Molecular characterization of gliomas and glioneuronal tumors amid Noonan syndrome: cancer predisposition examined

Introduction In the setting of pediatric and adolescent young adult cancer, increased access to genomic profiling has enhanced the detection of genetic variation associated with cancer predisposition, including germline syndromic conditions. Noonan syndrome (NS) is associated with the germline RAS pathway activating alterations and increased risk of cancer. Herein, we describe our comprehensive molecular profiling approach, the association of NS with glioma and glioneuronal tumors, and the clinical and histopathologic characteristics associated with the disease. Methods Within an institutional pediatric cancer cohort (n = 314), molecular profiling comprised of paired somatic disease–germline comparator exome analysis, RNA sequencing, and tumor classification by DNA methylation analysis was performed. Results Through the implementation of paired analysis, this study identified 4 of 314 (1.3%) individuals who harbored a germline PTPN11 variant associated with NS, of which 3 individuals were diagnosed with a glioma or glioneuronal tumor. Furthermore, we extend this study through collaboration with a peer institution to identify two additional individuals with NS and a glioma or glioneuronal tumor. Notably, in three of five (60%) individuals, paired genomic profiling led to a previously unrecognized diagnosis of Noonan syndrome despite an average age of cancer diagnosis of 16.8 years. The study of the disease-involved tissue identified signaling pathway dysregulation through somatic alteration of genes involved in cellular proliferation, survival, and differentiation. Discussion Comparative pathologic findings are presented to enable an in-depth examination of disease characteristics. This comprehensive analysis highlights the association of gliomas and glioneuronal tumors with RASopathies and the potential therapeutic challenges and importantly demonstrates the utility of genomic profiling for the identification of germline cancer predisposition.

Introduction: In the setting of pediatric and adolescent young adult cancer, increased access to genomic profiling has enhanced the detection of genetic variation associated with cancer predisposition, including germline syndromic conditions.Noonan syndrome (NS) is associated with the germline RAS pathway activating alterations and increased risk of cancer.Herein, we describe our comprehensive molecular profiling approach, the association of NS with glioma and glioneuronal tumors, and the clinical and histopathologic characteristics associated with the disease.
Methods: Within an institutional pediatric cancer cohort (n = 314), molecular profiling comprised of paired somatic disease-germline comparator exome analysis, RNA sequencing, and tumor classification by DNA methylation analysis was performed.
Results: Through the implementation of paired analysis, this study identified 4 of 314 (1.3%) individuals who harbored a germline PTPN11 variant associated with NS, of which 3 individuals were diagnosed with a glioma or glioneuronal tumor.Furthermore, we extend this study through collaboration with a peer institution to identify two additional individuals with NS and a glioma or glioneuronal tumor.Notably, in three of five (60%) individuals, paired genomic profiling led to a
The World Health Organization (WHO) 2021 guidelines (13) have expanded the defining molecular features associated with gliomas and, in concert with histopathologic findings, can provide additional supporting data for tumor classification in the setting of an integrated diagnosis.For example, RGNT is a rare, indolent, glioneuronal WHO grade 1 tumor (13), characterized by biphasic neurocytic and glial architecture, most commonly arising in the midline, especially in the fourth ventricular region (14)(15)(16), with rare cases reported in the spine or cerebral hemispheres (17,18).Molecularly, RGNT shares a characteristic methylation signature and FGFR1 alterations, co-occurring with alterations involving PIK3CA or PIK3R1 and/or NF1 (12,19,20).Diffuse leptomeningeal glioneuronal tumor (DLGNT) is an example of a glioma that was introduced as a provisional entity in the 2016 WHO Classification of CNS Tumors (21).DLGNT is a rare glioneuronal neoplasm, composed of oligodendrocyte-like cells and molecularly characterized by chromosome arm 1p deletion with MAPK pathway gene alteration (13).
Herein, we present a series of five individuals with NS diagnosed with glioma or glioneuronal tumors, along with molecular characterization of these tumors.The integrated diagnosis presented draws from the clinicopathologic, histologic, and molecular features of the tumors from these individuals.

Cohorts
At the Nationwide Children's Hospital (NCH; Columbus, OH), 314 individuals with solid tumors or hematologic malignancy/ disease consented as part of an institutional translational protocol from January 2018 to April 2022.Under this Institutional Review Board (IRB)-approved protocol (IRB17-00206), comprehensive molecular profiling, including paired exome sequencing of tumor tissue and a germline comparator, RNA sequencing of the tumor, and when possible, DNA methylation array-based tumor classification, was performed.A histopathologic review of a hematoxylin and eosin-stained slide was used to estimate tumor content among the assayed tumor specimens.This protocol allowed for the identification of somatic cancer-associated genomic alterations to include single nucleotide variants (SNVs) and small insertion-deletion (indel) events, copy number alterations, gene fusions, and importantly, germline findings associated with cancer predisposition.Individuals 1-3 were profiled as part of the NCH cohort methodology.
To expand the results and analysis of the NCH cohort, the Seattle Children's Hospital (Seattle, WA) contributed data from two adolescent young adult individuals who underwent molecular profiling as part of routine clinical testing in the setting of a CNS malignancy at the University of Washington (Seattle, WA).The contribution of this dataset was conducted under an IRB-approved protocol (#14449).Individuals 4 and 5 were profiled as part of the Seattle cohort methodology.

