR phosphorylation (FRAP_2443_2455) is upregulated in 1.5 meningiomas (Fig 1C; Supplementary Table S7). The literature reports that RB1 pathway inactivation results in mTOR overexpression (45). mTOR may directly regulate AKT phosphorylation, by decreasing AKT S473 (Fig 4A) aiding mTORC1 upregulation (ref. 46; Fig 4). Although our data reveal a decrease in both AKT S473 and T308 and no change in RPS6KB1 T389, the relationship to mTOR is unsettled (Fig. 2A). This will be the focus of future work. The increase in RB1 S780 phosphorylation may be driven by CDK4/6 overexpression (Figs. 2E and 4), which results in the dissociation of E2F-1, allowing E2F-1 to activate the transcription of genes required for DNA synthesis and cell-cycle progression (44).

Figure 4.

Proposed meningioma grade 1.5 mechanism of aggressiveness: Gene names are shown at the approximate positions where their encoded proteins function in the pathway. Colored molecules were significantly altered. Gray molecules: affected with no significant changes in regulation. Red molecules: significant upregulation. Green molecules: significant downregulation. Continuous arrows: activation. Dashed arrows: inactivation.

Figure 4.

Proposed meningioma grade 1.5 mechanism of aggressiveness: Gene names are shown at the approximate positions where their encoded proteins function in the pathway. Colored molecules were significantly altered. Gray molecules: affected with no significant changes in regulation. Red molecules: significant upregulation. Green molecules: significant downregulation. Continuous arrows: activation. Dashed arrows: inactivation.

Close modal

Phosphorylation of RB1 S780 proved to be a real-time biomarker for identifying the 1.5 meningiomas. Hyperphosphorylation of RB1 S780 correlated with increased recurrence/progression in a comprehensive cohort of patients. This was not the case in WHO grade I, II, or III tumors, illustrating the specificity for grade 1.5 meningiomas.

Although all grade I tumors, in the discovery set, were first-time surgery samples, three of five of the grade 1.5 tumors were initial surgical samples. None of the discovery set samples had been treated with radiotherapy prior to surgery. To understand if RB1 S780 is predictive, we organized a TMA cohort of samples including WHO grade I, 1.5, II, and III meningiomas. Samples were from initial and subsequent (second–fifth operations) interventions and sometimes before or after radiotherapy treatments. RB1 S780 staining was specific for the grade 1.5 group of meningiomas. Patterns for high RB1 S780 staining at the first or second through fifth surgery remained high. This suggests that RB1 S780 is phosphorylated at an early stage when WHO grade I histology is reflected, predicting the clinical activity seen with these aggressive tumors. RB1 staining did not have this predictive ability in postradiotherapy cases. RB1 dephosphorylation upon radiotherapy treatment occurs (47) and may be involved with meningioma escape from treatment. It is very interesting that the levels of RB1 S780 staining are also decreased (to levels seen in WHO grade I tumors) in the more aggressive WHO grade II and III tumors, before and after radiotherapy. This likely reflects the complexity and heterogeneous nature of meningiomas, deeper than the WHO grading scheme. The potential for therapeutic targeting of the RB1 pathway may lead to better outcomes for this particular subset of grade 1.5 tumors. To this end, the in vitro characterization of the Rb S780 site is under investigation.

Other authors have identified this group of histologically benign meningioma with clinically aggressive behavior (3, 4, 5, 6, 8, 12). So far, proposed molecular grading schemes for 1.5 meningioma have been based mainly on cytogenetics. The identification of chromosomal abnormalities (12, 6) with potential prognostic value has been reported in histologically benign meningiomas. A recent study shed light onto several meningioma subgroups defined by genomic methylation status (13). Compared with WHO grading, classification by individual and combined methylation classes more accurately identifies patients at high risk of disease progression in tumors with WHO I histology (13). Although mutational analysis and genome methylation are promising, our data demonstrate the potential of RB1 S780 staining for the diagnosis of 1.5 meningiomas and a marker for recurrence. Nonirradiated grade 1.5 samples had significant hyperphosphorylation at RB1 S780 versus WHO I. In resequencing NF2, SMO, AKT, KLF4, and TRAF7, we found that mutations did not correlate with RB1 S780 staining. Although there may be trends for TRAF7 and KLF4 mutations in benign and aggressive courses, respectively, these alterations did not identify grade 1.5 meningiomas. Mechanisms driving the clinical phenotype and oncogenesis of meningiomas are very complicated. Prognostic biomarkers and treatments are likely to draw from genetic, transcriptional, epigenetic, and proteomic mechanisms. The time has come for meningioma pathologic specimens to be analyzed with several biomarkers for the most accurate identification and prediction of clinical course. This will result in the best outcomes for patients with meningioma.

