The Current Landscape of Molecular Pathology for the Diagnosis and Treatment of Pediatric High-Grade Glioma

Abstract

Pediatric high-grade glioma (pHGG) is a devastating group of childhood cancers associated with poor outcomes. Traditionally, diagnosis was based on histologic and immunohistochemical characteristics, including high mitotic activity, presence of necrosis, and presence of glial cell markers (e.g., GFAP). With advances in molecular tumor profiling, these tumors have been recategorized based on specific molecular findings that better lend themselves to prediction of treatment response and prognosis. pHGG is now categorized into four subtypes: H3K27-altered, H3G34-mutant, H3/IDH-WT, and infant-type high-grade glioma (iHGG). Molecular profiling has not only increased the specificity of diagnosis but also improved prognostication. Additionally, these molecular findings provide novel targets for individual tumor-directed therapy. While these therapies are largely still under investigation, continued investigation of distinct molecular markers in these tumors is imperative to extending event-free survival (EFS) and overall survival (OS) for patients with pHGG.

1. Introduction

Pediatric high-grade gliomas (pHGG) are a group of highly aggressive tumors and are the leading cause of cancer-related death in children [1]. It has only been in the last two decades that pHGG has been recognized as a distinct, biologically unique entity from adult gliomas [2]. Until recent years, pathologists and oncologists relied on histologic and immunohistochemical findings in combination with clinical symptomology and tumor location to diagnose patients with a pHGG that was distinct from a pediatric low-grade glioma (pLGG). Despite the ability to generalize tumors based on these findings, pHGG and pLGG frequently have overlapping histopathologic findings making it difficult to predict response to treatment and clinical outcomes. Characteristic imaging findings have also historically played a role in diagnosis and differentiation, with pHGGs primarily appearing as heterogenous masses on MRI with or without enhancement, in comparison to the well-circumscribed lesions seen in pLGG; however, imaging alone is not adequate for distinguishing between types of pHGG [3]. It is only with advances in genetic and molecular profiling of pHGG that more specific diagnoses have become widely accepted. pHGG is now grouped into four subtypes as defined by the WHO CNS5 2021 guidelines with specific associated molecular and genetic changes (

Table 1. Summary of the four pHGG subtypes as defined by the WHO CNS5 2021 guidelines and their associated molecular and genetic changes.

Tumor Type	Key Molecular and Genetic Changes
H3K27-altered glioma	TP53, ACVR1, PDGFRA, EGFR, EZHIP
H3G34-altered glioma	TP53, ATRX, PDGFRA, MYCN
H3/IDH-WT glioma	TP53, PDGFRA, EGFR, MYCN, ID2, IDH-WT, H3-WT
iHGG	NTRK, ALK, ROS, MET
NF1/HGAP	MAPK, ATRX, CDK2A, TP53

). Additionally, pHGG has historically been treated with surgical resection followed by radiation and/or chemotherapy, typically temozolomide, as this agent has been proven effective in adult high-grade glioma (aHGG) [4]. Further profiling has expanded the current knowledge of oncogenesis and provides new avenues for treatment of these devastating tumors that align with the knowledge that pHGG is its own unique entity from aHGG [5]. This review describes the molecular subtypes of pHGG and how an understanding of these molecular differences allows for identification of novel targets for therapy.

2. pHGG Diagnoses and Therapeutic Targets

2.1. Diffuse Midline Glioma (DMG), H3K27-Altered

Perhaps one of the best examples of the need for molecular profiling in the classification and risk stratification of pHGG is seen with diffuse midline gliomas (DMG). Histologically, this tumor type is characterized by hypercellular, diffuse astrocytic morphology with high mitotic activity and varying amounts of necrosis [6]. The immunophenotype is consistent with its glial cell origin, with positive staining for GFAP and OLIG2. Historically, the degree of mitotic activity and necrotic tissue was used to diagnose and risk-stratify these tumors. Molecular profiling has shown that certain mutations place this tumor type in a high-risk category (CNS WHO grade 4) despite histologic findings [7]. The most well-known tumor in this group is the diffuse intrinsic pontine glioma (DIPG). DIPGs were not historically biopsied and were instead diagnosed based on classic MRI findings of at least 50% involvement of the pons associated with T1-hypointensity and T2-hyperintensity, so it was not possible to molecularly understand these tumors [8]. In recent decades as biopsy has become more common and safer, a specific histone mutation was found to be commonly associated with DIPGs in addition to other midline HGGs such as primary spinal cord or thalamic tumors, thus the DMG group was formed. This emphasizes the importance of molecular findings when diagnosing pHGGs.