Enhanced exome sequencing-NCH cohort
Libraries were prepared using 100 ng of input DNA beginning with enzymatic fragmentation, followed by end repair, 5′ phosphorylation, A-tailing, and platform-specific adapter ligation using NEBNext Ultra II FS reagents (New England Biolabs, Ipswich, MA).Target enrichment by hybrid capture was performed using the xGen Exome Research Panel v1.0 enhanced with the xGenCNV Backbone and Cancer-Enriched Panels-Tech Access (Integrated DNA Technologies, Coralville, IA).Paired-end 151-bp reads were generated on the Illumina HiSeq 4000 or NovaSeq 6000 (Illumina, San Diego, CA).Secondary analysis was performed using Churchill, a comprehensive bioinformatic workflow that takes raw sequencing reads from alignment through variant identification through the incorporation of the following analytical steps (32).Reads were aligned to the human genome reference sequence (build GRCh38) using Burrows-Wheeler Aligner (BWA) (v0.7.15) and refined according to community-accepted guidelines for best practices (https://gatk.broadinstitute.org/hc/en-us).Duplicate sequence reads were removed using samblaster-v.0.1.22,and base quality score recalibration was performed on the aligned sequence data using the Genome Analysis Toolkit (v4.1.9)(33).The average depth of coverage obtained for the comparator germline samples and disease-involved tumors is presented in Supplementary Table 1.Germline variants were called using GATK's HaplotypeCaller (34).Somatic SNV and indel detection was performed using MuTect2 (35).Variants were filtered for those >5% variant allele fraction (VAF), within a coding region of the genome, or in proximity to an exon/intron junction (splice site region ≤ 3 bp), with a parameter defined as passing (MuTect2 PASS), and according to population frequency (gnomAD FAFpopmax95 <= 0.0001).Additionally, germline and somatic variation in cancer-associated genes was filtered using a gene set curated from the published literature and genomic databases including those described by Zhang et al., as well as genes with strong or emerging evidence of germline or somatic cancer association as documented in the Cancer Gene Census (36,37) (Supplementary Table 2).Variants flagged as pathogenic/likely pathogenic or with conflicting interpretations in ClinVar were also reviewed.Reportable variants were manually reviewed using the Integrated Genomics Viewer.Variants reported as part of this translational research protocol met the aforementioned criteria and were evaluated in the setting of the tumor type under study.Variants of uncertain significance were not reported.Copy number variation (CNV) was assessed using a combination of GATK (v4.2.4.1) and VarScan2 (38).

RNA sequencing-NCH cohort
In parallel with enhanced exome sequencing, tumor-derived RNA, input at 500 ng, was subject to DNase treatment and ribodepletion prior to library preparation using the NEBNext Ultra II Directional RNA Library prep kit with 5-10 min of chemical fragmentation for individuals 1, 2, and 3. Independent cDNA libraries were diluted for whole transcriptome sequencing.Paired-end 151-bp reads were generated on the Illumina HiSeq 4000 or NovaSeq 6000 (Illumina, San Diego, CA), and reads were aligned to the human genome reference sequence (GRCh38).The resultant output represented 141,857,623 (individual 1), 236,186,711 (individual 2), and 185,159,431 (individual 3) uniquely mapped reads for the tumor samples.
Normalized read counts (TPM) were calculated for all samples from an internal NCH cohort of CNS cancers (n = 235) using the Salmon package.An external cohort of CNS cancers (n = 791) from the Treehouse Childhood Cancer Initiative at the University of California Santa Cruz Genomics Institute (v.9 and v.11, University of California Santa Cruz, Santa Cruz, CA) was combined with the internal cohort to improve clustering.Expression counts of proteincoding genes were log2(x + 1)-transformed and quantile normalization was performed.The 500 protein-coding genes with the highest variances were used to perform a principal components analysis.The data dimensionality was further reduced by performing Uniform Manifold Approximation & Projection (UMAP) using the umap package.This analysis used the first 34 principal components, as they account for >80% of the variance among the samples.The UMAP plot of UMAP1 and UMAP2 was generated using the R packages ggplot2 and plotly.