Taken together, these data provide a basis for the concept that RB1 S780 phosphorylation may play a role in the progression/recurrence phenotype in grade 1.5 meningiomas, and suggest it may be a potential predictive biomarker. Staining of RB1 S780 is a promising biomarker for risk stratification and diagnosis, and should be validated in a larger prospective cohort.

No potential conflicts of interest were disclosed.

Conception and design: C.A. Parada, J.W. Osbun, J. Zhang, M. Ferreira Jr

Development of methodology: C.A. Parada, J.W. Osbun, Y. Karasozen, S. Kaur, M. Shi, C. Pan, L.F. Gonzalez-Cuyar, D.E. Born, J. Zhang, M. Ferreira Jr

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): C.A. Parada, J.W. Osbun, T. Busald, Y. Karasozen, P.J. Cimino, C. Pan, L.F. Gonzalez-Cuyar, R. Rostomily, D.E. Born, J. Zhang, M. Ferreira Jr

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): C.A. Parada, J.W. Osbun, Y. Karasozen, M. Shi, J. Barber, W. Adidharma, P.J. Cimino, C. Pan, L.F. Gonzalez-Cuyar, J. Zhang, M. Ferreira Jr

Writing, review, and/or revision of the manuscript: C.A. Parada, J.W. Osbun, Y. Karasozen, M. Shi, W. Adidharma, P.J. Cimino, L.F. Gonzalez-Cuyar, R. Rostomily, J. Zhang, M. Ferreira Jr

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): C.A. Parada, T. Busald, W. Adidharma, C. Pan, D.E. Born, M. Ferreira Jr