Subtypes of DMG include DMG H3K27-altered, EGFR-mutant, and H3-WT with EZHIP overexpression [9]. The most common of these is DMG H3K27-altered, which is most commonly seen in young adults with studies reporting a median age of 14 [6]. These tumors are exclusively found along the midline and have a wide variety of histologic appearances. The majority exhibit hypercellularity and have an astrocytoma-like morphology with extensive mitotic activity and varying degrees of necrosis [6]. Given its histologic diversity, further molecular characterization is beneficial for diagnosis. The H3K27-altered group of tumors is a histone mutation-driven tumor, the first of which were identified in pHGG [1]. It is the result of missense mutations on the H3F3A gene affecting either histone 3.3 (DMG H3.3K27) or histone 3.1 (DMG H3.1K27) leading to replacement of lysine with methionine [10]. This results in downregulation of H3K27me3 and resultant amplification of platelet-derived growth factor (PDGFRA) and ACVR1 mutations, respectively [1]. These mutations typically co-occur with mutations in either tumor suppressor protein 53 (TP53) or PIK3R1, resulting in additional impacts to cell-cycle regulation and growth factor signaling [11]. H3K27-altered tumors have been shown to be associated with generally poorer prognosis in comparison to wild-type variants of pHGG, with a reported 99% 5-year mortality rate and median survival of approximately 11 months [6,10]. This is perhaps due to its low level of MGMT promoter methylation, which is associated with poorer response to traditional chemotherapy treatment with temozolomide [4]. However, the subset of these with ACVR1 mutations have exhibited slightly longer survival than other subsets [12]. Identification of these mutations and molecular markers also provides multiple avenues for targeted therapies.

Several clinical trials have been completed while others are actively enrolling to identify appropriate therapeutic targets for DMG, H3K27-altered tumors. The current standard of care is radiation therapy which prolongs progression-free survival without significant effects on overall survival (OS). In most pediatric oncology centers, an early phase clinical trial is considered as upfront therapy either concurrent with or post radiation. The multitude of trials either target molecularly altered pathways, genetic alterations in the tumor, or attempt to harness the immune environment as a mode of therapy [13]. For example, histone deacetylase (HDAC) is an enzyme involved in the removal of acetyl groups from histone tails. HDAC inhibitors such as Panobinostat have been shown to restore histone 3 (H3) acetylation, reducing cellular proliferation; however, clinical effectiveness has been suboptimal most likely due to poor penetration through the blood–brain barrier [14,15]. For the subset of DMG H3K27-altered tumors that overexpress PDGFRA, tyrosine kinase inhibitors have been shown to be effective at inhibiting PDGFRA and may be useful as a therapeutic option [14,16]. Other trials are also focusing on dual target therapy, such as CUDC-907, a HDAC and PI3K inhibitor, for the treatment of H3K27-altered DMG [14]. There are also several ongoing immunotherapeutic trials including the use of CAR-T cells such as those at Stanford and at Seattle Children’s [17,18].