DNA methylation array-NCH cohort
DNA methylation profiling was performed on nucleic acid extracted from disease-involved tissue.For each tumor studied, 250 ng of input DNA was bisulfite-converted (catalog # D5006, Zymo Research, Irvine CA) and, if applicable, treated using the Illumina formalin-fixed paraffin-embedded (FFPE) restoration process (catalog # WG-321-1002, Illumina, San Diego CA).Bisulfite-converted DNAs, including methylated human DNA controls (catalog # D5014, Zymo Research, Irving CA), were hybridized to the Infinium Methylation EPIC BeadChip (catalog # WG-317-1001, Illumina, San Diego, CA) following the Illumina Infinium HD Methylation protocol.Beadchips were imaged on the Illumina iScan System, and the resulting raw IDAT files were processed through a local installation of the German Cancer Research Center (DKFZ) DNA Methylation Brain Tumor Classifier, version 12.5 (47).The Classifier algorithm produces a score that indicates the similarity of the queried sample against previously characterized CNS tumor types, wherein a score ≥0.9 is strongly supportive of that type.

Targeted next-generation sequencing-Seattle cohort
Paired tumor-germline testing was performed on diseaseinvolved and normal sample DNA from individuals 4 and 5 using the University of Washington-OncoPlex version 7 (OPXv7), a DNA-based targeted next-generation sequencing panel, as previously described (48).In brief, DNA was extracted from peripheral blood using Qiagen DSP DNA Midi Kit, while total nucleic acid (TNA) was extracted from FFPE tissue using the Qiagen DNA/RNA AllPrep Kit followed by DNA extraction using the Qiagen GeneRead DNA FFPE Kit (Qiagen, Valencia, CA).After shearing, libraries prepared using KAPA HyperPrep reagents (Roche, Wilmington, MA) were hybridized to a set of custom probes (xGen Lockdown Probes, Integrated DNA Technologies, Coralville, IA) designed to target a panel of 377 genes (Supplementary Table 3) chosen for their relevance in cancer diagnosis, prognosis, and/or treatment.In addition to identifying SNVs, indels, select gene fusions, and CNVs, OPXv7 was also validated to detect microsatellite instability and tumor mutational burden.The libraries were subsequently sequenced on an Illumina NextSeq500 system (Illumina, San Diego, CA), and sequences were processed through an automated, custom-designed bioinformatics pipeline developed by the University of Washington Next Generation Sequencing Laboratory and Analytics group before analysis by a board-certified molecular pathologist (48).The average depth of coverage obtained for the comparator germline samples and disease-involved tumors is presented in Supplementary Table 1.

Results
Amid the NCH cohort, 4 of 314 individuals (1.3%) who underwent genomic profiling as part of an institutional translational cancer protocol harbored a germline variant in PTPN11 associated with NS.A diagnosis of NS was not previously known in two of the four (50%) individuals within this cohort until comprehensive molecular profiling was performed due to a cancer diagnosis.Three of these individuals were diagnosed with CNS malignancies including RGNT (n = 2; individuals 1 and 2) and a diffuse leptomeningeal glioneuronal tumor with worrisome molecular features, not elsewhere classified (NEC) (n = 1; individual 3).A fourth individual with NS in the NCH cohort was diagnosed with B-cell acute lymphocytic leukemia and is therefore excluded from further in-depth examination in this CNS-focused analysis.The paired exome sequencing used in this translational protocol testing was the first manner of identification of NS-associated PTPN11 variation in individuals 2 and 3 and subsequently allowed for orthogonal confirmation by targeted Sanger sequencing in the clinical laboratory.The observation of CNS malignancies and NS in the NCH cohort served as the foundation for an extended analysis.To further enhance our dataset, collaboration with a peer institution identified two additional individuals diagnosed with gliomas or glioneuronal tumors and NS (Seattle cohort-individuals 4 and 5).Given the unique molecular profiles associated with these CNS tumors, we discuss the utility of paired sequencing and the clinicopathologic, histologic, and molecular features of these five individuals.The average age at cancer diagnosis among this combined cohort of five individuals with NS was 16.8 years (range 9-22 years).The cohort was predominantly male patients (one female patient, four male patients) with variable clinical features associated with NS, including short stature (five of five), dysmorphic facial features (five of five), intellectual disability or delay (four of five), and cardiac abnormalities (three of five) (Supplementary Table 4).