Study supervision: C.A. Parada, M. Ferreira Jr

Other (IRB approval): T. Busald

Other (provided tumor samples and patient care): R. Rostomily

This study was supported by University of Washington Department of Neurosurgery, Goertzen Foundation, and Kapogiannatos family funds (to M. Ferreira). The funding sources had no role in the design and conduct of the study, collection, management, analysis and interpretation of the data, preparation, review or approval of the manuscript, or decision to submit the manuscript for publication.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1.
Wiemels
J
,
Wrensch
M
,
Claus
EB
. 
Epidemiology and etiology of meningioma
.
J Neurooncol
2010
;
99
:
307
14
.
2.
Louis
DN
,
Perry
A
,
Reifenberger
G
,
von Deimling
A
,
Figarella-Branger
D
,
Cavenee
WK
, et al
The 2016 World Health Organization classification of tumors of the central nervous system: a summary
.
Acta Neuropathol
2016
;
131
:
803
20
.
3.
Pramesh
CS
,
Saklani
AP
,
Pantvaidya
GH
,
Heroor
AA
,
Naresh
KN
,
Sharma
S
, et al
Benign metastasizing meningioma
.
Jpn J Clin Oncol
2003
;
33
:
86
8
.
4.
Nakasu
S
,
Fukami
T
,
Jito
J
,
Nozaki
K
. 
Recurrence and regrowth of benign meningiomas
.
Brain Tumor Pathol
2009
;
26
:
69
2
.
5.
Kotapka
MJ
,
Kalia
KK
,
Martinez
AJ
,
Sekhar
LN
. 
Infiltration of the carotid artery by cavernous sinus meningioma
.
J Neurosurg
1994
;
81
:
252
55
.
6.
Kakkar
A
,
Kumar
A
,
Das
A
,
Pathak
P
,
Sharma
MC
,
Singh
M
, et al
1p/14q co-deletion: a determinant of recurrence in histologically benign meningiomas
.
Indian J Pathol Microbiol
2015
;
58
:
433
38
.
7.
Marciscano
AE
,
Stemmer-Rachamimov
AO
,
Niemierko
A
,
Larvie
M
,
Curry
WT
,
Barker
FG
 2nd
, et al
Benign meningiomas (WHO Grade I) with atypical histological features: correlation of histopathological features with clinical outcomes
.
J Neurosurg
2015
;
14
:
1
9
.
8.
Pollock
BE
,
Stafford
SL
,
Link
MJ
,
Garces
YI
,
Foote
RL
. 
Single-fraction radiosurgery of benign cavernous sinus meningiomas
.
J Neurosurg
2013
;
119
:
675
82
.
9.
Perry
A
,
Scheithauer
BW
,
Stafford
SL
,
Lohse
CM
,
Wollan
PC
. 
"Malignancy" in meningiomas: a clinicopathologic study of 116 patients, with grading implications
.
Cancer
1999
;
85
:
2046
56
.
10.
Osbun
J
,
Tatman
PD
,
Kaur
S
,
Parada
CA
,
Busald
T
,
Gonzalez-Cuyar
L
, et al
Comparative proteomic profiling using two- dimensional gel electrophoresis and identification via LC-MS/MS reveals novel protein biomarkers to identify aggressive subtypes of WHO Grade I meningioma
.
J Neurol Surg
2017
;
78
:
371
79
.
11.
Kumar
S
,
Kakkar
A
,
Suri
V
,
Kumar
A
,
Bhagat
U
,
Sharma
MC
, et al
Evaluation of 1p and 14q status, MIB-1 labeling index and progesterone receptor immunoexpression in meningiomas: adjuncts to histopathological grading and predictors of aggressive behavior
.
Neurol India
2014
;
62
:
376
82
.
12.
Barbera
S
,
San Miguel
T
,
Gil-Benso
R
,
Muñoz-Hidalgo
L
,
Roldan
P
,
Gonzalez-Darder
J
, et al
Genetic changes with prognostic value in histologically benign meningiomas
.
Clin Neuropathol
2013
;
32
:
311
17
.
13.
Sahm
F
,
Schrimpf
D
,
Stichel
D
,
Jones
DTW
,
Hielscher
T
,
Schefzyk
S
, et al
DNA methylation-based classification and grading system for meningioma: a multicentre, retrospective analysis
.
Lancet Oncol
2017
;
18
:
682
94
.
14.
Ostasiewicz
P
,
Zielinska
DF
,
Mann
M
,
Wiśniewski
JR
. 
Proteome, phosphoproteome, and N-glycoproteome are quantitatively preserved in formalin-fixed paraffin-embedded tissue and analyzable by high-resolution mass spectrometry