2.2. Diffuse Hemispheric Glioma (DHG), H3G34-Mutant

The H3G34-mutant pHGGs are most commonly seen in young adults, with one study noting a median age of 19 [19]. These tumors typically arise in supratentorial spaces—most often the temporal and parietal lobes [10,19]. Similar to DMG H3K27-altered tumors, the DHG H3G34-mutant tumors are typically characterized histologically by high-grade astrocytic morphology. A subset of these have an embryonal-like appearance with a high nucleus to cytoplasm ratio, with associated vascular proliferation and necrosis [6,7]. Like the H3K27-altered tumors, GFAP is expressed with immunohistochemical staining, though conversely OLIG2 is not often expressed in these tumors [10]. Given their histopathologic and immunohistochemical similarities, the H3G34-mutant was not identified as a separate entity from the H3K27-altered tumors until much more recently following identification by molecular profiling [10]. Based on these findings alone, H3G34-mutant tumors were previously diagnosed as pHGG or as a primitive neuroectodermal tumor, resulting in minimally effective treatment selection and poor ability to predict outcomes [20]. Current treatment of these tumors primarily consists of surgical resection, followed by radiation therapy and chemotherapy, though a standardized chemotherapeutic regimen for pediatric-type H3G34-mutants does not currently exist [20]. Median OS is most often reported to be between 14 and 17 months, which is slightly improved from that of the H3K27-altered tumors; this may be secondary to ease of surgical resection based on accessibility of hemispheric versus midline gliomas [10,20].

The DHG H3G34-mutant is another example of a histone mutation-driven tumor, with its defining mutation on the H3-3A gene with resultant replacement of glycine with arginine or glycine with valine (H3G34R/V), with H3G34R the most common mutation found in DHG H3G34-mutants [10,20]. This replacement results in decreased SETD2 methyltransferase activity, leading to decreased levels of H3K36me3 [7,20]. Decreased H3K36me3 has been shown to be involved in the methylation of H3K27, resulting in increased levels of H3K27me3, in contrast to the downregulation of this post-translational modification seen in H3K27-altered gliomas [1,20]. These tumors are frequently found to have methylation of the MGMT promoter, despite global DNA hypomethylation resulting from decreased interaction with and redistribution of DNA methyltransferase 1 noted in these tumors [6,7,20]. These molecular findings can be beneficial in risk stratification; for example, tumors with MGMT promoter methylation have been shown to have a better prognosis than those characterized by amplification of oncogenes (e.g., EGFR, MDM2) [7]. Decreased H3K36me3 has also been implicated in defects in DNA mismatch repair (MMR), possibly contributing to tumorigenesis [20]. H3G34-mutants are additionally associated with mutations in TP53, a tumor suppressor protein, and ATRX, a protein involved in chromatin remodeling important for genetic stability; PDGFRA amplification is also commonly found [6]. These tumors are also associated with amplification of MYCN [6].

Identified molecular changes allow for various treatment targets. One of these targets is Aurora Kinase A (AURKA), a regulator of MYCN; AURKA inhibitors have shown selectivity against cells of DHG H3G34-mutant tumors [21]. The PDGFRA pathway has also been identified as a therapeutic target in these tumors; PDGFRA inhibitors in combination with MET inhibitors or mTOR inhibitors showed improved OS in vivo [22]. Cell-cycle components have also been noted to be more frequently found in H3G34-mutants, opening the possibility for cyclin-dependent kinase inhibitors as potential targets for treatment [10]. There are few clinical trials focusing on molecular targets of DHG H3G34-mutant tumors when compared to those for the DMG H3K27-altered tumors, requiring further investigation [10].

2.3. Diffuse Pediatric Type HGG, H3-Wild Type and IDH-Wild Type (pHGG H3/IDH-WT)

Akin to the H3G34-mutant, the pHGG H3/IDH-WT subgroup of pHGG was recently identified as a distinct tumor class. Its definition is reliant upon the lack of mutations in both IDH, common in pHGG, and the histone mutations found in the DMG H3K27-altered and DHG H3G34-mutants [23]. These tumors are histologically diverse. Hong et al. showed these tumors to be associated with TP53 and ATRX mutations, as well as mutations of MMR genes, with those harboring TP53 mutations most specifically associated with adverse outcomes [23]. Given the presence of MMR genes, these tumors have also been shown to be associated with familial cancer syndromes such as Lynch syndrome and Li-Fraumeni [9,23]. These tumors have been subcategorized into the following methylation profiles: RTK1, RTK2, and MYCN. The MYCN methylation subtype is characterized by resultant amplification of MYCN and often the amplification of ID2 [24]. These tumors are the most common subtype associated with TP53 mutations, as well as pHGG occurring in the context of Li Fraumeni [6,24]. In comparison to other subtypes, these tumors may be more likely located infratentorially, particularly in the brain stem [24]. Tumors with the RTK1 methylation profiles have been shown to be located either supratentorially or infratentorially, whereas the RTK2 profile tumors are most often supratentorial [6,24]. These subtypes are associated with amplification of PDGFRA and EGFR, respectively, resulting in disturbances in cellular signaling leading to oncogenesis; the RTK1 profile is most likely to represent those with pHGG in the context of Lynch syndrome [6,25]. The RTK2 methylation profile has also been shown to have a higher frequency of TERT-promoter mutations, and is associated with highest OS rate in comparison to other pHGG H3/IDH-WT tumors, with a 5-year OS of approximately 50% [24]. Other studies have not found a correlation between methylation profiles and their associated amplifications and OS [23,24,26]. The MYCN subgroup OS has been shown to be closer to that of the H3K27-altered pHGG with median OS of 17.2 months; this may be secondary to the more frequent association of this methylation profile with mutations in TP53 [24,26].