Individual 1
A 22-year-old male patient with NS diagnosed during infancy presented with progressive headaches, vomiting, and visual disturbances.A brain magnetic resonance imaging (MRI) demonstrated an extensive suprasellar mass involving the third ventricle, extending through the floor into the prepontine cistern with obstructive hydrocephalus (Figures 1A, B).Spinal MRI showed no evidence of drop metastasis.Subtotal surgical resection was performed.Histologic examination of the tumor (Table 1) showed a low-grade glioma of moderate cell density, composed of mostly small bland round neurocytic-type cells, with fewer bipolarspindled piloid cells, consistent with a partial pilocytic  astrocytoma component; focally, ganglion-like cells were also seen.
There was a partial angiocentric (perivascular rosette-like) pattern, with small round neurocytic cells surrounding perivascular neuropil-rich regions, which showed staining for synaptophysin (Figures 1C-E).In some areas, rosettes with eosinophilic neuropil cores surrounded by completely circumferential neurocytic tumor cells were obvious.Occasional eosinophilic granular bodies were seen focally, but no Rosenthal fibers were present.There was prominent glomerular-type microvascular proliferation and focal microcalcification.Comprehensive genomic profiling using paired exome sequencing confirmed the presence of a germline missense variant in PTPN11 p.Gly60Ala (classified as pathogenic in ClinVar, variation ID: 40493) (Table 2).Additionally, the tumor (estimated at 100% tumor content) was found to harbor an in-frame deletion in PIK3R1 p.Ile442_Thr454del and a recurrent hotspot missense variant in FGFR1 p.Lys656Glu.No clinically actionable gene fusions were identified.No CNV or copy-neutral loss of heterozygosity (cnLOH) was identified (Supplementary Figure 1A).DNA methylation array-based classification analysis (v12.5) demonstrated a high-confidence family score (0.99907) for a lowgrade glioneuronal tumor, with the highest methylation class associated with RGNT (0.56372).
The final integrated diagnosis was consistent with RGNT (WHO grade 1).A year following tumor resection, focal proton beam irradiation was performed for local progression, which was complicated by severe somnolence syndrome and managed with steroid therapy.There has been no evidence of tumor progression 38 months after irradiation.

Individual 2
A 19-year-old male patient was reported with complex medical history including prematurity, bradycardia, and cyanosis at birth requiring cardiopulmonary resuscitation, reported neonatal strokes, developmental delays, facial dysmorphism, short stature, and intellectual disability.He presented with chronic migraine headaches, and brain imaging at the age of 16 years displayed multiple abnormal signals in the pineal region, thalami, and midbrain of unknown etiology (Figures 2A, B).Spine MRI and cerebrospinal studies were negative.An MRI-guided stereotactic biopsy was consistent with RGNT; bland small round neurocytic cells of moderate density, with no significant pleomorphism, were found completely surrounding synaptophysin-positive neuropil cores, with no Rosenthal fibers or eosinophilic granular bodies seen (Table 1 and Figures 2C-E).
Paired exome sequencing (estimated 70% tumor content) identified a pathogenic missense hotspot variant in PTPN11 p.Asn308Asp (classified as expert panel pathogenic in ClinVar, variation ID: 13326) (Table 2).This variant was subsequently confirmed by targeted Sanger sequencing in our clinical laboratory and reported in the patient's medical record.Molecular characterization was therefore consistent with NS, a diagnosis which was previously unrecognized in this individual.The somatic analysis revealed three alterations within PIK3R1.
These include the missense variant, p.Val695Met, an in-frame indel p.Glu451_Asn453delinsAsp, and lastly, an in-frame deletion p .T h r 4 5 4 _ P h e 4 5 6 d e l .T h e t w o i n -f r a m e i n d e l s (p.Glu451_Asn453delinsAsp and p.Thr454_Phe456del) did not occur within the same read based on visualization of the aligned sequence (Supplementary Figure 2).Additionally, a recurrent hotspot missense variant in FGFR1 p.Asn546Lys was also detected.No clinically actionable gene fusions were identified.No CNV was identified; however, cnLOH involving chromosome 19p was detected (Supplementary Figure 1B).DNA methylation analysis (v12.5) demonstrated a high-confidence family score (0.99999) for a low-grade glioneuronal tumor, with the highest methylation class associated with RGNT (0.99998).
The final integrated diagnosis was consistent with RGNT.Now 36 months since diagnosis, initial slow but progressive growth of the tumor has been described, with subsequent stabilization.
RNA-sequencing studies were performed to better understand how these individuals cluster based on similarities in gene expression profiles.A UMAP data structure, as derived from RNA-sequencing studies, was generated from individuals with CNS tumors studied as part of the NCH cohort (n = 235) and the UCSC Treehouse Initiative (n = 791).Individuals 1 and 2 clustered closely together amid other LGG and glioneuronal tumors.This suggests similarity in gene expression between each tumor which may reflect the shared mutational profile, including activating FGFR1 variants, in addition to PIK3R1 variants predicted to activate the PI3K signaling pathway, in concert with the germline PTPN11 variant (Supplementary Figure 3, Table 2).