.
J Proteome Res
2010
;
9
:
3688
700
.
15.
Pan
C
,
Olsen
JV
,
Daub
H
,
Mann
M
. 
Global effects of kinase inhibitors on signaling networks revealed by quantitative phosphoproteomics
.
Mol Cell Proteomics
2009
;
8
:
2796
808
.
16.
Parada
CA
,
Kaur
S
,
Tatman
P
,
Shi
M
,
Busald
T
,
Pan
C
, et al
Phosphoproteomic signature of aggressive meningiomas reveals the A kinase (PRKA) anchor protein (gravin) 12 as a central regulator of migration, invasion, and cell cycle in meningioma malignancy
.
Sci Rep
2018
;
8
:
2098
.
17.
Gulmann
C
,
Sheehan
KM
,
Conroy
RM
,
Wulfkuhle
JD
,
Espina
V
,
Mullarkey
MJ
, et al
Quantitative cell signaling analysis reveals down-regulation of MAPK pathway activation in colorectal cancer
.
J Pathol
2009
;
218
:
514
19
.
18.
Gembitsky
DS
,
Lawlor
K
,
Jacovina
A
,
Yaneva
M
,
Tempst
P
. 
A prototype antibody microarray platform to monitor changes in protein tyrosine phosphorylation
.
Mol Cell Proteomics
2004
;
3
:
1102
18
.
19.
Guo
L
,
Xiao
Y
,
Fan
M
,
Li
JJ
,
Wang
Y
. 
Profiling global kinome signatures of the radioresistant MCF-7/C6 breast cancer cells using MRM-based targeted proteomics
.
J Proteome Res
2015
;
14
:
193
201
.
20.
Klammer
M
,
Kaminski
M
,
Zedler
A
,
Oppermann
F
,
Blencke
S
,
Marx
S
, et al
Phosphosignature predicts dasatinib response in non-small cell lung cancer
.
Mol Cell Proteomics
2012
;
11
:
651
58
.
21.
Wibom
C
,
Mörén
L
,
Aarhus
M
,
Knappskog
PM
,
Lund-Johansen
M
,
Antti
H
, et al
Proteomic profiles differ between bone invasive and noninvasive benign meningiomas of fibrous and meningothelial subtype
.
J Neurooncol
2009
;
94
:
321
31
.
22.
Barkhoudarian
G
,
Whitelegge
JP
,
Kelly
DF
,
Simonian
M
. 
Proteomics analysis of brain meningiomas in pursuit of novel biomarkers of the aggressive behavior
.
J Proteomics Bioinform
2016
;
9
:
53
7
.
23.
Abdi
F
,
Quinn
JF
,
Jankovi
J
,
McIntosh
M
,
Leverenz
JB
,
Peskind
E
, et al
Detection of biomarkers with a multiplex quantitative proteomic platform in cerebrospinal fluid of patients with neurodegenerative disorders
.
J Alzheimers Dis
2006
;
9
,
293
48
.
24.
Shi
M
,
Bradner
J
,
Bammler
TK
,
Eaton
DL
,
Zhang
J
,
Ye
Z
, et al
Identification of glutathione S-transferase pi as a protein involved in Parkinson disease progression
.
Am J Pathol
2009
;
175
:
54
5
.
25.
Lin
X
,
Shi
M
,
Masilamoni
JG
,
Dator
R
,
Movius
J
,
Aro
P
, et al
Proteomic profiling in MPTP monkey model for early Parkinson disease biomarker discovery
.
Biochim Biophys Acta
2015
;
1854
:
779
87
.
26.
Datta
A
,
Qian
J
,
Chong
R
,
Kalaria
RN
,
Francis
P
,
Lai
MK
, et al
Novel pathophysiological markers are revealed by iTRAQ-based quantitative clinical proteomics approach in vascular dementia
.
J Proteomics
2014
;
99
:
54
7
.
27.
Datta
A
,
Chai
YL
,
Tan
JM
,
Lee
JH
,
Francis
PT
,
Chen
CP
, et al
An iTRAQ-based proteomic analysis reveals dysregulation of neocortical synaptopodin in Lewy body dementias
.
Mol Brain
2017
;
10
:
36
.
28.
Xue
Y
,
Liu
Z
,
Cao
J
,
Ma
Q
,
Gao
X
,
Xue
Y
. 
GPS 2.1: enhanced prediction of kinase-specific phosphorylation sites with an algorithm of motif length selection
.
Protein Eng Des
2011
;
24
:
255
60
.
29.
Chartier
M.
,
Chenard
T
,
Barker
J
,
Najmanovich
R
. 
Kinome Render: a stand-alone and web-accessible tool to annotate the human protein kinome tree
.
PeerJ
2013
;
1
:
e126
.
30.
Blom
N
,
Sicheritz-Ponten
T
,
Gupta
R
,
Gammeltoft
S
,
Brunak
S
. 