Though advances in molecular diagnostics have allowed for further characterization of these tumors, PDGFRA, EGFR, and/or MYCN amplification are nonspecific findings and have been seen in other subcategories of pHGG. Additionally, tumors have been identified as not having IDH or H3 mutations and do not fall into any of the methylation profiles previously described [7]. Further research into H3/IDH-WT tumors is necessary for accurate subcategorization and risk stratification.

2.4. Infant Type Hemispheric Glioma (iHGG)

While the exact definition of what constitutes an infantile brain tumor remains up for debate, most agree that these are tumors occur within the first year of life. iHGG, histologically, is typically described as having astrocytic-like cells, as well as occasional spindle cells [9]. Advances in molecular profiling have allowed them to be differentiated from other neoplasms of infancy, and other histologically similar tumors [27]. Alterations in RAS/MAPK seen in tumors in this age group are most often categorized as pLGG, whereas those containing mutations in RTK oncogenes are considered pHGG. The patients are therefore diagnosed with iHGG despite the latter group often exhibiting methylation profiles similar to pLGGs [28,29]. These tumors are histologically characterized by sheets of astrocytic-type cells and focal spindled cells with a definitive separation between neoplastic and non-neoplastic tissue [12,27].

iHGG is currently defined by the presence of receptor tyrosine kinase (RTK) gene fusion, with reports of over 70% of iHGG containing these fusions [30]. These tumors can be molecularly subdivided based on type of RTK gene fusion: NTRK-altered, ALK-altered, ROS-altered, and MET-altered [7,9]. While these fusions may be associated with both pLGG and pHGG, over 80% (24/29) of tumors harboring these fusions have been shown to be enriched for pHGG [28]. Clarke et al. described an “intrinsic set” of infant gliomas (n = 130) based on fusion gene analysis and methylation profiling, 21 of which were considered iHGG. Of these, 10/21 (48%) were noted to contain ALK-fusions, 12/21 (57%) were noted to have NTRK-fusions, with ROS- and MET-fusions representing 9% (2/21) and 5% (1/21), respectively [29]. Smaller studies have confirmed the higher frequency of these mutations, reporting 27% (3/11) of an iHGG cohort containing NTRK-fusions [31]. Numerous chromosomal rearrangements are also noted in iHGGs. Most are associated with dysregulation of genes involved in development, including WNT signaling pathways. Molecular profiling by next generation sequencing has also contributed to better predication of OS, with the subset of iHGG containing chromosomal fusions having increased OS in comparison to the smaller subgroup without fusions [29]. The methylation profile of these tumors has been shown to be more akin to that of pLGGs which may also related to improved OS [27,32].

Due to RTKs being heavily implicated in tumorigenesis of iHGG, RTK inhibitors have been shown to have promising anti-tumor activity in pediatric patients [27]. ALK inhibitors have specifically shown to be an effective target of treatment amongst this subgroup of pHGG; for example, a combined ALK and ROS1 inhibitor has been developed as viable treatment option for these tumors, though clinical trials are ongoing [27]. A recent case study additionally identified a novel fusion of GAB1-ABL2 not previously reported in iHGG or any other cancer in a patient with a tumor methylation profile consistent with iHGG [33]. Based on this molecular analysis, dasatinib, an ABL inhibitor commonly used to treat CML, was trialed and found to be highly effective. This study highlights the impact of molecular profiling on treatment selection and outcomes. More recent studies have also shown that for these tumors, the traditional regimen, including radiation therapy, may not be needed depending on the effectiveness of molecular-targeted treatment options [31]. This is especially beneficial for overall survival and quality of life in the infant population, as molecular therapies have been shown to have fewer short- and long-term side effects.