Individual 3
A 9-year-old female patient presented with a transient headache and left-sided weakness which led to a brain and spine MRI being performed.Imaging revealed diffuse dural enhancement and nodular thickening of unknown etiology (without a focal primary mass) for which an oncologic process could not be ruled out.Cerebrospinal fluid cytology analysis following lumbar puncture was negative for malignant cells.Full body positron emission tomography (PET)/computed tomography (CT) scan did not demonstrate obvious extraneural disease.She presented 1 month later and repeat imaging showed persistent linear and nodular leptomeningeal enhancement with progressive increased fluidattenuated inversion recovery (FLAIR) signal abnormalities in the periventricular and deep white matter of bilateral frontal and parietal lobes (Figure 3A).At this time, a dural and leptomeningeal biopsy was performed.Neuropathologic evaluation revealed an unusual glioneuronal tumor with morphologic features suggestive of DLGNT.The tumor was composed of moderately cellular nodules of small bland cells with oval to elongate nuclei embedded in a desmoplastic to collagenous stroma (Figure 3B).No definite neuronal component was identified by morphology.Mitoses were not found.The tumor cells were immunoreactive for Olig2 (Figure 3C) and GFAP.Synaptophysin immunostain highlighted very rare small cells.Paired exome analysis of disease-involved tissue (estimated 30% tumor content) and a comparator sample revealed a germline missense variant in PTPN11 p.Lys70Arg (classified as expert panel pathogenic in ClinVar, variation ID: 44603) (Table 2).The tumor harbored a somatic hotspot activating variant in KRAS p.Gly12Asp.The somatic copy number analysis demonstrated a focal amplification of 4q12 (encompassing the PDGFRA, KIT, KDR, CHIC2, and FIP1L1 loci), a loss of chromosome arm 1p, and a gain  of 1q (Supplementary Figure 1C).Targeted fusion analysis was negative, including for KIAA1549::BRAF.In support of a DLGNT diagnosis, IDH1 and IDH2 were considered wild type with no evidence of mutation.DNA methylation analysis (v12.5) failed to classify confidently to a specific tumor type, with a family score of 0.39560 for meningioma and a class score of 0.35455 for meningioma, benign.The low scores may be attributed to low tumor cellularity in the sample under study.UMAP visualization of gene expression data demonstrated that the tumor from individual 3 clustered uniquely from those of individuals 1 and 2 and instead clustered with a separate group of LGGs (Supplementary Figure 3).
Given the unusual molecular findings and methylome features not entirely consistent with typical DLGNT (yet displaying MAPK pathway alteration and 1p loss, with Olig2 positivity and rare cells with synaptophysin immunoreactivity), the tumor was ultimately diagnosed as a glioneuronal neoplasm with worrisome molecular features, NEC.Following the identification of the germline PTPN11 variant in the translational study, sequencing of this locus was performed at an outside clinical laboratory which confirmed the presence of the variant.A review of the patient's clinical characteristics and history revealed that she had previously unrecognized features consistent with NS, including short stature, concerns for poor growth, dysmorphic facial features, easy bruising, and atrial septal defect.Notably, identification of this variant led to cascade testing of family members, and this variant was found to be de novo in our proband.
The patient was treated with the oral MEK inhibitor trametinib, with an initial decrease in disease burden (at 6 weeks), but then rapid progression of leptomeningeal disease and a new, growing right frontal nodule 2.5 months later.At this time, the decision was made to proceed with proton craniospinal irradiation (CSI; 41.4 Gy CSI with 54 Gy boost to right frontal nodule) given early disease progression on targeted therapy and concern for higher-risk molecular features [1q gain and 4q12 amplification (containing PDGFRA and other loci)].The patient is now 18 months postcompletion of proton CSI, with concern for disease progression based on radiographic findings, and has started maintenance chemotherapy with temozolomide and lomustine.

Individual 4
An 18-year-old male patient with a history of short stature, cryptorchidism, gastroesophageal reflux, and minor cardiac leaflet abnormalities presented with progressive headaches.Maternal history was positive for benign brain tumors, epilepsy, and mild pulmonary stenosis; however, neither the mother, patient, nor other family members were previously known to have a diagnosis of NS.Brain MRI/magnetic resonance angiography (MRA) found a primary hemorrhagic lesion arising from the left thalamus and filling a portion of the left lateral ventricle, along with two distinct lesions involving the left insular cortex and the right temporal fossa (Figures 4A, B).Spinal imaging was negative.Subtotal resection of the primary thalamic lesion and ventriculoperitoneal (VP) shunt placement were performed.Histologic examination of the tumor showed a low-grade glial neoplasm with abundant calcifications (Figures 4C-E).No rosettes or pseudorosettes were identified.Foci of microvascular proliferation were noted along with areas of acute hemorrhage.
Targeted molecular profiling of disease-involved tissue and a peripheral blood sample using OPXv7 identified a heterozygous germline pathogenic gain-of-function variant in PTPN11 p.Asn308Asp (classified as expert panel pathogenic in ClinVar, variation ID: 13326) (Table 2), prompting a new diagnosis of NS.The tumor tissue was also positive for well-characterized activating alterations in FGFR1 p.Asn546Lys and two variants in PIK3CA p.His1047Arg and p.His1047Tyr (Supplementary Figure 4).No clinically actionable gene fusions or copy number alterations were otherwise identified.The tumor did not demonstrate any evidence of microsatellite instability using the mSINGs method (49), and the tumor mutational burden was low, estimated to be approximately 1 mutation per megabase (50).
The final integrated diagnosis of the tumor was a low-grade glioneuronal tumor, NEC.Although this tumor had histologic and molecular features suggestive of an RGNT, that diagnosis could not be established given the lack of rosettes in an otherwise wellsampled tumor.The additional incidentally found lesion in the insular cortex was determined to be consistent with benign LGG, and biopsy was not indicated.The temporal fossa lesion was determined to be most compatible with a vascular malformation, as has been reported in NS and other RASopathies (2).Postoperatively, this individual developed right-sided hemiparesis and was found to have a venous infarct in the left thalamus and basal ganglia.Routine imaging 3 months following resection found ventriculomegaly with trapping of the temporal horn of the left lateral ventricle, necessitating a secondary VP shunt placement.Now 6 months since diagnosis, he has no evidence of tumor progression.