Prediction of post-translational glycosylation and phosphorylation of proteins from the amino acid sequence
.
Proteomics
2004
;
4
:
1633
49
.
31.
Varjosalo
M
,
Keskitalo
S
,
Van Drogen
A
,
Nurkkala
H
,
Vichalkovski
A
,
Aebersold
R
, et al
The protein interaction landscape of the human CMGC kinase group
.
Cell Rep
2013
;
3
:
1306
20
.
32.
Hanks
SK
,
Hunter
T
. 
Protein kinases 6. The eukaryotic protein kinase superfamily: kinase (catalytic) domain structure and classification
.
FASEB J
1995
;
9
:
576
96
.
33.
Clark
VE
,
Erson-Omay
EZ
,
Serin
A
,
Yin
J
,
Cotney
J
,
Ozduman
K
, et al
Genomic analysis of non-NF2 meningiomas reveals mutations in TRAF7, KLF4, AKT1, and SMO
.
Science
2013
;
339
:
1077
80
.
34.
Bi
WL
,
Greenwald
NF
,
Abedalthagafi
M
,
Wala
J
,
Gibson
WJ
, et al
Genomic landscape of high-grade meningiomas
.
NPJ Genom Med
2017
;
2
:
pii:15
.
35.
Goutagny
S
,
Yang
HW
,
Zucman-Rossi
J
,
Chan
J
,
Dreyfuss
JM
,
Park
PJ
, et al
Genomic profiling reveals alternative genetic pathways of meningioma malignant progression dependent on the underlying NF2 status
.
Clin Cancer Res
2010
;
16
:
4155
64
.
36.
Lindblom
A
,
Ruttledge
M
,
Collins
VP
,
Nordenskjöld
M
,
Dumanski
JP
. 
Chromosomal deletions in anaplastic meningiomas suggest multiple regions outside chromosome 22 as important in tumor progression
.
Int J Cancer
1994
;
56
:
354
57
.
37.
Nigro
JM
,
Takahashi
MA
,
Ginzinger
DG
,
Law
M
,
Passe
S
,
Jenkins
RB
, et al
Detection of 1p and 19q loss in oligodendroglioma by quantitative microsatellite analysis, a real-time quantitative polymerase chain reaction assay
2001
;
158
:
1253
62
.
38.
Combs
ES
,
Rieken
S
,
Wick
W
,
Abdollahi
A
,
von Deimling
A
,
Debus
J
, et al
Prognostic significance of IDH-1 and MGMT in patients with glioblastoma: one step forward, and one step back?
Radiat Oncol
2011
;
6
:
115
.
39.
Ge
NL
,
Elferink
CJ
. 
A direct interaction between the aryl hydrocarbon receptor and retinoblastoma protein. Linking dioxin signaling to the cell cycle
.
J Biol Chem
1998
;
273
:
22708
13
.
40.
Weinberg
RA
. 
The retinoblastoma protein and cell cycle control
.
Cell
1995
;
81
:
323
30
.
41.
Di Fiore
R
,
D'Anneo
A
,
Tesoriere
G
,
Vento
R
. 
RB1 in cancer: different mechanisms of RB1 inactivation and alterations of pRb pathway in tumorigenesis
.
J Cell Physiol
2013
;
228
:
1676
87
.
42.
Adams
PD
. 
Regulation of the retinoblastoma tumor suppressor protein by cyclin/cdks
.
Biochim Biophys Acta
2001
;
1471
:
M123
33
.
43.
Knudsen
ES
,
Wang
JY
. 
Differential regulation of retinoblastoma protein function by specific Cdk phosphorylation sites
.
J Biol Chem
1996
;
271
:
8313
20
.
44.
Connell-Crowley
L
,
Harper
JW
,
Goodrich
DW
. 
Cyclin D1/Cdk4 regulates retinoblastoma protein-mediated cell cycle arrest by site-specific phosphorylation
.
Mol Biol Cell
1997
;
8
:
287
301
.
45.
El-Naggar
S
,
Liu
Y
,
Dean
DC
. 
Mutation of the RB1 pathway leads to overexpression of mTor, constitutive phosphorylation of Akt on serine 473, resistance to anoikis, and a block in c-Raf activation
.
Mol Cell Biol
2009
;
29
:
5710
17
.
46.
Breuleux
M
,
Klopfenstein
M
,
Stephan
C
,
Doughty
CA
,
Barys
L
,
Maira
SM
, et al
Increased AKT S473 phosphorylation after mTORC1 inhibition is rictor dependent and does not predict tumor cell response to PI3K/mTOR inhibition
.
Mol Cancer Ther
2009
;
4
:
742
53
.
47.
Haapajärvi
T
,
Kivinen
L
,
Pitkänen
K
,
Laiho
M
. 
Cell cycle dependent effects of u.v.-radiation on p53 expression and retinoblastoma protein phosphorylation
.
Oncogene
1995
;
11
:
151
9
.