2.5. NF1-Associated HGG and HGAP

Neurofibromatosis 1 (NF1) is an autosomal dominant disorder caused by a mutation in the NF1 tumor suppressor gene, classically associated with characteristic skin patches (café-au-lait macules) and neurofibromas, as well as Lisch nodules and freckling [34]. Pediatric patients with NF1 are most likely to develop low-grade gliomas, especially optic nerve gliomas, whereas NF1-associated tumors arising in adulthood are more likely high-grade gliomas [35]. Unsurprisingly, these tumors almost all harbor germline NF1 mutations, but further molecular classification of these tumors have identified other mutations that group these tumors into a distinct entity from other low-grade or high-grade gliomas [35]. Importantly, NF1-related tumors arising in both childhood and adulthood largely lack IDH mutations, unlike most pLGGs, and additionally do not contain histone 3 mutations, unlike many pHGGs [35,36]. A certain subset of NF1-associated tumors is also distinct in that they contain numerous CD3 and CD8-positive T-lymphocytes, indicating that this particular subtype of NF1-related tumors may respond favorably to immunotherapies [36].

High-grade gliomas occurring in NF1 typically fall between H3/IDH-WT tumors and a newly defined entity, high-grade astrocytoma with piloid features (HGAP), a subset of gliomas with an intermediate prognosis between low-grade astrocytomas and H3/IDH-WT pHGGs [37]. These tumors most commonly arise in adulthood and are most often located in the posterior fossa. In addition to NF1 germline and somatic mutations, this group is associated with alterations in MAPK pathway genes; as well as loss of function of ATRX, CDKN2A, and TP53; likening them to both pLGGs and pHGGs, respectively [37]. What truly distinguishes HGAP from other pLGGs or pHGGs is its methylation profile. NF1-associated HGAP is associated with dysregulation of RNA processing secondary to both hyper- and hypomethylation [37]. Improved understanding of the molecular and epigenetic profiles of these tumors has opened new avenues for treatment of gliomas in NF1. Inhibition of MEK, a component of the MAPK pathway, has been shown to be effective in some patients, as well as other MAPK and mTOR pathway inhibitors [38].

3. Discussion

As advances in molecular profiling of tumors become more widely available, differentiating pHGG into its molecular subtypes becomes increasingly crucial for both prediction of outcomes and treatment selection. The reliance on molecular identification is evident in the release of the WHO CNS5 guidelines in 2021, whose primary focus was on the addition of molecular profiling as part of the diagnostic criteria for these tumors. WHO CNS5 allowed for the identification of the H3G34-mutant, H3/IDH-WT, and iHGG as distinct from one another and H3K27-altered tumors for the first time in the WHO diagnostic criteria [39].

While traditional methods of histologic and immunohistochemical classification are beneficial in differentiating pHGG from pLGG, these differences are inadequate predictors of treatment response and long-term outcomes. Incorporation of next-generation sequencing (NGS) has become a crucial component of the diagnostic process at multiple institutions as a way to glean insight into the particular molecular and genetic alterations in individual tumors as a guide for both diagnosis and treatment [14]. NGS is both more time and cost efficient in addition to requiring smaller tissue samples sizes when compared to whole genome sequencing. However, this is not necessarily widely available at all institutions, and therefore more work is needed to identify the most cost effective and efficient way to identify molecular targets. Many institutions now use their own targeted CNS neoplasm NGS panels while other institutions submit tissue to third party vendors or outside hospitals. It will be essential that NGS and further updated technology is available for all patients to ensure the most up to date diagnostic information.