Individual 5
A 16-year-old male patient had a remote history of growth delay which ultimately led to a workup and diagnosis of NS.He was treated with growth hormone replacement but was otherwise asymptomatic until he developed seizures.A brain MRI identified a left parietal lobe mass measuring 2.2 × 1.8 cm with increased T2/FLAIR signal intensity (Figure 5A).The MRI also identified small foci of signal abnormality in the thalamus.Gross total surgical resection of the main tumor mass was performed.Histologically, the tumor was largely composed of small oligodendroglial-like cells with round nuclei with occasional nuclear halos along with scattered interspersed cytologically normal neurons (Figures 5B-D).The background had a myxoid appearance in areas though no definitive "floating" neurons were identified.Delicate chicken wire-like vessels were noted in the background.
Molecular analysis of tumor tissue and a peripheral blood sample using OPXv7 identified a heterozygous germline pathogenic gain-of-function variant in PTPN11 p.Asn58Lys (classified as pathogenic in ClinVar, variation ID: 40488) and a germline heterozygous EGFR nonsense variant p.Arg531Ter along with a somatic FGFR1 p.Asn546Lys alteration (Table 2).The PTPN11 variant conferred a diagnosis of Noonan syndrome.This individual also harbored a germline inactivating variant in EGFR, for which he would be a carrier for autosomal recessive neonatal inflammatory skin and bowel disease (51).Clinical methylation profiling of the tumor demonstrated no high confidence match but was suggestive of DNET.The tumor did not demonstrate any evidence of microsatellite instability using the mSINGs method (49), and the tumor mutational burden was low, estimated to be approximately 1 mutation per megabase (50).
The final integrated molecular diagnosis of the tumor was consistent with DNET.Following surgical resection, this individual did well and had no subsequent seizures.Follow-up imaging studies have been stable with no new areas of signal abnormality in radiographic studies.

Discussion
Herein, we report on the clinical features, histopathologic characteristics, and clinical course associated with molecularly characterized gliomas and glioneuronal tumors among a group of individuals with NS.Part of the RASopathies spectrum, NS is associated with a heightened risk of pediatric malignancies, with a standardized incidence ratio of 8.1 in childhood cancer registries (6,8).A spectrum of CNS tumors has been reported in association with NS, including LGG, medulloblastoma, HGG, and others (52, 53).PTPN11 is the most commonly altered gene in NS and encodes the SH2 domain-containing tyrosine phosphatase 2 (SHP-2), modulating the RAS/MAPK and the PI3K/AKT pathways, with resultant control in cell growth and differentiation (Figure 6) (54,55).Dysregulation of these pathways leads to altered cellular proliferation, invasion, and inhibition of apoptosis, contributing to tumorigenesis (56, 57).In this report, we describe the molecular characteristics of gliomas or glioneuronal tumors in five individuals with NS associated with PTPN11.All five individuals also had somatic alterations which led to activation of the RAS/MAPK and PI3K/AKT signaling pathways (Table 2, Figures 6, 7).
In 2021, the WHO released guidelines for CNS tumors (13), which emphasized the use of an integrated diagnosis, in which the histopathologic features combined with molecular findings support a final diagnosis.Within these guidelines, the use of comprehensive molecular testing, in particular DNA-based methylation classification, can aid in accurately classifying specific tumors or tumor subtypes.Notably, the WHO identifies numerous tumor types, for which methylation profiling has high diagnostic utility (13).DNA-based methylation classification for CNS tumors was described by Capper et al., in which a machine learning-based classifier was developed to calculate the probability of an individual tumor sample's similarity to one of the 91 methylation classes included in the classifier's training dataset (47).Among >1,000 tumors evaluated by methylation classification, 76% were in agreement between histopathology and predicted methylation classification.Nevertheless, methylation analysis may not provide diagnostic certainty in all cases due to an atypical methylation pattern in the studied tumor or underrepresented methylation profiles for a given entity within the applied classifier.
The utility of methylation profiling is limited in tumors with diminished disease content.High tumor content (>60%) is often  In NS, most brain tumors are described as glial or glioneuronal, with DNETs accounting for 40% of the published cases.Other reported tumors include low-grade glioneuronal tumors, HGGs, and medulloblastoma (5-7, 52, 53, 60-78).Karafin et al. (5) described an early report of RGNT in NS, demonstrating strong pERK immunoreactivity, suggesting MAPK/ERK pathway activation playing a role in tumor development.Given the clinical, histopathologic, and molecular overlap between LGGs, diagnostic characterization can be challenging.As evidence of such, El-Ayadi et al. (52) reported a total of 24 LGGs in the setting of NS, of which only one was diagnosed as RGNT (5).Within this cohort, numerous morphologic mimics of RGNT were documented, including pilocytic astrocytoma (n = 4), pilomyxoid astrocytoma (n = 1), oligodendroglioma (n = 2), low-grade mixed glial-neuronal tumor (n = 1), other low-grade astrocytoma (n = 3), and DNET (n = 9).Histopathologic diagnosis may well be confounded by small biopsies with a high potential for sampling error.A piloid/pilocytic to oligodendroglial-like morphologic component predominates in most classic RGNT, while the characteristic/diagnostic neurocytic rosettes of RGNT may be quite subtle or only focal/sparse.Similarly, molecular characteristics, such as FGFR1 alterations, may frequently occur in other LGGs (11).Furthermore, the majority of NS-associated LGGs in the cohort described by El-Ayadi et al. presented in the now well-known typical midline location of RGNT and at least onethird were multifocal and/or disseminated, a generally much less frequent, if not rare, phenomenon in other LGG diagnoses.Thus, it appears highly plausible that some of the other LGGs previously described may well represent additional examples of RGNT occurring in NS and, with updated molecular characterization/ analysis, might be subject to reclassification.
Molecular profiling in the absence of a paired germline sample can confound variant interpretation, necessitating confirmatory testing for suspected germline findings.If a variant is inferred to be germline, published guidance from the scientific community supports laboratory-recommended follow-up confirmatory testing (79,80).In the setting of a paired analysis, such as that performed in this case series, the etiology of called variants is clearly delineated and may prompt genetic counseling and cascade testing.Importantly, a paired analysis can identify germline cancer predisposition in those with previously unrecognized clinical features.As a result of genomic profiling following the cancer diagnosis, two of four PTPN11-positive individuals within our NCH pediatric/adolescent and young adult (AYA) cancer cohort (individuals 2 and 3) received an NS diagnosis following their cancer diagnosis as a result of paired exome sequencing.Similarly, among the Seattle cohort, individual 4 remained undiagnosed for NS until paired genomic profiling was performed as a result of their CNS malignancy.
While surgical resection remains the standard of care for individuals with pediatric LGG (22,23,25,28), over the past decade, risk-adapted approaches and molecular-directed therapies have refined the management of pediatric LGGs (11, 28), most notably in patients with progressive or relapsing disease.Additionally, Ryall et al. (28) proposed a risk-stratified approach that integrates clinical, imaging, and molecular data to guide appropriate tumor management.In cases of low-risk stratification, conservative approaches would be suitable, while high-risk stratification would necessitate more aggressive interventions (28).
There is increasing experience demonstrating that molecularly targeted therapies, such as RAF (81)(82)(83), MEK (84)(85)(86)(87), and mTOR inhibitors (88)(89)(90), are effective in patients with pediatric LGGs who have failed first-line chemotherapy; these studies have early results with reassuring short-term safety data.The results of both phase I and II clinical trials have led the efforts to evaluate selumetinib as frontline therapy for pediatric LGGs with (NCT03871257) or without neurofibromatosis-1 (NCT04166409).Additionally, the combination of the BRAF inhibitor dabrafenib and the MEK inhibitor trametinib has recently gained FDA approval as a firstline treatment for pediatric patients with LGG (29), and tovorafenib was recently FDA-approved for use in BRAF-altered relapsed or refractory pediatric LGG (91, 92).
Somatic alterations in FGFR1 and PI3K pathway genes in LGG resulting in upregulation of the RAS/MAPK and the PI3K/AKT/ mTOR pathways provide attractive therapeutic targets in the setting of progressive disease.A recent case report describes the use of the mTOR inhibitor everolimus in RGNT which demonstrated radiographic response, after failing first-line treatment with chemotherapy (93).A single-center study observed objective clinical and radiographic responses to the oral FGFR1 inhibitor Debio1347 in three pediatric patients with progressive LGGs, with an acceptable toxicity profile (94).Interestingly, the authors identified specific FGFR1 alterations that could be associated with improved response, as well as concomitant co-occurrence of NF1 deletion as a potential mechanism for acquired resistance to therapy.Currently, the FGFR1 inhibitor erdafitinib is under evaluation in children with advanced solid tumors (NCT03210714).In the setting of NS and glioma, the utility of targeted therapy may be challenging, given the activation of the RAS/MAPK and PI3K/AKT pathways at multiple levels throughout the pathway (Figure 6).Combinatorial approaches or targeting downstream of SHP-2 (protein product of PTPN11) and the PI3K complex, by using MEK and/or mTOR inhibitors, respectively, may be necessary.Targeting RAS genes has proven more difficult due to the lack of binding pockets on the surface of these proteins for targeted therapies to bind; however, studies are ongoing to overcome these challenges (95,96).Recently, in the setting of lung cancer, advances in RAS variant-specific targeting have been achieved with FDA approval of sotorasib and adagrasib in patients with locally advanced or metastatic non-small cell lung cancer harboring a KRAS p.Gly12Cys variant (97,98).
Our study has several limitations.First, the cohort size and composition limits the generalizability of our findings.For the NCH cohort, individuals with rare or treatment-refractory cancers were nominated for translational research testing by a clinician engaged in their care (99).Thus, selection bias may have influenced those nominated for enrollment on the basis of clinical interest and timing of clinical care.Furthermore, those with low-grade CNS tumors that respond well to surgery or conventional chemotherapy may not have been considered for paired exome sequencing at the time of the study.While we were able to extend the NCH cohort findings through the inclusion of two individuals clinically tested as part of the Seattle cohort, there were fundamental differences between each cohort, and it was not possible to glean an overall frequency of NS in the setting of glioma or glioneuronal tumor amid the available data.Future studies with larger cohorts of individuals with CNS malignancies undergoing comprehensive molecular testing are needed to validate and extend our findings.
In the setting of cancer, comprehensive molecular profiling has demonstrated utility in aiding diagnosis, identifying treatment options, and prognostication.Importantly, utilization of paired tumor and germline samples allows for the identification of germline cancer predisposition variants, the frequency of which has been documented in the range of 10-18% amid pediatric/AYA cancer cohorts (100)(101)(102).In a CNS cancer-focused cohort comprised of pediatric/AYA individuals, germline genetic variation associated with cancer susceptibility was identified in 16% (99).Continued broad molecular characterization will extend our understanding of germline cancer susceptibility, particularly the breadth of cancer diagnoses, disease risk, correlation with age, and other genotype-phenotype relationships.Such knowledge potentiates improvement in surveillance recommendations and extension of treatment paradigms.Our study shows that amid this cohort, the characterization of disease was strengthened through the rendering of an integrated diagnosis, supported by multiple molecular testing modalities.This study extends the benefit of comprehensive profiling to aid in both germline and cancer diagnoses, as well as treatment considerations of pediatric/AYA gliomas and glioneuronal tumors.