Current clinical trials seek to target these molecular differences such as histone mutations, specific methylation profiles, and gene fusions. This presents an opportunity to both tailor therapy to specific tumor subtypes and better understand prognosis. Further molecular classification of these tumors may provide insight as to appropriate methods of drug delivery for specific tumors, a significant barrier to efficacy of therapy in pHGG. Additionally, advances in targeted therapy for pHGG may allow for more limited use of chemotherapy and local radiation. While the current standard of treatment for pHGG remains resection and radiation therapy, upfront use of targeted therapy remains under investigation, particularly for its radiation sparing implications in iHGG [31,32]. Use of existing and novel targeted therapies, therefore, becomes an increasingly crucial component of pHGG treatment, either alone or in combination with traditional therapy. Further investigation into which patients and tumor subgroups may benefit most from targeted therapies is needed to develop a more specific standard of care for these tumors. Current studies have shown an overall improvement in outcomes with the addition of individualized therapies based on molecular and genetic findings. For example, a patient with iHGG who was treated with NTRK-inhibitors as first-line treatment in comparison to radiation has been shown to have stable disease control 3 years after therapy, resulting in the ability to avoid the adverse effects radiation and generalized chemotherapy altogether [40]. In H3G34-mutant tumors, a combination of PDGFRA inhibitors and mTOR inhibitors was shown to improve OS in these patients in comparison to traditional therapy [41]. An example of some ongoing clinical trials based on molecular alterations can be seen below (

Table 2. Ongoing clinical trials regarding pHGG.

Clinicaltrials.gov ID	Brief Title	Phase/Status	pHGG Type	Drug Type Target
NCT06838676	ACT001 for the Treatment of Diffuse Intrinsic Pontine Gliomas and H3K27-altered High Grade Gliomas	Phase II	H3K27-altered glioma	PAI-1 inhibitor, inhibits glioma cell proliferation [42]
NCT06333899	Lorlatinib for Newly Diagnosed High-Grade Glioma With ROS or ALK Fusion	Early Phase I	H3K27-altered glioma, iHGG	ALK inhibtor
NCT06946680	IL-8 Receptor-modified CD70 CAR T Cell Therapy in CD70+ Pediatric High-grade Glioma (HGG)	Phase I	New diagnosis pHGG	CD70+ HGG immunotherapy
NCT06712875	MAPK Inhibition Combined with Anti-PD1 Therapy for BRAF-altered Pediatric Gliomas	Phase I/Phase II	pLGG, pHGG	MEK1/2 inhibition, PD-1 inhibition, BRAF kinase inhibitor
NCT05610891	Novel Targeted Radiotherapy in Pediatric Patients With Inoperable Relapsed or Refractory HGG	Phase I	Relapsed, refractory pHGG
NCT04911621	Adjuvant Dendritic Cell Immunotherapy for Pediatric Patients With High-grade Glioma or Diffuse Intrinsic Pontine Glioma	Phase 1/Phase 2	pHGG

). This is continuously being updated with new data emphasizing the importance of ongoing pre-clinical work and its translation to clinical trials.

4. Conclusions

As the importance of individualized molecular, genetic, and epigenetic profiling in pediatric high-grade gliomas grows, it is crucial to understand the limitations of this new approach to diagnosis. Despite the improved ability to identify molecular targets in these tumors, the challenge of developing accurate pre-clinical models, as well as therapies that can effectively target the identified pathways and adequately cross the blood–brain barrier remains an issue [43,44]. There are also known disparities in access, making the inclusion of molecular profiling of tumors part of the new standard of care more difficult to achieve on a global scale. Studies have shown that patients of lower socioeconomic status are less likely to be offered or undergo methylation profiling, despite its emerging importance in both diagnostic and treatment decisions [45].

Regardless of these barriers, it is clear that identification of unique molecular and genetic targets in pHGG will allow for improvement in therapies and continued hope for a cure [46]. The emerging field of radiogenomics—using machine learning to predict molecular changes from a large database of common imaging findings—provides another avenue for future growth in the field [47].

While the prognosis for pHGG remains poor, a continued focus on directing pharmaceutical development towards specific molecular markers offers hope for improvement in OS and quality of life for patients with pHGG.