FIGURE 3 Individual 3 -
FIGURE 3 Individual 3-radiologic and histopathologic studies of the tumor.(A) Axial T1-weighted, post-contrast images in two planes demonstrate a growing, enhancing right frontal nodule (yellow arrow) as well as nodular leptomeningeal enhancement and thickening throughout (green arrows).(B) H&E shows a low-grade-appearing glioneuronal tumor with features suggestive of diffuse leptomeningeal glioneuronal tumor.(C) Immunohistochemical stain shows tumor cell immunoreactivity for Olig2 [Olig2, ×40 ocular mag.].

FIGURE 4 Individual 4 -
FIGURE 4 Individual 4-Radiologic and Histopathologic Studies of the Tumor.(A, B) Axial T2 contrast-weighted images demonstrate a mass lesion ~3.6 cm in greatest dimension adjacent to the left thalamus within the ventricle.The left insular cortex contains a separate lobular lesion approximately 1 cm in diameter in within the right temporal fossa there is also a soft tissue lesion measuring approximately 3 cm in maximum dimension.(C) H&E shows a low-grade appearing glioneuronal tumor with abundant calcifications.(D) Immunohistochemical staining for GFAP is positive.(E)A subset of cells appears positive for synaptophysin.

FIGURE 5 Individual 5 -
FIGURE 5 Individual 5-radiologic and histopathologic studies of the tumor.(A) A coronal T2 post-contrast image contains a nodular cortically based left frontal mass.(B) H&E staining reveals a low-grade glioneuronal tumor with small oligodendroglial-like cells admixed with cytologically normalappearing neurons.(C) A GFAP stain highlights reactive astrocytes but appears negative in the small oligodendroglial-like cells.(D) A NeuN stain highlights scattered neurons between the oligodendroglial-like tumor cells.

TABLE 1
Comparative pathologic findings among the described tumors in this cohort.

TABLE 2
Clinical and molecular characteristics among this cohort.